Whole-crown 13C-pulse labelling in a sub-arctic woodland to target canopy-specific carbon fluxes

Whole-crown 13C-pulse labelling can target tree canopy C fluxes in regions with dense understorey cover and investigate how increased photosynthetic C inputs may affect whole-ecosystem C fluxes. Climate change-driven increases in plant productivity have been observed at high northern latitudes. These trends are driven, in part, by the increasing abundance of tall shrub and tree species in arctic ecosystems, and the advance of treelines. Higher plant productivity may alter carbon (C) allocation and, hence, ecosystem C cycling and soil C sequestration. It is important to understand the contributions that the newly established canopy forming overstorey species makes to C cycling in these ecosystems. However, the presence of a dense understorey cover makes this challenging, with established partitioning approaches causing disturbance and potentially introducing measurement artefacts. Here, we develop an in situ whole-crown 13C-pulse labelling technique to isolate canopy C fluxes in areas of dense understorey cover. The crowns of five mountain birch (Betula pubescens ssp. czerepanovii) trees were provided with a 13CO2 pulse using portable field equipment, and leaf samples were collected from neighbouring con-specific trees and hetero-specific understorey shrubs on days 1–10 and 377 post-crown labelling. We found effective and long-term enrichment of foliage in labelled trees, but no evidence of the 13C-signal in con- or hetero-specific neighbouring trees or woody shrubs. This method is promising and provides a valuable tool to isolate the role of canopy tree species in ecosystems with dense understorey cover.


Introduction
Trends of increasing plant productivity, or "greening", have been observed in recent decades across large parts of the northern hemisphere (Epstein et al. 2012(Epstein et al. , 2016Reichle et al. 2018). At high latitudes, these trends are partly driven by the increasing abundance of tall shrubs in tundra ecosystems (Myers-Smith et al. 2011;Bjorkman et al. 2018) and the northern advance of treelines (Hagedorn et al. 2014;Rees et al. 2020) with the presence of trees increasing at higher latitudes and up slopes (Hofgaard et al. 2013). This has the potential to alter the relative contribution of tall and short-stature plants to ecosystem carbon (C) cycling. With 'greening' and an increase in primary productivity, inputs of recently fixed C to Arctic and northern ecosystems are expected to increase (Pearson et al. 2013). However, we have limited understanding of the potential fate of this increased C, or how it impacts on below-ground processes, which may alter ecosystem C cycling and soil C sequestration Street et al. 2018Street et al. , 2020. Therefore, understanding the knock-on effects of increased C inputs and changes to plant community composition in northern ecosystems necessitates the ability to partition canopy and understorey C fluxes to forecast and understand the drivers of C fluxes at ecosystem level, including various plant, microbial, and soil C pools. One way to partition canopy and understorey C fluxes is through targeted stable isotope labelling of C assimilated by canopy forming trees. Communicated by Zhu. Labelling of plants using stable ( 13 C) or radioactive ( 14 C) isotopes has been used to investigate C cycling in terrestrial ecosystems and to trace C allocation and remobilisation in plants (Reviewed by Brüggemann et al. 2011). Using 14 CO 2 in pulse labelling is advantageous, because very small amounts of 14 CO 2 can be supplied to the foliage due to the very weak environmental background for 14 C compared with 13 C in the atmosphere and in the plant-soil system. However, using 13 CO 2 for pulse labelling is much more cost-effective than using 14 CO 2 , and avoids deployment of radio-isotopes in the environment (Epron et al. 2012). Numerous studies have focused on C dynamics in plants and soils using 13 CO 2 pulse labelling as a powerful tool in ecophysiology to analyse C fluxes (Dawson et al. 2002;Bowling et al. 2008;Epron et al. 2011), estimate turnover of soil organic matter (e.g., Lichtfouse et al. 1995;Klumpp et al. 2007;Theis et al. 2007), partition C fluxes and pools (e.g., Hoch and Keel 2006;Plain et al. 2009;Subke et al. 2012;Street et al. 2013Street et al. , 2018, and evaluate C fluxes to soil biota (e.g., Simard et al. 1997;Teste et al. 2010;Epron et al. 2011) and through food-webs (e.g., Leake et al. 2006;Pollierer et al. 2007Pollierer et al. , 2012Högberg et al. 2010). The technical complexity of pulse labelling experiments means that such experiments are challenging to implement in the field. Pulse labelling of short-stature Arctic shrub species has been accomplished using relatively small 13 C labelling chambers (Deslippe and Simard 2011;Street et al. 2018); however, pulse labelling larger shrubs and trees at remote high-latitude field locations present greater challenges and have rarely been attempted (Masyagina et al. 2016).
Methods for C-pulse labelling of trees have been extensively reviewed by Epron et al. (2012). Pulse labelling of whole areas of forest can be achieved using free-air CO 2 enrichment (FACE) techniques (e.g., Pepin and Körner 2002;Klein et al. 2016). However, this method does not provide insight at the level of individual trees, nor does it partition canopy and understorey fluxes. Pulse labelling of whole trees has been achieved using large labelling chambers housing individual trees in temperate (Plain et al. 2009;Drake et al. 2019;Gao et al. 2021), tropical (Glaser et al. 2012, and arctic (Masyagina et al. 2016) regions, or whole patches of boreal forest (Högberg et al. 2008). Such approaches have provided valuable insight into tree C dynamics, but due to the size of labelling chamber, these methods involve substantial investment in infrastructure and personnel. Whole-ecosystem labelling presents challenges for tracing the short-term fate of newly fixed C from plants to soil. Physical diffusion of 13 CO 2 into soil pores during the pulse period complicates the detection of biological signal from assimilate transfer to roots (Subke et al. 2009). Furthermore, large chambers generally label both trees and forest-floor species, making it harder to examine the significance and fate of tree-mediated inputs in isolation from understorey communities. At higher latitudes, with dense understorey cover contributing significantly to total leaf area index ) and ecosystem gross primary productivity (Kulmala et al. 2019), mechanistic understanding requires approaches that identify the relative contribution of tree versus understorey C inputs to the ecosystem C cycle. This distinction cannot be made using methods which label both canopy and understorey communities together.
One approach to tackle this challenge is to pulse-label a fraction of a whole tree, such as a single branch (Hoch and Keel 2006;Masyagina et al. 2016). Labelling only a fraction of a whole tree inevitably results in a high level of dilution of the 13 C signal by the 12 CO 2 fixed by the rest of the tree. Another approach is to inject 13 C-carbonate into tree xylem (Powers and Marshall 2011). This approach avoids the need for large chambers, but a large proportion of the C label delivered to the tree xylem has been found to diffuse primarily out of stem tissues (Żelawski et al. 1970), thereby limiting the value of this technique for tracing photosynthates into roots, mycorrhizal fungi, and soil.
Once isotopic labelling of recently fixed C in trees has been achieved, the isotopic signal can be used to trace C allocation into various plant tissues (e.g., leaves, stems, or roots), within which the main C sinks are growth and storage in the form of non-structural carbohydrates (NSCs), respiration, symbiotic associations with mycorrhizal fungi, and exudation (Ericsson et al. 1996). This makes isotopic labelling of trees in high-latitude regions a valuable method, not only to understand C cycling at the ecosystem level, but also to improve our limited understanding of how long-term storage pools (NSCs) buffer against seasonal variation in C supply through extended winter periods, maintain important mycorrhizal relationships, and how these C pools respond to environmental change (Hartmann and Trumbore 2016).
Here, we present a method for in situ whole-crown 13 C-pulse labelling of short-stature (2-3 m) trees, using portable field equipment and delivery of 13 CO 2 to individual trees. This provides a tool to trace the fate of carbon fixed by canopy forming trees, which create the forest overstorey, isolated from the short-stature understorey plant community.

Materials and methods
For full methods, see Supplemental Materials.

Site description and plot set-up
All studied plots were in the sub-Arctic mountain birch (Betula pubescens Ehrh. ssp. czerepanovii (Orlova) Hämet Ahti) treeline forest at Nissunsnuohkki, south of Abisko, Sweden (ca. 68°18'56.2"N 18°49'18.2"E). In these forests, mountain birch trees form the canopy, defined here as the upper vegetation layer formed by mature tree crowns, and the majority of the understorey is formed by short-stature woody shrubs, predominantly <50 cm in height.
Five single-stemmed trees (to be 13 C labelled, henceforth referred to as 'target trees') 2-3 m tall, with mean basal area of 37.0 ± 29.2 cm 2 (mean ± SD), were selected ≥ 50 m apart. Each plot included three selected trees: the 'target' tree, a 'trenched' control tree, and an 'un-trenched' control tree. The average distance between the target tree and selected control trees was 3.85 ± 0.56 m (mean ± SD). An approximately 25 cm-deep trench severing roots, and mycorrhizas was cut in a semi-circle at the mid-point between the target tree and one control tree (this became the 'trenched' control tree, henceforth referred to as such). The other control tree in the plot remained undisturbed (this became the 'un-trenched' control, henceforth referred to as such). Both trenched and un-trenched controls were included to account for any potential 13 C translocation via common mycelial networks (CMNs), below-ground, which has previously been observed in arctic Betula species (Deslippe and Simard 2011;Deslippe et al. 2016).
Plots were selected to have a representative understorey with three common species [Betula nana L., Vaccinium vitis-idaea L., and Empetrum nigrum L. ssp. hermaphroditum (Hagerup) Böcher] present. Four Betula nana plants (50-560 cm from target tree), and four patches of Vaccinium vitis-idaea (40-440 cm from target tree) and Empetrum nigrum (15-520 cm from target tree), were marked for leaf sampling, two on either side of the trench.

Isotope labelling
The crown of the target tree, defined here as all branches and leaves of the individual tree, was covered in a large (140 × 178 × 230 cm) clear plastic bag (Gardener's Dream Ltd, Glasgow, UK) and sealed to the main stem at the base of the crown, below the lowest branches. Two battery-operated fans inside the crown bag facilitated air movement during labelling. A temperature probe (Traceable ® , Webster, USA) was secured to a branch and shielded from direct solar radiation to monitor temperature inside the labelling bag throughout the labelling process (Online Resource Fig. S1).
Immediately after sealing the crown bags, each of the five target trees was labelled using 13 CO 2 produced in situ by adding 90 ml 2% HCl to 2 g Ca 13 CO 3 ; thus, 0.257 g 13 C (99 atom % enriched with 13 C, Sigma-Aldrich Ltd, Dorset, UK) contained in a beaker outside the crown bag (Fig. 1). The 13 CO 2 was led into and out of the crown bag via infrared gas analysers (PP Systems International, Amesbury, MA, USA) monitoring the concentration of CO 2 produced (Online Resource Fig. S1) and creating a closed loop (Fig. 1b). Crown 13 CO 2 labelling was conducted for approximately 1 h 15 min on 20 July 2018 between 08:00 and 17:00 h, with diffuse high clouds allowing sufficient light for high photosynthetic rates without long periods of direct sunlight.

Foliar and air sampling
10-15 mature leaves were sampled from the selected trees and shrubs prior to labelling and on days 1, 3, 6, 10, and 377 post-labelling. All foliar samples were frozen in the field within 1 h of sampling (Waeco Dometic CDF-26 21 L Portable Compressor Fridge Freezer) to reduce metabolic processes, and remained frozen at − 20 °C for ≥ 24 h and dried at 60 °C for 72 h before milling to ensure homogenisation.
Air samples, each of which was 24 ml, were taken from within the crown bag during 13 C-pulse labelling. During the 13 C-pulse labelling period, air samples were also taken from within the canopy of the control trees to detect any leaks from the labelling bag. After 13 C-pulse labelling, air samples were collected during foliar sampling, within the canopy of all three trees at each plot.

Statistical analyses
All analyses were carried out using R Version 3.4.0 (R Core Team 2017). δ 13 C enrichment of trenched and un-trenched control trees over time was modelled using a linear mixedeffects model 'nlme::lme' (Pinheiro et al. 2012; R Core Team 2017) with sampling day as a fixed variable and 'plot' as a random variable. Exponential decay of δ 13 C enrichment over time in target trees was modelled using a self-starting non-linear asymptotic regression (stats::SSasymp; (R Core Team 2017)). Pre-pulse data and day 377 data are presented as summarised raw data and tested using a paired Student's t test within each plot. δ 13 C enrichment of the understorey over time is presented as summarised raw data. The difference in δ 13 C enrichment between trenched control and untrenched control samples was tested using a paired Student's t test within species, with Bonferroni corrections.

Results and discussion
Using a whole-crown bag and δ 13 C-enriched CO 2 (Fig. 1), we successfully labelled five replicate mountain birch trees in situ in a Swedish sub-Arctic treeline forest (Fig. 2). The foliage of target trees assimilated 0.18 ± 0.06 g (mean ± SD, n = 5) 13 C based on enrichment of day 1 samples (Supplemental Materials Eq. 1).
There was no difference in δ 13 C values between trenched and un-trenched control trees either pre-or post-13 C labelling (P = 1.00 on all sampled days) suggesting no (or undetectable) tree-to-tree transfer of 13 C via CMNs. Although 13 C enrichment was not detected in the foliage of control trees, it is possible that the 13 C signal could have been transferred to the roots or stem tissues (Klein et al. 2016); however, these were not sampled here. More targeted experiments, examining contrasting plant and fungal tissues, are needed to draw any firm conclusions.
Leaves of the target tree had significantly higher δ 13 C enrichment throughout the initial 10-day post-pulse period than natural abundance samples and both the trenched and un-trenched control trees (P < 0.001) (Fig. 2). This demonstrates successful 13 C labelling of the target tree with minimal or undetectable leakage or transfer of the 13 C signal to surrounding mountain birch trees. Air sampling during the 13 C-pulse labelling shows very high δ 13 C enrichment inside the crown bag but none outside the bag near control trees (P < 0.001) (Fig. 3). This pulse labelling method can therefore be deployed with a high degree of accuracy, labelling only selected target trees. Air samples from the subsequent 10 days post-13 C-pulse trended higher in δ 13 CO 2 in crowns of target trees, but were not significantly enriched compared to neighbouring trees (P > 0.05) (Fig. 3), which means that any 13 CO 2 respired from target trees was undetectable in free-mixing treeline forest air. This also strongly suggests that any leaf enrichment detected stemmed from the original 13 C-pulse, and not significantly from fixed and subsequently respired 13 CO 2 from the target tree.
We found no evidence of the 13 C signal in any of the surrounding understorey shrubs on days 1-10 or 377 (Fig. 4) relative to natural abundance samples (P > 0.1). δ 13 C values on day 377 were significantly higher than on day 10 for all three understorey species (B. nana: P < 0.001, E. nigrum: P <0.001, V. vitis-idaea: P = 0.039) (Fig. 4). This small but Fig. 1 a Field pulse labelling set-up being deployed in situ. b Schematic diagram of 13 CO 2 pulse labelling set-up. A clear plastic bag fitted around target tree with two fans (F), one thermometer (T) in the crown space, and gas flow to and from the bag. Gas flows through two infrared gas analysers (IRGA) either side of 13 CO 2 generated by adding HCl to Ca 13 CO 3 using a syringe. Dashed lines indicate direction of cyclic gas flow statistically significant increase in 13 C enrichment is consistent in all understorey shrub species >1 year post-pulse. However, because the response was observed in shrubs from both trenched and un-trenched areas, it is unlikely that the 13 C enrichment is caused by the pulse itself (e.g., re-fixation of 13 CO 2 respired from target tree roots), and more likely reflects interannual variation. Interannual variation in 13 C natural abundance may occur due to interannual variation in abiotic factors such as precipitation, snow depth, and temperature (Blok et al. 2015;Gamm et al. 2018), which can influence stomatal aperture and alter C fractionation patterns.
These results show that this pulse labelling technique is able to target specific members of the plant community in isolation, thereby facilitating the partitioning of understorey and canopy tree C fluxes (not directly measured here). The Fig. 2 δ 13 C of leaf biomass for each treatment type (target, trenched, un-trenched, n = 5) pre-, 1-10 days, and 377 days post-pulse labelling. Pre-pulse samples are taken as natural abundance δ 13 C. Error bars and ribbons are ±1 standard deviation, and results within sample day are off-set to aid comparison. The inset magnifies the y-axis range for trenched and un-trenched control samples as well as target tree samples pre-pulse and on day 377, to aid visualisation of differences between trees Fig. 3 Mean (points) δ 13 C values in air samples taken at all three trees (target, trenched, un-trenched, n = 5) during and after 13 C-pulse labelling. Error bars are ±1 standard error; n = 5. Mean values within sample day are off-set to aid comparison. Note the contrast in y-axis scales during and post-13 C-pulse (the latter are expanded to aid data visualisation) success of this method in isolating canopy C fluxes in forests with dense understorey cover is promising, with considerable scope for application in other forest ecology contexts. Understanding of C cycling and forest function could be advanced substantially by: (1) determining the fate of C fixed by canopy trees through tracing the 13 C into rhizosphere, mycorrhizal, microbial, or soil C pools (Brüggemann et al. 2011), and (2) identifying linkages between individual plants within forests by tracing C fixed by one tree into other plants (Simard et al. 2012).
One timely application could be to investigate the effects of climate change-driven Arctic 'greening', expansion of shrubs, and northward advance of treelines (Myers-Smith et al. 2011;Rees et al. 2020), which increases above-ground plant productivity and C inputs below-ground. As different species allocate recently fixed C between plant and soil pools differently (Street et al. 2018), it remains largely unknown how shifts in species abundance will alter below-ground processes and affect soil C sequestration Parker et al. 2021). Most research in this region, to date, has dealt with the mountain birch forest as a whole (e.g., Sjögersten and Wookey 2009;Hartley et al. 2010Hartley et al. , 2012Parker et al. 2015), or focussed on canopy species Parker et al. 2020). However, to understand changes in C pathways between plants, microbes and soil organic matter, we need methodologies that can selectively target contributions of key species without physically disturbing understorey plants or below-ground connections. At high latitudes, where biomass is more evenly distributed between canopy and understorey (Kulmala et al. 2019), and climate change may alter the relative dominance of the two, it is increasingly important to disentangle the contribution of these two plant community components to whole-ecosystem C fluxes to forecast the potential effects of such a shift in dominance. The specificity of this labelling technique is complementary to manipulative techniques such as 'stem girdling' (Högberg et al. 2001;Parker et al. 2020) or trenching (e.g., Bond-Lamberty et al. 2004;Street et al. 2020), but offers advantages in terms of minimal disturbance and the opportunity to determine the medium-term fate of specific C inputs. Thus, it will be valuable in partitioning the relative influence of canopy and understorey C fluxes within plant communities with potentially interactive effects on belowground processes.
Stable isotope pulse labelling creates, by definition, a transient signal, and by day 377 post-13 C labelling, the foliage of target trees was no longer significantly δ 13 C-enriched relative to both the trenched (P = 0.14) and un-trenched (P = 0.079) control trees (Fig. 2). When comparing day 377 samples with the natural abundance pre-pulse samples, there was no significant difference in 13 C enrichment between natural abundance and day 377 for both trenched (P = 0.35) and un-trenched (P = 0.24) control trees. However, target trees were significantly enriched on day 377 compared to natural abundance δ 13 C pre-pulse samples (P = 0.027) which suggests that a fraction of the 13 C label may have been preserved in the target trees >1 year after the 13 C-pulse. This indicates that a fraction of the 13 C fixed during pulse labelling was retained in long-term C storage pools of target trees, likely as reservoirs of non-structural carbohydrate (NSC) (Martínez-Vilalta et al. 2016). The detection of the 13 C within new leaves more than 1 year after labelling suggests that the pulse likely generated a strong signal in NSCs. As this is a deciduous species, detection of enhanced 13 C in annual foliage provides compelling evidence of translocation of NSC from stem and/or root tissues to leaves in the year following Fig. 4 Mean δ 13 C values in understorey species up to 10, and 377 days, post-pulse labelling of the target tree. Colour illustrates whether there was a trench severing roots and mycorrhizas between the pulse labelled tree and where the understorey was sampled. Error ribbon and bars are ±1 standard deviation, n = 5. Mean results for day 377 are off-set to aid comparison assimilation (Hartmann and Trumbore 2016). Isolating specific compound classes may open up new applications of this method over multiple growing seasons in the future, and thus, the chance to improve understanding of how C reserves are used in arctic tree and shrub species adapted to surviving extended winters.

Conclusions
We present a method for in situ whole-crown 13 C-pulse labelling mountain birch trees in a sub-arctic treeline forest. Using this method, we successfully labelled the foliage and long-term C storage pools (>1 year post-13 C-pulse) of five mountain birch tree crowns, whilst detecting none of the 13 C signal in the foliage of neighbouring trees or in any of the sampled understorey shrubs. This method of whole-crown pulse labelling provides a valuable tool to trace the fate of recently fixed C from individual canopy trees into various plant, microbial, and soil pools, and to investigate how increased photosynthetic C inputs and shifts in canopy and understorey dominance in northern ecosystems may affect whole-ecosystem C fluxes in a changing world.
Author contribution statement NLF, TCP, J-AS, and PAW contributed to the study conception and design. Material preparation, data collection, and analysis were performed by NLF. Stable isotope analysis was carried out by HKG.
Funding The research leading to these results has received funding from the European Union's Horizon 2020 project INTERACT,under Grant Agreement No 730938. Stable isotope analysis of samples was funded by the UK Natural Environment Research Council (NERC) Life Sciences Mass Spectrometry Facility (LSMSF CEH_L_113_05_2018) awarded to NLF. Equipment for measurement of CO 2 concentrations and environmental monitoring were funded by NERC via projects NE/ K000284 and NE/P002722 awarded to PAW and J-AS.

Availability of data and materials
The data that support the findings of this study will be freely accessible in due course, but in the meantime, please contact the corresponding author for access to data.

Code availability
The code used for statistical analysis in the current study are available from the corresponding author on request.

Conflict of interest
The authors declare that they have no conflict of interest.
Open Access This article is licensed under a Creative Commons Attribution 4.0 International License, which permits use, sharing, adaptation, distribution and reproduction in any medium or format, as long as you give appropriate credit to the original author(s) and the source, provide a link to the Creative Commons licence, and indicate if changes were made. The images or other third party material in this article are included in the article's Creative Commons licence, unless indicated otherwise in a credit line to the material. If material is not included in the article's Creative Commons licence and your intended use is not permitted by statutory regulation or exceeds the permitted use, you will need to obtain permission directly from the copyright holder. To view a copy of this licence, visit http:// creat iveco mmons. org/ licen ses/ by/4. 0/.