Plant size, leaf economics, and their diversity are strong controls of tundra carbon cycling

1. The functional composition and diversity of plant communities are globally applicable predictors of ecosystem functioning. Yet, it is unclear how traits influence carbon cycling. This is an important question in the tundra where vegetation shifts are occurring across the entire biome, and where soil organic carbon stocks are large and vulnerable to environmental change. 2. To study how traits affect carbon cycling in the tundra, we built a model that explained carbon cycling (above-ground and soil organic carbon stocks, and photosynthetic and respiratory fluxes) with abiotic conditions (air temperature and soil moisture), plant community functional composition (plant size described with average plant height, and leaf economics with leaf dry matter content and specific leaf area), and functional diversity (weighted standard deviations of the traits). 3. The explanatory power of the models was relatively high, but a large part of variation in soil organic carbon stocks remained unexplained. Plant size was the strongest predictor of all carbon cycling variables except soil carbon stocks. Communities of larger plants were associated with larger CO2 fluxes and above-ground carbon stocks. Communities with fast leaf economics had higher photosynthesis, ecosystem respiration, and soil organic carbon stocks. 4. Diversities on axes of size and leaf economics affected ecosystem functions differently. Leaf economic diversity increased CO2 fluxes and soil organic carbon stocks, while size diversity increased the above-ground carbon stock. The contributions of functional diversity metrics to ecosystem functioning were about as important as average leaf economic traits. 5. Synthesis: Community composition and diversity on the two axes of the global spectrum of plant form and function have clear and separate effects on ecosystem functioning. The importance of functional diversity highlights a potentially important mechanism controlling the vast carbon pools that should be better recognized. More research on root traits and decomposer communities is needed to understand the below-ground mechanisms regulating carbon cycling in the tundra.


Introduction
The Arctic is warming two to four times faster than the world on average (Holland and Landrum 2015;Post et al. 2019), leading to changes in plant growth (Myers-Smith et al. 2015), height (Bjorkman et al. 2018), and species distributions (Steinbauer et al. 2018). It is unclear how such changes in vegetation will influence the carbon stocks and balance of these ecosystems. On one hand, the expansion of larger plants might increase net carbon uptake (Cahoon et al. 2012;Sørensen et al. 2019). On the other hand, vegetation-driven changes in soil conditions and microorganisms might accelerate decomposition and, in turn, increase carbon losses to the atmosphere (Parker, Subke, and Wookey 2015;Vowles and Björk 2018).
Arctic soils contain more than half of the global soil organic carbon stock (Hugelius et al. 2014), thus changes in tundra vegetation and carbon cycling have the potential to influence atmospheric CO 2 concentrations at the global scale.
Plant functional traits provide a quantitative and globally applicable means to study the interactions among vegetation and carbon cycling in the tundra (Thomas et al. 2020).
Above-ground traits of plant species and the functional composition of plant communities have been found to vary primarily along two axes (Díaz et al. 2016;Bruelheide et al. 2018).
The first axis describes the trade-offs between resource acquisitive (fast) and conservative (slow) strategies, and has often been called the leaf economics spectrum (Wright et al. 2004) that is measured with traits such as specific leaf area (SLA), leaf nitrogen content, and leaf dry matter content (LDMC) (Bruelheide et al. 2018). The other axis is related to plant size, characterizing the trade-offs between investments to light competition, and photosynthesis and reproduction, and can be measured with, for example, plant vegetative height or stem specific density (Díaz et al. 2016). The plant size axis, and in particular plant height, has been observed to increase with climate change over the past decades across the tundra, whereas changes in leaf economic spectrum have been more variable (Bjorkman et al. 2018).
These trait axes explain differences between species' vital rates that define population processes (Adler et al. 2014), and ecosystem functioning such as carbon cycling from local to global scales (Diaz et al. 2004;Michaletz et al. 2014). In the tundra, photosynthesis has been shown to correlate positively with fast leaf economics, indicated by, for example, high average SLA (Sørensen et al. 2019). This is because fast communities have more photosynthetic machinery relative to leaf area compared to slower communities (Shipley et al. 2006;Wright et al. 2004). Communities with fast traits often have higher ecosystem respiration as well because maintaining photosynthetic capacity is costly (Cavaleri, Oberbauer, and Ryan 2008). Plant size is often described using biomass or its proxies, such as total plant cover or leaf area index (LAI) (Oberbauer et al. 2007;Street et al. 2007;Marushchak et al. 2013). Studies have shown that larger biomass leads to higher photosynthesis, net carbon uptake, and sometimes also soil organic carbon stocks (Lafleur and Humphreys 2018;Gagnon, Domine, and Boudreau 2019).
The relationships of the two main axes of the functional composition of plant communities with soil organic carbon stocks can be more complex. This is because these stocks might be more indirectly linked to traits, via, for example, the quantity and quality of litter inputs , by affecting microbial communities or soil microclimate (Parker, Subke, and Wookey 2015;Kemppinen et al. 2021), or by correlating with traits that more directly affect soil processes, such as root chemical composition (G. T. Freschet et al. 2010;Bergmann et al. 2020). Earlier research indicates that faster communities such as meadows often accumulate large amounts of carbon into the soil whereas shrubs, and in particular deciduous shrub communities have smaller soil organic carbon stocks and high soil respiration (Sørensen et al. 2018;Vowles and Björk 2018;Parker, Subke, and Wookey 2015). However, the effects of the two axes of above-ground traits of plant species on carbon cycling has not been comprehensively studied yet.
Climate change-driven changes in tundra vegetation are also predicted to influence the diversity of communities (Niittynen, Heikkinen, and Luoto 2020) which also influences carbon cycling (Tilman, Isbell, and Cowles 2014;Duffy, Godwin, and Cardinale 2017). In the tundra, some evidence points out that more diverse shrub communities located in transition zones might be less productive than communities dominated by one shrub species due to competition (Fletcher et al. 2012). However, in general, the effects of tundra biodiversity on carbon cycling remain poorly understood. Using functional traits allows partitioning the diversity effect to contributions from complementarity along different niche axes and helps to understand how carbon cycling is influenced by changes in functional diversity, which represents one component of biodiversity (Cadotte, Carscadden, and Mirotchnick 2011).
In this study, we examine the effect of plant community functional composition and diversity on growing season ecosystem CO 2 fluxes and above-ground and soil organic carbon stocks in a tundra ecosystem. We use a systematic study setting of up to 129 intensively measured plots and regression models while controlling for the effects of key microclimate variables of soil moisture and air temperature explaining carbon cycling (Poyatos et al. 2014;Nobrega and Grogan 2008;Siewert 2018;López-Blanco et al. 2017).

Study site
The field observations were collected in 2016-2018 in a subarctic tundra environment in Kilpisjärvi, northwestern Finland (Fig. 1). The study area is located on an elevational gradient between two mountains, Saana (1029 m.a.s.l) and Korkea-Jehkats (960 m.a.s.l), and the valley in between (ca. 600 m.a.s.l.). The study area is above the mountain birch (Betula pubescens var. pumila (L.) Govaerts) forest line, and is predominantly dwarf-shrub heath, with graminoid-and herb-rich meadows concentrated around meltwater streams. Empetrum nigrum L., Betula nana L., Vaccinium myrtillus L., Vaccinium vitis-idaea L. and Phyllodoce caeruleae (L.) Bab. are highly abundant in the area with graminoids and herbs such as Deschampsia flexuosa (L.) Trin. and Viola biflora L. dominant along the streams. The main herbivores in the area are reindeer (Rangifer tarandus tarandus L.), and voles and lemmings (Cricetidae). The soils in the area are mostly poorly developed leptosols with shallow organic layers and occasional podzolization; however, the meadows have soils with thicker organic layers. The mean annual air temperature and annual precipitation at a nearby meteorological station (Kilpisjärvi kyläkeskus, 480 m.a.s.l, 1.5 km from the study area, 1981-2010) are -1.9°C and 487 mm, respectively (Pirinen et al. 2012). The annual air temperature was -0.4°C in 2016, -1.5°C in 2017, and -0.9°C in 2018, and precipitation 552 mm in 2016, 551 mm in 2017, and 405 mm in 2018(Finnish Meteorological Institute 2021. The study design consisted of 129 locations in a 1.5 x 3.0 km area (Fig. 1). The distance between two adjacent locations was a minimum of 23 m (average 101 m). Individual study locations were thus in separate vegetation patches. Not all variables were measured in all locations, and the number of replicates for each analysis varied between 119 and 129 (Table   2). At the sampling locations where all variables were collected, organic and mineral layer depths and longer-term (i.e., a three-year record of) soil moisture and air temperature were measured at the central plot of the measurement scheme. Plant functional traits, above-ground carbon stocks, and CO 2 fluxes measured together with instantaneous chamber air temperature and soil moisture were measured in another plot 1-3 meters away from the central plot to avoid perturbing these permanent plots; however, we made sure that the environmental conditions of this plot were similar to the central plot to guarantee that these plots are comparable with each other. Finally, soil samples were collected from the immediate vicinity of the trait-carbon plot. A schematic illustration of the sampling scheme is depicted in Fig.   S1. Further information on the measurement of different variables is depicted below. To ease reading, the different measurement instruments are listed separately in Table S1.
Environmental predictors of tundra carbon cycle variables Microclimate variables: air temperature and soil moisture Soil moisture was measured during the growing seasons of 2016-2018 3-6 times per year between June 6 and August 23 as volumetric water content (Happonen et al. 2019;Kemppinen et al. 2018). In each soil moisture plot, three measurements were taken, and the average was used in further analyses. Temperature loggers were installed in 2016 to monitor air temperature 10 cm above-ground at 2-4 h intervals. These air temperature measurements were aggregated to monthly averages. We used July air temperature in further analyses, because it is the warmest month in the area.
To quantify longer-term mean summer air temperature and soil moisture conditions (i.e., not representing just one year), we averaged July air temperature and growing season soil moisture levels over 2016-2018 for each location using random effects models with the R package lme4 (Bates et al. 2015). Air temperature were modelled using location and year as random effects (y ~ 1|location + year). In each location, soil moisture was measured from five 1 m 2 plots: one central plot and one plot 5 m away in each cardinal direction to reflect fine-scale heterogeneity (Fig. S1). Soil moisture was modelled using plot nested in location, and year as random effects (y ~ 1|location/plot + year). Soil moisture was log-transformed before modelling. Finally, the environmental variables were predicted for the central plot of the soil moisture measurement scheme, averaging out yearly variation.
Longer-term air temperature data were available for 86 plots with carbon stock measurements. To increase this number to 119, we interpolated average July air temperature using kriging with a spherical variogram model (R package gstat, Pebesma and Heuvelink 2016). Visual inspection revealed no effects of interpolation on the covariance and partial covariance structures of our data.
We further measured the air temperature inside the chamber and soil moisture directly next to the collar during the CO 2 flux measurements to reflect instantaneous microclimate conditions during the flux measurement. We recorded soil moisture as the average of three measurements using the same devices as for the longer-term measurements; the air temperature measurements are described in section CO 2 flux data.
Vegetation data: trait and community data We quantified vascular plant community species composition in the majority of the plots where CO 2 flux measurements were conducted using the point-intercept method during the peak growing season in 2017 (Happonen et al. 2019). A circular frame (20 cm diameter) with 20 evenly spaced pinholes was placed over the soil collar (see section CO 2 flux data) after CO 2 flux measurements. The total number of times each species touched pins (3 mm in thickness) lowered into the vegetation through the frame was counted and used as a measure of species abundance. While cryptophytes might have important effects on soil processes, their biomass is considerably lower than that of vascular plants in subarctic heaths (M.C. Press et al. 1998). Furthermore, there is still no consensus on which quantitative traits of bryophytes and lichens best capture their effects on ecosystem functioning. For these reasons, the present study focuses on the effects of vascular plants on carbon cycling.
We measured plot-specific plant height to indicate plant size, and LDMC as well as SLA to represent the leaf economics spectrum for all species (Happonen et al. 2019). These three traits have been shown to be good proxies of the axes of above-ground plant trait variation (Bruelheide et al. 2018;Díaz et al. 2016). Trait measurements were done separately for all locations, and always on plant individuals within the soil collars in order to account for intraspecific trait variation. Height was quantified as the height of the highest leaf on two random ramets within the collar. LDMC and SLA were measured from two leaf samples taken from two different ramets. The leaf samples were taken at the time of community surveys, put in re-sealable plastic bags with moist paper towels, and transported to the lab to be stored at 4℃ for up to three days before scanning for leaf area and weighing for fresh mass. The leaves were then dried at 70℃ for 48 h and weighed for dry mass. A precision scale with a resolution of 0.001 g was used for weighing. Location-specific trait values for each of the 70 observed vascular plant species were quantified as the average of the two individual trait values.
As a measure of functional composition, we used community weighted means (CWM) of these three traits following the mass-ratio hypothesis (Grime 1998), which states that ecosystem processes are determined by the trait values of dominant species in the community.
CWM of traits has been shown to correlate both with environmental conditions and ecosystem functioning (Garnier et al. 2004). We chose abundance-weighted standard deviation as the measure of functional diversity. Functional composition and diversity were calculated separately for all traits.
Carbon cycle variables CO 2 flux data We measured CO 2 exchange using a static, non-steady state non-flow-through system (Livingston and Hutchinson 1995) composed of a transparent acrylic chamber (20 cm diameter, 25 cm height). Measurements were done during the growing season in 2017 during which the weather conditions in the near-by meteorological station (10.8°C and 104 mm in July) were rather close to the average climatic conditions (11.2°C and 72 mm in July over . Measurements were done during the growing season as it is the most active season for plants with the largest net CO 2 uptake (Belshe, Schuur, and Bolker 2013). The chamber included a small ventilator, a carbon dioxide probe, an air humidity and temperature probe, and a measurement indicator. In the chamber, CO 2 concentration and air temperature were recorded at 5-s intervals for 90 s. PAR was logged manually at 10-s intervals during the same period using a quantum sensor with a hand-held meter.
Steel soil collars (21 cm in diameter and 6-7 cm in height) were inserted in the soil at least 24 h before the measurements to avoid potential CO 2 flush from soil due to the disturbance caused by the installation of the collars. The soils in the study area are relatively rocky and the plants have long horizontal roots, thus our collars could be embedded only ca. 2 cm into the soil. To guarantee an air-tight seal, we sealed the edges of the collar using inert quartz sand. The chamber was placed on top of the collar and ventilated after each measurement. We progressively decreased the light intensity of NEE measurements from ambient conditions to ca. 80%, 50% and 30% PPFD by shading the chamber with layers of white mosquito net (n = 7-10). ER was measured in dark conditions (0 PPFD), which were obtained by covering the chamber with a space blanket (n = 3). Each plot was measured at midday during the peak season in 2017 between the 26 th of June and 27 th of July.
To convert concentration measurements into flux estimates, we deleted the first and last 5 s of each measurement series to remove potentially disturbed observations. The final measurement period was thus 80 seconds for each plot. Fluxes were calculated using linear regression, and converted to µmol CO 2 m⁻² s⁻¹.

Carbon stock data
Stocks reflect the balance of carbon uptake and losses over a longer period of time. We measured the depth of the soil organic and mineral layers from three points on each central plot up to a depth of 80 cm using a metal probe. In the analyses, we used the mean value of the three point measurements to represent the organic and mineral layer depths in each sampling location. We collected samples of roughly 1 dl from the soil organic and mineral layers with metal soil core cylinders. The organic samples were collected from the top soil, and mineral samples directly below the organic layer. Organic samples were collected from all plots and mineral samples from a subset of plots. The soil samples were taken on the 1 st to 31 st of August 2016 and 2017, and were pre-processed and analysed at the Laboratory of Geosciences and Geography and the Laboratory of Forest Sciences, University of Helsinki.
Soil samples were first freeze-dried (ISO 11464:1994E). We analysed the bulk density from all organic samples, as well as the total carbon content (C%) or the soil organic matter content (SOM%). We estimated the bulk density (kg m -3 ) by dividing the dry weight by the sample volume. We used C% for the majority of the organic layer stock calculations, but for plots that were missing C% data we used SOM% (n = 24). We further analysed bulk density from 70 mineral samples and C% from 54 samples, respectively. We used the median bulk density and C% of these samples (820 kg m -3 for bulk density, range 470-1400 kg m -3 ; 3% for C%, range 0.4-6.5%) to estimate the carbon stock of the mineral layer. We used the median values, as we did not have data on the mineral soil conditions from all plots. Before C% analysis, mineral samples were sieved through a 2 mm plastic sieve. Organic samples were homogenized by hammering the material into smaller pieces. We analysed C% using elemental analysers, and SOM% using the loss-on-ignition method according to SFS 3008. SOC for organic and mineral layer was estimated using the following equations: Organic layer carbon stock = organic layer C% / 100% ⨉ organic layer bulk density (kg m -3 ) ⨉ organic layer depth (m) (Eqn 1) Mineral layer carbon stock = mineral layer C% / 100% ⨉ mineral layer bulk density (kg m -3 ) ⨉ mineral layer depth (m) (Eqn 2) Finally, organic and mineral layer stocks were summed together to calculate the total SOC stock up to 80 cm. For plots that were missing organic layer C%, SOM% was used instead of C% in Eq. 1 to calculate the soil organic matter stocks, which we converted into organic layer carbon stocks based on a relationship between C% and SOM% (carbon fraction in the soil organic matter). We calculated the carbon fraction in the soil organic matter as 0.54 (R² = 0.97), similar to Parker, Subke and Wookie (2015). For more details, see . Locations with no soil were excluded from the analysis.
Above-ground vascular plant biomass was collected between the 1 st and 10 th of August, 2017 from the collars. Biomass samples were oven-dried at 70°C for 48 h and weighed after drying. Above-ground carbon stocks (AGC) were estimated by multiplying the total biomass by 0.475 (Schlesinger, 1991).

Statistical analyses
Light response model and flux normalization All models were run in R version 3.5.3 with the package brms (Bürkner, 2018), which is an interface to the Bayesian modelling platform Stan (Carpenter et al., 2017).
To account for the effects of variation in light levels and air temperature on CO 2 fluxes, we fitted plot-specific light-response curves using a non-linear hierarchical bayesian model. We used the Michaelis-Menten equation to model the n instantaneous observations of NEE as a function of k plot-specific ER, maximum photosynthetic rate GPP max , and the half-saturation constant K (Eqns 1 & 5). ER, GPP max and K were allowed to have plot-specific averages (Eqns 2-4 & 6-8), and ER also had an exponential air temperature (T) response (Eqn 2).
The Michaelis-Menten parameters GPP max and half-saturation constant K sometimes identify weakly, so that the data would be consistent with infinitely increasing photosynthesis. This is especially true when CO 2 fluxes are small, which is frequently the case in tundra ecosystems.
To counter this, we set weakly informative priors on the plot-specific intercept terms based on visual inspection of the scale of variation in our data and typical parameter values reported in Williams et al., (2006) (Eqns 9-11). A weakly informative prior was also set for the air temperature effect on respiration β Air temperature (Eqn 12). Priors for the variance parameters were left as the weakly informative brms defaults (Eqn 13).
We used this model to predict the rate of NEE at dark (0 PPFD) and average light (600 PPFD) conditions, and an air temperature of 20℃ at each plot. Normalizing removes the effects of variation in instantaneous light and temperature conditions from the resulting variables, making them more comparable across the landscape. We refer to the flux normalized to dark conditions as ecosystem respiration (ER). We then subtracted ER from the NEE normalized to average light conditions to arrive at an estimate of normalized GPP.
Modeling tundra carbon cycling We modelled GPP, ER, AGC and SOC as functions of functional composition and functional diversity while adjusting for soil moisture and air temperature using bayesian generalized additive models (GAMs). All response variables were log-transformed before analyses to satisfy the assumption of homoscedasticity. Functional diversities were square root We then calculated the values of the tensor product smoothers at these trait values to visualize the effects of covarying SLA and LDMC. We repeated the procedure for the square root transformed functional diversities. These axes of leaf economic composition and diversity were scaled to lie between zero and one.
Data and scripts used in this study are deposited into Zenodo ).

Variability in environmental conditions and carbon cycle variables
Average values and within-landscape variabilities of carbon cycling, air temperature, soil moisture, traits and their diversities are presented in Table 2.

Model diagnostics
The diagnostics for all bayesian models were satisfactory: no MCMC chain showed erratic behaviour under visual inspection, rank-normalized potential scale reduction factors for all parameters stayed below 1.01, and visual inspection of the posterior predictive distributions showed no large systematic differences between distributions of predicted and observed data.
The median Bayesian R² (Gelman et al. 2019) for the light-response model was 0.95.
Bayesian R² values for models of stocks and fluxes are presented in Table 2 and show that the model explaining soil organic carbon stocks had the lowest explanatory power.
Effects of air temperature and soil moisture on carbon cycling Microclimate conditions and carbon cycle variables were in general positively linked to each other. In the light-response model, a one degree increase in air temperature resulted in a 2.9% (SE: 0.3%) increase in ER. In the GAM, air temperature and soil moisture increased GPP by 40% from the driest and coldest to the warmest and wettest parts of the gradient. AGC was predicted to be higher than average in warm and moist conditions, but lower than average in cool and moist environments. SOC stocks corresponded strongly with soil moisture and air temperature; average SOC stocks almost doubled along the gradient from cool and dry to warm and moist plots (Fig 2).
Trait effects on CO 2 fluxes GPP and ER had very similar positive responses to plant community functional composition and diversity (Fig. 3 Trait effects on carbon stocks AGC increased with plant height and decreased with faster leaf economic traits (Fig. 3). On the observed gradient of plant height, average AGC increased 23-fold. The gradient from slow to fast leaf economic traits caused a 70% reduction in AGC. There was also a 95% probability that AGC increases along with increasing height diversity, with the difference in AGC between highest and lowest height diversity being about 4-fold on average. Above-ground carbon stocks were smaller in moist and cold environments, which might reflect a general scarcity of vascular plant vegetation in these environments.

Plant size is the most important trait for most of the carbon cycle variables
Plant size had a strong and positive relationship with all growing season fluxes and above-ground carbon stocks, but not with soil organic carbon stocks. The relationship between plant size and photosynthesis was stronger than that of plant size and ecosystem respiration. Our study agrees with the results of Lafleur and Humphreys (2018), who discovered that taller plants (and in particular, shrubs) had larger growing season net CO 2 uptake than lower plants in the low Arctic. They argued that this was likely due to colder soils limiting heterotrophic respiration and taller plants having a larger total leaf area to capture sunlight, which likely also explains the stronger relationship that we observed for photosynthesis.
However, our data indicate that communities with high growing season productivity likely also suffer higher yearly soil carbon losses, as we found no clear links with plant size and soil organic carbon stocks that reflect the long-term annual carbon accumulation. This was an interesting finding because we hypothesized the relationship between plant size and soil organic carbon stocks to be positive as larger communities might produce larger carbon inputs to soils . The lack of this relationship suggests that in this ecosystem, other mechanisms have a stronger control over soil organic carbon stocks. Such mechanisms can be linked to carbon cycling processes during the shoulder and winter seasons or lateral transport of carbon, for example. One such mechanism is that during the winter, taller communities have warmer soils (Kropp et al. 2020), which accelerate soil respiration and increase soil carbon losses (Natali et al. 2019). Or, tall plant communities might inhabit environments where carbon leaching is high (Ma et al. 2019). In any case, our results show that tundra soil organic carbon stocks might be weakly linked to changes in growing season ecosystem CO 2 fluxes, above-ground carbon stocks, and plant size, similar to the findings by Sørensen et al. (2018).
Evidence suggests that plant size is increasing around the Arctic in response to climate change. Sometimes this is in part due to "shrubification", the increase in the biomass, cover, and abundance of woody species (Myers-Smith et al. 2011;Berner et al. 2020), but the height of non-woody species is increasing as well. Bjorkman et al. (2018) reported an average increase of about 1 cm in plant height over 1989-2015 years across the tundra. Our results suggest that such an increase would have resulted in 8-16% larger growing season photosynthesis and 6-11% times larger ecosystem respiration (assuming a linear response to height, 90% CI). From our results it thus follows that the documented changes in plant height across the tundra might have already caused large shifts in the functioning of Arctic ecosystems, especially on growing season CO 2 fluxes and above-ground carbon storage, but the effects on total ecosystem carbon storage require further investigation. New measurements in taller plant communities are likely important to address this as our study design did not cover the full plant size spectrum observed across the tundra; community-weighted plant height varied between 1 and 20 cm in this study whereas it varied between 0 and 40 cm in Bjorkman et al. (2018).
Earlier studies in the tundra have often characterized the effects of vegetation functional composition on carbon cycling with LAI Shaver et al. 2007;Dagg and Lafleur 2011;Marushchak et al. 2013). Measuring LAI does not require information on species assemblages, making it useful for landscape-scale or temporal predictions of carbon cycling Cahoon, Sullivan, and Post 2016). However, LAI is a composite of several trait axes as, for example, increases in plant size and SLA can both increase LAI (Fletcher et al. 2012). Consequently, understanding the mechanisms driving the relationships between the environment and LAI, or LAI and carbon cycling, can be challenging. Our study adds to the results of these earlier studies by partitioning plant community functional composition to community position on the two main axes of trait variation, thus improving our mechanistic understanding about vegetation-carbon linkages in the tundra.
Communities with fast traits are associated with higher soil organic carbon stocks The effects of the leaf economics spectrum on carbon cycling were more variable and in general weaker compared to the relationships between plant size and carbon cycling. Our study supports the well-established strong positive relationship between faster economic traits and photosynthesis (Williams et al. 2006;Street et al. 2007;Shaver et al. 2007), but suggests that the connection to ecosystem respiration is not as strong. The relationship between ecosystem respiration and economic traits is more complex due to the various sources of ecosystem respiration (above-ground plant tissues, roots, symbionts, and heterotrophs) which all have different drivers (Segal and Sullivan 2014;Barba et al. 2018). We speculate that there are four main mechanisms that explain higher ecosystem respiration associated with faster communities. First, higher levels of photosynthesis of fast communities also accelerates plant respiration (Lund et al. 2010). Second, faster growing plants often produce more litter that is rich in nutrients and more easily broken down by soil microbes, resulting in higher soil respiration rates (Cornwell et al. 2008;G. T. Freschet, Aerts, and Cornelissen 2012). Third, faster plants produce more root exudates, which are, together with root respiration, suggested to be a main source of soil respiration in the tundra (Illeris, Michelsen, and Jonasson 2003). Even though plant size did not have a clear effect on soil organic carbon stocks, leaf economic traits had a relatively strong relationship with it. This might indicate that differences in organic matter quality (associated with leaf economics) instead of quantity (associated with larger plants) controls soil organic carbon stocks in this landscape (Hobbie 1996). Similar to Sørensen et al. (2018), largest soil organic carbon stocks in our study were located in graminoid and forb-dominated meadow communities with fast traits whereas smallest soil organic carbon stocks were found in slow communities. Therefore, although it is generally hypothesized that fast communities often produce litter that decomposes rapidly, and that slow communities associated with evergreen species produce recalcitrant litter that decomposes slowly and accumulates into the soil (Hobbie et al. 2000), these hypotheses did not explain our results. Instead, our results suggest that mechanisms related to roots or microorganisms likely have a stronger control on soil organic carbon stocks. For example, communities with fast traits might have deeper-reaching roots and larger amounts of biomass below-ground (Ylänne et al. 2018;Iversen et al. 2015), and conditions deeper in the soil might make root litter relatively resistant to microbial decomposition compared to leaf litter and root litter in higher soil layers (De Deyn, Cornelissen, and Bardgett 2008;G. T. Freschet et al. 2013). Further, as speculated by Sørensen et al. (2018), microorganisms could retain more carbon in the soil due to the arbuscular mycorrhiza associated with faster communities whereas slower communities with mycorrhizal fungi and ectomycorrhiza could decompose organic matter faster (Väre, Vestberg, and Eurola 1992;Becklin, Pallo, and Galen 2012;Parker, Subke, and Wookey 2015). Moreover, communities with slow leaf economics often have evergreen leaves that produce relatively small carbon inputs to the soil compared to faster communities that lose their leaves each year (Hobbie 1996).
Nevertheless, a large part of variation in soil organic carbon stocks remained unexplained, even when soil moisture and air temperature conditions were considered. It is possible that these weaker links are partly due to practical reasons, as we measured soil organic carbon outside the vegetation collar (see Fig. S1). However, our study design captured the entire continuum of fast-slow communities representative of the entire tundra biome as well as the global spectrum of leaf economics spectrum (Thomas et al. 2020), thus our results should provide relatively robust results describing the direction and magnitude of the relationship between leaf economic traits and carbon cycle variables.

Functional diversity of leaf economics increases ecosystem functioning
All fluxes and soil organic carbon stocks responded positively to the functional diversity of leaf economics, but not to the diversity of plant size. Thus, within-community variability on the slow-fast strategy axis increased ecosystem functioning, supporting the niche complementarity hypothesis (Cadotte, Carscadden, and Mirotchnick 2011). Similar findings of a positive correlation with plant diversity and soil organic carbon stocks have also been made in observational (Chen et al. 2018) and experimental (Lange et al. 2015;Fornara and Tilman 2008) studies. There are several mechanisms for how soil organic carbon stocks could increase with plant functional diversity. One such mechanism is that increased productivity could lead to higher litter inputs from leaves and roots Chen et al. 2018). This is plausible, since leaf economic diversity also increases photosynthesis and the ratio of photosynthesis to ecosystem respiration. Niche complementarity pertaining to nutrient-use strategies could also lead to more efficient use of below-ground niche space and thus higher total investments in acquiring below-ground resources. Other mechanisms are related to soil microbes. High plant diversity might enhance the diversity and activity of soil microbial communities and decrease carbon losses from microbial decomposition (Chen et al. 2018).
Functional diversity of plant size was positively if somewhat uncertainly associated with above-ground carbon stocks, indicating that multi-layered vegetation stores more carbon compared to less structurally diverse plant communities. Similar effects of size-related trait diversity on above-ground biomass have been reported for temperate semi-natural grasslands (Schumacher and Roscher 2009). On the other hand, Conti and Díaz (2013) reported a negative correlation between structural diversity and above-ground carbon stocks in subtropical forests, but note that this is probably due to confounding factors. Our results provide more evidence that structurally layered plant communities might contain more carbon than vertically homogeneous vegetation patches, although the effects were uncertain enough that further research is needed to verify this relationship.
Characterizing the underground community should be a priority Our results showed that plant functional community composition and diversity were good predictors of above-ground carbon stocks and ecosystem fluxes whereas a large part of variation in soil organic carbon stocks remained unexplained. This is an issue as most of the carbon in tundra ecosystems is located in the soil (Hugelius et al. 2014;Mishra et al. 2021).
Field and laboratory evidence suggests that this carbon pool is vulnerable to warming and permafrost thaw (Voigt et al. 2017;Schädel et al. 2016), but it is not well known how it responds to the wide-spread shifts in tundra vegetation (Bjorkman et al. 2018;Berner et al. 2020;Parker et al. 2021). This is because soil organic carbon stocks are a result of various carbon inputs from above-and below-ground plant tissues, outputs from ecosystem and soil respiration, and lateral transport, which all respond differently to trait variation. Therefore, interpreting the effects of above-ground traits on soil organic carbon stocks is challenging as the links might be weak and indirect (e.g., via litter decomposition and accumulation).
Further, soil organic carbon stocks are strongly driven by abiotic environmental conditions as well, some of which might have been missing from our model. For example, soil organic carbon stocks can be influenced by geomorphological processes such as cryoturbation, or aeolian and fluvial transportation of litter (Klaminder, Yoo, and Giesler 2009), or soil development and carbon accumulation in the past (Palmtag et al. 2015).
Most importantly, our results highlight that we need more measurements and understanding on belowground community composition, diversity and functioning of plant roots, litter, and microorganisms (Orwin et al. 2010). With recent advancements such as a new a root ecology handbook (G. Freschet et al. 2020) and the identification of major axes of variation in below-ground traits (Bergmann et al. 2020), some of these measurements can soon be done in a comparable way across the globe. Such measurements will help us understand the mechanisms influencing soil organic carbon stocks and vegetation-carbon feedbacks in a changing climate.

Conclusions
We found that carbon cycling in the tundra was well explained by plant community functional composition and diversity. Average plant size was the strongest predictor of most

Photosynthesis
The process by which plants build carbohydrates from light and CO 2 .
GPP Gross primary productivity normalized at a common irradiance (600 PPFD, partly sunny conditions)

Ecosystem respiration
The process by which plants, soil animals and microbes convert sugars into energy, thus releasing CO 2 to the atmosphere.

ER
Ecosystem respiration normalized at a common air temperature (20°C, average chamber air temperature) Above-ground carbon stock The carbon content in the living plant biomass. AGC Above-ground carbon stocks were estimated by multiplying the above-ground vascular plant biomass by 0.475.

Soil organic carbon stock
Soil organic carbon stocks in the soil horizon (up to 80 cm) excluding the largest roots.

SOC
Soil organic carbon stocks were estimated using the information of soil organic and mineral layer depth, bulk density, and carbon or organic matter content.