Carbon allocation to root exudates is maintained in mature temperate tree species under drought

allocated to root exudation during early summer by collecting root exudates from mature Fagus sylvatica and Picea abies exposed to experimental drought, and combining above-and belowground C ﬂuxes with leaf, stem and ﬁne-root surface area. (cid:1) Exudation from individual roots increased exponentially with decreasing soil moisture, with the highest increase at the wilting point. Despite c . 50% reduced C assimilation under drought, exudation from ﬁne-root systems was maintained and trees exuded 1.0% ( F. sylvat-ica ) to 2.5% ( P. abies ) of net C into the rhizosphere, increasing the proportion of C allocation to exudates two-to three-fold. Water-limited P. abies released two-thirds of its exudate C into the surface soil, whereas in droughted F. sylvatica it was only one-third. (cid:1) Across the entire root system, droughted trees maintained exudation similar to controls, suggesting drought-imposed belowground C investment, which could be beneﬁcial for ecosystem resilience.


Introduction
In recent years, important processes controlling ecosystem carbon (C) dynamics and plant susceptibility to drought have been identified in the rhizosphere -the interface between plant roots and the soil environment (Finzi et al., 2015;Joseph et al., 2020;Williams & de Vries, 2020).In this narrow zone, plants interact with their environment by releasing root exudates, which fulfill fundamental roles in the regulation of microbial growth (de Graaff et al., 2010), the liberation of C from protective associations with minerals (Keiluweit et al., 2015), maintenance of soil hydrological properties (Carminati et al., 2016) and communication with plants and other organisms (Bais et al., 2006).Collectively, these interactions facilitate water and nutrient acquisition (Coskun et al., 2017;Williams et al., 2021), microbiome selection (van Dam & Bouwmeester, 2016), and plant species interactions (Ehlers et al., 2020) that can alleviate plant stress (Vives-Peris et al., 2020).Potential shifts in C allocation to exudates in drought-exposed ecosystems can affect many of the processes influenced by root exudates.However, although drought is a major natural risk that threatens the functionality of long-living ecosystems such as forests in the 21st century (IPCC, 2018), we do not know how shifts in C allocation to root exudates in response to soil water limitation are related to tree C budgets.
Trees respond to reduced water supply by modifying their belowground C allocation (Rühr et al., 2009;Hagedorn et al., 2016;Hommel et al., 2016) and potentially increase root exudation rates (Karst et al., 2017;Karlowsky et al., 2018;Preece et al., 2018;de Vries et al., 2019;Jakoby et al., 2020).However, most studies only use single root branches -defined as ephemeral terminal branch orders -to describe a plant´s exudation behavior, which does not consider changes in root growth, distribution, and longevity that can also be significantly altered under drought (Nikolova et al., 2020;Zwetsloot & Bauerle, 2021).Allometric scaling of root exudates from a single root branch to the entire root system, while accounting for changes in root production and longevity, can advance our understanding of species-specific belowground C allocation patterns during periods of drought and improve terrestrial biosphere models (Fatichi et al., 2019).In combination with an assessment of aboveground net-C assimilation, calculating the balance of belowground C allocation dynamics can identify whether trees "invest" in the production of root exudates under drought.
Belowground C allocation has been assessed in pot experiments with small annual or perennial species (Kaštovská et al., 2015;de Vries et al., 2019) and tree saplings (Hagedorn et al., 2016;Preece et al., 2018).However, findings from these experiments cannot be easily translated to mature forest ecosystems.Soil water dynamics not only deviate drastically between homogenized and naturally developed field soils but also between surface soil and subsoil.Consequently, it is difficult to simulate exudation dynamics in artificial setups and field-based studies are required to understand how an entire root system responds to drought.Previous studies addressing the impact of drought on root 4 exudation failed to include measurements across different soil depths, although general vertical variations in exudation rates were identified (Finzi et al., 2015;Tückmantel et al., 2017).However, altered root distribution patterns with depth may affect root-system level exudation and consequently whole-tree C budgets.Stable-isotope labeling studies have allowed C-flux integration over the entire rooting zone but this was usually achieved by tracing belowground C allocation via microbial activity (Joseph et al., 2020;Gao et al., 2021).Since microbial respiration is hampered under drought (Moyano et al., 2013), tracing C via microbial activity may hide potential increases in exudation, particularly if vertical variations occur.To scale root exudates to C-allocation dynamics in a forest ecosystem, vertically separated in situ exudate capture, combined with belowground root abundance is needed.
Root growth and exudation responses to water limitation may vary among tree species according to their drought susceptibility.Shallow-rooting species can be particularly vulnerable to drought; for example, when exposed to seasonal drought, Picea abies (L.) Karst., one of Central Europe's most abundant and economically important tree species (Caudullo et al., 2016) had a five-fold higher mortality rate compared to Fagus sylvatica L. (Pretzsch et al., 2020), a broadleaf species representing the widespread natural vegetation in Central Europe (Fang & Lechowicz, 2006).Each species exhibited different root responses to drought, with F. sylvatica having an inherently deeper root system (Schmid & Kazda, 2002), reduced fine-root diameter, and increased specific root area to improve water uptake (Comas et al., 2013;Hertel et al., 2013;Nikolova et al., 2020).By contrast, P. abies did not respond to soil moisture deficit by growing new, deeper roots but instead prolonged existing fine-root lifespan (Zwetsloot & Bauerle, 2021).It is likely that earlier seasonal transpiration by P. abies compared to deciduous F. sylvatica results in lower soil moisture under P. abies throughout the year (Grams et al., 2021).Thus, the potential lack of access to water from deeper soil and overall lower soil moisture may amplify the susceptibility of P. abies to drought.Given the potentially crucial role of root exudates in response to water limitation, greater root exudation by both F. sylvatica and P. abies would be anticipated at root branches located in dry soils.In P. abies, prolonged root-system lifespan in dry surface soils may imply higher exudation across a larger proportion of P. abies root systems.By contrast, for the more dynamic root system of F. sylvatica, overall exudation amounts are harder to predict.
In this study, we utilized a novel throughfall-exclusion experiment in a mature temperate forest, which imposed five years of severe drought during the entire growing season, to test if the allocation of photosynthates to root exudation increases under drought.We combined vertically distributed in situ root exudation measurements with fine-root surface area observations throughout the soil profile of mature P. abies and F. sylvatica trees to identify C partitioning at the whole-tree level.We hypothesized that 1) roots in dry surface soils exude more C than roots in deeper moist soils and root Therefore, allocation of C to exudates will be greater for the more drought-susceptible P. abies than for F. sylvatica.We further hypothesized that 2) at the tree level, the proportion of C exuded by roots increases relative to net-photosynthetic C assimilation, which could be considered as a greater investment into root exudation in water-limited trees.

Site description
Sampling occurred at the 'Kranzberg Forest Roof' (KROOF) long-term drought experiment located in southern Bavaria, Germany (N 48° 25.2'; E 11° 39.7').Drought was imposed on six throughfall exclusion plots (sizes between 110 and 200 m 2 ; Grams et al. (2021)) via automated understory roofs that withheld throughfall during the growing season (April to November).On average, roof closure withheld c. 70% of total annual precipitation during five years of simulated drought (Grams et al., 2021).Six additional plots without roofs served as non-droughted controls.The mixed stands comprised large groups of F. sylvatica (90 ± 4 years old) surrounded by P. abies (70 ± 2 years old) trees.Each plot consisted of an F. sylvatica and a P. abies cohort with 3-6 individuals each (Grams et al., 2021).The soil at the site originated from Loess over Tertiary sediments and was classified as haplic Luvisol (FAO Classification) with moder type humus.Sediments form a loamy dense layer at c. 50 cm depth that is difficult for roots to penetrate, so that > 90% of roots are found between 0-50 cm depth (Häberle et al., 2012).Soil pH was between 3.8-4.6(P.abies: 4.1, F. sylvatica: 4.5) and C:N ratios typically decreased with depth and were higher under P. abies (14.4 ± 0.6) compared to F. sylvatica (12.5 ± 0.4; Table S1).

Root exudate collection and analysis
We sampled intact root branches in each of three drought and three control plots in previously installed root window boxes (40 cm long, 40 cm wide, c. 50 cm high; n = 3 per plot) that allowed access to roots without disturbing the experimental site.Root branches, comprising 1 st -3 rd order roots attached to a single transport root, were randomly selected for sampling (Figure S1).Sampled root branches had an average weight of 0.20 ± 0.02 g, an average fine-root (≤ 2 mm diameter) surface area F o r P e e r R e v i e w 6 of 17.15 ± 1.83 cm 2 and 23.9 ± 4.5 tips per cm 2 root surface area (Table S2).We sampled exudates from root branches growing in surface soils at the interface between the organic layer and mineral soil (0-7 cm depth) and the mineral soil (7-30 cm depth) according to Phillips et al. (2008).Briefly, root branches were carefully excavated, and the soil was gently removed with tweezers and by rinsing with a nutrient solution to limit osmotic stress (0.5 mM NH4NO3, 0.1 mM KH2PO4, 0.2 mM K2SO4, 0.15 mM MgSO4, 0.3 mM CaCl2).We excluded dead roots and roots that did not pass a vitality check (i.e.no lateral roots present or black tissue color) from sampling and evaluation.Afterwards, root branches were left to recover for 48 hours in a 1:1 mixture of sand and native soil from the site, cleaned again, and placed into 30-ml glass syringes containing sterile glass beads simulating a physical soil environment.Syringes were flushed three times with the nutrient solution and then equilibrated for 48 hours, flushed again, and left wrapped in aluminum foil and covered with leaf litter.After another 48 hours, we extracted root exudates trapped in the syringes using a membrane pump after adding 30 ml nutrient solution.
We sampled 36 root branches in total, 18 from F. sylvatica and 18 from P. abies at either 0-7 cm or 7-30 cm soil depth (Table S3).Blank syringes (n = 4) with glass beads, flushed with nutrient solution but without root branches, served as a reference.Root exudates were filtered through sterile syringe filters (0.22 µm, ROTILABO® MCE, Carl Roth GmbH + Co. KG, Karlsruhe, Germany) and stored at 4° C until analysis.All consumables were acid-washed in 1% HNO3 before use.Root exudation below 30 cm soil depth was estimated from minirhizotron and soil water content data (see 2.4.4.).
Exudate samples were quantitatively analyzed for total non-purgeable organic carbon concentration (TOC) with a multi N/C 2100 S (Analytik Jena GmbH, Jena, Germany).The method included the removal of total inorganic carbon by adding 50 µl 2 M HCl and flushing with synthetic air (180 s).The detection limit was 69.8 µg C L -1 .

Root characteristics
All root branches were harvested after exudate collection and scanned at 1200 dpi (Epson Perfection 4990 Photo, SEIKO Epson CORPORATION, Tokyo, Japan).Root-surface area and the number of root tips were determined using WinRhizo (WinRHIZO Pro 2016a, Regent Instruments Inc., Quebec, Canada).
Root branches were dried and total dry biomass was recorded.Measured exudate TOC was expressed per root-surface area with a diameter ≤ 2 mm (henceforth: fine roots) of each branch, to correspond to sampled roots from soil coring (see 2.4.4).We also related exudation rates to the dry biomass of the branches (Figure S2) and to absorptive-root density (Figure S3), calculated as the number of root tips per unit of total surface area of the root branches (Table S2).Similar trends with treatment and depth were observed regardless of which parameters were used for normalization.1500 µmol m -2 s -1 ) gas exchange rates (Asat) were determined at 400 ppm carbon dioxide (CO2) concentration for two trees per species and plot using an open gas-exchange system (LI-6800, Li-Cor Inc., Lincoln, NE, USA) over two weeks in June 2019.Gas exchange rates were modeled for leaves in the shade crown for both species and six different needle ages for P. abies (see supplement).Light response curves were derived for leaves in the sun and shade crowns of F. sylvatica and P. abies, assuming steady assimilation at respective light saturation points (Larcher, 2001;Matyssek, 2010), a linear decrease between light saturation and light compensation and leaf respiration below light compensation (see supplement).Assimilation rates were derived from light response curves during each 10-min interval when PAR was measured during exudate sampling.Daily assimilation rates were calculated assuming constant light conditions within these 10-min intervals.The total leaf area for F. sylvatica and P. abies was calculated using allometric equations determined individually for both species based on tree diameter and tree height (Patzner (2004); Table S7).No reduction in the leaf area was detected for F. sylvatica or the shade crown of P. abies in drought plots, while the leaf area in the sun crown of P. abies trees in drought plots was c. 50 % lower compared to trees on control plots (data not shown) and the reduction was considered in our calculations accordingly.To obtain daily C assimilation per tree, leaf areas of the shade and sun crown were multiplied with assessed assimilation rates.Daily C assimilation was summed for all trees per species and plot and divided by plot size (Grams et al., 2021) to obtain assimilation per species and m 2 and day, assuming each species occupied 50 % of the plots as species distribution was uniform (Grams et al., 2021).

Stem respiration
Stem respiration (µmol CO2 m -2 stem area and s -1 ) was measured on two F. sylvatica and two P. abies trees per plot using custom-built chambers (60-204 cm 2 ) that were sealed to the stem at 1-m height with Terostat-IX (Henkel AG & Co. KGaA, Duesseldorf, Germany).Respired CO2 was measured with a Delta Ray Isotope Ratio Infrared Spectrometer (Thermo Fisher Scientific, Waltham, MA, USA) in 5-min intervals during day and night when C assimilation measurements took place.The cumulative daily stem respiration was calculated for each tree on days when C assimilation was measured, and scaled to the total tree stem area based on tree diameter and height (based on a conical tree shape; McDaniel et al. (2012), Rance et al. (2012); Table S7), assuming unchanged respiration rates along the stem (see supplement).

Soil and root respiration
Soil respiration rates were used to estimate the microbial response to drought and to calculate root respiration.Soil CO2 efflux (µmol m -2 plot area and s -1 ) was measured via permanent soil collars (PVC pipe, 20-cm inner diameter, 12 cm height), which were inserted c. 2 cm deep into the soil and sampled every 30 min to 1 h per tree species for seven days in each plot (n = 1-3 per species and plot) using a multiplexed automated soil-chamber system (LiCor-8100M, LiCor Biosciences, Lincoln, NE, USA), and the daily sum was calculated per plot (see supplements).We averaged the daily sums of seven-day measurement periods per plot and species to calculate the contribution of root respiration to total soil respiration using estimates from the site, i.e.50 % for F. sylvatica and control P. abies trees and 40 % for P. abies trees on drought plots (Nikolova, 2007).

Exudation at the root system and tree level
Fine-root biomass, surface area, and number of tips per plot were assessed using two soil cores (34 mm diameter) per species and plot in October 2018.Cores were taken randomly within the rooting zones of each species and divided into two depth increments (0-7 cm and 7-30 cm mineral soil depth; Nickel et al. (2018).Fine roots (≤ 2 mm) were extracted from cores by washing with tap water and separated by species under a stereomicroscope.Fine roots were scanned and analyzed for surface area and the number of tips using WinRhizo (WinRHIZO Pro 2016a, Regent Instruments Inc.), and subsequently dried to assess dry fine-root biomass.Fine-root surface area (Frsa) per m 2 for each species and soil depth was calculated from fine-root surface area per soil core (Frcore), using the core volume (Vcore) and the respective thickness of the soil depth increment (7 cm for 0-7 cm and 23 cm for 7-30 cm soil depth): The total number of fine-root tips per m 2 was calculated using the same function, i.e. by dividing root tips per soil core by core volume and multiplying by soil increment thickness.
Although most fine roots of both species were in the upper 30 cm (Zwetsloot et al., 2019), we estimated fine-root surface area at 30-50 cm soil depth to integrate over the entire rooting zone (Häberle et al., 2012).Since no soil cores were taken to this depth, we analyzed images from minirhizotron tubes (six per plot, capturing roots of both species and each reaching a vertical depth of 50 cm), taken every two weeks during the growing season, and once a month during the winter months with a minirhizotron camera (BTC-100X Camera, Bartz Technology, Carpinteria, California; Zwetsloot et al., 2019; see supplements).We analyzed the number of root tips from minirhizotron images for the 7-30 cm and 30-50 cm depth layers, respectively, and calculated their ratio to estimate fine-root surface area below 30 cm.There were 1.9 times more tips at 7-30 cm than at 30-50 cm for F. sylvatica, F o r P e e r R e v i e w 9 and 12.4 times more tips for P. abies.Using these factors, the total number of root tips for the 30-50 cm soil was calculated from the number of root tips obtained from cores: ! "# $# = !% "# 1.9 12.4 ⁄ A non-linear regression between the number of fine-root tips and fine-root surface area ( = 8.1 * !#." , R 2 =0.4,p < .001) was then used to estimate fine-root surface area at 30-50 cm depth.

Soil water content
Volumetric soil water content (SWC) was lower in drought plots compared to control plots for both species but the difference was only significant at 0-7 cm depth (Table 1).Under drought, P. abies trees tended to have the lowest SWC across all soil depths and 0-7 cm soils were significantly drier than the deeper 7-30 cm and 30-50 cm soils under both species (Table 1).In the control plots, SWC at 0-7 cm depth was lower than SWC below 30 cm but neither differed from SWC at 7-30 cm (Table 1).species and treatments within each soil depth increment (0-7 cm, 7-30 cm, and 30-50 cm, respectively).Capital letters indicate significant differences between soil depths within the same species and treatment.Values are given as means with standard errors for n = 3 plots per treatment.

Exudation rates of single root branches
Neither biomass nor fine-root surface area of root branches differed between species, treatments or depths, whereas root tip abundance and estimated absorptive-root density were overall higher in F.
sylvatica than in P. abies (Table S2).Exudation rates were significantly higher in the dry 0-7 cm soil than in the more moist 7-30 cm soil, for both species in drought plots (Figure 1, Figure S1 and Figure S2).Exudation rates per fine-root surface area were 3.8 ± 2.1 µg C cm -2 d -1 in 0-7 cm depth and 0.6 ± 0.4 µg C cm -2 d -1 in 7-30 cm depth for F. sylvatica (p = .1)and 5.8 ± 2.5 µg C cm -2 d -1 in 0-7 cm and 0.9 ± 0.3 µg C cm -2 d -1 in 7-30 cm for P. abies (p < .01; Figure 1).In the control plots, where the vertical SWC distribution was more homogeneous, exudation rates did not differ across soil depths for either species.Average exudation rates per fine-root surface area did not differ between drought plots and control plots.However, in the drought plots, there was a strong trend towards increased exudation in the 0-7 cm depth and decreased exudation in 7-30 cm depth compared to controls (Figure 1).Exudation rates of root branches per fine-root surface area declined with increasing SWC across treatments and soil depths in P. abies.Although a similar trend of declining exudation with increasing SWC was detected in F. sylvatica, the relationship was not statistically significant (Figure 2).Overall, root branches exuded more C at lower SWC than at higher SWC under drought (Figure 2, Figure S4).In both species, a single root branch in the driest 0-7 cm soil exuded substantially higher amounts of C than all other root samples (Figure 2).However, there were no distinctive features to these rootsother than being in the driest soils -that would justify removing them from the dataset.Interestingly, expressing exudation rates per number of root tips (Figure S5) brought the exudation rate in the F. sylvatica root branch with the highest exudation rate closer to the mean values of the other root branches, supporting our assumption that the high exudation rates were reliable.Due to the high variability in a few data points, we also ran the regression analyses without the two high-exuding branches in the driest soils and obtained a similar relationship between root exudation and SWC regardless of whether or not these two datapoints were included in the model (Figure S6).We identified a SWC threshold (the maximum curvature of the power function) at which exudation rates increased, which was similar for both species: 9.1 vol-% SWC for P. abies and 8.3 vol-% for F. sylvatica

Root exudation and carbon allocation at the root system and the tree level
Fine-root surface area did not differ between treatments (Table 2).However, for both species there was a trend towards a smaller proportion of fine-root surface area at 0-7 cm depth in the drought plots, while the proportion of fine-root surface area at 7-30 cm and 30-50 cm soil depth was greater compared to the controls (Table 2).Numbers right of the dotted line give the fine-root distribution (as % of the total fine-root surface area) across the soil profile in three depth increments.Note that fine-root abundance at 30-50 cm depth was modeled from minirhizotron regression data (see methods).There were no significant differences between treatments.Values are given as means with standard errors for n = 3 plots per treatment.
Scaled to the root-system level, fine-root exudation across all soil depths did not differ between species or treatments (Figure 3A).Fine-root exudation of F. sylvatica trees was 0.099 ± 0.023 g C m -2 d -1 in control plots and 0.106 ± 0.037 g C m -2 d -1 in drought plots, whereas fine-root exudation of P. abies amounted to 0.091 ± 0.021 g C m -2 d -1 in control and 0.119 ± 0.044 g C m -2 d -1 in drought plots (Figure 3A).The amount of C exuded at the root-system level did not change with soil depth for F. sylvatica, but there was a trend towards higher exudation rates below 30 cm depth in drought (0.022 ± 0.003 g C m - 2 d -1 ) compared to control plots (0.013 ± 0.002 g C m -2 d -1 , Figure 3A, Figure 4).In drought plots, P. abies tended to exude more at 0-7 cm and 30-50 cm depth (0.079 ± 0.050 g C m -2 d -1 and 0.016 ± 0.005 g C m -2 d -1 , respectively) than in control plots, whereas exudation at 7-30 cm depth (0.024 ± 0.014 g C m -2 d -1 ) was lower than in control plots (0.047 ± 0.025 g C m -2 d -1 , p > .05; Figure 3A, Figure 4).
During early summer, both, F. sylvatica and P. abies trees in drought plots assimilated less than half the C of trees in control plots.Assimilation of F. sylvatica was 25.5 ± 4.8 g C m -2 d -1 in control and 12.7 ± 3.9 g C m -2 d -1 in drought plots (p = .05),whereas P. abies assimilated 22.5 ± 2.4 g C m -2 d -1 in control and 8.3 ± 0.7 g C m -2 d -1 in drought plots, respectively (p < .05).At the tree level, stem respiration did not differ between species but there was a trend towards higher stem respiration in F. sylvatica in control (3.0 ± 0.5 g C m -2 d -1 ) compared to drought plots (0.8 ± 0.2 g C m -2 d -1 ; p = .07)and stem respiration also tended to be higher in control P. abies (4.6 ± 1.0 g m -2 d -1 ) than in P. abies in drought 14 plots (2.8 ± 0.7 g m -2 d -1 , p = .1;Figure 4).Root respiration of F. sylvatica in control (3.8 ± 1.1 g m -2 d -1 ) was significantly higher than root respiration in drought plots (1.4 ± 0.5 g m -2 d -1 , p < .05)and somewhat higher than of P. abies.Roots of P. abies in control plots (2.9 ± 0.9 g m -2 d -1 ) tended to respire more than roots in drought plots (0.7 ± 0.1 g m -2 d -1 , p = .1,Figure 4).Net assimilation was higher in control than in drought plots in both F. sylvatica (18.7 ± 4.0 g C m -2 d -1 in control and 10.6 ± 3.5 g C m -2 d -1 in drought plots; p = .1)and in P. abies trees (15.1 ± 2.2 g C m -2 d -1 in control and 4.8 ± 0.4 g C m -2 d -1 in drought plots; p = .07;Figure 4).The proportion of net-C assimilation allocated to root-system exudation (Exfra) during early summer in F. sylvatica trees was 0.5 ± 0.1 % in control plots and doubled to 1.0 ± 0.1 % of net assimilation in drought plots (p = .1,Figure 3B, Figure 4).In P. abies trees, 0.7 ± 0.2 % of net-C assimilation was allocated to root exudates in control plots, whereas in drought plots the proportion of net-C assimilation allocated to fine-root exudation increased more than threefold (2.5 ± 1.0 %, p < .05, Figure 3B, Figure 4).

Discussion
Our study aimed to investigate whether tree species increased C allocation to root exudation in response to drought, both at the individual root and at the whole-tree level.Consistent with our first hypothesis, P. abies root exudation rates increased with decreasing soil water content and root exudates in F. sylvatica showed a similar trend, indicating increased exudation rates of root branches in dry surface soils.When scaled to the whole-tree level, fine-root exudation did not differ between the control and drought treatment.However, the proportion of net-C assimilation partitioned to root exudation was significantly higher for trees under drought, supporting our second hypothesis that the belowground investment increases when water becomes limited.We found stronger evidence to support both hypotheses in the more drought-susceptible P. abies, but F. sylvatica showed similar trends.

Lower soil water content promotes C exudation of root branches
Various studies have found elevated exudation when roots were exposed to dry soil (Karlowsky et al., 2018;Preece et al., 2018;de Vries et al., 2019;Jakoby et al., 2020).Accordingly, we hypothesized that exudation rates would be highest from roots exposed to the lowest soil water content (SWC).
Supporting this hypothesis, we found significantly higher exudation rates for both species in the drier surface soil under drought, whereas exudation rates in the moister control plots, where vertical differences in SWC were less distinct, did not differ across soil depths (Figure 1, Table 1).These trends persisted regardless of whether exudation was normalized by root biomass or absorptive-root density (Figure S1, Figure S2).However, and in contrast to previous studies (Finzi et al., 2015;Tückmantel et al., 2017), root exudation tended to increase with depth under control conditions, which may reflect site-specific soil texture characteristics (Grams et al., 2021).We found a threshold at low SWC where root exudation rates increased sharply (9.1 vol-% SWC for P. abies and 8.3 vol-% for F. sylvatica; Figure 2), which corresponded to the wilting point in the loess-dominated silty soil at the study site (Grams et al., 2021), suggesting that trees were stimulated to release exudates when water availability became severely limiting.However, it is unlikely that exudation rates increase indefinitely with decreasing SWC, as there is evidence that root exudation is eventually reduced under severe drought (Williams & de Vries, 2020), e.g. when roots lose contact to the soil.However, given that the SWC in the rhizosphere is less dynamic and likely higher under drought than the SWC of non-rooted soil (Carminati, 2013;Holz et al., 2018), the SWC of the rhizosphere may differ from the bulk soil measurements captured by the TDR method used in this study.Thus, exudation may already be stimulated at higher rhizosphere SWC than the observed threshold indicates.Fine-scale spatio-temporal measurements in the rhizosphere could further elucidate the relationship between SWC and root exudation.Although, we found no changes in absorptive-root density with drought (Table S2), further studies are necessary to identify whether and how root morphology interacts with root exudation under drought (Wen et al., 2022).
Since we did not sample root exudates from dead roots and excluded roots that did not pass the vitality check, the presented exudation rates only reflect those in vital tree roots.Nonetheless, the in situ exudate capture approach provides a reasonable measure of soluble C input to the rhizosphere under drought, a fraction of C that is disregarded when belowground C allocation is solely traced via respiratory losses from soil.As soil microbial activity declined under drought (indicated by reduced soil CO2 efflux, Table S6), increased exudation under low SWC might not be captured by measurements of respiratory losses or microbial biomass.For example, Joseph et al. (2020) reported that C mineralization was strongly reduced in soils below 15 % SWC, which was close to the threshold at which we measured the highest exudation.Consequently, elevated exudation may contribute to C accumulation in the dry surface mineral soil, where large increases in C stocks at 0-5 cm depth were measured (Brunn et al., unpublished data).4.2.Belowground C allocation at the root-system level is maintained under drought Despite aboveground growth reduction and declining photosynthesis rates, several studies have reported increased belowground C allocation to roots under drought (Poorter et al., 2012;Hagedorn et al., 2016;Hommel et al., 2016;Jakoby et al., 2020).Although the opposite has also been shown (Rühr et al., 2009), these studies mostly measured C allocation as root growth or exudation at the rootbranch level but did not assess whether C exudation at the root-system and the tree level also increased.Extending root C exudation to larger scales helps to identify processes related to the upand down regulation of exudation at the whole-tree level and the linkage to rhizosphere characteristics (Prescott et al., 2020;Schnepf et al., 2022).Given the potential ecological benefits of belowground C allocation in the forest´s capacity to recover from drought (Hagedorn et al., 2016) and for tree drought tolerance (Carminati et al., 2016), we hypothesized that trees would increase the partitioning of C from net photosynthesis into root exudation under drought.
We found an overall reduction in net-C assimilation with drought for both species, > 40 % in F. sylvatica and > 60 % in P. abies.However, in contrast to declining aboveground C assimilation, belowground C release through fine-root exudation at the root-system level remained constant with drought (Figure 3A, Figure 4), suggesting that the reduced fine-root surface area at 0-7 cm depth and increased fineroot surface area at 7-30 cm depth (Table 2) were compensated by higher exudation in surface and lower exudation in deeper soils.Nevertheless, the fraction of net-C assimilation allocated to root exudates doubled for drought-stressed F. sylvatica trees and tripled for P. abies (Figure 3B, Figure 4), supporting our second hypothesis that trees under drought partition relatively more available C to root exudation at the tree level.
In our study, the proportion of net-C assimilation allocated to root exudation was only 0.6 ± 0.1 % and 1.8 ± 0.6 % in control and drought plots, respectively, which was below the 3-30 % previously reported at other study sites for multiple species (Kuzyakov & Domanski, 2000;Jones et al., 2009;Finzi et al., 2015;Abramoff & Finzi, 2016;Gougherty et al., 2018).Our observed exudation rates from root branches are in line with modeled or measured root exudation rates of diverse vegetation types (Finzi et al., 2015;Dror & Klein, 2021;Rog et al., 2021;Sell et al., 2021), although they are at the lower end of reported values from comparable temperate forests (Tückmantel et al., 2017;Meier et al., 2020) and other ecosystems (summary provided by Gougherty et al. (2018).Discrepancies in exudate estimates across studies may arise due to methodological differences such as filter size variations (0.2 µm vs. 0.7 µm) or the use of C-free materials (Gougherty et al., 2018), bedrock characteristics (Meier et al., 2020), or potential reuptake during longer collection periods (Oburger & Jones, 2018).In this study, we targeted low molecular weight substances of vital roots and thus excluded other rhizodeposits or volatile compounds (Delory et al., 2016), which might account for a large fraction of 18 previously reported root C deposition rates.Low root-system level exudation could also be related to physiological conditions varying throughout seasons, as exudates may not peak in early summer when we sampled, but in the late summer and autumn (Jakoby et al., 2020), when fine-root production is higher (Abramoff & Finzi, 2016;Zwetsloot et al., 2019).As net-C assimilation is lower in autumn, the proportion of total C assimilation allocated to root exudates might therefore be substantially higher towards the end of the growing season.Thus, the presented C fluxes may not reflect whole year dynamics but give an accurate approximation of relative and absolute exudation patterns of mature trees during early summer.We did not measure exudation or fine-root surface area in the deepest soil increment, but we ensured high scaling accuracy to the whole-tree level by observing and modelling C fluxes of different soil depths and entire above-and belowground compartments (see supplementary methods S1 for further discussion on accuracy).Exudate C may have partially originated from tree C storage pools that were reduced under drought (Hesse et al., 2021).However, there is indication of rapid belowground allocation of recently fixed C (Gorka et al., 2019;Fossum et al., 2022) and exudates at the experimental site contained at least 65-90% newly assimilated C (Hikino et al., unpublished data).
Our approach did not allow us to account for potential C fluxes to mycorrhizal fungi.However, root exudation in ectomycorrhizal trees under drought can be twice as high as under well-watered conditions (Liese et al., 2018) suggesting preferential C allocation to exudation than to mycorrhizae.
Although the rate of colonization for our exclusively ectomycorrhizal trees was comparable between control and drought plots, the number of vital ectomycorrhizal tips declined by >70% after three years of drought at the experimental site (Nickel et al., 2018).This decline was accompanied by changes in ectomycorrhizal species composition, suggesting a relative increase in more C-demanding ectomycorrhizal types able to forage long distances (Nickel et al., 2018).Thus, it is unclear whether drought altered the partitioning of belowground C to exudates or mycorrhizae.Nonetheless, the presented rates reflect the soluble C that enters the rhizosphere.Although the proportion of netassimilated C allocated to root exudation seems negligible in forest C budgets, after entering the soil, root exudate C can accumulate in dry soil and facilitate ecosystem functions (e.g.soil water storage or C sequestration; Sokol et al. (2019), thereby contributing to the belowground C sink strength of forests and acting as a component of drought resilience (Körner, 2015;Hagedorn et al., 2016).The composition of exudates can also change with drought (Gargallo-Garriga et al., 2018) and specific compounds in root exudates have been associated with complex and diverse roles, e.g.changing the quantity of osmolytes that maintain cell turgor under water stress, developing the soil structure (Ahmed et al., 2014;Baumert et al., 2018;Guhra et al., 2022) and enabling microbial recruitment or selection (van Dam & Bouwmeester, 2016), which may ensure survival during periodic stresses (Huang et al., 2019).Such changes in the metabolite composition of root exudates under drought could 4.3.Drought-susceptible P. abies has a greater belowground C allocation under waterlimitation than F. sylvatica Although both species showed similar patterns in exudation rates from individual root branches (Figure 1) and at the root-system level (Figure 3), we found relatively higher C allocation to root exudation in P. abies than F. sylvatica under drought (Figure 3B).Greater tree-level exudation was mostly a result of the stronger decline in net-C assimilation in P. abies (>60 %) than in F. sylvatica (>40 %) under drought.Although both species maintained root-system level exudation at comparable rates throughout the soil profile, they showed different vertical distribution patterns: in F. sylvatica, rootsystem level exudation was homogeneously distributed through the soil profile, whereas P. abies released two-thirds of the allocated C into the surface soil under drought (Figure 3).In addition, although both species reduced fine-root surface area in the surface soil, the decline in P. abies roots was less pronounced (Table 2), and exudation rates per fine-root surface area of root branches tended to be higher (Figure 1).The decreased assimilation, respiration (Figure 4, Table S6), and reduced growth (Pretzsch et al., 2020;Grams et al., 2021) of P. abies indicates that this species was more strongly affected by drought than F. sylvatica.It is therefore striking that P. abies allocated a relatively greater proportion of C belowground (Figure 3B).However, our findings agree with the theory of Williams and de Vries (2020) that fast-growing species like P. abies increase relative exudation, while slower growing species like F. sylvatica maintain root exudation under drought (Williams & de Vries, 2020).Although the proportion of net-C assimilation allocated to root exudation in P. abies was greater than in F. sylvatica, assessing the benefits to the water balance or the ecosystem resilience of these species due to exudates was beyond the scope of this study.Whether tree-level C investment into root exudation is an active or passive process calls for finer-scaled manipulative experiments to identify mechanistic underpinnings.Alongside lower SWC, it should finally be noted that there may be several additional reasons for higher root exudation from P. abies in the surface soil.For example, soil-root nutrient concentration gradients may increase concentration-related diffusion under water limitation and contribute to elevated exudation (Canarini et al., 2019;Butcher et al., 2020).The low variation in absorptive-root density in P. abies compared to F. sylvatica (Table S2) further suggests limited morphological adaptation of P abies roots to drought.Together with the observed prolonged lifespan of P. abies roots in the surface soil (Zwetsloot & Bauerle, 2021), overall root functionality might have been reduced (Vetterlein & Doussan, 2016;Nikolova et al., 2020)

Conclusion
Root-system and whole-tree level exudation during the study period in early summer were small compared to other assessed C fluxes and seemed negligible in the overall C budget of the forest.
negatively correlated with soil water content across root-accessible soil depths.
of C fluxes and parameters for scaling to the rooting zone and the tree level 2.4.1.C assimilation To quantify C assimilation, light-saturated (Photosynthetically active Photon Flux Density: All statistical analyses were conducted in R (version R 3.6.3,R Development Core Team 2020) in the RStudio environment(version 1.2.1335,RStudio Team, 2019).We used linear mixed-effects models (lme function in the nlme package; version 3.1-137,Pinheiro et al. (2018)) with plot as random effect to test the relationship between dependent variables (exudation, assimilation, respiration, root characteristics) and independent variables (soil depth, treatment (control or drought) and species).The significance of individual terms and interactions of independent variables were determined by likelihood ratio tests using the anova function.Pairwise post-hoc testing of significant terms and interactions was performed using the emmeans function (emmeans package version 1.5.2-1,Searle et al. (1980)).Differences were considered as significant at p < .05.We checked if the model assumptions of homoscedasticity (leveneTest function in the car package, version 2.1-2,Fox and Weisberg (2019) and normal distribution of residuals (shapiro.test)were met and transformed dependent variables, where necessary.We performed a non-linear regression (nls) to fit a power function for the relationship between root exudation rates and soil water content.The coefficient of determination and p-value for the regression were estimated from power transformation and linear regression of the data.Finally, we assumed that the maximum curvature of the power function represented the highest increase in exudation with SWC.Therefore, we calculated the first derivation of the power function and, using the optimize function (stats package, version 4.0.4),assessed the maximum curvature of the power function as a threshold for increased exudation with SWC.Results are presented as mean values ± 1 standard error (1 SE) for n = 3 plots per treatment and species.

Figure 1 :
Figure 1: Fine-root exudation rates (branch-level exudation) per fine-root surface area in F. sylvatica, P. abies and average values over both species in control (blue) and drought (red) plots in the KROOF experiment.Significant differences between 0-7 cm and 7-30 cm soil depths for the drought plots are indicated with red asterisks, where (*) is p = .1,and ** is p < .01.Symbols and whiskers indicate means ± standard errors for n = 3 plots per treatment.

Figure 2 :
Figure 2: Relationships between exudation rate per fine-root

Figure 3 :
Figure 3: Fine-root exudation of F. sylvatica and P. abies trees integrated over three rooting depths in the KROOF experiment as A) total fine-root exudation (root-system level exudation) in g C m -2 plot surface area and day -1 , and B) as a fraction of net assimilation of the trees (tree-level exudation: Exfra, in %).Significant differences are highlighted, with (*) indicating p = .1,and * indicating p < .05.Bars and whiskers indicate means ± standard errors for n = 3 plots per treatment.Note that values for 30-50 cm soil depth were modeled from minirhizotron and soil water content data (see methods).Exudation data were integrated over a two-week period in early summer.

Figure 4 :
Figure 4: Carbon (C) fluxes in A) F. sylvatica and B) P. abies on control (left) and drought plots (right) after 5 years of repeated summer drought.Numbers next to the arrows show C fluxes in g C m -2 plot surface area and day -1 (net assimilation, stem respiration, root respiration, and root exudation).Respiration fluxes are shown in grey boxes.Numbers next to the roots give the fine-root exudation separated by soil depth increments (dark brown: 0-7 cm, brown: 7-30 cm, and light brown: 30-50 cm).Total exudation of the entire rooting zone and the proportion of net-C assimilation allocated to total exudation (assimilation -stem respiration -root respiration; see methods) are given next to the brackets.Note that values for 30-50 cm soil depth were modeled from minirhizotron and soil water content data (see methods).Bold numbers and asterisks indicate significant differences (p < .05) in scaled root respiration and proportion of net assimilation allocated to exudation between control and drought plots.Values are given as means with standard errors for n = 3 plots per treatment.All data represent a two-week period in early summer.
increased belowground C allocation we measured here, presenting an intriguing avenue for further research.
However, the observed elevated belowground C partitioning under drought may play a crucial role in ecosystem functioning and maintaining tree vitality, with the drought-susceptible P. abies investing more C belowground under water limitation than F. sylvatica.Our findings pave the way for future work integrating the chemical composition of exudates, microbial, and plant functional processes to evaluate the fate of root exudate C entering the soil, its spatio-temporal stability, and its role in forest ecosystem drought resilience.Our findings encourage future studies to record belowground C allocation even under low microbial activity by including 1) in situ exudate collection during drought experiments, 2) spatially explicit exudation measurements in naturally developed soil profiles, and 3) calculations of tree-level exudation in mature forest.By integrating across different soil depths and using allometric scaling of the unique empirical dataset of the KROOF experiment, our study demonstrates that trees can maintain root exudation by increasing the proportion of net-C assimilates allocated to exudates under water-limitation, suggesting novel strategies of up-and downregulating belowground C partitioning under drought.Given that there is large variation in how models estimate belowground C allocation under changing climate, our data provide valuable information about how temperate tree species partition assimilates into individual soil layers as well as to the entire rhizosphere under water limitation.
Plant root exudation under drought: implications for ecosystem functioning.New Phytologist 225(5): 1899-1905.Williams A, Langridge H, Straathof AL, Muhamadali H, Hollywood KA, Goodacre R, de Vries FT. 2021.Root functional traits explain root exudation rate and composition across a range of grassland species.Journal of Ecology n/a(n/a).Zwetsloot MJ, Bauerle TL. 2021.Repetitive seasonal drought causes substantial species-specific shifts in fine-root longevity and spatio-temporal production patterns in mature temperate forest trees.New Phytologist n/a(n/a).Zwetsloot MJ, Goebel M, Paya A, Grams TEE, Bauerle TL. 2019.Specific spatio-temporal dynamics of absorptive fine roots in response to neighbor species identity in a mixed beech-spruce forest.Tree Physiology 39(11): 1867-1879.
SWC was measured on 27 May, before exudate sampling.Lowercase letters indicate significant (p < .05)differences between

Table 2 :
Fine-root (≤ 2 mm) surface area and depth distribution of F. sylvatica and P. abies trees on control and drought plots in the KROOF drought experiment.
and P. abies might have lost its capability to control C release to a greater extent than F. sylvatica.Although the relationships between root exudation, root morphology, and root lifespan (both in general and under drought) require further findings indicate that drought stress will have a greater impact on rhizosphere processes in P. abies than F. sylvatica.