Effects of Tree Species on Root Exudation and Mineralization of Organic Acids in a Tropical Forest

Aims Root exudation of organic acids is one of strategies for tropical trees to facilitate nutrient 23 uptake from the highly weathered soils. However, paradoxical relationship remains that root 24 exudation also stimulates microbial activities to consume organic acids in the rhizosphere (root- 25 soil interface). Plant-specific root exudation might shape different rhizosphere carbon (C) cycles 26 between tree species. We test whether root exudation and rhizosphere C fluxes of organic acids 27 and sugars differ between dominant dipterocarp trees and pioneer trees ( Macaranga spp.). 28 Methods We measured (1) root exudation from mature trees, (2) soil solution concentrations of 29 organic acids and monosaccharides, and (3) mineralization kinetics of 14 C-radiolabelled substrates 30 in the rhizosphere and bulk soils of the Dipterocarp and Macaranga trees. aluminum detoxification, and lignin degradation in acidic soils.


Introduction 46
In the tropical forests, long-term weathering and acidification of soils generally result in a 47

Soil solution extraction and chemical analysis 122
The centrifugation-drainage technique was used to extract soil solution (Giesler and Lundström, 123 1993). Without addition of water, the rhizosphere and bulk soil fractions were centrifuged for 30 124 min at a speed of 8,800 rpm (10,560 g; ~1.5 MPa; Hitachi centrifuge) within 36 h of sampling, 125 respectively. The soil solution extracts were filtered through a 0.6 m filter (GF/C, Whatman) and 126 frozen at 24ºC prior to analysis. The monosaccharide concentrations were determined using 127 periodate oxidation (Burney and Sieburth, 1977; Johnson and Sieburth, 1977) and glucose 128 standards. The concentrations of LMWOAs were determined by high performance liquid 129 chromatography (HPLC, Shimadzu, Japan) using the method by Van Hees et al. (1999). Organic 130 acids were separated on a Supelcogel C610-H ion exclusion column using 0.1% H3PO4 as the 131 mobile phase at operating temperatures of 60ºC for citric acid and 30ºC for oxalic and malic acids 132 with UV detection at 210 nm. were estimated for the rhizosphere and bulk soil fractions. 14 C-radiolabelled glucose or organic 137 acid solution (100 µL; specific activity: 0.17 kBq mL 1 ; pH 4.5) was added to 1 ± 0.02 g of field-138 moist soil in 50-mL polypropylene vials. 14 C-glucose (U-14 C; American Radiolabeled Chemicals, 139 Inc., 0.4 GBq mmol 1 ) and four organic acids, 14 C-acetic acid (1,2-14 C; 2.2 GBq mmol 1 ), 14 C-140 oxalic acid (1,2-14 C; 0.2 GBq mmol 1 ), 14 C-malic acid (1,2-14 C; 0.2 GBq mmol 1 ), and 14 C-citric 141 acid (1,5-14 C; 2.2 GBq mmol 1 ), were used in the mineralization assays. The initial solution 142 concentrations of each substrate were 50, 250, 500, and 1000 µM. Following addition, the soil 143 was gently shaken to ensure mixing and incubated at 25°C in sealed vials. 14 C-CO2 produced by 144 mineralization of the added substrate was collected in a plastic scintillation vial containing 1.0 145 mL of 1 M NaOH placed on top of the soil, separated by a spacer. The 14 C-CO2 concentrations 146 trapped in NaOH were determined by liquid scintillation using alkali-compatible scintillation 147 fluid (Hionic-Fluor; Perkin Elmer). 14 C-CO2 production was measured during the initial linear 148 phase of decomposition (1 h), which was confirmed by the pilot experiment. 149 The data of mineralization kinetics were fitted to a single Michaelis-Menten equation: 150 where V is the mineralization rate (nmol g 1 h 1 ), C is the substrate concentration (M) in the soil 152 solution, Vmax is the maximum mineralization rate (nmol g 1 h 1 ), and KM is the concentration at 153 which the half-maximal mineralization rate occurs ( 2 1 V max; M). Michaelis-Menten plots of 154 organic acids were constructed using the equilibrium organic acid concentrations in soil solution, 155 assuming complete mixing of the organic acid with the intrinsic soil water and the sorption 156 reaction (see the following section). 157 158

Sorption isotherms of organic acids 159
Glucose is not adsorbed onto the solid phase due to a lack of charge, whereas negatively charged 160 carboxylic acids (acetic, oxalic, malic, and citric acids) are strongly adsorbed onto the solid phase (Jones and Brassington, 1998). To estimate the equilibrium concentrations of the organic acids in 162 the soil solution after adding organic acid in the kinetic experiments, sorption isotherms were 163 obtained by the method of Fujii et al. (2019). In each tube, 2.5 mL of 14 C-radiolabelled organic 164 acid solution (170 Bq mL 1 ; pH 4.5) was added to 0.50 g of chloroform-fumigated (48 h) field-165 moist soil in 6-mL plastic vials with a soil to solution ratio of 1:5 (w/v). The initial organic acid 166 concentration ranged from 100 to 1000 µM. Following addition, the samples were shaken for 10 167 min on a reciprocating shaker at 320 rpm. The samples were subsequently centrifuged (16,000 × 168 g for 5 min) and the supernatant was recovered. The equilibrium 14 C concentrations in solution 169 were determined by liquid scintillation counting (Aloka Liquid Scintillation System, LSC-3050; 170 Hitachi) using Optiphase HiSafe 2 scintillation fluid (Perkin Elmer, Japan). 171 Then, the sorption isotherm data were fitted to the Freundlich equation: 172 where A is the quantity of organic acid adsorbed (nmol g 1 ), C is the equilibrium solution 174 concentration (µM), k is the Freundlich's constant related to soil ability to sorb organic acid, and 175 1/n is the constant related to sorption intensity. The quantity of anion adsorbed (A) can be 176 calculated using the following equation: 177 where Ctot is the total quantity of organic acid added to the soil (nmol cm 3 ), C is the equilibrium 179 soil solution concentration (µM),  is the volumetric water content (cm 3 cm 3 ), and is the soil 180 bulk density (g cm 3 ). 181

Rhizosphere effects of substrate mineralization rates 182
The mineralization rates of LMWOSs at their actual substrate availability were estimated using 183 Eq. 1, assuming that LMWOSs in soil solutions are utilized by soil microbes as described by

Soil and root properties of Dipterocarp and Macaranga forests 221
Under the Macaranga trees, soil pH and available P were higher than under the Dipterocarp trees 222 due to the ash inputs in the past fires (Table 1). There was no difference in microbial biomass C 223 between Dipterocarp and Macaranga forest sites (Table 1). Although Dipterocarp tree roots had 224 larger root surface areas and more root tips (

Low molecular weight organic substance concentration in rhizosphere and root exudation 230
Compared to the bulk soil, rhizosphere soil solution displayed significantly higher concentrations 231 of malic and oxalic acids in the Dipterocarp soil ( Fig. S1; Fig. 1

). No enrichment effects of 232
LMWOAs in the rhizosphere were observed in Macaranga soil ( Fig. S1; Fig. 1). There was no 233 significant difference in acetic acid and monosaccharide concentrations in rhizosphere and bulk 234 soil solutions between Dipterocarp and Macaranga sites ( Fig. S1; Fig. 1). 235 Monosaccharides, acetic, malic, and citric acids were detected in root exudates, but the 236 composition and rates differed between Dipterocarp trees (Dipterocapus cornutus and Shorea 237 laevis) and Macaranga trees. The organic acid exudation rates were positively correlated with root 238 surface areas (Fig. 2a) and with root tips, respectively. Root exudation rates of malic and acetic 239 acids by the Dipterocarp trees were greater than the Macaranga trees (Table 2). This contrasts with the higher monosaccharide exudation of the Macaranga trees (Table 2). There was no 241 significant difference between Dipterocapus cornutus and Shorea laevis, except for in oxalic acid 242 exudation (Table 2). 243 Using the exudation rates and the fine root biomass in the soil profile (0-10 cm; Tables 1 and  244 2), root exudation rates were transformed into the C fluxes in the in the forest ecosystem (Table  245 4). In the Dipterocarp and Macaranga forests, the C fluxes of multivalent organic acid (citric, 246 malic, and oxalic acids) exudation corresponded to 4.8-6.1% and 2.1% of net primary production 247 (NPP) [5.1 mol C m -2 month -1 (Toma et al., 2000) and 6.2 mol C m -2 month -1 (Gamo, 2003), 248 respectively] ( Table 2). When the published data (Aoki et al., 2012) and those from the present 249 study were included in correlation analysis, the proportions of organic acid exudation relative to 250 NPP were negatively correlated with soil pH (KCl) (Fig. 2b). 251 252

Organic acid sorption reactions 253
To estimate organic acid sorption in the mineralization kinetics experiment, the data of organic 254 acids sorption and equilibrium concentration were fitted well to a Freundlich equation (R 2 > 0.95; 255 Fig. 2; Table 3). Among the four organic acids, the degree of organic acid sorption followed the 256 order: malate > citrate > oxalate > acetate (Fig. 3). There were no differences in sorption of 257 respective organic acids between the soils under Dipterocarp and Macaranga trees (Fig. 3). 258 We compared mineralization kinetics to assess the specialization of substrate mineralization by 261 rhizosphere and bulk soil microbes. The mineralization rates of both soils varied between 262 substrates and followed the order: malate > citrate, oxalate > acetate, glucose (Fig. 4). The higher 263 mineralization activities in the rhizosphere, compared to the bulk soil, were observed for malate 264 and oxalate in the Dipterocarp soil, but only for malate in the Macaranga soil (Fig. 4). There were 265 no significant differences in mineralization rates of glucose, acetate, and citrate between the 266 rhizosphere and bulk soil fractions under the Dipterocarp and Macaranga trees, respectively (Fig.  267

4). 268
The data of LMWOS mineralization rates were fitted well to the single Michaelis-Menten 269 kinetic equation (R 2 > 0.98; Fig. 4, Table 4). Michaelis-Menten kinetic parameters (Vmax and KM) 270 describe microbial capacity to mineralize substrate and microbial response to substrate availability, 271 respectively. An increase in malate and oxalate mineralization activity in the Dipterocarp 272 rhizosphere (Fig. 4) was caused by the higher Vmax values, compared to the bulk soil (Table 4). 273 An increase in malate mineralization activity in the Macaranga rhizosphere (Fig. 4)

Carbon fluxes of low molecular weight organic substance mineralization in the rhizosphere 277
Using the bulk density, the volumes of the rhizosphere and bulk soil fractions, the mineralization 278 kinetics (Table 4), we quantified soil C fluxes of LMWOS mineralization to test whether root 279 exudation and rhizosphere C fluxes are quantitatively important relative to the bulk soil C cycles. 280 Monosaccharides, malate, and citrate were major substrates for microbial LMWOS 281 mineralization in the bulk soil, while malate accounted for the majority of mineralization C fluxes 282 by in the rhizosphere (Fig. 5). When MRTs of LMWOAs and monosaccharides were calculated 283 by dividing the amount of LMWOS-C in soil solution by mineralization C flux, MRTs were short 284 ranging from 0.2 h to 5.8 h for LMWOAs and 10.9 h to 15.0 h for monosaccharides, respectively 285 (Table 5). Malate exhibited the shorter MRTs among LMWOSs (Table 5). When the 286 mineralization C fluxes per soil mass in the rhizosphere were compared to the bulk soil, the 287 rhizosphere effects differed between tree species and between LMWOSs (

Effects of tree species on root exudation rates 293
Judging from the fact that LMWOSs are rapidly consumed by microbial uptake and sorption, 294 LMWOS dynamics could be strongly regulated by soil types, substrate charges, and root and relative to the bulk soil suggests that the pool size of organic acids could also vary between tree 299 species, depending on the supply of organic acids (Fig. 1). Consistent with the hypothesis, the composition and rates of root exudation differ between Dipterocarp and Macaranga trees (Table  301 2; Fig. 5a). The greater rates of organic acid exudation from Dipterocarp trees (Table 2; Fig. 2a) 302 are consistent with the finding that ectomycorrhizal fungi promote mineral weathering by 303 releasing LMWOAs from the hyphae (Jongmans et al., 1997). The lower soil pH under 304 Dipterocarp trees also increases allocation of photosynthate to organic acid exudation (Fig. 2b). 305 Soil pH is used as a proxy of P deficiency and Al toxicity in our study, because P solubility 306 decreases and Al 3+ solubility increases along with soil acidification (Jones, 1998;Fujii et al., 307 2018). For plants' survival on acidic soils, the high levels of divalent organic acids in the 308 rhizosphere need to be maintained to solubilize recalcitrant P from Al or Fe oxides or to detoxify 309 Al 3+ (Van Hees et al., 2005). Soil acidification leads to a shift of vegetation towards tree species 310 with capacity to release divalent organic acids from roots, as seen by succession from pioneer tree 311 species to dipterocarp tree species in our study (Fig. 2b). Recently, Mn concentrations in the plant 312 leaf is postulated as a proxy for organic acid exudation activities (Pang et al., 2018;Lambers et 313 al., 2020). This pattern has also been confirmed in pour study, Dipterocarp leaf displays the higher 314 Mn concentrations (1.06 mg g -1 ) than Macaranga leaf (0.34 mg g -1 ) (Fujii et al., 2020). Both soil 315 acidity and organic acid exudation are considered to increase Mn solubility and root Mn uptake 316 in our study. 317 supported by the higher concentrations and C fluxes of malate mineralization, compared to the 321 bulk soil in our study (Figs. 1 and 6). This is primarily due to the substrate inputs via root 322 exudation Butler et al., 2003), but microbial potentials to mineralize exudates 323 are also elevated ( Fig. 4; Fujii et al., 2013). This is evidenced by the higher Vmax values of malate 324 mineralization in the Dipterocarp rhizosphere (Table 4). It has been shown that microbial 325 mineralization potentials (transporter activity) could vary, depending on C sources that microbes 326 have grown on . Microbial community grown on malate could have the higher 327 activity of malate transporter that takes the malate into the cell, compared to microbial community 328 grown without malate . In our study, root exudation of malate might increase 329 malate preference and mineralization activity of rhizosphere microbial community (Table 4). This 330 contrasts with the low availability of organic acids in the volcanic soils leading to the low 331 microbial mineralization activities (Fujii et al., 2019). Both higher substrate availability in the 332 rhizosphere and the associated higher mineralization activities of microbes shape the hotspots of 333 organic acids in the Dipterocarp rhizosphere (Fig. 5). 334 335

Compound-specific rhizosphere effects 336
Tropical trees require rhizosphere processes to acquire P and protect roots in the highly weathered 337 and acidified soils (Fujii et al., 2018). Due to the highest sorption of malate and its shortest MRTs 338 among organic acids ( Fig. 3; Table 5), the efficacy of exuded malate on P mobilization and Al 3+ 339 detoxification in the rhizosphere could be reduced . There should occur the mechanisms to maintain the pool size of malate in rhizosphere soil solution (Van Hees et al.,  341   2005b). This could be partly accounted for by the higher rates of malate exudation from the 342 Dipterocarp roots (Table 2), but malate-C inputs by root exudation can not account for the whole 343 C fluxes of malate mineralization in the rhizosphere (Figs. 5 and 6). This indicates that malate           (a) Relationship between root surface area and root exudation rate of organic acids, (b) relationship between root exudation of organic acids relative to net primary production (NPP). Acetate, oxalate, malate, and citrate were counted in Fig. 2a, while oxalate, malate, and citrate were counted in Fig. 2b to compare with the previous study [1, 2 = tropical montane forest, 3 = tropical forest (Aoki et al., 2012)]. Bars indicate standard errors (N = 5). Root Concentration-dependent sorption of citrate, oxalate, malate, and acetate in soils. Symbols denote experimental points, while the curves represent Freundlich sorption isotherms tted to the experimental data. Bars indicate standard errors (N = 5).