Introduction

Volcanic soils derived from volcanic ash deposits have a high adsorption capacity of soil phosphorus (P) (Soil Survey staff 1999; Takahashi and Dahlgren 2016). This is because the non-crystalline Al and Fe minerals which are produced during weathering process are abundant in volcanic soils. These minerals have many reactive sites to combine P with ligand binding, causing decreased soil P availability (Borie and Rubio 2003; Escudey et al. 2001, 2007; Harsh et al. 2002; Parfitt 2009). The major characteristics of volcanic soils (Andisols) are high concentrations of total P and organic matter but plant-available P is low (Amano 1980; Jorquera et al. 2008).

Forest ecosystems occur on such volcanic soils and they show evidence of enhanced P cycling. Mukai et al. (2016) showed that the P-use efficiency of forest productivity (which is an index of P deficiency) was decreased on soils with ample non-crystalline minerals indicating that soil P deficiency was less severe on volcanic soils with greater influence of volcanic ash. Moreover, the amount of cycling P in a forest was greater on soils with greater non-crystalline minerals. These patterns indicate the possibility that trees use P that is absorbed to non-crystalline minerals in volcanic soils. However, mechanisms of P acquisition by trees on volcanic soils are not precisely known.

To clarify tree strategies to acquire P in volcanic soils, we focused on the flux of root exudates from fine roots. Fine roots are known to release various organic compounds into the ambient soil environments (i.e. the rhizosphere soils) including sugars, amino acids, low-molecular-weight organic acids (LMWOAs), and phenolics (Uren 2000). All are derived from photosynthetic metabolites and incur a cost for plants although they comprise a few percent of net primary productivity in terms of carbon (C) budget at a stand level (Yin et al. 2014). Root exudates are constantly released from fine roots and are considered as one of the investments by trees to acquire soil nutrients effectively (Haichar et al. 2014). LMWOAs consist a major fraction of root-exudate organic compounds (Uren 2000) and are believed to play a significant role in nutrient acquisition. The following two mechanisms of LMWOAs might be involved in the soil P acquisition strategies of trees on volcanic soils; (1) direct chemical reaction via ligand exchange and soil acidification, and (2) priming effects.

First mechanism is that LMWOA exudates chemically release P bound to non-crystalline minerals as inorganic P through ligand exchange and through lowering pH. LMWOAs as anions in soils can replace the phosphates which form complexes on ligand exchange surfaces of Fe and Al, prevent the released phosphates from binding with absorption sites on soil particles, and then effectively release the phosphates as available P (Bais et al. 2006; Chen et al. 2008; Jones 1998). LMWOAs can also mobilize phosphates by reducing rhizosphere soil pH and dissolving non-crystalline minerals (Hedley et al. 1994). In addition, LMWOAs contributed to the dissolution of Al and Fe from organo-metal complexes and release organic as well as inorganic P in an experiment with the artificial addition of LMWOAs in the laboratory (Keiluweit et al. 2015). These mechanisms were mainly clarified in laboratory experiments or in agricultural fields using crop cultivars, and knowledge on forest ecosystems is very limited.

Second mechanism is that LMWOAs cause priming effects in the rhizosphere by supplying labile organic matter to microorganisms and stimulate microbial growth and activity. Priming effects are defined as extra decomposition of organic C after the addition of easily decomposable organic substances to soils (Dalenberg and Jager 1989; Kuzyakov et al. 2000). Due to the priming effects, organic P as microbial biomass which were produced by immobilization of soil inorganic P increase in the rhizosphere. Activated microbial metabolism and/or higher microbial biomass turnover produce mineralized P and enhance P availability for trees. Organic P in soils is mineralized by phosphatase enzymes through hydrolysis and the availability of inorganic P is enhanced. Root exudates are utilized by soil microbes to produce microbial biomass as well as extracellular phosphatase enzymes as easily decomposable energy source, and therefore microbial biomass and extracellular phosphatase enzymes must be enhanced in the rhizosphere (Chen et al. 2002). Thus, we hypothesize that the input of LMWOAs as an easily decomposable energy source can cause the acceleration of P mineralization in the rhizosphere soils.

The aim of this study was to investigate the above two mechanisms of P acquisition by trees on volcanic soils. Firstly, we investigated the composition of LMWOAs that were released to soils from fine roots of dominant tree species, and if the composition changed with the concentration of non-crystalline minerals. Secondly, we investigated the above two mechanisms of P acquisition with the following two working hypotheses: (i) the concentrations of both non-crystalline minerals and P bind to non-crystalline minerals are lower in the rhizosphere soils than in non-rhizosphere soils because non-crystalline minerals and bound P are dissolved with LMWOAs; and (ii) the concentration of organic P is higher in the rhizosphere soils than in non-rhizosphere soils reflecting the proliferation of soil microbes due to the priming effects of LMWOAs. To test these hypotheses, we compared three evergreen broadleaved forests on volcanic soils with different concentrations of non-crystalline minerals but in the same vegetation zone on Yakushima Island, Japan.

Materials and methods

Study sites

Yakushima Island is located in southern Japan (30° N, 130° E) with an area of 504 km2. Mt. Miyanoura is the highest point (the summit 1936 m a.s.l.). Mean annual temperature (MAT) is around 19.4 °C at the metrological station in 37 m a.s.l. (Japan Metrological Agency 2018). The geology of Yakushima is mainly Miocene granite except for the northeastern to the southern periphery, where the Kumage group of sandstone and shale are distributed. Distinct elevational zones of vegetation are observed: lower subtropical evergreen broad-leaved forest (0–100 m), warm-temperate evergreen broad-leaved forest (100–800 m), evergreen broad-leaved/conifer mixed forest (800–1200 m), Cryptomeria japonica (Cupressaceae) coniferous forest (1200–1700 m), and Pseudosasa owatarii grassland (1700 m–). In higher elevations, dominant conifers are Cryptomeria japonica (Cupressaceae), Abies firma and Tsuga sieboldii (Pinaceae) (Okada and Ohsawa 1984; Ohsawa 1984).

Yakushima was strongly affected by the eruption of Kikai Caldera about 7300 years ago. The eruption consisted of two episodes: pyroclastic flow (Koya pyroclastic flow) at the first stage and ashfall (Akahoya volcanic ash) at the second stage. The Koya pyroclastic flow deposited more than 3-m thickness and Akahoya volcanic ash deposited more than 50-cm thickness on Yakushima (Geshi 2009; Ui 1973). The Akahoya volcanic ash mantles extensive areas from Kyusyu to central Honshu and the fallout area is estimated to be approximately 1 × 106 km2 and the volume of tephra was more than 150 km3 (Machida and Arai 1978; Machida 1999). Geochemical effects of Akahoya volcanic ash are still evident in extant ecosystems because the concentration of soil P was high and the concentration of acid-oxalate extractable Al and Fe (an index of the abundance of non-crystalline minerals derived from volcanic ash) was also greater than a threshold of volcanic ash soil (Mukai et al. 2016). The characteristics of volcanic ash soils are more evident in the lowland ecosystems than in the upland (Mukai e al. 2016).

Study design

We used three study plots in the lowland, warm-temperate evergreen broad-leaved forest zone (Aiba et al. 2007, 2013) (Table1). Their elevations are 170, 200 and 280 m a.s.l., respectively, within the range of mean annual temperature (MAT) from 18.1 to 18.9 °C (Aiba et al. 2013). Mean annual precipitation (MAP) of the three study sites ranges from 2155 to 5883 mm year−1 (Mukai et al. 2016). The topography of all plots is a gentle slope and the area of the plots range from 0.25 to 0.5 ha. These three forests are all old-growth stands without significant anthropogenic disturbances for more than 100 years (Aiba et al. 2013). Broad-leaved trees are dominant in all study sites.

Table 1 List of study plots and their climatic conditions and soil chemical properties
Full size table

Soil chemical analysis of study plots

As properties characterizing volcanic soils, we measured pH(NaF) and soil P sorption by using the air-dried soils collected in July 2013. The method of collecting surface soil was the same as Mukai et al. (2016). Briefly, in each plot, four 30-m transects were laid out at intervals of 10 m, parallel to contour lines. Four cores (37 mm diameter and 20 cm depth) were collected at intervals of 10 m along each transect and were combined as a composite sample. For collection of each soil core, the litter layer was removed and then the top mineral soil from the depth of 0–10 cm was extracted. Subsequently, the subsoil of the 10–20 cm depth was sampled in the same way.

Soil pH(NaF) was determined following the CSIRO Linked Data Registry (2016). pH(NaF) is an indicator of the presence of non-crystalline minerals which are produced during weathering process and/or active Al (Fieldes and Perrot 1966). 1-g of air-dried soils were mixed with 50 mL 1 M NaF (pH 7.0) and stirred for 1 min. pH electrode was placed in the suspension and pH value was read after 2 min of swirling the suspension.

P sorption ratio was determined following the Robertson et al. (1999). 1-g of air-dried soils were added to the 10 mL solution of 0.01 M KCl containing 0 and 200 µg P mL−1, respectively, as potassium dihydrogen phosphate (KH2PO4) with one drop of chloroform in a 50 mL centrifuge tube. The soil solution mixture was shaken for 24 h and centrifuged at 3000 rpm for 30 min. The concentration of P in supernatant fluid was determined by colorimetric method following Murphy and Riley (1962) using a microplate reader (SH-9000, CORONA, Ibaraki, Japan). P sorption ratio was determined as the adsorption ratio of P to soil when 2000 µg P was added to 1-g dried soil following the equation;

$$Xs = \{ (s - c) \times F\} - E$$

where Xs was sorbed P concentrations in P-added solution (µg P g−1 soil); s was 200 µg P mL−1 soil of the original working solution; c was µg P mL−1 in equilibrium solution; F was mL working solution/g soil dry soil (10 mL 1 g−1 dry soil); E was exchangeable P as µg P g−1 soil.

Alo + 1/2 Feo, which are values of oxalate-extracted Al and Fe, is an index of non-crystalline materials derived from volcanic ash, per study plots was cited from Mukai et al. (2016).

Collecting root exudates

Four common dominant species (Distylium racemosum, Quercus salicina, Castanopsis sieboldii, and Schefflera octophylla) of the three study plots were selected as target trees for collecting root exudates. We selected five tree individuals per species in each study plot, i.e. a total of 20 individual trees in each site. Root exudate were collected from intact lateral fine roots of each tree using a non-soil syringe system modified from Phillips et al. (2008). Briefly, roots on each tree in the top 5 cm of the A horizon were carefully excavated. They remained attached to the target trees during the entire procedure until harvest. We gently washed the intact fine roots with a spray bottle, using deionized water to remove soil particles and other possible contaminants. The fine roots were carefully placed into a syringe filled with glass beads and a C-free nutrient solution (see detail of the solution in Phillips et al. 2008). Triplicates of syringes were installed on each individual tree. Control syringes were prepared only with beads and nutrient solutions to correct for the C originating from non-exudate sources or contamination during operation. Two control syringes were prepared for each tree and left in situ in this study. Finally, the syringes including controls were covered with aluminum foil, a wet paper towel, and leaf litter to block sunlight and heat. After 24 h, the nutrient solution was collected from each syringe system. An additional 5 mL of the nutrient solution was flushed twice through the syringe system to obtain a representative C recovery. All the recovered solutions were filtered immediately through a 0.45-µm sterile syringe filter made of mixed cellulose ester (Advantec, DISMIC, Japan) in the field.

All fine roots placed in the syringes were harvested after the recovery of the solution and transferred to the laboratory in an insulated cooler box for further measurements. Targeted fine roots were scanned and analyzed for morphological traits using WinRHIZO software (Regents Instruments Inc., Quebec City, Canada). Then, they were oven dried and weighed. The root-exudation rate of LMWOAs was calculated by dividing the total C of detected LMWOAs in the trap solution with the residence time and the root length or root weight. We determined eight kinds of LMWOAs (oxalic, citric, tartaric, malic, succinic, formic, acetic, fumaric acid) with a high-performance liquid chromatography (HPLC, Shimadzu, Japan). These eight LMWOAs represent the most common LMWOAs for tree species in previous studies (e.g. Chen et al. 2001; Jiang and Hu 2017; Smith 1976). LMWOAs were separated by a Supelcogel C610-H ion exclusion column with the mobile phase of 0.1% H3PO4 using the same method as Aoki et al. (2012). Column was kept at 35 °C during the analysis and the concentrations of LMWOAs were determined at 210 nm. Identification and quantitation were made by comparing with known standards.

Collecting rhizosphere and non-rhizosphere soils and pre-treatment

We collected rhizosphere soils and non-rhizosphere soils, respectively, for each individual tree, from which we sampled root exudates. As for rhizosphere soils, we followed the methodology suggested by previous studies (Phillips et al. 2011; Yin et al. 2014). We collected fine roots with rhizosphere soils attached to them in the surface horizon (down to 5 cm depth) within 2-m diameter from the stem of a target tree. The collected fine roots were sealed in plastic bags and immediately stored in a cooler box in the field and then in a refrigerator at 4 °C in the laboratory. Within one day after the collection, we collected rhizosphere soils adhering to fine roots with a brush after shaking off the soils that were loosely attaching to the roots in the laboratory. We also collected non-rhizosphere soils as bulk soils not attaching to fine roots in the same area. Collected non-rhizosphere soils were sieved (< 2 mm) in the field. Rhizosphere soils and non-rhizosphere soils were, respectively, well mixed for each tree and stored in sealed plastic bags in a refrigerator at 4 °C until analysis.

Subsamples of field-moist soils were oven dried at 70 °C for 48 h to determine gravimetric soil water contents. Soil pH was determined from a slurry of 1:10 fresh soil to H2O.

Determination of acid oxalate extractable Fe and Al

To quantify non-crystalline inorganic forms of Al, Fe and Si, we weighed 0.5-g subsamples of each air-dried soil (both rhizosphere and non-rhizosphere) and extracted with 50 mL 0.2 M acid ammonium oxalate adjusted to pH 3 (McKeague et al. 1971; Ross and Wang 1993). The soil extraction was conducted under dark conditions. The soil-solution mixture was shaken for 4 h and filtered through Advantec 5C filter paper. The concentrations of Al, Fe and Si (Alo, Feo and Sio) in the filtrates were determined on an inductively coupled plasma spectrometer (ICPS-7510, Shimadzu, Japan). We also determined P concentration from the same extraction and treated as oxalate-extracted P, which was considered the P bound to non-crystalline minerals.

Soil P extraction

Soil P fractions (NaHCO3-Pi, organic P, total P) were sequentially extracted following the modified method of Tiessen and Moir (1993). A known amount of field-moist soil (about 0.8 g) from each composite sample was sequentially extracted by a series of successively stronger reagents in 50 mL polypropylene centrifuge tubes as follows: 0.5 M NaHCO3, 0.1 M NaOH, and concentrated H2SO4. Extracts of the NaHCO3 and NaOH solutions contained both inorganic P (Pi) and organic P (Po); the Pi in these extracts was determined after precipitating organic matter by acidifying subsample solutions to pH 1.5 with 0.9 M H2SO4. The Pi concentration was determined following the Murphy–Riley (1962) methods using a microplate reader (SH-9000, CORONA, Ibaraki, Japan). The concentration of total P in each fraction was analyzed by an inductively coupled plasma spectrometer (ICPS-7510, Shimadzu, Japan). The concentration of organic P in each extract was determined as total P minus Pi. All organic fractions (NaHCO3 and NaOH) were aggregated as organic P. The Pi extracted with NaHCO3 (NaHCO3-Pi) was considered labile Pi. Soil total P was derived as the sum of all fractions.

Statistics analysis

Significant differences in the mean flux rates of LMWOAs among tree species across study sites were tested by using ANOVA followed by Tukey’s HSD test. A principal component analysis (PCA) was conducted on the composition of LMWOAs released from fine roots. Mean concentration of each LMWOA for each individual was calculated based on triplicated samples and then absolute values of mean LMWOAs were derived for each individual both on a root length basis and on a root weight basis. Tree individuals were ordinated on a two-dimensional coordinate with the PCA analysis. Significant differences in axis values among species and among sites were tested with one-way PERMANOVA. Furthermore, a pairwise PERMANOVA was applied to test significant differences for a given combination of sites or species.

We investigated the relationships between the composition of LMWOAs and external soil factors using a canonical correspondence analysis (CCA). The compositions of LMWOAs were expressed both on a root length basis and on a root weight basis. External soil chemical factors included in the CCA were concentrations of P fractions (inorganic P, organic P, oxalate extracted P, and total P) and non-crystalline minerals (Alo + 0.5 Feo), and pH. All external soil factors represent non-rhizosphere soils.

A paired t-test was conducted to examine significant differences between rhizosphere soils and non-rhizosphere soils for pH, non-crystalline minerals and oxalate-extracted P for testing hypothesis 1, and for inorganic P, organic P and total P for testing hypothesis 2. All statistical analyses were conducted using the R statistical program, version 3.0.1 (R Development Core Team 2013).

Results

The flux rates of organic acids from fine roots and their composition

The flux rates of LMWOAs identified from the target species in each study site are shown in Table 2 (root length basis) and Table SI1 (root weight basis), respectively. Oxalic acid was released from all target species in all sites and its flux rate was consistently greatest in S. octophylla among the four species. Citric acid was also released from almost all species in all sites except for Q. salicina in the 170-m study site. Tartaric acid was also released from almost all species and sites except for Q. salicina in the 170-m and D. racemosum in the 200-m and 280-m site. As for the flux rates of malic, succinic, formic and acetic acid, there are no consistent trends among tree species and study sites. Succinic, formic, acetic acid were not released from all tree species in the 200-m site. Although all species in the 200-m site and three species in the 170-m site released fumaric acid, just one species (Q. salicina) in the 280-m site released fumaric acid.

Table 2 Mean flux rates (nmolC m−1 root h−1) of LMWOAs identified from the target tree species in each study site (± standard deviation)
Full size table

The composition of LMWOAs in root exudates on a root-length basis (nmolC m−1 root h−1) varied among tree species but did not vary among study sites (PERMANOVA, Fig. 1 a and b). S. octophylla demonstrated a significantly different composition of LMWOAs from the other species (D. racemosum and Q. salicina) in terms of PCA axis values (pairwise PERMANOVA, Table 3). Results of PCA analysis for LMOWAs on a root-weight basis (nmolC g−1 root h−1) were similar with those on a root-length basis (Figs. 1, SI1) but there was no significant difference among tree species (Table SI2).

Fig. 1
figure 1

The result of PCA analysis of LMWOAs (root length basis, nmolC m−1 root h−1) released from fine roots. Dots indicate replicated individuals for study sites (a) and for tree species (b), respectively. In this PCA analysis, the first two axes explained 29.5% and 22.6% of the variation of LMWOA composition. Different colors indicated different site (a, red 170 m, green 200 m, blue 280 m) and different tree species (b, pink Distylium racemosum; orange, Quercus salicina; green, Castanopsis sieboldii; blue, Schefflera octophylla). R2 and p-value indicate the results of PERMANOVA analysis, which show a significant difference among tree species (a) but no difference among study sites (a)

Full size image
Table 3 Results of pairwise PERMANOVA for testing a significant difference of a given combination of study sites (170, 200, 280) and tree species (Distylium racemosum, Quercus salicina, Castanopsis sieboldii, Schefflera octophylla) in flux rates of LMWOAs on a root length basis
Full size table

The relationship between soil environmental factors and the flux of LMWOAs

The results of CCA to investigate the relationship between the composition of LMWOAs and environmental factors (i.e. chemical properties of non-rhizosphere soils) are shown in Fig. 2 (root length basis) and Fig. SI2 (root weight basis), respectively. Citric and malic acids tended to increase with increasing concentration of non-crystalline minerals (Alo + 0.5 Feo) because the locations of the two LMWOAs in the two-dimensional coordinate of the CCA correspond with the direction of the arrow of non-crystalline minerals in Fig. 2. The concentration of inorganic P and total P did not relate with the relative release rates of any LMWOAs. The results of CCA were consistent between flux rates on a root-length basis and on a root-weight basis (Figs. 2, SI2).

Fig. 2
figure 2

The results of CCA analysis to investigate the relationship between non-rhizosphere soil environmental factor and the flux of LMWOAs (root length basis) for all tree species investigated in this study. In this CCA analysis, the first two axis explained 62.9% and 26.6% of variation, respectively. Light blue arrows indicate the environmental condition in nonrhizosphere soils. Black letters indicate soil chemical properties (pH, soil pH; TP, soil total P; Po, soil organic P; Pi, soil NaHCO3-Pi; Oxa-P, soil oxalate extracted P; Alo + 0.5 Feo, soil oxalate ammonium extracted Al and Fe, respectively). The length of arrows indicates the impact strength of each chemical properties in the non-rhizosphere soils. The flux of LMWOAs released from the fine roots were written in red empty diamond (Oxa oxalic acid, Cit citric acid, Tar tartaric acid, Mal malic acid, Suc succinic acid, For formic acid, Ace acetic acid, Fum fumaric acid)

Full size image

Chemical properties of non-rhizosphere and rhizosphere soils

Mean soil pH (H2O) of non-rhizosphere soils and rhizosphere soils was 5.03 and 4.95 in the 170-m site, 5.33 and 5.18 in the 200-m site and 4.91 and 4.88 in the 280-m site, respectively (Table 4). Mean soil pH (H2O) of rhizosphere soils was significantly lower than that of non-rhizosphere soils in both 170-m and 200-m sites. Mean concentration of oxalate-extracted Al, Fe and Si (Alo, Feo and Sio) was significantly lower in rhizosphere soils than in non-rhizosphere soils in the 170-m sites. Alo, Feo and Sio did not differ between non-rhizosphere soils and rhizosphere soils in the 200-m and 280-m sites (except for Feo in the 200-m sites). The mean concentration of oxalate-extracted P for non-rhizosphere soils and rhizosphere soils was 378.8 µg g−1 and 331.6 µg g−1 in the 170-m site, 218.6 µg g−1 and 261.3 µg g−1 in the 200-m site and 196.2 µg g−1 and 198.4 µg g−1 in the 280 m site, respectively (Table 4). There was no consistent difference in oxalate-extracted P between rhizosphere and non-rhizosphere soils across sites.

Table 4 Mean (standard deviation) soil chemical properties (non-rhizosphere soils and rhizosphere soils) of three study sites
Full size table

The mean concentration of NaHCO3-Pi for non-rhizosphere soils and rhizosphere soils was 23.9 µg g−1 and 38.5 µg g−1 in the 170-m site, 12.6 µg g−1 and 23.1 µg g−1 in the 200-m site and 23.1 µg g−1 and 40.9 µg g−1 in the 280-m site, respectively (Table 5). The mean concentration of soil organic P concentration for non-rhizosphere soils and rhizosphere soils was 443.7 µg g−1 and 516.6 µg g−1 in the 170-m site, 256.3 µg g−1 and 348.0 µg g−1 in the 200-m site and 420.2 µg g−1 and 496.7 µg g−1 in the 280-m site, respectively (Table 5). The mean concentration of soil total P for non-rhizosphere soils and rhizosphere soils was 883.2 µg g−1 and 1033.0 µg g−1 in the 170-m site, 605.6 µg g−1 and 817.4 µg g−1 in the 200-m site and 917.1 µg g−1 and 1112.3 µg g−1 in the 280-m site, respectively (Table 5). Mean concentrations of NaHCO3-Pi, organic P and total P for rhizosphere soils was significantly higher than for non-rhizosphere soils consistently across all three study sites (Table 5).

Table 5 Mean (standard deviation) soil P fraction concentrations (non-rhizosphere soils and Rhizosphere soils) of three study sites
Full size table

Discussion

The composition of released LMWOAs is different among tree species but not among study sites (Fig. 1a, b), suggesting that the composition of exudated LMWOAs is phylogenetically constrained. Earlier studies also reported that the composition of LMWOAs as root exudates differed across species (e.g. Boldt-Burisch et al. 2019; Chen et al. 2001; Jiang and Hu 2017; Sun et al. 2017; Tuason and Arocena 2009). LMWOAs in root exudates are mainly secreted as secondary metabolites through the TCA cycle in the cytosols, and its composition is probably affected by various traits relating to photosynthetic activities. Photosynthetic traits are dependent on phylogeny (Reich et al. 1994, 1998) and the similarity in LMWOA composition seems explained by phylogeny also in this study. Quercus salicina and Castanopsis sieboldii (both Fagaceae) and Distylium racemosum (Hamamelidaceae) belonging to Rosids showed similar compositions of LMWOAs, whereas Schefflera octophylla (Araliaceae) belonging to Asterids showed an outstanding composition (Table 3).

Citric and malic acids tend to be positively related to the concentration of non-crystalline minerals in the non-rhizosphere soils (Figs. 2, SI2), although the composition of LMWOAs does not significantly differ among sites. Both citric and malic acids are known to be exudated frequently in herb roots with a relatively high amount (Kpomblekou and Tabatabai 2003). Citric acid is generally more effective than other LMWOAs in mobilizing inorganic P (Johnson and Loeppert 2006; Jones 1998; Kpomblekou and Tabatabai 2003; Staunton and Leprince 1996) because citric acid has three carboxyl groups while others have fewer carboxyl groups. Additionally, citrate is an effective chelator particularly for trivalent metal ions such as Fe3+ and Al3+, which are dominant in acidic soils (Hayes et al. 2000). Therefore, citric acid would be effective in mobilizing bound P in volcanic soils with a relatively low C investment. In addition, the primary product of carboxylate biosynthesis in the TCA-Cycle in cytosols under P limitation is oxaloacetate, which is readily converted to malate by the malate dehydrogenase reaction in cytosols (Uhde-stone 2003). An earlier study reported that the concentration of malate in root exudates tended to be high in P diminished soils (Fujii et al. 2012). Our study suggests that citric and malic acids in root exudates slightly increase within phylogenetic constraints in response to P limitation.

Mean concentrations of non-crystalline minerals and the P bound to non-crystalline minerals (oxalate extracted-P) were lower in the rhizosphere than in non-rhizosphere soils in the 170-m site only (Table 4). Thus, the first hypothesis was supported in the 170-m site only. The reason why non-crystalline minerals and the P bound to non-crystalline minerals (oxalate extracted-P) are not reduced in rhizosphere soils in the 200-m and 280-m sites is not known. Probably, desorption and adsorption of Fe/Al simultaneously occur in rhizosphere soils, which may obscure the effects of organic acids. Alternatively, organic-acid exudates cannot solubilize in-situ recalcitrant non-crystalline mineral fractions that are extracted with acid oxalate solution. We used acid-oxalate solution with 0.2 M ammonium oxalate, which is a standard method to quantify non-crystalline minerals in soil science. Organic-acid exudates in-situ may function to solubilize more weakly sparingly available soil P. Thus, analytical procedures should be improved in the future.

On the other hand, the concentration of organic P was significantly higher in rhizosphere than in non-rhizosphere soils in all study sites (Table 5), which consistently supported the second hypothesis. Probably, the growth of soil microbes was enhanced by the addition of labile organic matter in the rhizosphere soils through root exudates as previous studies reported (e.g. Phillips et al. 2011; Phillips and Fahey 2006; Yin et al. 2014). Consequently, the concentration of organic P as live and dead microbial P must have increased in the rhizosphere soils by capturing soil inorganic P to enhance their biomass. Organic P in rhizosphere soils also contains the P that originates from exfoliated root tissues; but such organic P must be quickly mineralized and resorbed by plants.

Organic P in rhizosphere soils must be mineralized by extracellular microbial phosphatases and/or root phosphatases to increase inorganic P (NaHCO3-Pi) in the rhizosphere. This rhizosphere effect was observed in all study sites so that the activities of soil microbes and enzymes are important to P acquisition by trees on volcanic soils. Portion of mineralized P (NaHCO3-Pi) can be acquired by trees, which would be a reward for trees in exchange for the cost of organic acid exudation. Reflecting these P dynamics, a greater concentration of total P, the sum of inorganic and organic forms of P, in the rhizosphere soils was observed. Exudation of organic acids, which causes rhizosphere priming effects and enhances the mineralization of P (Cheng 2009; Phillips and Fahey 2006), thus has an adaptive significance for trees growing on volcanic soils. However, we did not determine the soil microbial biomass P and soil enzyme activities. The detailed mechanism involving mineralization of organic P in rhizosphere soils should be studied in the future.

Conclusions

Our results indicate that the composition of LMWOAs in root exudates is species-dependent and phylogenetically determined. The roles of LMWOAs released from the roots are not to directly solubilize the P bound to the non-crystalline minerals, which are extractable with acid oxalate solution, but to increase organic-P fraction through priming effects. Priming effects by LMWOAs would enhance soil microbial biomass and probably their activities in rhizosphere soils. Secreting LMWOAs exudates is adaptive for trees to enhance microbial rhizosphere activities, which in turn benefits the trees to acquire inorganic P in volcanic soils. Further study is needed to clarify the contribution of soil microbes and enzymes in the rhizosphere.

Author contribution statement

MM designed the study and performed data collection, laboratory work and data analysis. SA performed fieldwork. KK contributed to the writing of the manuscript. All authors reviewed, revised and approved the articles.