Root response of soybean genotypes to low phosphorus availability from juvenile to adult vegetative stages

ABSTRACT Due to the rapid exhaustion of global phosphorus (P) resources, P-efficient crops are required. In this study, we used various soybean genotypes collected from around the world and investigated the morphophysiological responses of their roots to low-P conditions at the cotyledon emergence (VE), unifoliate leaves emergence (VC), and fourth trifoliate leaflet emergence (V4) growth stages. First, we compared the growth of 81 soybean genotypes under different P conditions at the VC stage. The root morphology of most genotypes did not differ according to P conditions. However, GmWMC138 showed increased root weight under low-P conditions at the VC stage, and was therefore selected for further comparative analysis with genotypes with similar seed weights. Four selected genotypes were compared in terms of their seed storage P content and responses of plant growth and phytase and acid phosphatase activities to low-P conditions at the VE and VC stages. The inorganic-P (Pi) levels and shoot growth at the VE and VC stages of GmWMC138 were less affected by low-P conditions compared to the other genotypes. In this genotype, root fresh weight at the VC stage, phytase activity in roots at the VE stage, and acid phosphatase activity in roots at the VC stage increased under low-P conditions. The differences in these enzyme activities may have led to the maintenance of root Pi content and subsequent increase in the root fresh weight at the VC stage under low-P conditions. In addition, the low-P responses of growth and P content at the V4 stage were compared among the selected genotypes. In GmWMC138, fine-root length increased and total P content was maintained under low-P conditions compared to normal-P conditions. These results imply that in soybeans morphological changes in roots in response to low-P conditions at juvenile growth stages, such as the VE and VC stages, may contribute to P-deficiency tolerance in subsequent growth stages, such as the V4 stage.


Introduction
Phosphorus (P) is an essential nutrient for plant growth, with important roles in ATP synthesis and photosynthesis (Chu et al. 2018). However, it is mostly fixed in the soil, and is unavailable for plants. Therefore, farmers fertilize the soil with excess P to enhance its uptake by plants and ensure optimal growth. However, as it is anticipated that rock phosphate, the natural source of P, will be depleted within the next few hundred years (Dawson and Hilton 2011), it is crucial to develop P-efficient crop genotypes to achieve sustainable agricultural practices.
Plants have developed various strategies to acquire poorly soluble P from the soil (van de Wiel, van der Linden, and Scholoten 2016). The organic acids released by plants stimulate P uptake by mobilizing the inorganic P (Pi) bound to soil particles, e.g., Fe-P and Al-P (Wang and Lambers 2020). Furthermore, various plant species, including Lupinus albus and Caustis blakei, secrete phosphatases to mobilize organic Pi (Johnson, Vance, and Allan 1996;Playsted et al. 2006). To obtain isolated P, roots adopt various strategies, including symbiosis with arbuscular mycorrhizal fungi (Smith, Smith, and Jakobsen 2003) and changes in the root architecture to increase soil exploration.
Root architecture is an important factor in P-acquisition efficiency. Generally, plants resistant to low P levels have short taproots, long lateral roots, and shallow root architecture (Niu et al. 2013;Lynch 2011). Lateral root branching is considered beneficial for the exploration of soil domains that are not reached by axial roots (Postma, Dathe, and Lynch 2014). Plants show phenotypic plasticity in their roots, which is defined by the ability to change root traits in response to the environment (Sandhu et al. 2016). Schneider and Lynch (2020) reviewed a number of studies concerning root phenotypic plasticity in various plant species under different types of stress. Genotypes with high plasticity are expected to have adaptability to P-deficient environments (McLachlan et al. 2020). Architectural plasticity of roots, in response to low-P conditions, has been observed in various crops, including rice (Oryza sativa), maize (Zea mays), sorghum (Sorghum bicolor), and common bean (Phaseolus vulgaris) (Vejchasarn, Lynch, and Brown 2016;Zhu, Kaeppler, and Lynch 2005;Zhang et al. 2013;Parra-Londono et al. 2018;Liao et al. 2001). A common bean genotype resistant to low P levels was found to have high root architecture-plasticity: its roots became shalloweunder low-P treatment than under high-P treatment (Liao et al. 2001). Under low-P conditions, soybean relies on plasticity in terms of both root architecture and root exudation (Lyu et al. 2016). This is in contrast to Cicer arietinum, which relies mainly on plasticity in terms of root exudation, and to Z. mays, Triticum aestivum, and Brassica napus, which rely on root architectural plasticity (Lyu et al. 2016). It is not yet clear whether there is genotypic variation in root architectural plasticity among soybean genotypes. Miguel et al. (2013) and White et al. (2018) reported that in common bean and potato, genotypic differences in root growth in the early growth stages under low-P conditions affect growth at later stages. They noted that early root development plays an important role in P acquisition and overall yield under low-P conditions. However, most studies have assessed low-P resistance at later growth stages, such as the flowering and harvesting stages, when P-deficiency symptoms are evident (Li et al. 2010;He et al. 2017;Zhang et al. 2009). Therefore, it is important to perform assessments at earlier growth stages.
During the early growth stages, the growth of many plant species is strongly affected by seed weight and seed nutrient content (Milberg and Lamont 1997;Hanley et al. 2007). Vandamme et al. (2016) reported that the weight of soybean seeds was strongly correlated with growth in the juvenile and flowering stages under both low-and high-P conditions, and a stronger correlation was observed under low-P conditions than high-P conditions. However, the effects of seed size and seed nutrition on root morphological plasticity in response to low-P conditions are poorly understood.
Intracellular phytases and acid phosphatases play major roles in P metabolism. Phytate is the main form of P storage in the seeds of cereals and legumes (Reddy et al. 1989). Plant roots also contain phytate as well as seeds (Campbell et al. 1991). Plants use phytases to convert phytates into more mobile orthophosphate anions. In several plant species, phytase is thought to play a more important role in internal plant P metabolism than in external P acquisition (Hübel and Beck 1996;Richardson, Hadobas, and Hayes 2000). Acid phosphatase liberates Pi from phosphate esters and insoluble soil minerals both within and outside cells. Intracellular acid phosphatase enhances Pi salvage from old cells for P remobilization (Plaxton and Carswell 1999). Acid phosphatase activity increases in germinating seeds and plays an important role in seed germination (Biswas and Cundiff 1991). Hayes, Richardson, and Simpson (1999) reported that low-P conditions increase phytase and acid phosphatase activity in 21-or 22-day-old temperate pasture grass and legume species, Trifolium subterraneum, Medicago polymorpha, Phalaris aquatica, and Danthonia richardsonii. They reported that the increased phytase activity in response to P deficiency was more apparent than that of acid phosphatase. However, the phytase and acid phosphatase activities of juvenile soybeans and their effects on root plasticity, with respect to adaptation to low-P soil conditions, remain unclear.
Soybean is an essential legume crop that is widely used in the food and oil industries. Demand for soybeans is increasing with the world's population, and it is therefore critical to improve the efficiency of soybean production. Legumes, including soybeans, generally require more P than other crops, partly because N 2 -fixing root nodules are strong P sinks (Pang et al. 2018). As it has recently been estimated that P will become exhausted by the end of this century (Dawson and Hilton 2011), P-efficient soybean genotypes will be needed. However, little is known about the root plasticity of soybeans in the early growth stages in response to low-P soil conditions, or the effects of such changes on growth and P acquisition in the later growth stages. Therefore, to identify P-efficient soybean genotypes, we focused on the juvenile growth stages when P deficiency symptoms are not expected to be evident in shoots. We screened 81 soybean genotypes for low-P tolerance during the unifoliate leaves emergence (VC) stage. The selected genotypes were used to investigate traits that may be involved in tolerance to low-P conditions. Root architecture, chemical forms of P in seeds, and phytase and acid phosphatase activities at the cotyledon emergence (VE) and VC stages were compared among the selected genotypes. Furthermore, the shoot and root growth and P acquisition of these genotypes were compared at the adult vegetative stage, the fourthtrifoliate-leaflet-emergence stage (V4), when P deficiency symptoms begin to appear.

Experiment 1: evaluation of root plasticity in response to low-P conditions at the VC stage
Eighty genotypes from the World Soybean Mini-core Collection (Kaga et al. 2012) were obtained from the Genebank of the National Institute of Agrobiological Sciences, Tsukuba, Japan (NARO Genebank Project, Japan). This set includes genotypes from global geographical locations and with a wide range of single nucleotide polymorphisms (SNPs). The seeds of the World Soybean Mini-core Collection genotypes were harvested from an experimental field of Osaka Prefecture University, Sakai, Japan (34°32′36.6′N, 135°30ʹ22.4′E). In addition to the 80 genotypes from the World Soybean Mini-core Collection, we also included the common Japanese Fukuyutaka genotype.
For the −P and +P treatments, 0 and 500 mg kg −1 Ca(H 2 PO 4 ) 2 ·H 2 O were applied, respectively. N (120 mg kg -1 of [NH 4 ] 2 SO 4 ) and K (120 mg kg -1 of K 2 SO 4 ) were applied for both treatments. For the -P-treatment conditions, Ca was supplemented by adding 341 mg kg -1 CaSO 4 · 2H 2 O to adjust the Ca in Ca(H 2 PO 4 ) 2 ·H 2 O which was added as fertilizer under the +P treatment-conditions. The available P (Truog-P) values in soils in the +P and -P treatment-conditions were 6.1 mg P 2 O 5 /100 g and 1.8 mg P 2 O 5 /100 g, respectively.
Seven seeds were sown in vinyl pots, each 12 cm in diameter and 11 cm in height filled with 640 g andosol. Plants were grown in a growth chamber (Biotron NC350; Nippon Medical & Chemical Instruments Co., Ltd., Osaka, Japan) at 25°C/18°C, with a 14 h photoperiod and a photosynthetic photon flux density of 422 µmol m −2 s −1 . After 8-12 days of germination, three to five plants were sampled when the unifoliate leaves emerged (VC stage). Then the plants were rinsed with tap water and each plant was separated into shoots and roots. Lateral root numbers were counted manually, and taproot lengths were measured using a ruler. After oven drying at 70°C for 48 h, the dry weights of shoots and roots were determined. The average lateral-root weight (g per lateral root) was calculated by dividing the lateral-root weight by the number of lateral roots.

2.2.
Experiment 2: comparison of root lengths classified by root thickness, seed content of different forms of P, and enzyme activity among selected genotypes in two juvenile growth stages Four of the World Soybean Mini-core Collection genotypes (GmWMC138, GmWMC150, GmWMC165, and GmWMC170) were selected for further analysis based on the results of Experiment 1. To compare P storage among the genotypes, total P and Pi contents of the seeds were measured. About 20 seeds of each genotype were dried in an oven at 70°C for 24 h and ground. To determine total P concentration, three replicates of 0.1 g of each seed sample were digested with HNO 3 and H 2 O 2 . Total P was estimated by colorimetric vanadomolybdate assay (Ma and McKinley 1953), based on absorbance measured at 440 nm (A 440 ) using a spectrophotometer (ASV11D; As One Corp., Osaka, Japan). The Pi concentrations in seeds were measured according to the method of Raboy and Dickinson (1984) with minor modifications. Briefly, 0.1 g of each ground seed sample was extracted twice with 4 mL 12% (w/v) trichloroacetic acid containing 25 mM MgCl 2 using a vortex mixer (GENIE2; Electro Scientific Industries, Inc., New York, NY, USA). Each extract was centrifuged at 10,000 × g for 10 min and the supernatant was collected. The collected supernatant samples were combined, and Pi was estimated using the molybdenum-blue reaction (Murphy and Riley 1962) based on measurement of the absorbance at 710 nm (A 710 ) using a spectrophotometer (U-1900; Hitachi High-Technologies Corp., Tokyo, Japan). The total P and Pi contents of seeds were calculated by multiplying the seed weight by the respective concentration.
Pot tests of the selected genotypes were performed using the same soil as in Experiment 1. For -P and +P-treatment conditions, 0 and 2200 mg kg -1 of KH 2 PO 4 were applied, respectively. N (120 mg kg -1 of [NH 4 ] 2 SO 4 ) and Ca (341 mg kg -1 of CaSO 4 · 2H 2 O) were applied in both treatments. In the -P-treatment condition, K was supplemented by adding 1200 mg kg -1 KCl. Two seeds per pot were sown in plastic pots, each 7.5 cm in diameter and 7.5 cm in height and filled with 150 g andosol. To avoid root interactions among plants, the seedlings were later thinned to one plant per pot. The plants were grown in a growth chamber under similar environmental conditions to those in Experiment 1. Sampling was conducted at two growth stages: on cotyledon emergence after 6 days of germination (i.e., at the VE stage) and at the VC stage. To avoid deactivation of enzymatic activity, all sampling operations were performed within 30 min and the sample temperature was maintained at approximately 4°C by immersion in ice-cold tap water. Sampled plants were rinsed with ice-cold tap water and transferred into a clear tray filled with water before being scanned (Epson Perfection V750 Pro; Epson America Inc., Los Alamitos, CA, USA). The scanned images were analyzed using WinRHIZO software (Regent Instruments Inc., Québec, QC, Canada) to estimate the total root length and classify root length by thickness. Then the scanned plants were separated into shoots and roots, and their fresh weights were determined after blotting off any excess water with a paper towel. The weighed samples were frozen rapidly in liquid nitrogen and stored at -20°C until enzyme extraction.
Enzyme solution was prepared as described by Zhou et al. (2018), with minor modifications. The shoot and root samples were ground with liquid nitrogen, and ice-cold sodium acetate extraction buffer (100 mmol L −1 , pH 5.4) was added at a ratio of 1:8 (fresh weight:extraction buffer volume). Then the samples were centrifuged at 12,000 × g for 20 min at 4°C, and the supernatants containing the enzymes were collected. Phytase activity was determined as described by Hayes, Richardson, and Simpson (1999), with minor modifications. The enzyme assay mixture contained 40 µL substrate solution (2 mmol L -1 sodium phytate and 1 mmol L -1 CaCl 2 in 100 mmol L -1 sodium acetate buffer, pH 5.4) and 10 or 20 µL enzyme solution from shoots and roots, respectively. After 60 min of incubation at 37°C, the reaction was terminated by adding 50 µL ice-cold 10% trichloroacetic acid. The control mixture was prepared by adding trichloroacetic acid to the sample-buffer mixture before adding the substrate solution. The amount of inorganic phosphate liberated in the assay mixture was estimated based on measurement of A 710 with a spectrophotometer (U-1900; Hitachi High-Technologies Corp.) using the molybdenum-blue reaction.
Acid phosphatase activity was measured according to the protocol of Hayes, Richardson, and Simpson (1999) with minor modifications. The enzyme solution was diluted with sodium acetate buffer at a ratio of 1:24 (shoot) or 1:1 (root). The enzyme assay mixture contained 123.5 µL 10 mM p-nitrophenylphosphate solution in sodium acetate buffer and 3 µL diluted enzyme solution. After 20 min incubation at 27°C, the reaction was terminated by adding 123.5 µL 0.25 M NaOH. To determine the concentration of p-nitrophenol, the absorbance at 415 nm (A 415 ) was measured using a filter-based microplate photometer (MTP-310Lab; Corona Electric Co., Ltd., Ibaraki, Japan).
The protein content of the enzyme solution was determined using Pierce 660 nm Protein Assay Reagent (Thermo Fisher Scientific K.K., Tokyo, Japan) with bovine serum albumin (BSA) as a standard. The supernatants (5 µL) were diluted by adding 20 µL sodium acetate buffer. The diluted solution (20 µL) was added to 200 µL assay reagent and the mixture was left for 5 min to react. Then the absorbance at 660 nm (A 660 ) was measured using a filter-based microplate photometer. One phytase activity unit (U) was defined as 1 µmol Pi released from sodium phytate per minute and one acid phosphatase activity unit (U) was defined as 1 µmol p-nitrophenol released from p-nitrophenylphosphate per minute. In both cases, enzyme activity was expressed as mU mg -1 protein.

Experiment 3: phenotypic variation in the root system and growth at the V4 stage in response to low-P conditions
This experiment was conducted in a glasshouse at Osaka Prefecture University, Sakai, Japan (34°32′36.6′N, 135°30′22.4′E), from September to November 2019. The average temperature during the experimental period was 24°C. The four genotypes selected in Experiment 2 were used in this experiment. Plastic pots, each 12 cm in diameter and 12 cm in height were filled with 640 g andosol. Considering the longer growth period of this experiment, P fertilization was applied in both low-P (LP) and high-P (HP) treatments (220 and 2200 mg kg -1 of KH 2 PO 4 were applied, respectively). N (120 mg kg -1 of [NH 4 ] 2 SO 4 ) was applied in both treatments. To ensure K supplementation, 1100 mg kg -1 KCl was added to the LP treatment. To mimic natural conditions, commercial inocula of the N 2 -fixing bacterium, Bradyrhizobium japonicum ('Konryukin Mame-Zo'; Tokachi Nokyoren, Hokkaido, Japan), were added at 5 mL per pot of inoculum suspension at a ratio of 10 g:1000 mL (inocula weight:tap water volume). One seedling was grown per pot. At the V4 stage (26-50 days after sowing), each plant was sampled and separated into shoots and roots. The roots were scanned, and the images were used for analysis. Total root length and the ratio of root length to root diameter were measured using image-analysis software, as described previously. The shoots and roots were dried in an oven at 70°C for 48 h and then weighed. Shoots and roots were ground and digested separately with HNO 3 to analyze P. The P concentration was measured as A 440 using the colorimetric vanadomolybdate assay (Ma and McKinley 1953) with a spectrophotometer (ASV11D; As One Corp.). The P contents of the shoots and roots were calculated by multiplying the dry weight by the P concentration. The sum of the P levels of individual shoots and roots was considered the total P content. The root P ratio was calculated by dividing the root P content by the total P content.

Data analysis
All data were analyzed using IBM SPSS statistics v. 26.0.0.1 (IBM, Chicago, IL, USA). In all experiments, the mean values of each treatment were compared using Student's t test. Pearson's simple correlation analysis was used to analyze the relationships between seed weight and dry weight under -P and +P treatment conditions at the VC stage in Experiment 1. The values of the root parameters under -P/+P treatment conditions in Experiment 1 were standardized and used as explanatory variables in principal component analysis. In Experiment 2 and Experiment 3, the data were examined by analysis of variance (ANOVA), and genotypic variation was compared using Tukey's multiple comparison test at the 0.05 probability level.
The plasticity index was calculated as described by Sandhu et al. (2016) using the data of root P distribution from Experiment 3, according to the following formula: where x LP represents a single replicate value from the LP treatment and � x HP is the average value from the HP treatment.

Experiment 1: evaluation of root plasticity in response to low-P conditions at the VC stage
Eighty of the World Soybean Mini-core Collection genotypes and Fukuyutaka, a Japanese common genotype, were evaluated for genotypic variation and root architecture under the different P conditions. Genotypic variation was observed in the dry weights of shoots and roots ( Figure S1). Most shoot weights were in the range 0.075-0.100 g plant -1 in both treatments, and most root weights were in the range 0.015-0.025 g plant -1 under +P-treatment conditions and 0.020-0.025 g plant -1 under -P treatment conditions. Overall, histograms showed that -P-treatment conditions did not change the frequency distributions in either shoot or root growth at the VC stage ( Figure S1) but did affect the dry weight of the roots of some genotypes. To examine the effects of P status on the growth of individual genotypes, the dry weights of shoots and roots under the -P and +P-treatment conditions were compared. The shoot dry weights of most genotypes plotted very close to the y = x line ( Figure S2). However, the plots of root dry weight were distinct, with some genotypes plotted far from the y = x line. These observations indicated that the root growth of some genotypes changed in response to -P soil conditions. There were significant positive correlations between seed weight and the dry weights of both shoots and roots, regardless of the P treatment ( Figure S3). However, the correlation coefficient between root dry weight and seed weight was weaker than that between shoot dry weight and seed weight in both P treatments. These results indicate that growth at the VC stagewas markedly affected by seed weight, although root growth than on shoot growth.
To select characteristic genotypes, principal component analysis was performed using the -P/+P ratio of root traits as explanatory variables. PC1 explained 61.7% of the variability in root traits, while PC2 explained 23.6% (Figure 1). As the first two PCs accounted for more than 80% of the variability, only they were used in further analysis. PC1 was positively related to root weight; PC2 was positively affected by taproot length and the number of lateral roots. In the plot of the PC scores of each genotype, many genotypes were plotted near the origin. However, some genotypes were plotted far from the origin. These were considered characteristic genotypes in which the roots displayed plasticity under low-P conditions. Among them, GmWMC138 had the highest PC1 score, and was plotted furthest from the origin (Figure 1(b)). Therefore, GmWMC138 was used in subsequent experiments.
Furthermore, three genotypes with seed weights similar to GmWMC138, i.e., GmWMC150, GmWMC165, and GmWMC170, were selected as contrast genotypes. Judging from the PC scores of these genotypes, GmWMC150 and GmWMC165 had low plasticity, GmWMC170 had medium plasticity, and GmWMC138 had high plasticity with regard to P-free fertilization conditions (Figure 1(b)).

Experiment 2: comparison of root lengths classified by root thickness, seed content of differnt forms of P, and enzyme activity among selected genotypes in two juvenile growth stages
To assess the genotypic differences associated with seed P storage, the total P and Pi contents of the seeds were compared among the four selected genotypes (Table 1). Although there were no significant differences in seed weight, genotypic variation was observed in total P and Pi contents. GmWMC165 had the highest P concentration, seed P content, and seed Pi content, and all three were lowest in GmWMC170. GmWMC138 and GmWMC150 had intermediate seed P concentrations, but their seed Pi contents were lower than that of GmWMC138. The ratio of Pi content in seeds to total P content of all selected genotypes was around 11%.
The growth of the selected genotypes was investigated at the VE and VC stages ( Figures S4, S5). Only GmWMC138 showed an increase in root fresh weight at the VC stage under the -P-treatment conditions; it did not show any other significant differences in growth parameters according to P status. On the other hand, under the low-P-treatment conditions, the shoot fresh weight of GmWMC150 decreased significantly even at the VE stage, and all growth parameters decreased at the VC stage. In GmWMC165, the shoot and root fresh weights at the VE stage also decreased under -P-treatment conditions. However, this genotype recovered from the growth delay at the VC stage. GmWMC170 did not show any changes in growth under -P-treatment conditions at either growth stage.
Root length, classified by root diameter, was examined to investigate changes in root architecture under different P conditions. At the VE stage, the root systems of all genotypes were composed of taproots and first lateral roots. Roots > 0.8 mm in diameter were mainly taproots, and those with a diameter < 0.8 mm were first lateral roots. In the VC stage, the root systems of all genotypes consisted of taproots, first lateral roots, and second lateral roots ( Figure S5). In the VC stage, the taproot was generally classified by a diameter > 0.8 mm, and the relatively thick first lateral root was classified by a diameter of 0.4-0.8 mm. In addition, the relatively thin first lateral root and the second lateral root were classified by a diameter < 0.4 mm. At the VE stage, GmWMC138 and GmWMC170 had longer roots compared to the other genotypes, with diameters of 0.2-0.4 mm, regardless of the P condition ( Figure S6). However, the differences in root length among the genotypes were reduced at the VC stage (Figure 3). Comparrison of the treatments in terms of root growth at the VC stageshowedthat under the low-P-treatment conditions, the root lengthof GmWMC138 roots with a diameter > 0.8 mm increased significantly ( Figure 3). By contrast, the length of GmWMC150 roots with diameters of 0.2-0.4 mm and 0.6-0.8 mm decreased under -Ptreatment conditions.
The shoot and root Pi contents of GmWMC138did not differ significantly between treatments in either the VE or VC stage ( Table 2). The shoot and root Pi contents of GmWMC150 also did not differ significantly between the treatment conditions at the VE stage; however, significant decreases were observed under -P-treatment conditions at the VC stage. The shoot Pi content of GmWMC165 decreased significantly under the -P-  treatment conditions at the VE stage, while root Pi content decreased under the -P-treatment conditions at the VC stage. Under the Low-P-treatment conditions, the root Pi content of GmWMC170 decreased significantly, while the shoot Pi content was maintained, at both growth stages.
To assess the relationship between internal enzyme activity and growth response to low-P conditions, phytase and acid phosphatase activities were determined. Overall, phytase activity in shoots was maintained from the VE stage to the VC stage, and was higher at the VC stage than at the VE stage in all genotypes (Tables 3, 4). On the other hand, the phytase and acid phosphatase activities in roots tended to be lower at the VC stage than at the VE stage in all genotypes. Genotypic differences were observed in phytase activity in roots; in GmWMC170, activity was higher at the VC stage. In GmWMC138, there was a significant decrease in shoot phytase activity and a significant increase in root phytase activity under -P-treatment conditions at the VE stage. On the other hand, in GmWMC150 and GmWMC165, root phytase activity decreased significantly under -P-treatment conditions at the VE stage. Genotypic differences were observed in acid phosphatase activity in both shoots and roots. As observed for phytase activity, acid phosphatase activity in shoots the same uppercase or lowercase letters within respective valuables are not significantly different at P < 0.05 under +P-or -P-treatment conditions, as determined by Tukey's multiple comparison test. § Significant difference between +P-and -P-treatment conditions at *P < 0.05; **P < 0.01; ns not significant as determined using Student's t test. SEM: standard error of the mean. 4.3 ± 0.33 a 4.4 ± 0.21 a 5.5 ± 0.46 a 3.9 ± 0.80 a † Values represent the means of five replicates ± SEM. ‡ One phytase activity unit (U) was defined as 1 µmol Pi released from sodium phytate per minute. § Values with the same uppercase or lowercase letters within respective valuables are not significantly different at P < 0.05 in the +P-or −P-treatment conditions, as determined by Tukey's multiple comparison test. ¶ Significant difference between the +P-and −P-treatment conditions at *P < 0.05; **P < 0.01; ns not significant as determined by Student's t test. SEM: standard error of the mean. .7 ab †Values represent the means of five replicates ± SEM. ‡ One acid phosphatase activity unit (U) was defined as 1 µmol p-nitrophenol released from p-nitrophenylphosphate per minute. § Values with the same uppercase or lowercase letters within respective valuables are not significantly different at P < 0.05 in the +P-or −P-treatment conditions, as determined by Tukey's multiple comparison test. ¶ Significant difference between the +P-and −P-treatment conditions at *P < 0.05; **P < 0.01; ns not significant as determined by Student's t test. SEM: standard error of the mean. and roots was higher in GmWMC170 than in the other genotypes. At the VE stage, a significant increase in root acid phosphatase activity under -P-treatment conditions was observed only in GmWMC170.At the VC stage, root acid phosphatase activity under the -P-treatment conditionsincreased significantly in GmWMC138 and GmWMC165.

Experiment 3: phenotypic variation in the root system and growth at the V4 stage in response to low-P conditions
At the V4 stage, all genotypes exhibited symptoms of P deficiency, such as yellow bottom leaves. The dry weights of shoots of GmWMC138 were unaffected bythe P-treatment conditions but the dry weights of roots and total root length increased significantly under LP conditions (Figure 4(b)). GmWMC170 exhibited similar behavior to that of GmWMC138 in the V4 stage. By contrast, GmWMC150 showed a significant decrease in shoot dry weight under LP conditions, while the root dry weight and total root length were unaffected. Furthermore, GMWMC165 showed no significant differences in the dry weights of shoots or roots between conditions. The R/S ratios of GmWMC150, GmWMC165, and GmWMC170 increased under LP conditions but were unaffected by low-P conditions in GmWMC138.
The total root lengths of GmWMC138 and GmWMC170 were significantly longer under LP conditions (Figure 4(c)). Image analyses of roots showed that roots in all diameter ranges were longer under LP conditions than under HP treatment in GmWMC138 ( Figure 5). GmWMC170 also showed a similar tendency except for roots with a diameter of 0.4-0.2 mm. These results indicate that under LP conditions, the root length of GmWMC138 increased to a significantly greater extent compared to the other genotypes ( Figure 6). By contrast, the lengths and weights of the roots of GmWMC150 and GmWMC165 did not change under LP conditions. Shoot P content decreased in all genotypes under LP conditions (Figure 7(a)). A similar trend was observed in total P content, but no significant differences in total P content were observed in GmWMC138, and the decrease in total P content under LP conditions was limited to approximately 30% (Figure 7(c)). By contrast, in the other three genotypes, the total P contents under LP conditions were reduced to approximately half or less of the values obtained under HP conditions. The plasticity index was calculated by dividing the difference between the average of values under HP conditions and each value under LP conditions by the average under HP conditions (Figure 7(d)). The plasticity index values of the root P ratio were positive in all genotypes, showing that the rate of P distribution to roots increased under LP conditions in all genotypes. The plasticity index value of the root P ratio in GmWMC138 was 1.56, significantly higher than that of GmWMC150 (0.53) and slightly higher than those of GmWMC165 (0.83) and GmWMC170 (1.06).

Responses of roots to low-P conditions vary among genotypes in the VC stage
Large variations in shoot and root growth were observed among genotypes in Experiment 1, which were largely unaffected by the P treatment conditions ( Figure S1). Due to the short growth period, the effects of the P-treatment conditions on the parameters analyzed in this study were small in most genotypes. However, some genotypes showed marked differences in root dry weight between the -P and +P treatment conditions ( Figure S2). These observations corroborate the results reported by Liao et al. (2001), who compared root growth in five common bean (P. vulgaris) genotypes at 6 days after transplantation under low-P conditions. They reported that only some genotypes exhibited changes in root architecture , and the response remained active even at 4 weeks after transplantation. Consistent with these findings, we observed that some genotypes exhibited more pronounced root responses to low-P conditions, even in the very early growth stages. In principal component analysis, the PC1 value of GmWMC138 was particularly high, corresponding to an increase in traits related to the weights of taproots, lateral roots, and total roots under low-P conditions (Figure 1(b)).

Seedling growth at the VC stage is affected by seed weight but there is no relationship between root response to low-P conditions and P levels of seeds
The growth of soybeans in the period after sowing is generally associated with seed weight. Vandamme et al. (2016) observed an association between seed weight and shoot/ root growth from the first trifoliate leaflet emergence stage to the flowering stage among 42 soybean genotypes. They showed that early growth was strongly correlated with seed weight, and that this relationship was maintained even at the flowering stage. In the present study, the shoot and root dry weights were positively correlated with seed weight under both +P and -P treatment conditions among the 81 genotypes examined ( Figure S3). We observed a weaker positive correlation of root dry weight with seed weight than that of shoot dry weight in both treatment conditions. This may have been because the dependence on seed P is different between shoots and roots. Nadeem et al. (2013) reported that up to 71% of P from the seed was relocated to the shoot in the early growth stage of maize. They suggested that the root is less dependent on seed P than the shoot during the early growth stage.
GmWMC138 is a genotype with small seeds. The average seed weight of the World Soybean Mini-core Collection genotypes used in Experiment 1 was 153 mg seed −1 , while that of GmWMC138 was 77 mg seed −1 . Therefore, we selected contrasting genotypes with similar seed weights. However, the results of seed P analysis revealed genotypic differences in P concentration, total P content, and Pi content (Table 1).
As Pi is a free form of P and does not need to be metabolized before use, we considered that the seed Pi-rich genotypes would show better growth in the early growth stages under low-P conditions. However, no relationship was evident between seed Pi content and growth under low-P conditions in this study.

Phytase and acid phosphatase activities may be related to plant Pi content and growth in the VE and VC stages
The relationships between internal phytase and acid phosphatase activities and soil P have been studied in several plant species (Hayes, Richardson, and Simpson 1999;Richardson, Hadobas, and Hayes 2000;Araújo, Plassard, and Drevon 2008). Hayes, Richardson, and Simpson (1999) compared the internal phytase and acid phosphatase activities of roots under adequate and deficient P conditions using 21-or 22day-old pasture grass and legume species: T. subterraneum, Trifolium repens, M. polymorpha, P. aquatica, and D. richardsonii. They found that P deficiency led to significant increases in phytase activity in all of these species except Trifolium repens, and lesser increases in acid phosphatase activity in all five species.
In the present study, GmWMC138 showed significant increases in root phytase activity at the VE stage and in root acid phosphatase activity at the VC stage under -P-treatment conditions (Tables 3, 4). These increases may have contributed to the maintenance of root Pi content and increased the root fresh weight at the VC stage (Table 2, Figure 2(b)). GmWMC150 did not show increases in either enzyme activity under -Ptreatment conditions in either growth stage, and showed significantly decreased shoot and root Pi contents in the VC stage.
GmWMC165 did not maintain its Pi content in either growth stage, although this genotype showed increased acid phosphatase activity in roots under -P-treatment conditions in the VC stage. Acid phosphatase activity increased in GmWMC170 under -P-treatment conditions at the VE stage, and Pi content decreased under -P-treatment conditions at both the VE and VC stages. Therefore, it is likely that acid phosphatase in roots may not have been responsible for the increase in internal Pi in this genotype. In summary, GmWMC138 may have enhanced P circulation as a result of increased root phytase activity in the VE stage and increased root acid phosphatase activity in the VC stage.

Root plasticity at the VC stage partially contributes to growth up to the V4 stage
Under LP conditions, GmWMC138 and GmWMC170 developed greater numbers of fine roots (0.2-0.4 mm in diameter) in the VE stage than the other genotypes ( Figure S6), which resulted in significantly increased dry weight and total length of roots under LP conditions in the V4 stage (Figures 4 and 7). Fine and long roots increase the surface area and enhance the ability of P exploration (Kochian, Hoekenga, and Piñeros 2004). Strock, Morrow de La Riva, and Lynch (2018) compared shoot growth under low-P conditions among common bean genotypes with reduced or advanced radial thickening of the roots, and found that genotypes with reduced thickening had greater shoot growth compared to those with advanced thickening. In the present study, GmWMC138 and GmWMC170 may have also had superior P exploration ability due to the increased length of thin roots. On the other hand, compared to GmWMC150 and GmWMC165, the total P content and plant biomass of GmWMC138 and GmWMC170 were not higher under HP conditions in this study, implying that these genotypes require less P for growth than the others. The development of greater numbers of fine roots in juvenile growth stages may represent a survival strategy to obtain a competitive advantage for P acquisition. GmWMC138, which exhibited root-system changes earlier than did the other genotypes, showed superior root development at the V4 stage (Figures 4 and 6); its plasticity index values for P distribution to roots were particularly high (Figure 7). Using a greenhouse experiment and a simulation root model of Lupinus angustifolius, Chen et al. (2013) observed that a genotype with high root plasticity had superior P uptake compared to genotypes with low plasticity. Sandhu et al. (2016) showed that rice genotypes with high root plasticity achieved more stable yields under both drought and low-P conditions. Therefore, the promoted root growth observed in GmWMC138 would contribute to increasing P uptake and shoot growth under low-P conditions.
The shoot dry weights of GmWMC165 at the V4 stage were similar under the LP and HP-treatment conditions in this study (Figure 4), although its growth was significantly suppressed under low-P conditions at the VE stage ( Figure S4). Its roots did not show morphological changes but it had the highest plant biomass and total P content at the V4 stage (Figure 7). Efficient P acquisition by GmWMC165 may depend on traits other than root plasticity.

Conclusion
We used World Soybean Mini-core Collection genotypes to investigate the morphophysiological responses of soybean to low-P conditions at the VE and VC stages, when soybean relies on seed nutrition, and the V4 stage, when it becomes independent from its seed nutrition. Under low-P conditions in the VC stage, the number of roots of the soybean genotype GmWMC138increased. The shoot growth of this genotype was also constant under low-P conditions, even in the V4 stage, implying that the changes in roots in the early growth stage in response to low-P conditions represent a strategy of this genotype to acquire P. Further studies are required to evaluate whether root plasticity in the early growth stage can directly increase P acquisition and growth in later growth stages.

Acknowledgments
We are deeply grateful to Lixing Yuan (China Agricultural University) for providing many constructive comments and kind support to the study. . Genotypic variation in (a) shoot dry weight, (b) root dry weight, (c) total root length, and (d) R/S ratio at the V4 stage with five replicates. Bars with the same uppercase or lowercase letters are not significantly different at P < 0.05 under HP or LP treatment conditions, as determined by Tukey's multiple comparison test. *P < 0.05; **P < 0.01; ns: not significant, as determined using student's t test. Error bars represent standard error of the mean (SEM). . Genotypic variation in (a) shoot, (b) root, (c) total P content, and (d) root P content plasticity index values at the V4 stage with five replicates. The plasticity index values were calculated by dividing the value of the difference between the average of HP and each value in LP by the average of HP. Bars with the same uppercase or lowercase letters are not significantly different at P < 0.05 under the HP-or LP-treatment conditions, as determined by Tukey's multiple comparison test. *P < 0.05; **P < 0.01; ns: not significant, as determined using student's t test. Error bars represent standard error of the mean (SEM).

Disclosure statement
No potential conflict of interest was reported by the author(s).

Funding
This work was partly supported by JSPS KAKENHI under Grant Number 18K05599; Cabinet Office, Government of Japan, Moonshot Research and Development Program for Agriculture, Forestry and Fisheries (funding agency: Bio-oriented Technology Research Advancement Institution) under Grant Number JPJ009237; and the Yanmar Environmental Sustainability Support Association (K30057).