Phosphorus (P) is an indispensable macronutrient critical for plant growth and development [1], participating in a wide array of cellular processes including metabolic regulation, photosynthesis, respiration, and the biosynthesis of membrane phospholipids [2,3]. Soybeans (Glycine max L.) exhibit a high demand for phosphorus to achieve optimal seed yield and quality [4]. Phosphorus deficiency frequently manifests during early developmental stages rather than at maturity [4,5], leading many researchers to focus on the seedling phase as the key period for assessing phosphorus tolerance [6,7]. In soybean, phosphorus deficiency during early growth can significantly reduce seedling biomass, phosphorus concentration [8], leaf area [9], the number of nodes and seeds [10], and these detrimental effects may persist even after phosphorus availability is restored [11].

The primary source of phosphorus for plants is inorganic phosphate (Pi) [12], mainly in the form of H₂PO₄⁻ at neutral pH, which is actively transported into plant cells [2]. Despite its low concentration in soil solution (typically 1–10 µM), the internal Pi concentration in plant tissues is considerably higher, approximately 5–20 mM [13]. In alkaline soils, Pi tends to form complexes with calcium, whereas in acidic soils, it is often bound to aluminum or iron [13,14]. Furthermore, soil organic components, such as fertilizers and crop residues, particularly phytic acids (inositol compounds), can also sequester phosphates [15].

Phosphorus efficiency (PE) is a complex trait that can be dissected into two major components: phosphorus uptake efficiency (PUpE, the capacity to acquire inorganic phosphate from the soil) and internal phosphorus utilization efficiency (PUtE, the ability to produce biomass or yield per unit of tissue phosphorus) [14]. Phosphorus deficiency tolerance is a broader agronomic concept that encompasses the overall ability of a genotype to maintain growth and yield under low-P stress; it can be achieved through superior PUpE, superior PUtE, or a combination of both. To cope with phosphorus deficiency, plants have evolved an array of physiological and morphological adaptations that enhance phosphorus acquisition (i.e., phosphorus uptake efficiency, PUpE) and improve internal utilization efficiency (PUtE) [16,17,18]. These strategies, which primarily enhance PUpE, include forming symbiotic associations, modifying root architecture, and exuding organic acids and acid phosphatases from roots [19]. In this context, the cultivation of soybean varieties with high PE—defined here as the ability to sustain high biomass or grain yield under limited phosphorus availability—holds significant potential for enhancing soybean production [20,21]. Genotypic variation in both PUpE and PUtE has been documented across multiple crops [7], including soybean [4,21,22,23], rice (Oryza sativa L.) [24,25,26], Brassica (Brassica napus L.) [27], and wheat (Triticum aestivum L.) [28]. Vandamme et al. [21] observed that phosphorus uptake (a measure of PUpE) varied little among soybean genotypes under high phosphorus availability but showed a two-fold difference under low phosphorus conditions. While enhanced PUpE is recognized as a key adaptive trait, particularly in soils with high phosphorus retention capacity [29,30], superior PUtE may also contribute to reduced soil phosphorus extraction, especially in low-input agricultural systems [31,32]. Early root growth and the relative amount of seed-borne phosphorus reserves are potential determinants of phosphorus deficiency tolerance(i.e., the ability to maintain growth and yield under low P stress) [33,34].

Various stress indices have been proposed to quantify crop responses to environmental constraints, such as the stress susceptibility index (SSI) and the stress tolerance index (STI) [35]. These indices typically incorporate the population mean into their calculation and are thus sensitive to the composition of the evaluated panel. In the present study, we adopted the percent tolerance to phosphorus deficiency (PTPD) as the primary screening metric. PTPD expresses the response of each genotype solely as a proportion of its own control performance, making it more intuitive and independent of the group of accessions under evaluation. Although direct comparisons with STI or SSI were not a primary objective, this approach emphasizes relative stability and avoids the confounding effect of population-dependent ranking. Since individual growth parameters often respond inconsistently to low-P stress across genotypes, reliance on a single trait may lead to misclassification. This underscores the need for an integrated metric. We hypothesized that PTPD, derived from multi-trait PCA, could effectively rank soybean genotypes for P efficiency, and that shoot-based traits would serve as reliable proxies under P deficiency.

This study aims to: (1) screen for soybean genotypes exhibiting high phosphorus efficiency, and (2) identify key PE-related traits that can facilitate the development of high-efficiency soybean varieties through breeding programs.

Developing crop genotypes that maintain productivity with reduced phosphorus (P) inputs is a critical goal for sustainable agriculture [36]. In the present study, we proposed a multi-stage screening pipeline that integrated percent tolerance to phosphorus deficiency (PTPD), principal component analysis, and a comprehensive evaluation index to identify reliable selection traits and phosphorus-efficient soybean genotypes.

The broad variation in PTPD and comprehensive evaluation (CE) values observed among 98 genotypes is consistent with extensive genetic diversity in soybean response to low-phosphorus stress, consistent with previous reports in various crops [33,37]. The majority of growth traits declined significantly under P deficiency, with photosynthetic rate being the most severely affected (average −32.4%) and displaying the highest diversity index. This observation reinforces the central role of photosynthesis in plant responses to P limitation [38], as P is essential for ATP synthesis, RuBP regeneration, and chloroplast integrity. A small number of positive PTPD values were observed, indicating better trait performance under low phosphorus than under control conditions. These cases may reflect compensatory growth triggered by mild stress [39,40] or the well-documented acceleration of reproductive development under phosphorus limitation [41,42]. Such positive values should not be taken to suggest that phosphorus deficiency is advantageous; rather, they point to genotype-specific adaptive responses that merit further investigation.

A key challenge in breeding for phosphorus efficiency is the identification of easily measurable, high-throughput traits that reliably reflect overall phosphorus efficiency. Through PCA, we reduced the dimensionality of ten growth parameters and identified five core traits—SPAD values at V3 and R1, photosynthetic rate at R1, shoot dry weight at R8, and seed number per plant at R8—that repeatedly emerged as dominant contributors in both Experiment I and Experiment II. The consistency of these results across two independent screenings strengthens the argument for their use in routine selection. SPAD values, in particular, are rapid and non-destructive measures of leaf chlorophyll content, which declines under P stress due to reduced protein synthesis and chloroplast development [43,44]. Similarly, shoot dry weight and seed number at maturity integrate the cumulative effects of P availability throughout the growth cycle and are agronomically relevant endpoints.

Although root architectural traits are often emphasized in P efficiency studies [45], their measurement is labor-intensive and poorly suited for large-scale screening. Moreover, Vandamme et al. [46] noted that increases in root-to-shoot ratio under low P often reflect stunted shoot growth rather than enhanced root proliferation, which further complicates the interpretation of root morphology alone as a selection criterion. Our results suggest that shoot-based physiological traits, particularly those linked to photosynthetic capacity and reproductive output, may serve as pragmatic selection criteria for identifying genotypes with superior internal phosphorus utilization efficiency. However, it should be noted that root traits were not directly assessed in this study, and the contribution of the root system to phosphorus efficiency remains unevaluated. The absence of root phenotyping is therefore acknowledged as a limitation, and future studies integrating both shoot and root phenotyping are warranted to validate these findings.

Among the six phosphorus-efficient genotypes identified in Experiment II, CN 15 was unique in achieving the highest comprehensive evaluation (CE) value in both experiments: 0.856 (rank 1/98) in Experiment I and 1.06 (rank 1/12) in Experiment II. This consistency across independent screening stages suggests that CN 15 integrates multiple tolerance-related traits more effectively than other genotypes tested. In Experiment III, CN 15 maintained significantly higher ET₀/RC, φE₀, and PIABS than SN 22 under all phosphorus levels (p < 0.05), and exhibited a 26.2% increase in GPUE under 0 µM KH₂PO₄ treatment. These results indicate that CN 15 sustains photosynthesis under low P through more efficient electron transport and light energy utilization [47,48] and superior internal P-utilization efficiency (PUtE). Although not every pairwise trait difference among the six efficient genotypes reached statistical significance, CN 15 uniquely combined the highest composite CE ranking across both experiments with the strongest physiological validation in Experiment III, supporting its selection as a promising candidate for further field evaluation.

It should be acknowledged that the reduced set of traits identified by PCA, though consistently contributing to the comprehensive evaluation, does not by itself establish their predictive value for independent genotype panels. The secondary screening employed two phosphorus levels (0 and 100 µM Pi), which better reflect gradual soil P depletion, and ranking genotypes by the sum of CE values across both stress levels helps avoid selecting lines that perform well only under extreme starvation. Furthermore, the three experiments were conducted in different years (2020–2023). While the screening pipeline was sequential and year-to-year environmental variation could not be statistically separated from the screening stage, the greenhouse conditions were managed consistently to minimize environmental fluctuations. The strong physiological contrasts observed between the two extreme genotypes (CN 15 and SN 22) provide encouraging biological support for the selected indicators; however, because only the extremes were validated, the resolution of these traits for discriminating intermediate phosphorus efficiency remains to be tested in a broader population.

In addition to the above considerations, sand culture differs from field soils in several key aspects that may affect phosphorus efficiency rankings. First, sand lacks the strong P-sorption capacity of many agricultural soils (e.g., Fe/Al oxides in acidic soils or Ca-bound P in calcareous soils), which can drastically reduce P availability and alter the relative importance of P uptake versus internal utilization traits. Second, the absence of a complex rhizosphere microbiome in sand eliminates potential plant–microbe interactions, such as mycorrhizal colonization or P-solubilizing bacteria, which are known to contribute to P acquisition under field conditions. Third, sand provides a homogeneous rooting environment with uniform P distribution, whereas field soils exhibit pronounced spatial heterogeneity in both P concentration and root accessibility. Finally, the controlled greenhouse environment does not capture fluctuations in temperature, moisture, and other abiotic stresses that can modulate plant responses to P limitation. Therefore, although sand-culture screening is effective for initial identification of physiological traits related to internal P utilization, the agronomic performance and ranking stability of the selected genotypes need to be validated under diverse field conditions before breeding programs can rely on these selections. Future multi-location and multi-year field evaluation is essential to confirm the stability of these genotypes’ phosphorus efficiency.

Beyond its immediate utility in phenotypic screening, the phenotyping pipeline—from PTPD standardization to CE calculation—may provide a foundation for integrating physiological screening with genomic tools. It integrates multiple tolerance-related traits into a single index. This makes it well-suited as a phenotypic target for genomic studies. In a typical genomic selection (GS) pipeline, the CE value can serve as the response variable, with genome-wide SNP markers (coded as 0, 1, 2) as predictors. Rather than building separate models for each of the ten traits—which may yield conflicting predictions—a single GS model based on CE can directly predict the overall low-P tolerance of breeding lines using marker data alone. Similarly, CE-based GWAS can detect genomic regions associated with comprehensive phosphorus efficiency, which may harbor regulatory genes that coordinate multiple downstream physiological responses rather than controlling a single trait. Although genomic data were not generated in the present study, the phenotyping framework established here—from PTPD standardization to CE calculation—provides a ready-to-use quantitative phenotype for future molecular dissection of phosphorus efficiency in soybean.

In conclusion, our multi-stage screening approach based on percent tolerance to phosphorus deficiency (PTPD) successfully identified photosynthetic rate, SPAD chlorophyll index at early and mid-reproductive stages, shoot dry weight at maturity, and seed number per plant as informative indicators for assessing soybean low-P tolerance under controlled conditions. The genotype CN 15 consistently exhibited superior phosphorus utilization efficiency and PSII functionality in sand culture, suggesting its potential as a candidate for further evaluation. However, as all experiments were conducted in a greenhouse using sand culture, the agronomic relevance of these traits and the selected genotypes remain to be validated under multi-location and multi-year field trials.