Plant material and growing conditions.

In the first study (study 1) two greenhouse experiments were conducted at the Louisiana State University Agricultural Center Sweet Potato Research Station (SPRS) in Chase, LA, USA (lat. 32°6′N, long. 91°42′W) (Table 1). The orange-flesh cultivar Beauregard was used in all experiments. This cultivar was released in 1987 and has been used to characterize the molecular and physiological mechanisms of SR initiation (Rolston et al. 1987; Villordon and LaBonte 2024). The current sweetpotato reference genome was also developed using this cultivar (Fei and Buell 2024). This work used a split-root system (Fig. 1A), which is a relevant tool for addressing the long-distance signaling and remobilization effects of nutrient heterogeneity in the rhizosphere (Flores et al. 2002; Schortemeyer et al. 1993). This growth system enables the separation of local and systemic responses, allowing for a better understanding of their respective molecular mechanisms (Torres et al. 2025). In both experiments, terminal cuttings measuring 25 to 30 cm in length, with six fully expanded leaves and evenly spaced nodes, were obtained from virus-tested plants maintained in the SPRS foundation seed greenhouse (Smith et al. 2009). The experimental setup used 10-cm-diameter polyvinylchloride (PVC) pots (height, 30 cm) with plastic detachable bottoms (Fig. 1A). Each PVC pot had four rows of side drain holes (diameter, 2 mm; spacing, 3 cm apart within rows). These drain holes were added to help reduce the incidence of a perched water table (Bilderback and Fonteno 1987). Washed river sand (diameter, 0.05–0.9 mm) was used as the substrate for all experiments. At planting, vegetative terminal cuttings were positioned to ensure that AR primordia were present in each compartment (Fig. 1B and C). Greenhouse growth conditions were maintained at 29 °C/23 °C (day/night), with a relative humidity of 60% to 80%, and supplemental light-emitting diode lighting to provide a 14 h photoperiod.

Table 1.Description of greenhouse experiments conducted on sweetpotato cv. Beauregard.

View Fullscreen

View Figure

• Download as Powerpoint
• Download Figure

Description of experimental treatments.

In this work, we simulated the progressive decline in Pi availability in the substrate using the following approach. First, we used half-strength Hoagland’s No.1 (Pi = 15 mg·L^–1) (+P = Pi sufficient control) solution (Hoagland and Arnon 1950). This nutrient solution concentration has been validated to optimize AR emergence and SR initiation at 5 and 15 d after planting (DAP), respectively, in ‘Beauregard’ (Villordon and Clark 2018; Villordon et al. 2012, 2013). Second, we reduced available Pi progressively (–P = declining Pi) by 25% (11 mg·L^–1), 50% (8 mg·L^–1), 75% (4 mg·L^–1), and 100% (0 mg·L^–1), at 7, 9, 11, and 13 DAP, respectively, corresponding to key AR developmental stages (Villordon et al. 2009a, 2009b). During the establishment phase (1–3 DAP), 50 mL of nutrient solution was applied twice daily to each compartment. This was followed by a 100 mL application to each compartment every other day for the remainder of the experiment. All experiments were arranged in a randomized complete block design, with four replicates per treatment (one pot = one replicate).

The first experiment (Expt. 1; Table 1) was conducted to standardize experimental procedures and to confirm prior findings that anatomic cues associated with the onset of SR initiation is localized in the APT region (2–5 cm from the base), with well-developed lateral branches (Firon et al. 2013; Villordon et al. 2012). Three tissue types were surveyed to assess SuSy gene expression: 1) immature leaf tissue, 2) a 2- to 3-cm section of the adventitious root tip (ART), and 3) the APT (Fig. 2). In this experiment, we focused on homogeneous +P/+P (sufficient Pi, 15 mg·L^–1 in both compartments) and declining –P/–P (declining Pi in both compartments as defined previously).

View Figure

• Download as Powerpoint
• Download Figure

The second experiment focused on tissue-specific SuSy gene expression in the APT region (Expt. 2; Table 1). Four Pi treatments were tested: homogeneous sufficient Pi (+P/+P; 15 mg·L^–1 in both compartments), homogeneous Pi omission (0P/0P; 0 mg·L^–1), homogeneous declining Pi (–P/–P), and heterogeneous Pi (+P/–P). The stepwise reduction in the –P treatment followed the same schedule as described in the first experiment. Based on the findings of the first experiment, which showed no differences between time points 8 and 12 DAP, the sampling time points were reduced to 5, 10, and 15 DAP.

Total RNA extraction and complementary DNA synthesis.

Approximately 100 mg of leaf, ART, and APT tissue were collected for RNA extraction. Tissue samples were ground in liquid nitrogen, and total RNA was extracted using the NucleoSpin^® RNA Plant (Macherey-Nagel, Bethlehem, PA, USA) according to the manufacturer’s instructions, with slight modifications described by Arce et al. (2024). RNA quality and quantity were assessed using a NanoDrop One Spectrophotometer (Thermo Fisher Scientific Inc., Waltham, MA, USA). First-strand complementary DNA (cDNA) synthesis was performed with the iScript cDNA Synthesis Kit (Bio-Rad Laboratories, Hercules, CA, USA) using 1 µg total RNA per 20-µL reaction. All cDNA samples were diluted to 100 ng/µL with nuclease-free water before setting up the quantitative polymerase chain reactions (qPCRs).

Primer design and reverse transcription–qPCR analysis.

The SuSy gene sequence used in this study was identified in prior work (Arce et al. 2024). The primer pair that was used was forward: 5′-CTGAAATTGAAAAGATGCACGCT-3′ and reverse: 5′-TTCGTGCACGGTTTGTTTGG-3′.

These primers were designed using Primer 3 Plus (Untergasser et al 2012) and were validated for specificity using the BLAST program. Amplification specificity was tested by melt curve analysis to ensure that only a single melting peak was observed. PCRs and gel electrophoresis were performed to verify the annealing temperature, specificity, and expected product size. qPCR was performed using Bio-Rad Universal SYBR Green Super Mix (Bio-Rad Laboratories) according to the manufacturer’s protocol. Each 20-µL reaction contained 2 µL cDNA (100 ng/µL), 0.5 µL of each primer (10 μM), 10 µL master mix, and 7 µL nuclease-free water. Reactions were run on a CFX Duet Real-Time PCR System (Bio-Rad Laboratories) using the following conditions: initial denaturation at 95 °C for 1 min, followed by 40 cycles of denaturation for 15 s at 95 °C, extension for 30 s at 58 °C, and fluorescence plate read at the end of extension step. Melt curve analysis was performed from 65 to 95 °C in increments of 0.5 °C, with a 5-s hold at each step. Negative controls were included with 2 μL nuclease-free water instead of the cDNA template. Each sample was analyzed using four biological and three technical replicates, with sweetpotato cyclooxygenase (Park et al. 2012) as the internal reference gene for normalization.

Plant height and RSA measurements.

A separate study (study 2) was conducted to generate plants for shoot and root system architecture (RSA) measurements (Table 1). Plant height and RSA measurements were performed at 5, 10, and 15 DAP. For RSA sampling, the detachable plastic bottoms of the pots were removed, the pots were immersed in water, and they were rinsed while holding the stem to prevent AR crossover between compartments. ARs were detached carefully, rinsed with deionized water to remove any remaining debris, and then placed in plastic trays filled with water before scanning. During scanning, intact ARs were transferred to a Plexiglas scanner tray (20 × 30 cm) containing water (depth, 3–4 mm). AR samples were staged carefully to minimize overlapping of LRs roots (Fig. 2). Measurements of AR architectural attributes followed the procedures described in prior work (Villordon et al. 2020). Imaging was performed using an Epson Perfection V850 Pro Photo Scanner (Epson Corp., San Jose, CA, USA). Image acquisition parameter was set to high accuracy (600 dpi; image size, ∼18 MB); analysis precision was also set to high. Measured parameters included AR length (in centimeters), AR number, lateral root count (LRC), and lateral root length (LRL). Image J software (Schindelin et al. 2012) was used to measure the length of the main AR (Fig. 2). RhizoVision explorer software (Seethepalli and York 2020) was used to analyze LRC and LRL.

Statistical analysis.

Sucrose synthase relative expression was calculated using the 2^–ΔΔCt method (Livak and Schmittgen 2001). Standard deviations were calculated from three replicates. Statistical significance was determined using Student’s t test (one tailed) at P ≤ 0.05. Root and shoot biomass were analyzed using analysis of variance (ANOVA) using the aov() function in R v. 2024.04.0-735 (R Foundation for Statistical Computing, Vienna, Austria). The ANOVA was followed by Tukey’s honestly significant test to identify significant differences using the TukeyHSD() function. Multiple comparisons were done using the HSD.test() functions in R package agricolae v. 1.3-5 (R Foundation for Statistical Computing, Vienna, Austria) (de Mendiburu 2021). There were no significant interaction effects among planting dates for either shoot or root traits; therefore, the data were combined. To assess the RSA response and variability patterns graphically within the RSA data, the R package ggplot2 (Wickham 2016) was used to generate boxplots using R.

Shoot and root biomass.

Representative plants for each treatment are shown in Fig. 3.

View Figure

• Download as Powerpoint
• Download Figure

Initially at 5 DAP, Pi-deprived plants (0P/0P) exhibited comparable shoot and root measurements relative to the +P/+P control (Figs. 4 and 5). However, by 10 DAP, 0P/0P reduced shoot growth and suppressed root development significantly, with latter reductions of 26%, 50%, and 37% in in AR number, LRC, and LRL, respectively, relative to +P/+P plants. In contrast, shoot growth with progressive Pi reduction (–P/–P) was comparable to that with +P/+P until 15 DAP. There were no differences in RSA traits across the time points. Heterogeneous Pi treatment (+P/–P) was associated with no changes in shoot or RSA traits relative to +P/+P at any time point. There were no differences in RSA traits between compartments for all treatments.

View Figure

• Download as Powerpoint
• Download Figure

View Figure

• Download as Powerpoint
• Download Figure

These responses are consistent with prior evidence in other crop species, in which short-term Pi deprivation does not affect growth immediately as a result of Pi remobilization from stored pools, such as seeds, tubers, or leaves (Barker and Pilbeam 2015; López-Arredondo et al. 2014). However, these internal Pi pools are finite, and external Pi availability is critical for sustained growth (López-Arredondo et al. 2014; Wissuwa 2003). The data also support the role of systemic Pi signaling, in which roots in Pi-rich zones compensate for Pi-deprived regions, enabling sustained growth under spatially variable conditions (Thibaud et al. 2010). Similar compensatory responses have been reported in maize and wheat (Triticum aestivum) under heterogeneous Pi supply (Zhu et al. 2005). As a mobile nutrient, Pi can be translocated from Pi-sufficient tissue to demand sites, mitigating localized deficiencies effectively (Barker and Pilbeam 2015; Khan et al. 2023). The shoot and root growth reductions under prolonged Pi deprivation are consistent with previous reports linking Pi deficiency to source–sink disruptions that impair photoassimilate transport and use, consequently limiting sink organ growth (Hermans et al. 2006; Wissuwa 2003). This disruption often leads to feedback inhibition of photosynthesis and a subsequent reduction in overall biomass accumulation (Wissuwa 2003).

In addition to its effects on photosynthesis, variation in Pi availability modulates RSA traits (Lynch 2011). Under Pi-limiting conditions, plants actively remodel their root systems to enhance nutrient acquisition, a response that includes altering LR development and increasing root hair density (López-Bucio et al. 2003; Lynch 2011; Péret et al. 2011). Consistent with our results, Villordon et al. (2020) demonstrated that Pi availability influences sweetpotato RSA significantly during the critical stage of SR initiation. Specifically, low Pi suppresses AR and LR development, which are key traits closely associated with SR initiation (Villordon et al. 2020).

SuSy activity increased in APT in response to declining Pi availability.

The findings in Expt. 1 of the study 1 align with prior work that anatomic cues associated with the onset of SR initiation are localized in the APT region (2–5 cm from the base) with well-developed, first-order lateral roots (Firon et al. 2013; Villordon et al. 2012). In the leaf tissue, SuSy expression in –P/–P plants did not vary relative to +P/+P plants until 15 DAP (Fig. 6A). In contrast, SuSy expression varied temporally in ART and APT regions (Fig. 6B and C). Initially, there were no differences in SuSy APT and ART expression in –P/–P relative to +P/+P at establishment (5–8 DAP) (Fig. 6B and C). At 10 DAP, when Pi was reduced by 50%, SuSy expression increased in both APT and ART (Fig. 6B and C). This increase coincided with active AR growth and LR development (Fig. 5E–H), suggesting enhanced sink activity, increased carbon requirement, and Pi foraging under nutrient-limited conditions (Li et al. 2024; Stein and Granot 2019). From 12 to 15 DAP, corresponding to the critical stage of SR initiation (Villordon et al. 2020), SuSy expression became predominantly localized in APT, with no further induction in the ART. This shift reflects a developmental transition from exploratory root growth toward carbon allocation for storage, consistent with the established role of SuSy in sink establishment (Gessler 2021; Li and Zhang 2003).

View Figure

• Download as Powerpoint
• Download Figure

The APT-specific upregulation of SuSy aligns with anatomic cues associated with the transition from ARs to storage roots (Firon et al. 2013; Villordon et al. 2012). This transition involves the formation of anomalous cambia around the protoxylem and secondary xylem, facilitating secondary thickening and starch accumulation (Togari 1950; Wilson and Lowe 1973). More important, such cambial activity is localized within the proximal 2 to 3 cm of ARs, with optimal LR development (Firon et al. 2013; Villordon et al. 2012). Our results corroborate this spatial specificity, with APT exhibiting increased SuSy activity relative to ART, which supports the hypothesized role of SuSy as a marker for enhanced sink strength during SR initiation (Li and Zhang 2003). Previous studies support these observations. For example, Firon et al. (2013) defined and used this AR region for transcriptomic analysis and measured gene expression of genes involved in sucrose and starch metabolism during SR initiation. Furthermore, Singh et al. (2021) showed the proliferation of starch-accumulating parenchymal cells associated with bulking in the APT region. Similar patterns have been reported in cassava (Manihot esculenta), in which elevated SuSy expression in basal root tissues correlates with early starch deposition (Huang et al. 2021).

Temporal SuSy activity in the APT.

SuSy expression in Expt. 2 of study 1 was constant across compartments with homogeneous Pi levels (+P/+P, 0P/0P, and –P/–P) at all time points (Supplemental Fig. 1). Based on these results, RNA samples from both compartments were pooled for analysis. Declining Pi availability (–P/–P) increased SuSy activity relative to the control (+P/+P) at 10 and 15 DAP across 2 years (Fig. 7B, C, E, and F). In contrast, homogeneous 0P/0P was not associated with increased SuSy activity across time points and both years (Fig. 7A–F). In the heterogeneous Pi treatment (+P/–P), SuSy expression initially did not vary across compartments at 10 DAP (Fig. 7B and E). However, at 15 DAP, SuSy activity increased in both compartments.

View Figure

• Download as Powerpoint
• Download Figure

Prior work has linked sucrose signaling in sweetpotato SR initiation, but scant information is available about what modulates sink strength (Ahn et al. 2010; Li and Zhang 2003; Zhang et al. 2017). Our findings demonstrate that declining Pi availability increases sink strength in ARs undergoing SR initiation, as evidenced by increased SuSy activity. The increased SuSy activity in ARs undergoing SR initiation is consistent with findings reported by Li and Zhang (2003). These findings are consistent with increased SuSy activity in turnip SRs during the early, rapid sink filling phase, accompanied by a concomitant reduction in leaf SuSy activity (Gupta et al. 2001). A similar pattern of increased SuSy activity at the early stage of potato tuber initiation has also been documented previously (Kaur and Das 2022). Our results are consistent with available evidence that supports the hypothesis that SuSy is a molecular marker for sink strength in starch-accumulating sink organs [reviewed in Fernie et al. (2020)].

Our results corroborate findings in the model plant Arabidopsis, in which sugars are redistributed preferentially to the root system in response to reduced Pi availability (Hammond and White 2008; Hermans et al. 2006). In addition, reduced Pi triggers the production of intracellular phosphatases and nucleases that facilitate Pi remobilization from internal pools, further supporting sustained metabolism in developing sinks (Hammond and White 2011; Müller et al. 2007; Soumya et al. 2022). These processes are believed to be coordinated with sugar signaling pathways. In this case, the increased root sucrose levels also regulate genes encoding RNases, phosphatases, and the phosphate starvation response (PSR) transcriptional cascade to remobilize internal Pi or alter root biochemistry and morphology for efficient Pi uptake (Fang et al. 2009; Karthikeyan et al. 2007; Ruan 2014; Yoon et al. 2021). In sweetpotato, Arce et al. (2024) reported that low Pi activates PSR transporters IbPHT1;4 and IbPHT1;5 in ‘Bayou Belle’, but not in ‘Evangeline’ or ‘Orleans’. This is consistent with the hypothesis that variation in SuSy activity may also reflect cultivar-specific differences in SR initiation and timing.

Our work used a split-root system. In the past, this system has been used to investigate the possible presence of local and systemic responses to the variation of Pi availability in root systems (Flores et al. 2002; Torres et al. 2025; Schortemeyer et al. 1993). Our morphological and molecular data using ‘Beauregard’ do not appear to show responses to local Pi availability. However, novel molecular data support the hypothesis of cultivar-specific Pi sensitivity (Arce et al. 2024). Previous data also show RSA variation among cultivars subjected to reduced Pi availability (Villordon et al. 2018, 2020). We hypothesize that cultivars adapted to low Pi availability may respond differently to heterogeneous Pi levels.

The results of our study have fundamental and applied implications. First, the findings help to bridge significant knowledge gaps in our current understanding of environmental and management cues that trigger the induction of SR initiation in sweetpotato ARs. Follow-up studies can focus on cultivar-specific sensitivity to Pi depletion. Accounting for this genotypic variability is essential, as specific cultivars may require a higher initial Pi rate, whereas others may respond optimally to sustained, moderate Pi levels. Second, the potential role of other nutrients cannot be ruled out. Third, the findings can be leveraged to optimize the management of Pi fertilizers in sweetpotato production. For example, our results underscore the importance of applying the optimum Pi fertilizer rate. Applying excessive amounts may potentially delay the induction of storage root formation in putatively low-Pi-requiring cultivars. Taken together, these findings underscore the potential for considering soil Pi levels and cultivar for developing site-specific management approaches for improved sweetpotato production efficiency.