Precision fertilizer management for germplasm maintenance of queen’s flower (Lagerstroemia speciosa)

Nitrogen management in container-grown ornamentals must balance plant quality against the risk of nutrient loss to surrounding water systems, a challenge that is especially pronounced in subtropical nursery environments. This study evaluated precision nitrogen strategies for Lagerstroemia speciosa ‘Big Pink,’ an increasingly popular landscaping tree, by integrating leachate monitoring with plant-based optical sensing. We evaluated the effects of six Azalea fertilizer (8-4–8 N-P-K) treatments: T0 (25 g), T1 (15 g + 10 g supplemental fertilizer, SF), T2 (20 g + 10 g SF), T3 (25 g + 10 g SF), T4 (30 g + 10 g SF), and T5 (35 g + 10 g SF), each with five replications, on container-grown Lagerstroemia speciosa ‘Big Pink’ under subtropical nursery conditions in southern Florida. Plant growth (leaf count), chlorophyll-based indices (SPAD, atLEAF), canopy reflectance (NDVI), and leachate chemical properties (Ca 2+, NO 3 −, K +, Na +, pH, salinity, and EC) were assessed monthly and analyzed across early, mid-, and late growth stages using ANOVA, Tukey’s HSD test, and Pearson correlation analysis. The two highest fertilizer treatments, T4 (30 g base rate plus a 10 g supplemental application) and T5 (35 g base rate plus a 10 g supplemental application), consistently produced greater NO 3 − and K + concentrations in leachate across all growth stages, indicating elevated nutrient mobility and leaching risk. Treatment (T5) further lowered substrate pH and raised salinity and EC at the late stage, reflecting salt accumulation in the root zone. Leaf production was maximized under intermediate fertilizer rates (T2-T4), while both insufficient (T0-T1) and excessive (T5) fertilization reduced growth. Chlorophyll indices (SPAD, atLEAF) reliably detected nitrogen deficiency under T0 from the mid-stage onward, whereas NDVI became a responsive indicator only at the late stage as canopy development increased. Correlation analysis showed progressively stronger associations among leachate variables over time, alongside consistent agreement between SPAD and atLEAF across all stages. Moderate fertilizer rates (T2-T3) optimized leaf growth and plant nitrogen status in container-grown L. speciosa ‘Big Pink’ while minimizing nutrient losses, supporting their adoption as a precision nitrogen management strategy for subtropical ornamental nursery production.

Ornamental tree production in the United States is increasingly influenced by two major challenges: growing water and climate related stresses and rising expectation for responsible nutrient management to protect water quality. In warm, humid regions such as southern Florida, high rainfall intensity, frequent irrigation, and coarse-textured soils increase these pressures and can accelerate nutrient movement beyond the root zone (Wilson et al., 2010; Yeager et al., 2010). Florida’s widely distributed sandy soils have relatively low water and nutrient-holding capacity (Harris et al., 2010), which can further increase leaching losses when fertilizer rates or irrigation scheduling exceed crop demand. In addition, ornamental production systems in South Florida face significant environmental threats such as high temperature and episodic saline intrusion (Costa et al., 2024). Among ornamental tree species used in subtropical landscapes, Lagerstroemia speciosa (L.) Pers. has emerged as an increasingly important plant for evaluating nutrient management strategies under these production constraints.

Lagerstroemia speciosa (L.) Pers. (queen’s crape myrtle) has received a significant amount of popularity from Europe, Asia, and the southern United States due to its wide variety assortment of inflorescence colors and a long summer flowering period (up to 120 days), along with an often-exceptional exfoliating bark (Cabrera, 2003; Meerow et al., 2015) The cultivar ‘Big Pink’ was released as an improved pink-flowering selection developed by USDA-ARS in Miami, Florida, specifically for enhanced landscape performance under subtropical conditions (Meerow et al., 2015). It is described to be a multi-stemmed, densely branched, deciduous large shrub to small tree growing 4.5-6.0 m (14.76-19.68 ft) tall in 5 years from a 0.9-m (2.95 ft) tall liner (Meerow et al., 2015). Due to its ability to adapt to a warm and humid environment, it gained popularity in urban streetscapes and container nursery systems across South Florida (Cabrera, 2003). In addition to its ornamental value, L. speciosa has high medicinal properties, with various plant parts historically used for managing metabolic and inflammatory conditions (Al-Snafi, 2019).

As with many ornamental woody crops, the growth and market quality of ‘Big Pink’ are strongly influenced by nitrogen (N) fertilization because N availability directly affects canopy development, leaf chlorophyll content, vegetative growth rate, and overall aesthetic value (Cabrera, 2003; Chen et al., 2005). Over-fertilization in containerized ornamentals has become critical agronomic and environmental concern. Excessive fertilization disrupts nutrient balance in the rhizosphere, increases osmotic pressure, and promotes salt accumulation, ultimately impairing root function and reducing substrate porosity (Khoddamzadeh et al., 2025, 2026). Moreover, in containerized nursery systems, excessive N application can increase nitrate losses in leachate and elevate soluble salt concentrations and electrical conductivity (EC). This creates a link between fertilizer practices and potential impact on surrounding water quality (Wilson et al., 2010; Yeager et al., 2010). As a solution, Best Management Practice (BMP) guidelines for Florida nursery production emphasize aligning fertilizer applications with the plant’s needs, improving irrigation efficiency, and adopting monitoring tools to track nutrient status. These strategies are designed to maintain plant quality while minimizing nutrient losses and reducing the risk of leaching and runoff (Yeager et al., 2010; Florida Department of Agriculture and Consumer Services, 2014).

A practical strategy for balancing crop productivity with environmental protection is precision N management, which synchronizes N inputs with crop demand using direct or indirect indicators of plant N status (Haque, 2026; Haque et al., 2026). Extensive research in precision agriculture demonstrates that proximal optical sensors such as normalized difference vegetation index (NDVI) derived from active canopy sensors (e.g., GreenSeeker) and leaf chlorophyll meters (e.g., Soil Plant Analysis Development; SPAD) provide rapid, non-destructive proxies of crop N status that can support more efficient and responsive fertilization strategies (Raun et al., 2002, 2005; Muñoz-Huerta et al., 2013; Nogueira Souza Costa and Khoddamzadeh, 2025; Khoddamzadeh et al., 2025). Chlorophyll meters have been repeatedly validated as reliable indicators of plant N sufficiency because leaf chlorophyll concentration is often strongly correlated with leaf N content when measurements are collected consistently and interpreted using crop-specific reference values (Follett et al., 1992; Muñoz-Huerta et al., 2013). Furthermore, the atLEAF chlorophyll meter has been shown to track chlorophyll and N status comparably to SPAD across multiple species, supporting its use as a lower-cost alternative (Zhu et al., 2012). More recent studies emphasize that integrating canopy sensing and field data can substantially improve N use efficiency and reduce N losses, particularly within variable-rate and adaptive nutrient management frameworks (Ali et al., 2024; Padilla et al., 2018; Sharma and Bali, 2017).

Beyond agronomic field crops, optical sensors have also been evaluated in woody ornamental systems, highlighting their potential for nursery fertilization programs (Demotes-Mainard et al., 2008). Similarly, for container-grown ornamentals, precision nutrient management benefits from integrating plant-based sensing with substrate and leachate monitoring. For collecting container leachate, pour-through method remains a widely used, non-destructive technique to measure pH, EC, and nutrient concentrations, allowing growers and researchers to track root zone nutrient status and potential leaching risk over time (Wright et al., 1990; Altland, 2021; Altland and Owen, 2024). This approach is particularly relevant in Florida, where nursery BMP guidance provides recommended leachate thresholds (including nitrate-N, potassium, calcium, pH, and EC) to support production decisions and minimize nutrient runoff (Florida Department of Agriculture and Consumer Services, 2014).

However, there remains a critical research gap in applying plant-based sensing and leachate monitoring approaches to optimize N management in container-grown L. speciosa subtropical condition. Hence, this study was conducted in southern Florida to evaluate precision N management for L. speciosa ‘Big Pink’ under subtropical ornamental nursery production. We integrated (i) leachate chemistry monitoring, including calcium (Ca² +), nitrate (NO 3 −), potassium (K +), sodium (Na +), pH, salinity, and EC, with (ii) plant growth and optical sensing measurements, including leaf count, SPAD chlorophyll index, atLEAF chlorophyll index, and GreenSeeker-derived NDVI. The objective of this study was to (i) quantify the effects of six fertilizer treatment rates on plant growth, chlorophyll-based sensor indices, and leachate chemical properties across growth stages, and (ii) examine relationships among leachate chemistry variables and plant-based sensor measurements. By linking fertilizer inputs with plant growth responses and root-zone nutrient dynamics, this study provides data-driven insights for optimizing N management in container-grown L. speciosa.

Experimental site and growing conditions

This study was conducted at the United States Department of Agriculture- Agricultural Research Service, Subtropical Horticulture Research Station (USDA-ARS-SHRS). In total, 30, 2-year-old L. speciosa individuals were used and cultivated from October 2023 to May 2024. Each of them was extracted from the SHRS germplasm collection in which they were part of the regenerative process to back-up USDA-National Plant Germplasm Resources. The 30 L. speciosa individuals were relocated into a single, uniformly shaded shade house to minimize block-to-block variation in light exposure, where they were studied in 18.9 L nursery pots (Viagrow, USA) with Azalea fertilizer (8-4-8; Sunniland, USA). The media consisted of 37.5% pinebark, 30.0% coir pith, 25% perlite, and 7.5% sand.

In addition to the fertilizer treatments described below, all plants received uniform cultural management throughout the study period. Irrigation was supplied through an overhead sprinkler system equipped with four nozzles per bench. Sprinklers operated twice daily for approximately 5 min per irrigation event (10 min day −1 total), independent of fertilizer treatment, to ensure uniform water availability and leaching conditions across all treatments. No pest or disease incidence was observed during the experiment.

Treatments and experiment design

There were six treatment groups in this study. Treatments included in the study were T0: 25g, T1: 15g+10g supplement fertilizer (SF) twice in January and March, T2: 20g+10g SF, T3: 25g+10g SF, T4: 30g+10g SF, and T5:35g+10g SF. Pots were arranged within the shade house in a completely randomized design, with each of the 30 individuals independently assigned a treatment using a random number generator prior to fertilizer application, resulting in 5 replications per treatment. Plants were irrigated uniformly across all treatments to prevent possible water-related stressors and to standardize leaching potential independent of fertilizer treatment.

Data collection

Data was collected every month for 8 months. Chlorophyll content was determined using soil plant analytical development value (SPAD) meter (SPAD-502PLUS, Konica Minolta Inc, Tokyo, Japan) and an atLEAF CHL Plus (FT Green LLC, Wilmington, DE, USA). The measurement was done in three different healthy and fully developed leaves of each individual plant. Normalized difference vegetation index (NDVI) was evaluated using GreenSeeker™ handheld canopy sensor (Trimble Agriculture, Sunnyvale, CA, USA) during which the sensor was held 45 cm above the plants’ canopy in order to retrieve accurate readings.

Leachate and salinity detection

Leachate collection was initiated on 14 days after treatment (November 1 st, 2023), where 18 out of the 30 individuals (three randomly selected individuals per fertilizer treatment) were used to evaluate each of the treatments’ salt accumulation and nutrient mobility. The leachate collection was done by collecting 100 mL leachate samples using the pour-through method. During the process, the containers were irrigated with 1 liter of deionized water to collect the solution from the substrate. LAQUAtwin-Na-11 Na + Ion Sensor, LAQUAtwin-K-11 Potassium K + Sensor, LAQUAtwin-NO3–11 NO 3 − Ion Sensor and LAQUAtwin-Ca-11 Ca 2+ Ion Sensor (LAQUATWIN-S050, Horiba Scientific Ltd., Japan), were used to quantify the concentration of these ions in the leachate. Electrical conductivity (µS cm −1) and salinity (ppm) of the leachate were measured in situ using a calibrated ExStik ® EC500 handheld multiparameter probe (EXTECH Instruments, FLIR Commercial Systems, Goleta, CA, USA).

Statistical analysis

For each growth stage, the monthly repeated measurements collected from an individual plant within that stage were averaged to yield a single stage-level value per plant, so that each plant (rather than each monthly observation) served as the experimental unit and independent replicate in the subsequent analysis. This approach avoided pseudoreplication that would otherwise arise from treating repeated monthly observations from the same plant as independent replicates. Data were analyzed using R (version 4.5.0, R Core Team, 2025) with one-way ANOVA, applied separately within each growth stage, to assess treatment effects on leaf count, sensor measurements (SPAD, atLEAF, NDVI), and leachate parameters (Ca 2+, NO 3 −, K +, Na +, pH, salinity, EC). Pairwise comparisons were performed using Tukey’s HSD test. Pearson’s correlation analysis was used to evaluate relationships between variables within each stage. All analyses were performed at a significance level of α = 0.05. Because Table 1 comprises 11 ANOVAs per growth stage and the correlation analyses in Figures 1, 2 involve additional significance tests, no correction for multiple comparisons was applied; consequently, borderline results close to α = 0.05 (e.g., late-stage Ca 2+ and late-stage pH) are interpreted with caution in the Discussion, and this is noted as a limitation of the analytical approach.

Leachate nutrient dynamics (Ca 2+, NO 3 −, K +)

At the early stage, NO 3 − and K + showed strong and significant treatment effects (NO 3 −: p = 0.0001; K +: p = 0.0001 (Table 1). Higher fertilizer treatments, particularly T4 and T5, resulted in greater NO 3 − concentrations compared with lower treatments (T0-T3) (Figure 3B). Potassium had a similar pattern, with T5 showing the highest K + concentration, while T0 remained significantly lower than the other treatments (Figure 3C).

During mid-stage, treatment effects for NO 3 − and K + remained same as early stage, although the differences among treatments were reduced for NO 3 −. Nitrate concentrations differed significantly (p = 0.0091), with T4 remaining higher than the lower treatments, while T5 was intermediate (Table 1 and Figure 3E). Potassium continued to show a clear and strong response to treatment (p = 0.0001), with T5 having significantly higher K + concentration than all other treatments (Table 1 and Figure 3F).

Treatment difference in Ca 2+ was significant at late stage (p = 0.0479), with T5 having the higher concentration than T0 (Table 1 and Figure 3G). By this stage, significant treatment effects were observed for NO 3 − and K +. Nitrate differed significantly (p = 0.0004), with T5 having the highest concentrations, followed by T4 (Table 1 and Figure 3H). Potassium showed the strongest treatment effect of all stages (p = 0.0001), with T5 having the highest concentration compared to all other treatments (Table 1 and Figure 3I).

Leachate chemical properties (pH, salinity, EC, Na +)

During the mid-stage (3–5 MAT), there was a significant treatment effect for pH and EC (Table 1). Substrate pH differed significantly among treatments (p = 0.0013), with T4 lower than T2 and T3 (Figure 4E). While EC showed strong treatment separation (p = 0.0216), with T1 significantly lower and T5 and T2 significantly higher (Table 1 and Figure 4G).

By the late stage (6–8 MAT), salinity and EC had significant treatment effects (salinity: p = 0.0164; EC: p = 0.0207), where higher fertilizer treatments particularly T5 had the highest values compared to T0 (Table 1 and Figures 4J, K). Differences in pH were significant (p = 0.0429), with T5 having the lowest value compared to T0 (Table 1 and Figure 4I).

Leaf count and sensor-based plant indices (SPAD, atLEAF, NDVI) response to fertilizer treatments

At the early stage (1–2 MAT), leaf count showed a strong and significant treatment effect (p = 0.0001) (Table 1). Treatments T2, T3, and T4 had significantly higher leaf numbers than T0, T1, and T5 (Figure 5A).

During the mid-stage (3–5 MAT), significant treatment effects were detected for leaf count, SPAD, and atLEAF. Leaf count strongly showed treatment effect (p = 0.0001), with T2-T4 maintaining higher leaf numbers than the lower fertilizer treatments (T0 and T1) (Table 1 and Figure 5B). Both SPAD (p = 0.0001) and atLEAF (p = 0.0001) differed significantly among treatments, with T0 consistently having lower values than other fertilizer treatments (Table 1 and Figures 6D, E).

By the late stage (6–8 MAT), treatment effects were significant across all measured variables. Leaf count had the strongest response (p = 0.0001), with T2-T4 producing the highest leaf numbers, while T0, T1 and T5 remained significantly lower (Table 1 and Figure 5C). Both SPAD (p = 0.0118) and atLEAF (p = 0.0007) continued to show significant treatment differences, with T0 having significantly lowest value compared to all other treatments (Table 1 and Figures 6G, H). Importantly, NDVI showed a significant treatment effect only at the late stage (p = 0.0015), with lower NDVI values under T0 and higher values under all other fertilizer treatments (Table 1 and Figure 6I).

Correlation among leachate chemical properties

Pearson correlation analysis revealed that relationships among leachate chemical properties varied across growth stages (Figure 1). At the early stage (1–2 MAT), the strongest positive association was observed between salinity and EC (r = 0.92, p < 0.001), indicating that increases in soluble salts were closely reflected in EC values (Figure 1A). Concentration of NO 3 − was positively correlated with K + (r = 0.54, p < 0.001). In contrast, pH showed significant negative correlations with NO 3 − (r = -0.52, p < 0.01) Ca 2+ (r = -0.47, p < 0.01), and salinity (r = -0.40, p < 0.01).

At the mid stage (3–5 MAT), correlations among leachate variables were generally weaker than at the early and late stages. Concentration of NO 3 − remained positively correlated with K + (r = 0.36, p < 0.01) (Figure 1B). Salinity and EC remained positively related, although the association was more modest than in the early stage (r = 0.34, p < 0.05).

By the late stage (6–8 MAT), correlations among leachate properties became stronger and more interconnected. Salinity and EC showed an almost perfect positive relationship (r = 0.98, p < 0.001) (Figure 1C). Likewise, NO 3 − was positively associated with K + (r = 0.67, p < 0.001), while pH was negatively correlated with both K + (r = -0.37, p < 0.01) and NO 3 − (r = -0.41, p < 0.01).

Correlation between leaf count and sensor-based plant indices

Pearson correlation analysis among leaf count and sensor-based plant indices also showed clear stage-dependent patterns (Figure 2). At the early stage (1–2 MAT), leaf count was only weakly related to the optical indices, with low positive correlations with SPAD, atLEAF, and NDVI (Figure 2A). In contrast, the sensor-based indices themselves were positively related with one another. There were significant positive correlations of SPAD with atLEAF (r = 0.36, p < 0.05) and NDVI (r = 0.35, p < 0.05), while atLEAF was also positively correlated with NDVI (r = 0.31, p < 0.05).

At the mid stage (3–5 MAT), the relationship between SPAD and atLEAF became much stronger (r = 0.72, p < 0.001), indicating close agreement between the two chlorophyll meters during active plant growth (Figure 2B). Leaf count showed a weak but significant positive correlation with NDVI (r = 0.29, p < 0.05). Greenseeker derived NDVI was only weakly related to SPAD and atLEAF at this stage.

By the late stage (6–8 MAT), the strongest correlation again occurred between SPAD and atLEAF (r = 0.73, p < 0.001), confirming strong consistency between these two leaf-level chlorophyll indices across later growth (Figure 2C). The NDVI became more closely related with atLEAF (r = 0.33, p < 0.01) and with leaf count (r = 0.39, p < 0.001). However, leaf count remained only weakly related to SPAD and atLEAF.

Fertilizer-driven nutrient dynamics in container leachate

The patterns of leachate nutrient concentrations in our study indicate that fertilizer rate strongly influenced the mobility of key nutrients in the container substrate over time. In our study, NO 3 − showed consistently higher leachate concentrations under the higher fertilizer treatments (T4 and T5) across all growth stages. This can be because of high solubility and mobility of NO 3 − in the substrate solution, which makes it susceptible to leaching when fertilizer application exceeds the crop demand (Pitton et al., 2022; Brown et al., 2025). Similar trend have been reported in a study by Wilson et al. (2010) in containerized nursery production systems, where NO 3 − concentrations in drainage water was substantial under nursery fertilization and irrigation regimes (Wilson et al., 2010).

There was a consistently high concentration of K + in the leachate under the highest fertilizer treatment (T5) across all stages. Potassium is also highly soluble and moves easily in the soil profile, hence elevated fertilizer inputs can result in greater potassium mobility and leaching losses in container substrates (Balík et al., 2019). This finding from our study highlights that higher fertilizer rates (T4-T5) in container-grown L. speciosa can increase nutrient losses through leachate with limited additional benefit to plant growth whereas lower fertilizer rates (T1-T3) can reduce the leaching losses while maintaining performance.

In our study during the late growth stage, substrate pH was significantly lower under the highest fertilizer treatment (T5) compared with the treatment with no SF (T0), while the remaining treatments showed intermediate values. This reduction in pH under higher fertilizer inputs is mechanistically consistent with microbial nitrification, in which ammonium (NH 4 +) supplied by the granular fertilizer is oxidized to NO 3 − via nitrite, releasing two protons (H +) per NH 4 + oxidized; because higher fertilizer treatments supply proportionally more NH 4 + -N, they generate a correspondingly larger acidifying load in the substrate solution over the growing season (Johnson et al., 2010; Kissel et al., 2020). Similarly, in our study, there was significantly higher salinity and EC under T5 compared with T0 during the late stage. This shows the increased concentrations of dissolved salts in the substrate solution as fertilizer rates increased. Given that EC reflects total soluble salts, these results confirm that higher fertilizer inputs promote salt buildup in the root zone (Caplan et al., 2017). Overall higher fertilizer rates, particularly T5, may promote substrate acidification and salt accumulation during the late growth stage of container-grown L. speciosa.

Plant growth responses and chlorophyll-based sensor measurements

Leaf count responded strongly to fertilizer treatments, with higher values observed under the intermediate fertilizer rates (T2-T4) and lower values under the lowest (T0, T1) and highest (T5) treatments. This pattern suggests that moderate fertilizer inputs were sufficient to support vegetative growth, whereas insufficient fertilizer reduce leaf development and higher nutrient inputs did not further improve growth. Similar responses have been reported in container-grown ornamental plants, where optimal growth is often achieved at intermediate fertilizer rates rather than at the highest nutrient inputs (Clark and Zheng, 2015).

We found during the mid-growth stage, SPAD and atLEAF values were significantly higher in all fertilizer treatments compared to T0, indicating improved plant N status with supplemental fertilizer. Chlorophyll meters such as SPAD and atLEAF are well-established nondestructive indicators of leaf N due to the close relationship between chlorophyll content and tissue N concentration (De Souza et al., 2019; Bijay-Singh and Ali, 2020; Kamarianakis and Panagiotakis, 2023). The consistently lower values observed in T0 therefore suggest N limitation in the absence of supplemental fertilizer.

A notable pattern in our results is the contrast between T0 and T3, which both received the same 25 g base fertilizer rate but differed in whether a 10 g supplemental application (SF) was applied in January and March. Despite sharing the same base rate, T0 consistently produced lower leaf counts and chlorophyll index values than T3, whereas T3 performed among the best treatments across growth stages. This difference is most plausibly attributable to the timing and split application of nitrogen rather than to the total seasonal rate: the mid-season supplemental application in T1-T5 likely resynchronized N supply with the plant’s active growth demand during the spring flush, whereas T0 received its full allotment only at the start of the study and may have experienced N depletion by the mid- and late-growth stages as the initial charge was progressively leached or immobilized in the substrate. This interpretation is consistent with the leachate data, in which T0 showed the lowest NO 3 − and K + concentrations across all stages (Figure 3), indicating that little residual fertilizer remained available in the root zone by the later stages. We emphasize that T0 should not be interpreted as an unfertilized control; rather, it represents the lowest-intensity fertilizer regime in this study (a single, undivided 25 g application), and its comparatively poor performance reflects a split-application and timing effect rather than the complete absence of nitrogen input.

We note that leaf count was the only direct growth parameter recorded in this study; measures such as plant height, canopy width, leaf area, or aboveground biomass were not collected. Leaf count is a practical and non-destructive metric for germplasm-collection material, for which destructive or highly disruptive sampling is undesirable, but it does not fully capture overall plant size, canopy architecture, or vigor. Because canopy development also affects NDVI, height and canopy-width measurements in particular would have helped validate the optical indices independent of leaf number. This is acknowledged as a limitation of the present growth assessment, and future studies on this germplasm collection should incorporate additional, non-destructive growth measurements (e.g., caliper-based height and canopy width) alongside leaf count.

Relationships among leachate chemistry variables

In this study, leachate chemical properties were more correlated during the late growth stage than the early and mid-stages, indicating progressive ion accumulation in the substrate as fertilization continued. The consistent positive correlation between NO 3 − and K + across all growth stages likely reflects their common fertilizer origin and high mobility, as similarly reported in container nursery system (Hoskins, 2014).

The negative relationships between substrate pH and nutrient concentrations during the early and mid-stages suggests progressive root-zone acidification with increase in nutrient availability. This is consistent with nitrification of fertilizer-derived nitrogen (particularly NH 4 +), which releases hydrogen ions and lowers pH (Goulding, 2016; Kissel et al., 2020; Hao et al., 2020; Zhang et al., 2024). The continued negative association between pH and nutrients such as NO 3 − and K + in later stages can be due to cumulative nutrient buildup in the substrate solution.

The strong positive relationship between salinity and EC across all growth stages in our study, particularly during the early and late stages confirm EC as an indicator of total dissolved salts in the substrate solution (Clean Water Team (CWT), 2004; Corwin and Yemoto, 2020).

Relationships among plant growth and optical sensor measurements

In our study there was consistently positive correlation between SPAD and atLEAF across all growth stages, particularly during the mid and late stages. Similar results were observed by Zhu et al., 2012 in their study where there was strong correlation between SPAD and atLEAF values (Zhu et al., 2012) supporting the use of atLEAF as a reliable and lower-cost alternative to SPAD for assessing plant N status.

A positive correlation between SPAD and NDVI was observed at the early growth stage, likely reflecting NDVI sensitivity to chlorophyll under low canopy density (Gamon et al., 1995; Raun et al., 2002). However, this relationship weakened with canopy development as NDVI increasingly reflected canopy structure and biomass rather than leaf-level chlorophyll (Padilla et al., 2018).

The positive correlation between NDVI and leaf count during the mid and late stages further highlights the importance of canopy structure in determining NDVI values, consistent with previous studies showing greater sensitivity of NDVI to plant biomass as canopy closure progress (Neupane et al., 2024).

The higher fertilizer treatments (30 g + 10 g and 35 g + 10 g supplemental fertilizer applications) consistently increased leachate NO 3 − and K + concentrations across all growth stages, indicating greater nutrient mobility and elevated leaching risk. Substrate pH, salinity, and EC remained relatively stable during early and mid-stages but changed significantly during the late stage as soluble salts accumulated in the substrate solution. Plant growth responses indicated that moderate fertilizer applications (20 g + 10 g, 25 g + 10 g, and 30 g + 10 g supplemental fertilizer) produced the highest leaf counts, whereas both insufficient and excessive fertilizer inputs reduced growth performance. Optical sensor measurements further confirmed improved N status in fertilized plants.

Although T4 (30 g + 10 g SF) produced leaf counts statistically comparable to T2 (20 g + 10 g SF) and T3 (25 g + 10 g SF), it is not recommended as an optimal treatment because it consistently generated higher leachate NO 3 − and K + concentrations than T2-T3 across growth stages, indicating a disproportionate increase in nutrient loss risk relative to the growth benefit gained. In contrast, T2-T3 achieved growth and plant N status equivalent to T4 while releasing appreciably less nitrate and potassium to the leachate, making them the more nutrient-efficient choice. Under the subtropical conditions of South Florida, these results suggest that moderate fertilizer rates (T2-T3) are sufficient to support optimal growth of L. speciosa ‘Big Pink’ while reducing excessive nutrient losses and are therefore recommended over T4 for precision nitrogen management of this germplasm collection.