Light Spectrum Strategies to Optimize Yield and Physiological Performance of Korean Thistle (Cirsium setidens (Dunn) Nakai) in a Closed-type Plant Production System

Experimental Materials

Seeds were obtained from a commercial supplier (Asia Seed, South Korea); the seed lot had been harvested in October 2024. The culture medium consisted of a 1:1:1 (v/v/v) mixture of non-fertilized substrate (Hanareumsangto, Shinsung Mineral, South Korea), 5-mm perlite (Ecolite Perlite, Homansaneob, South Korea), and 5-mm vermiculite (Verminuri, GFC, South Korea). The mixture was placed into pots measuring 6.5 × 6.5 × 6.5 cm (length × width × height), and three seeds were sown per pot. Sown pots were maintained under white LED lighting (4100 K). Photon flux density (PFD) at the substrate surface was adjusted to 100 μmol·m⁻²·s⁻¹ in the 350–800 nm range, as measured with a spectroradiometer (SpectraPen Mini, Photon Systems Instruments, Czech Republic). Prior to germination, environmental conditions were maintained at 20 ± 1°C and 56.3 ± 12.7% relative humidity. After germination, seedlings were thinned to one plant per pot once the first true leaf had fully expanded to ensure uniform growth.

LED Light Quality Treatments and Cultivation Environment

The experiment was conducted in a closed-type plant production system (6.4 × 5.2 × 2.4 m) located inside an experimental greenhouse at the Department of Environmental Horticulture, Sahmyook University. T5 LED lights (Zhong Shan Jinsung Electronic, China) were used as the artificial light source. Eight light-quality treatments were applied: monochromatic red (630 nm), green (520 nm), and blue (450 nm) LEDs; a purple LED consisting of red and blue wavelengths supplemented with approximately 17.6% far-red; and white LEDs with four correlated color temperatures (2100, 3000, 4100, and 6500 K) (Fig. 1). The 2100 K white LED treatment was produced by modifying a 4100 K white LED with an orange T5 LED cover (Zhong Shan Jinsung Electronic, China) to adjust its color temperature and spectral distribution. The PFD at the substrate surface was maintained at 100 μmol·m⁻²·s⁻¹ within the 350–800 nm range by adjusting the distance between the LEDs and plants weekly. The photoperiod was set to 14 hours of light and 10 hours of darkness. Each treatment plot was separated from adjacent plots using black polyethylene blackout curtains to prevent light contamination among treatments. Plants were grown for approximately six months under these conditions. The temperature and relative humidity within the closed-type plant production system were maintained at 20 ± 1°C and 58.1 ± 14.2%, respectively. Irrigation was supplied twice weekly via sub-irrigation. The nutrient solution was prepared by dissolving 750 mg·L⁻¹ of a premixed fertilizer (Masterblend 4-18-38, Masterblend International, USA) in water, supplemented with 375 mg·L⁻¹ magnesium sulfate (MgSO₄) and 750 mg·L⁻¹ calcium nitrate (Ca(NO₃)₂).

Fig. 1

Spectral distribution of different light-emitting diodes (LEDs) used in this study. Monochromatic LED spectra: (A) red; (B) green; and (C) blue; (D) a purple LED composed of red and blue wavelengths supplemented with far-red radiation; and white LEDs with correlated color temperatures of (E) 2100, (F) 3000, (G) 4100, and (H) 6500 K.

Plant Growth, Yield, and Quality Parameters

Growth and yield (biomass) characteristics were evaluated at the end of the experiment, six months after the treatments were applied. Measurements included shoot height, shoot width, stem diameter, root length, ground cover, number of leaves, leaf length, leaf width, leaf area, petiole length, and the fresh and dry weights of shoots and roots. Total fresh and dry weights were also determined. For quality-related parameters, chlorophyll content was measured using a portable chlorophyll meter (SPAD-502Plus; Konica Minolta, Japan). Leaf color was assessed with a spectrophotometer (CM-2600d; Konica Minolta, Japan) to obtain color coordinates in the Commission Internationale de l’Éclairage Lab (CIELAB) color space, expressed as L^*, a^*, and b^* values. CIELAB measurements followed the method of Lee (2023) using the D65/10° mode and the specular component included (SCI) setting. For each treatment, 20 measurements were taken for both chlorophyll content and leaf color. Indices used to quantify plant quality included relative moisture content (RMC; Lee and Nam, 2024; Equation 1), the shoot dry weight-to-root dry weight ratio (S/R; Equation 2), compactness (Jeong et al., 2020; Equation 3), and the Dickson quality index (DQI; Dickson et al., 1960; Equation 4).

(Abbreviations; SDW: shoot dry weight; RDW: root dry weight; FW: fresh weight; DW: dry weight; SH: shoot height; TDW: total dry weight; SD: stem diameter)

Remote Sensing Vegetation Indices

To analyze the spectral characteristics of C. setidens leaves, nine remote sensing vegetation indices were assessed using a portable spectroradiometer (PolyPen RP410; Photon Systems Instruments, Czech Republic). The indices included the normalized difference vegetation index (NDVI; Rouse et al., 1973), photochemical reflectance index (PRI; Gamon et al., 1992, 1997), modified chlorophyll absorption ratio index (MCARI; Daughtry et al., 2000), renormalized difference vegetation index (RDVI; Roujean and Breon, 1995), optimized soil-adjusted vegetation index (OSAVI; Rondeaux et al., 1996), transformed chlorophyll absorption in reflectance index (TCARI; Haboudane et al., 2004), structure-insensitive pigment index (SIPI; Peñuelas et al., 1995), carotenoid reflectance index 2 (CRI2; Gitelson et al., 2002), and anthocyanin reflectance index 2 (ARI2; Gitelson et al., 2001). Each index was calculated using the equations provided in Table 1. Measurements were taken randomly from leaf regions that did not intersect the main vein, with 20 measurements per treatment.

Table 1

Definitions and equations of remote sensing vegetation indices used in this study

OJIP Chlorophyll Fluorescence Analysis

An OJIP chlorophyll fluorescence analysis was performed using a portable fluorometer (FluorPen FP 110/D, Photon Systems Instruments, Czech Republic). Fully expanded leaves were dark-adapted for 15 minutes prior to measurement using a dark-adaptation leaf clip (PSI, 2025), and 20 measurements were taken per treatment. Following the methods of Lee et al. (2022b), Oh et al. (2014), PSI (2025), and Stirbet and Govindjee (2011), a set of chlorophyll fluorescence parameters was selected and applied in this study. The equations and detailed descriptions of these parameters are provided in Table 2. The analysis included basic fluorescence parameters (F_o, F_j, F_i, and F_p) as well as technical parameters (F_v, V_j, and V_i). In addition, the maximum quantum yield of photosystem II (PSII), expressed as F_v/F_m, was determined. Furthermore, the parameter M_o, which represents the initial slope of the OJIP induction curve (OJIP transient), was examined along with several quantum-yield-related indices (Φ_Po, Ψ_o, Φ_Eo, and Φ_Do). Specific energy fluxes per reaction center (RC), including ABS/RC, TR_o/RC, ET_o/RC, and DI_o/RC, were estimated. Finally, the performance index on an absorption basis (PI_ABS) was evaluated as an integrative parameter of overall photosynthetic efficiency of PSII.

Table 2

Definitions and equations of the chlorophyll fluorescence parameters used in this study

Experimental Statistical Analysis

The experiment was conducted using a completely randomized design (CRD), with each experimental unit consisting of a single plant grown in an independent pot. Fifteen plants were assigned to each treatment, resulting in a total of 120 plants used in this study. All statistical analyses were performed using SAS 9.4 (SAS Institute, USA). A one-way analysis of variance (ANOVA) was conducted to evaluate treatment effects, and mean comparisons were carried out using Duncan’s multiple range test (DMRT) at a significance level of p < .05. A principal component analysis (PCA) was performed on a correlation matrix after missing values resulting from parameter-specific differences had been imputed using the variable-wise mean. All parameters were then Z-standardized. Loadings (eigenvectors × √eigenvalues) were depicted as arrows in the PC1–PC2 biplot, while variable-principal component correlations were visualized using a correlation circle, facilitating interpretation of the relationships among measured variables. For better visualization, the lengths of the arrows were scaled accordingly. Furthermore, to examine the relationships among indices, Pearson correlation coefficients (r) were calculated to construct a correlation matrix. To quantify the similarity among indicators based on their correlation profiles (i.e., the rows or columns of the correlation matrix), Euclidean distances were derived from the correlation matrix. Hierarchical clustering was subsequently performed using the average linkage method.

Analysis of Growth Parameters

LED light quality has been reported to induce a wide range of morpho-physiological responses in plants, affecting photosynthetic efficiency, morphogenesis, and overall developmental processes (Hogewoning et al., 2010; Park et al., 2024). However, these responses can vary substantially depending on plant species or cultivars (Shin et al., 2024). Representative growth images of Korean thistle (Cirsium setidens (Dunn) Nakai) obtained in this study are shown in Fig. 2. Based on the growth parameter results, shoot height was greatest under the green LED treatment, reaching 14.73 cm. In contrast, the 4100 K white LED, blue LED, and 6500 K white LED treatments produced comparatively shorter shoots, with heights of 8.40, 7.43, and 6.69 cm, respectively (Table 3). Shoot width was also greater under the red and green LED treatments (36.56 and 36.36 cm, respectively) than under other treatments. In comparison, the blue LED and 6500 K white LED treatments resulted in narrower shoot widths of 24.08 and 22.38 cm, respectively. This trend suggests that a higher proportion of blue wavelengths in the spectral distribution may reduce overall shoot size. Although stem diameter did not differ significantly among treatments, root length tended to be longer under the 3000 K white LED treatment (15.13 cm), but this was not statistically different from the lengths observed under red, green, purple, or 4100 K white LEDs. Conversely, shorter root lengths were recorded under the blue LED and 2100 K white LED treatments, measuring 9.04 and 9.02 cm, respectively. Additionally, root thickness tended to increase under the white LED treatments compared with monochromatic LEDs (data not shown). Ground cover area was larger under the red and green LED treatments, with values of 1456 and 1392 cm², respectively.

Fig. 2

Representative images of Korean thistle (Cirsium setidens) grown under different light-emitting diode (LED) spectral qualities for six months. Scale bar = 5 cm. The 2100, 3000, 4100, and 6500 K treatments correspond to white LEDs (correlated color temperature).

Table 3

Growth parameters of Korean thistle (Cirsium setidens) after six months of cultivation under different LED spectral qualities

These results are consistent with previous findings that the spectral distribution regulates morphological development by controlling cell elongation and the division patterns of leaf and stem tissues (Folta and Childers, 2008). Specifically, blue light is known to suppress cell elongation, increase leaf thickness, and enhance tissue density, thereby restricting the expansion of shoots (Hogewoning et al., 2010). In contrast, the red and green wavelengths tend to promote stem elongation. Such wavelength-dependent responses can vary depending on species- or cultivar-specific genetic backgrounds and the sensitivities of photoreceptors, such as phytochromes and cryptochromes (Franklin and Quail, 2010; Park et al., 2024). Consequently, the red and green LED treatments were evaluated as more effective for promoting the expansion of C. setidens shoots, whereas the blue and 6500 K white LED treatments, which contain relatively higher proportions of blue light, were considered effective for suppressing shoot growth.

Regarding leaf-related growth parameters, the number of leaves did not differ significantly among treatments. In general, leaf appearance rate is determined primarily by temperature and photoperiod rather than by light quality (Clerget et al., 2008; Padilla and Otegui, 2005). In this study, the lack of significant differences in leaf number among treatments is presumed to reflect the species-specific characteristics of C. setidens. Leaf length was greater under the red and purple LED treatments, measuring 11.20 and 10.93 cm, respectively. Leaf width was also larger under the purple and red LEDs, at 7.20 and 6.43 cm, respectively; however, these values were not statistically different from those observed under the green and 2100–4100 K white LED treatments. Consistent with the trends found for shoot height and width, both leaf length and width showed slight reductions under the blue and 6500 K white LED treatments, indicating a partial expression of the dwarfing effect associated with blue light. Leaf area was greatest under the purple LED t reatment ( 80.7 cm²), although not significantly different from that under the red LED treatment ( 76.3 cm²). In contrast, petiole length was longest under the green (13.35 cm) and red (11.75 cm) LED treatments.

These results appear to reflect differences in photoreceptor activation patterns depending on spectral distribution, which selectively stimulate the elongation of leaf tissues and petioles (Folta and Childers, 2008). Specifically, red wavelengths are known to promote petiole elongation by stimulating the secretion of auxins and gibberellins through phytochrome-mediated pathways. In contrast, under light conditions with a higher proportion of blue wavelengths, cryptochrome- and phototropin-mediated signaling pathways suppress cell expansion, thereby restricting leaf length, width, and area (Franklin and Quail, 2010; Hernández and Kubota, 2014). These mechanisms are consistent with the results observed in this study. Moreover, leaf length and width—major parameters determining overall leaf size—were effectively increased under purple LED irradiation that included far-red wavelengths. This suggests that the inclusion of far-red light within the spectral composition may be advantageous for promoting leaf growth. This trend aligns with the findings of Park and Runkle (2017), who reported that far-red light induces leaf expansion and enhances leaf assimilation, ultimately stimulating plant growth. The observed increase in leaf area further underscores the importance of the red/far-red ( R/FR) ratio, a key determinant of the plant shade-avoidance response (Ciolfi et al., 2013). Therefore, these results highlight the necessity of developing customized spectral designs in LED cultivation systems tailored to specific target traits (Massa et al., 2008).

Biomass Analysis

Biomass, which represents plant yield, serves as a key indicator of the accumulation of photosynthates and the allocation of resources within plants (Poorter et al., 2012). According to the biomass analysis, the highest shoot fresh weight was observed under the red LED treatment, reaching 10.68 g. The purple LED treatment produced the second-highest value (8.07 g), although this was not statistically different from the values recorded under green, 3000 K white, and 4100 K white LEDs ( Fig. 3). The shoot dry weight, similar to the results for shoot fresh weight, was highest under the red LED treatment at 0.79 g; however, this value did not differ statistically from those obtained under the green, purple, and white LED treatments (2100–4100 K). In contrast, root fresh weight tended to be higher under the 3000 K white LED treatment (3.56 g), and root dry weight showed a comparable pattern, with the highest value also occurring under the same t reatment ( 0.36 g). Neither parameter, however, differed significantly from those measured under the 4100 K white LED. Total fresh weight was highest under the red LED treatment (12.79 g), whereas total dry weight was relatively higher under the 3000 and 4100 K white LEDs, measuring 1.07 and 1.06 g, respectively.

Fig. 3

Yield (biomass) of Korean thistle (C. setidens) after six months of cultivation under different LED spectral qualities. (A) fresh weights of shoot and root; (B) dry weights of shoot and root; (C) total fresh and dry weights. Data are shown as mean ± standard error (SE). Different lowercase letters indicate groups separated by Duncan’s multiple range test (DMRT) at p < .05; same lowercase letters denote no significant difference (n = 15).

As a result, the red-centered spectrum not only enhanced shoot growth in C. setidens but also significantly promoted the accumulation of photosynthates. This finding is consistent with Park et al. (2024), who reported similar outcomes in Coleus cultivars under varying light quality conditions. Previous studies have shown that white LEDs tend to promote root development (e.g., root length) and enhance root biomass accumulation (Lee et al., 2024b; Shin et al., 2024). This trend was also observed in this study. Meanwhile, the inclusion of far-red wavelengths within the light spectrum can modify biomass allocation patterns (Jang et al., 2023; Zhen and Bugbee, 2020). In this study, the purple LED treatment, which contained a far-red component, had a moderate effect on biomass accumulation. A relatively high proportion of far-red radiation appeared to have some effect on increasing leaf dimensions, such as leaf length and width, in C. setidens; however, its impact on total leaf area was minimal. Likewise, its influence on overall biomass accumulation was relatively limited. Given that plant species exhibit considerable variability in their responsiveness to far-red light (Zhou et al., 2025), further research is needed to clarify this species-specific sensitivity in C. setidens.

Analysis of Quality-related Parameters

In the analysis of quality-related parameters, chlorophyll content (SPAD units) tended to be higher under the 6500 and 3000 K white LED treatments, with values of 46.85 and 45.47 SPAD units, respectively (Fig. 4). In contrast, lower chlorophyll levels were observed under the green, blue, and red LED treatments, with values of 39.07, 38.78, and 38.56 SPAD units, respectively. This indicates a reduced chlorophyll content per unit leaf area under these conditions. These results are consistent with previous reports showing that spectral imbalance, particularly when dominated by monochromatic lighting, can reduce the relative density of pigments such as chlorophylls and carotenoids in leaf tissues and cells (Gao et al., 2022; Zheng and Labeke, 2017). Although earlier studies have reported a positive correlation between chlorophyll content and plant growth performance (Ghosh et al., 2020; Peng et al., 1996), the present results show an opposing pattern. However, as SPAD readings can be influenced by factors such as leaf thickness, further research is needed to clarify the underlying mechanisms.

Fig. 4

Quality-related parameters of Korean thistle (C. setidens) after six months of cultivation under different LED spectral qualities. (A) SPAD units; (B) L^* (lightness); (C) a^* (green-red axis); and (D) b^* (blue-yellow axis). Data are shown as mean ± SE. Different lowercase letters indicate groups separated by DMRT at p < .05; same lowercase letters denote no significant difference (n = 20).

The Commission Internationale de l’Éclairage Lab (CIELAB) color model is widely recognized as a robust tool for characterizing the external quality attributes of plants (Lee et al., 2024a) and has increasingly replaced the earlier Hunter Lab system in recent research applications. According to the CIELAB color space analysis, the L^* value, representing lightness, was 38.40, 37.21, and 36.79 in C. setidens grown under monochromatic green, blue, and red LED treatments, respectively. These results indicate that leaf lightness was greater under monochromatic LED lighting than under purple or white LED conditions. In contrast, the a^* value, which denotes the green-red chromatic axis, exhibited an opposite trend to L^*. Relatively showed more negative a^* values were observed under white LED treatments with correlated color temperatures of 2100, 6500, 3000, and 4100 K, recorded as −6.17, −6.23, −6.52, and −6.61, respectively. The b^* value, representing the blue-yellow chromatic axis, was highest (16.94) under green LED treatment, indicating that the leaves exhibited a comparatively more yellowish coloration, although this difference was not statistically significant relative to red or blue LED treatments. Conversely, plants grown under white LEDs at 2100, 6500, and 3000 K displayed lower b^* values of 12.56, 12.36, and 12.14, respectively.

These results suggest that variations in light quality can alter the distribution and density of leaf pigments, a finding that has also been emphasized in previous studies (Hogewoning et al., 2010; Olle and Viršile, 2013; Park et al., 2024). Regarding chlorophyll content, white LED lighting appears to promote chlorophyll biosynthesis and stability, likely due to the synergistic effects of its broad spectral composition, ultimately resulting in increased chlorophyll accumulation (Shin et al., 2024). Meanwhile, plants exposed to monochromatic LEDs—particularly green LEDs—tended to exhibit higher L^* and b^* values compared with those grown under white LEDs, indicating a more yellowish coloration of the foliage. This pattern was inversely associated with chlorophyll content, supporting the interpretation that reduced chlorophyll levels contribute to leaf yellowing under these conditions. Collectively, these findings highlight the need to strategically combine LED spectral distributions in horticultural lighting design to optimize plant quality attributes.

Analysis of Plant Quality Indices

Evaluating plant growth characteristics solely on the basis of morphological traits has inherent limitations. Therefore, integrated assessment approaches—such as plant quality indices—have been proposed to more comprehensively reflect multiple growth parameters and biomass components (Cha et al., 2017; Dickson et al., 1960). In this study, the shoot RMC of C. setidens was highest under the red LED treatment (92.5%), followed closely by the green LED treatment (92.3%), with no statistically significant difference b etween t he t wo ( Fig. 5). I n contrast, r oot RMC did not differ significantly among treatments. Total RMC, representing the overall plant water status, was similarly high under the red and green LED treatments (92.4 and 92.3%, respectively), but did not differ significantly from those observed under blue, purple, or 6500 K white LED treatments. The shoot-to-root ratio (S/R), defined as the ratio of shoot dry weight to root dry weight, was higher under the green and red LED treatments, reaching 7.1 and 6.5, respectively. These results indicate that shoot development relative to root development was markedly greater under these light conditions than under the other treatments. Compactness analysis revealed that plants grown under the four white LED treatments and the red LED treatment exhibited relatively high compactness values compared with those grown under blue and green LEDs, suggesting enhanced shoot density under these lighting conditions. Meanwhile, the Dickson Quality Index ( DQI), a composite indicator widely used to evaluate overall plant quality, was highest under the 3000 K white LED treatment (0.113), indicating comparatively superior plant vigor. However, no statistically significant differences were observed relative to the 4100 and 6500 K white LED treatments, both of which produced DQI values of 0.106.

Fig. 5

Plant quality indices of Korean thistle (C. setidens) after six months of cultivation under different LED spectral qualities. Relative moisture content of (A) shoots and (B) roots; (C) total relative moisture content; (D) S/R ratio; (E) compactness index; and (F) Dickson quality index. Data are shown as mean ± SE. Different lowercase letters indicate groups separated by DMRT at p < .05; same lowercase letters denote no significant difference (n = 15).

These results demonstrate that plant quality is determined not only by overall size but also, critically, by the allocation of biomass between shoots and roots. In particular, red and green LED treatments enhanced both S/R and RMC, indicating increased accumulation of photosynthates and improved water retention capacity in shoots relative to roots. Although the red LED treatment promoted quantitative growth parameters, it appeared to compromise balanced resource allocation and structural stability. In contrast, the 3000 K white LED spectrum effectively mitigated growth imbalances—such as disproportionate enlargement of specific organs—and contributed to improved overall plant vigor. The consistent patterns observed among DQI, S/R, compactness, and plant moisture status-related indices (shoot and total RMC) in this study indicate that the light quality resulting from LED spectral design modulates not only biomass accumulation but also structural stability and internal water retention. Furthermore, this approach utilizing multivariate quality indices can provide a practical framework for developing LED-based cultivation strategies tailored to the production goals of specific species or cultivars.

OJIP Chlorophyll Fluorescence Analysis

Chlorophyll fluorescence is widely used as a non-destructive indicator for evaluating the photochemical reactions and energy conversion efficiency of photosystem II (PSII) (Guidi and Calatayud, 2014; Jang et al., 2025; Woo et al., 2008). In particular, fluorescence induction kinetics (OJIP transients) provide a suite of parameters that enable quantitative interpretation of PSII energy trapping, electron transport, and energy dissipation by tracking the rise from the initial to the maximum fluorescence level when a dark-adapted leaf is exposed to saturating light (Stirbet and Govindjee, 2011). This technique is commonly employed for rapid assessment of plant stress responses through real-time monitoring of physiological status (Jang et al., 2023). In this study, analysis of the basic fluorescence parameters of C. setidens under different LED light qualities revealed that the initial fluorescence level (F_o) was highest under the green LED treatment (8096; Fig. 6 and Table 4). In contrast, the maximum fluorescence yield (F_m) reached its highest values under purple and red LEDs, at 42753 and 41506, respectively.

Fig. 6

OJIP induction of Korean thistle (C. setidens) after six months of cultivation under different LED spectral qualities. O: origin-step; J: jump-step; I: intermediate-step; P: peak of fluorescence (F_m). (A) monochromatic lights of red, green, blue LEDs and composite light of purple LED (red and blue with far-red); (B) 2100, 3000, 4100, and 6500 K white LEDs.

Table 4

Basic chlorophyll fluorescence parameters, technical fluorescence parameters, and maximum quantum yield (QY) of Korean thistle (C. setidens) after six months of cultivation under different LED spectral qualities

The analysis of technical fluorescence parameters showed that the variable fluorescence (F_v) was highest under purple LED light, reaching a value of 35814. The variable fluorescence at the J-step (V_j) was greatest under green and red LEDs, with values of 0.385 and 0.378, respectively. In contrast, the variable fluorescence at the I-step (V_i) tended to be higher under red, green, and 6500 K white LEDs, with values of 0.784, 0.781, and 0.779, respectively. The parameter M_o, which represents the initial slope of the OJIP fluorescence transient, was also relatively high under green and red LEDs (0.602 and 0.586, respectively) compared with the other treatments. An elevated M_o suggests a greater excitation pressure on the electron-accepting side of PSII, which may be interpreted as an early indication of abiotic stress and/or activation of photoprotective mechanisms. The maximum quantum yield of PSII (F_v/F_m) tended to be high—ranging from 0.832 to 0.837—under blue, purple, and 2100, 4100, and 6500 K white LEDs, whereas the green LED treatment exhibited the lowest value (0.794). The F_v/F_m ratio of physiologically healthy higher plants typically falls within 0.78–0.84 (Ahn et al., 2024; Lee et al., 2025b; Park et al., 2024; Shin et al., 2024; Stirbet and Govindjee, 2011). In this study, even the green LED treatment, which showed the statistically lowest F_v/F_m value, remained within this normal range, indicating that C. setidens did not experience severe photoinhibitory stress under any of the tested light-quality treatments. Nevertheless, the observed differences in overall growth among treatments appear to arise from light-quality-dependent variation in the distribution and density of photoreceptors within the leaves.

The maximum quantum efficiency (Φ_Po), analogous to F_v/F_m, was higher under blue, purple, and white LED light at correlated color temperatures of 2100, 4100, and 6500 K. The probability that trapped excitons transfer electrons beyond Q_A to the downstream electron transport chain (Ψ_o) reached its highest value under the 3000 K white LED treatment, with a recorded value of 0.661 (Fig. 7 and Table 5). In contrast, the quantum yield of electron transport (Φ_Eo) was highest under the 4100 K white LED condition, reaching 0.550. Meanwhile, the probability that an absorbed photon is dissipated as heat or fluorescence (Φ_Do) was 0.206 under green LED light.

Fig. 7

Quantum yields of photosystem II (PSII) and specific energy fluxes per reaction center (RC) of Korean thistle (C. setidens) after six months of cultivation under different LED spectral qualities. Monochromatic lights of (A) red; (B) green; and (C) blue LEDs; composite light of (D) purple LED (red, blue with far-red); white lights of (E) 2100; (F) 3000; (G) 4100; and (H) 6500 K LEDs (n = 20).

Table 5

F-statistics and p-values from one-way ANOVA for quantum yields of PSII (Φ_Po, Ψ_o, Φ_Eo, and Φ_Do) and specific energy fluxes per RC (ABS/RC, TR_o/RC, ET_o/RC, and DI_o/RC) of Korean thistle (C. setidens) after six months of cultivation under different LED spectral qualities (n = 20)

The absorption flux per reaction center (ABS/RC) exhibited elevated values of 1.971 and 1.876 under green and red LED lights, respectively. Correspondingly, the trapped energy flux per reaction center (TR_o/RC) was 1.561 under green LEDs and 1.544 under red LEDs. The electron transport flux from Q_A to Q_B per reaction center (ET_o/RC) was identical for the two treatments, with both recording a value of 0.958. In contrast, the dissipated energy flux per reaction center (DI_o/RC) reached 0.410 under green LED exposure. According to Shin et al. (2024), the ‘Asia Sugar’ cultivar of chicory displayed the highest ABS/RC values under red and green LEDs, which is consistent with the present findings. This suggests that red and green LEDs may promote the partial inactivation of PSII reaction centers. When a greater proportion of reaction centers in the denominator becomes inactivated, the average ABS/RC value tends to increase. Therefore, a relatively high ABS/RC value may indicate the partial inactivation of some reaction centers.

The performance index on an absorption basis (PI_ABS) reached its highest value of 6.39 under the 4100 K white LED treatment, with no significant differences relative to the 2100 and 3000 K white LED conditions. In contrast, the lowest PI_ABS value, 3.24, was recorded under green LED light treatment (Fig. 8). Taken together, the findings to this point indicate that the overall chlorophyll fluorescence displayed trends that contrasted with those of the plant growth parameters and biomass production. This divergence may be associated with the concept of plant adaptive stress, or eustress (Rouphael and Kyriacou, 2018), wherein moderate stress levels can stimulate growth while inducing mild stress responses (Villagómez-Aranda et al., 2022). Furthermore, stress symptoms such as relative reductions in F_v/F_m and PI_ABS, coupled with increases in M_o, may not necessarily reflect damage to the reaction centers. Rather, these responses may represent downregulation signals arising from excitation energy distribution between PSII and PSI, as well as from spectral acclimation (Minagawa, 2011).

Fig. 8

Performance index on absorption basis (PI_ABS) of Korean thistle (C. setidens) after six months of cultivation under different LED spectral qualities. Data are shown as mean ± SE. Different lowercase letters indicate groups separated by DMRT at p < .05; same lowercase letters denote no significant difference (n = 20).

Red and green wavelengths are characterized by relatively deep canopy penetration, enabling more uniform light distribution to leaves in the lower canopy layers. Thus, beyond the photochemical responses described earlier, total photosynthetic carbon gain was likely enhanced in parallel with increases in plant size. As a result, red and green light treatments promoted plant growth and biomass accumulation, even though increases in stress-related fluorescence indicators such as ABS/RC and DI_o/RC were also observed. Such trends may occur under spectral conditions in which the proportion of blue light is very low and red wavelengths predominate (Van Brenk et al., 2024). Spectra with a relatively dominant red wavelength are known to significantly enhance growth parameters such as leaf enlargement and shoot extension. However, improvements in photosynthetic capacity, stomatal function (opening and closing), and chlorophyll or nitrogen concentration per unit leaf area typically require sufficient excitation from blue wavelengths (Hoenecke et al., 1992; Hogewoning et al., 2010). From a comprehensive perspective, monochromatic red LED exposure likely promoted an expansion of the total light-receiving leaf area, resulting in relatively greater cumulative carbon gain. Nonetheless, the lack of photosynthetically active radiation from other wavelengths, such as blue light, appears to have adversely affected overall photochemical efficiency.

Analysis of Remote Sensing Vegetation Indices

The analysis of remote sensing vegetation indices provides a non-destructive approach for estimating physiological and structural characteristics—such as chlorophyll content, photosynthetic efficiency, and canopy or leaf-level coverage (Xue and Su, 2017). These indices can contribute to a comprehensive understanding of the physiological responses of C. setidens by linking plant traits to leaf-level spectral signatures. According to the analysis, the normalized difference vegetation index (NDVI)—an indicator widely used to assess overall plant vigor—exhibited high values ranging from 0.770 to 0.786 under four types of white LEDs and a purple LED. In contrast, NDVI values under red, green, and blue LEDs were relatively lower at 0.748, 0.744, and 0.751, respectively (Fig. 9). Since NDVI approaches 1 as physiological conditions improve, and values above approximately 0.6 typically indicate healthy vegetation (Kocur-Bera and Małek, 2024), even the lowest average NDVI observed in this study (0.744) falls within the normal range. Meanwhile, the photochemical reflectance index (PRI), which is associated with the epoxidation-de-epoxidation status of the xanthophyll cycle (Gamon et al., 1992) and inversely related to non-photochemical quenching (NPQ) (Ogawa et al., 2024), tended to be highest under the 2100 K white LED treatment. However, no statistically significant differences were observed among treatments. The modified chlorophyll absorption ratio index (MCARI) generally exhibits an inverse relationship with chlorophyll content (Lee et al., 2025a; Shin et al., 2024). In this study, the highest MCARI value was observed under the green LED (0.149), with relatively low mean values recorded under the 4100 and 6500 K white LEDs (0.079 and 0.075, respectively). These results corroborate the inverse relationship between MCARI and SPAD units, an indirect measure of chlorophyll content, underscoring their utility as complementary indicators for estimating relative chlorophyll levels.

Fig. 9

Remote sensing vegetation indices of Korean thistle (C. setidens) after six months of cultivation under different LED spectral qualities. (A) normalized difference vegetation index (NDVI); (B) photochemical reflectance index (PRI); (C) modified chlorophyll absorption ratio index (MCARI); (D) renormalized difference vegetation index (RDVI); (E) optimized soil-adjusted vegetation index (OSAVI); (F) transformed chlorophyll absorption in reflectance index (TCARI); (G) structure insensitive pigment index (SIPI); (H) carotenoid reflectance index 2 (CRI2); and (I) anthocyanin reflectance index 2 (ARI2). In the box plots, the line inside each box marks the median (Q2), and the ‘×’ denotes the mean value. The top and bottom of the box correspond to the third (Q3) and first (Q1) quartiles, respectively, and the whiskers span the minimum and maximum values. Different lowercase letters indicate groups separated by DMRT at p < .05; same lowercase letters denote no significant difference (n = 20).

The renormalized difference vegetation index (RDVI), a variant of the NDVI, increased to 0.603 under the 6500 K white LED treatment but reached its lowest value (0.575) under red LED. The optimized soil-adjusted vegetation index (OSAVI), which corrects for soil background effects, also exhibited a relatively high value of 0.718 under the 6500 K white LED, while its minimum value (0.692) was observed under the red LED treatment. Meanwhile, the transformed chlorophyll absorption in reflectance index (TCARI), an index sensitive to chlorophyll absorption, exhibited a value of −0.081 under the 6500 K white LED. Similarly, the structure-insensitive pigment index (SIPI), which is minimally affected by canopy structure, recorded its highest value (0.799) under the same treatment. Overall, indices including RDVI, OSAVI, TCARI, and SIPI tended to exhibit higher values under the 6500 K white LED, which provides a relatively greater proportion of blue wavelengths. However, these indices did not differ significantly from those observed under the other white LED treatments or the purple LED. Therefore, both white and purple LED lighting can generally be considered effective in enhancing the physiological performance of C. setidens.

The carotenoid reflectance index 2 (CRI2) and the anthocyanin reflectance index 2 (ARI2), which estimate the relative concentrations of carotenoids and anthocyanins, tended to increase under purple LED light, reaching 7.84 and 0.404, respectively. The addition of far-red light to combined red and blue spectra has previously been reported to promote leaf expansion and increase pigment density per unit weight, including carotenoids (Van Brenk et al., 2024), which is consistent with the results of the present study.

The vegetation index patterns observed in this study revealed distinct perspectives and weightings among parameters related to plant vigor (e.g., NDVI, RDVI, OSAVI), leaf pigment composition (e.g., MCARI, TCARI, SIPI, CRI2, ARI2), and photosynthetic light-use efficiency (i.e., PRI). Among these indices, NDVI, RDVI, and OSAVI tended to remain relatively stable and exhibited comparatively high values under white LED conditions. In contrast, indices strongly associated with pigment concentrations— such as MCARI, TCARI, SIPI, CRI2, and ARI2—were more responsive, reflecting physiological differences even among treatments exhibiting similar growth levels. Overall, these results indicate that relying on a single vegetation index is inherently limiting. A more comprehensive approach—considering multiple indices simultaneously—provides a clearer and more balanced assessment of physiological responses across treatments.

Principal Component Analysis (PCA)

Principal component analysis (PCA) is a widely used dimensionality reduction technique that summarizes intercorrelated multivariate variables into a small number of orthogonal components, thereby revealing the primary axes of variation in a dataset and the covariance structure among variables (Abdi and Williams, 2010). In this study, PCA was applied to major growth parameters, chlorophyll fluorescence variables, and vegetation indices. The first and second principal components (PC1 and PC2) explained 29.2 and 20.6% of the total variance, respectively, resulting in a cumulative variance of 49.8% (Fig. 10). When PC3 was additionally considered, the cumulative variance increased to 61.9% (data not shown). Based on the direction and relative length of the loading vectors, PC1 represented an axis contrasting chlorophyll fluorescence parameters associated with plant stress (e.g., M_o, ABS/RC, DI_o/RC) and indicators reflecting physiological vitality (e.g., F_v/F_m, PI_ABS). This trade-off pattern between photochemical efficiency indices and stress-related parameters has been observed in previous studies (Kalaji et al., 2014; Rapacz et al., 2019). Specifically, ABS/RC and DI_o/RC, together with the spectrally sensitive MCARI, were positioned on the positive side of PC1 within the same quadrant, whereas F_v/F_m and PI_ABS were oriented in the negative direction. This distribution clearly illustrates a trade-off whereby greater energy dissipation corresponds to lower photochemical efficiency. Although NDVI and SPAD units were also aligned on the negative side, their relatively short loading vectors indicated limited contributions, suggesting only a moderate association with the physiological efficiency axis. In contrast, S/R and total RMC, positioned in the positive direction of PC1 with minor contributions from PC2, indicated that water status and biomass allocation between shoots and roots were primarily associated with physiological indices rather than with the biomass axis per se.

Fig. 10

Principal component analysis (PCA) biplot of various parameters and indices of Korean thistle (C. setidens) after six months of cultivation under different LED spectral qualities. PC1 and PC2 explained 29.2 and 20.6%, respectively, accounting for 49.8% in total. SH: shoot height; WD: shoot width; SD: stem diameter; SDW: shoot dry weight; RDW: root dry weight; TDW: total dry weight; TMC: total relative moisture content; LA: leaf area.

PC2 was evaluated as a growth-related axis that integrates plant size, biomass accumulation, and plant quality attributes. Total dry weight, shoot dry weight, root dry weight, along with the DQI, compactness, leaf area, stem diameter, and shoot width exhibited relatively long and densely clustered loadings in the positive direction of PC2. Notably, compactness, DQI, SDW, RDW, and TDW formed small inter-vector angles, indicating strong positive correlations among these variables. This pattern suggests that these parameters are most important for assessing plant quality, reflecting not only increases in biomass but also concurrent enhancements in compactness and DQI. In contrast, PRI and SIPI were positioned near the origin with short loading vectors, indicating their relatively minor contributions to both PC1 and PC2.

These analytical results provide a foundational framework for understanding how plant morphophysiological traits—including chlorophyll-fluorescence responses and remote-sensing vegetation indices—are structurally organized and interrelated. Consequently, an assessment approach incorporating maximum quantum yield (F_v/F_m), the performance index on an absorption basis (PI_ABS), energy-flux parameters per reaction center (ABS/RC, DI_o/RC), and plant quality indices (compactness, DQI) provides essential baseline information for optimizing spectral regimes and producing high-quality C. setidens plants.

Correlation and Cluster Analysis

Correlation coefficients and cluster analysis are useful statistical approaches for elucidating relationships among multiple key indices (Eisen et al., 1998). Pearson correlation analysis combined with Euclidean distance-based average-linkage hierarchical cluster analysis clearly identifies the cluster distinctions and inter-index relationships (Fig. 11). Clustering based on correlation profiles (i.e., rows and columns of the correlation matrix) resolved three major groups. The first cluster comprised DI_o/RC, M_o, ABS/RC, S/R, TMC, SH, and MCARI. The second cluster grouped PI_ABS, F_v/F_m, SPAD units, SIPI, and NDVI. The third cluster included TDW, SDW, compactness, RDW, DQI, PRI, SD, LA, and WD. These cluster distinctions indicate that different sets of indices capture distinct morphological or physiological attributes.

Fig. 11

Heatmaps of Pearson correlation coefficients with hierarchical clustering of various parameters and indices of Korean thistle (C. setidens) after six months of cultivation under different LED spectral qualities. TMC: total relative moisture content; SH: shoot height; TDW: total dry weight; SDW: shoot dry weight; RDW: root dry weight; SD: stem diameter; LA: leaf area; WD: shoot width.

The most consistent trend observed was the inverse relationship between photochemical stress and photochemical efficiency. DI_o/RC showed a strong negative correlation with F_v/F_m (r = −0.92), ABS/RC was strongly negatively correlated with PI_ABS (r = −0.84), and M_o was negatively correlated with F_v/F_m (r = −0.69). This supports the view that greater reaction-center inactivation is associated with lower chlorophyll-fluorescence-based indices of photochemical performance. Conversely, ABS/RC and DI_o/RC displayed a strong positive correlation (r = 0.92), as did M_o and DI_o/RC (r = 0.86), highlighting tight coherence among stress-related indices. Meanwhile, DQI exhibited strong alignment with biomass-related indices. DQI positively correlated with TDW (r = 0.78), compactness (r = 0.78), and SDW (r = 0.61). In contrast, S/R was negatively correlated with DQI (r = −0.52), suggesting that disproportionate shoot development may reduce overall plant quality.

Correlation analysis of the remote sensing vegetation indices revealed a strong positive correlation between NDVI and SIPI (r = 0.94), whereas MCARI showed a negative correlation with NDVI (r = −0.62). These patterns indicate that greenness-related signals—reflecting overall plant health—as well as pigment consumption and chlorophyll-sensitive indices may vary either proportionally or inversely depending on the specific index considered. Meanwhile, PRI was largely uncorrelated with most indices, suggesting that its utility as an index of physiological status in C. setidens under varying light-quality conditions may be relatively limited.

In summary, the findings demonstrate partial linkages among three domains: plant growth (morphological parameters and biomass), photochemistry-based physiological status (chlorophyll fluorescence), and chlorophyll-related spectral signals (remote sensing vegetation indices). However, these domains were not fully congruent. Therefore, rather than relying on a single index, comprehensive observation of relative changes from a morphophysiological perspective and interpretation based on a balanced assessment are recommended.