The experiment was conducted in a growth chamber at Ghent University, Belgium. Three ornamental species were selected: C. australis “Red Star” (monocot), F. benjamina “Exotica” (dicot, evergreen leaves), and S. speciosa “Sonata Red” (dicot, deciduous leaves). Young plants were obtained from a commercial plant producer and transplanted into 0.3 L pots filled with peat-based potting soil (Van Israel nv, Belgium). The plants were acclimated for 1 week in broad spectrum light (100 μmol m⁻² s⁻¹) provided by SON-T high pressure sodium lamps (Philips Inc., Eindhoven, the Netherlands). Then for each species, 12 replicates per treatment were randomly allocated to four spectral light treatments. Air temperature of the growth chamber was set at 22 ± 2°C and plants received a photoperiod of 16 h. Irrigation and fertilization with a water-soluble fertilizer (N:P:K = 4:1:2, EC 1.5 ds m⁻¹, pH = 6.5) was applied once every 2 days.

Light intensity at the canopy level was set at 100 μmol m⁻² s⁻¹ by adjusting the distance of the light source and a photoperiod of 16 h per day was given. Light treatment sections were separated with curtains, four treatments were applied using different light qualities equipped with LED lighting, which were B (100% blue, peak at 460 nm), R (100% red, 660 nm), and W [white, 7% blue (400–500 nm), 16% green (500–600 nm), 75% red (600–700 nm), and 2% far red (700–800 nm)] (Philips Inc., Eindhoven, The Netherlands) as well as RB (75% R and 25% B, peak at 460 and 660 nm) by a CID-800 programmable LED lighting system (CID Bio-Science, USA), respectively. Light distribution was recorded using JAZ-ULM-200 spectrometer (Ocean Optics, FL, USA) and converted with Spectrasuite software (Ocean Optics) to μmol m⁻² s⁻¹ (Figure 1) and uniformity was verified by measuring the light intensity at five points of each light treatment at the canopy level (Table 1).

The plants were grown for 8 weeks and then the second or third leaf counting from the apex (fully expanded leaves that developed entirely under the given light quality) were selected for the measurements. All measurements were performed in four replications per treatment and per plant species.

Leaf segments of 2 × 2 cm of the central leaf blade next to main vein were excised and fixed for at least 24 h in a formaldehyde-based fixative (FAA). Then, leaf segments were dehydrated using a series graded concentration ethanol, embedded in paraffin, and sectioned at thickness of 12 μm with a microtome (R. Jung AG, Heidelberg, Germany). The sections were deparaffinized with xylene and rehydrated with graded ethanol, stained with safranin for 30 min and fast green for 30 s. Stained sections were sealed with Canadian balsam and examined with a bright field microscopy (IX81, Olympus, Japan) at magnification 400x. Images of the cross sections were taken and measured for widths of whole-leaf, palisade mesophyll, spongy mesophyll and abaxial and adaxial epidermal tissues with ImageJ (ImageJ 1.48v, NIH, USA).

The hydraulic conductance of whole leaves (K_leaf) was performed using a high pressure flowmeter method (HPFM) with slight modifications (Sack et al., 2002). The second or lower fully expanded leaf (depending on the species) was cut next to the petiole stem insertion and immediately placed in a water bath. The petiole was cut under water with a sharp blade to 1 cm length, then wrapped with parafilm (to ensure good seal between petiole and tubing) and inserted into the silicon tube which was connected to the HPFM hydraulic measurement system as described by Tyree (Tyree et al., 2005). Briefly high pressure water was pressed into the leaf vein, leaves were perfused at 0.3 MPa with distilled water for around 60 min until steady-state conditions (±5%), the flow rate was recorded and calculated to leaf hydraulic conductance (mmol m⁻² s⁻¹ MPa⁻¹). Leaf area was measured afterward with a leaf area meter (Li-Cor 3000, LiCor, USA) to normalize hydraulic measurements by leaf area.

Stomatal traits were analyzed using a nail polish print method on the leaf abaxial side as describe by Mott et al. (1991). The nail polish layer was removed with a transparent tape and pasted on a glass slide, the slides were observed with a bright field microscopy (IX81, Olympus, Tokyo, Japan) and stomatal density was calculated based on stomatal counts of 12 microscopic fields (25x, microscopic field 0.16 mm²) per leaf, ensuring a 95% confidence level of the results, as the number of stomata per mm². The stomatal index was calculated as number of stomata/(number of epidermal cells + number of stomata) × 100 (Kubinova, 1994). The stomatal aperture, width and length was defined as (Chen et al., 2012) and stomatal aperture area was calculated by assuming an oval pore shape. The total stomatal aperture area per unit leaf area (cm² m⁻²) was calculated as stomatal average density × stomatal aperture area.

Stomatal conductance (g_s) was measured using a leaf porometer (AP4 porometer, Delta-T Devices, Cambridge, UK). The second/third fully developed leaf (different according to the plant species) was chosen for measurements. Four positions on each leaf were measured and the average result was used as the stomatal conductance of this leaf. C. australis is characterized by narrow leaves, which did not allow a correct measurement of g_s by porometry, therefore g_s of was estimated based on stomatal characteristics as described by Franks and Farquhar (2001): gs=SDDa′V(l+π4a′π)

Where SD = stomatal density (N m⁻²), D = diffusivity of water in air (22°C, 24.5 × 10⁻⁶ m² s⁻¹), a' = stomatal aperture area (m⁻²), V = molar volume of air (m³ mol⁻¹), l = depth of stomatal pore (m, 12 × 10 ⁻⁶ m for C. australis, mean of 10 replicates).

The leaf chlorophyll fluorescence measurement was conducted using a PAM-2500 portable chlorophyll fluorometer (Walz, Effeltrich, Germany). The second fully expanded leaf of S. speciosa and the third leaf for C. australis and F. benjamina were selected for this measurement. Leaves were dark adapted with a leaf clip for 20 min, then a 0.6 s saturating light pulse (3,450 μmol m⁻² s⁻¹) was given to obtain the F_m and F₀. After that, the leaf was light adapted with 5 min continuous actinic light (100 μmol m⁻² s⁻¹, similar as the growing condition) with saturating pulses every 25 s, after that, the maximum light adapted fluorescence (F_m′) and steady state fluorescence (F_s) were recorded. The maximum photochemical efficiency of PSII (F_v/F_m) was calculated as F_v/F_m = (F_m−F₀)/F_m (Genty et al., 1989). Then, the actinic light was turned off and a far-red pulse was applied to obtain the minimal fluorescence after the PSI excitation (F₀′). PSII operating efficiency (Φ_PSII) was calculated as Φ_PSII = (F_m′ − F_s)/F_m′ and qP was calculated as qP = (F_m′ − F_s)/(F_m′ − F₀′); NPQ, which is proportional to the rate constant of the thermal energy dissipation, was estimated as NPQ = (F_m−F_m′)/F_m′ (Baker, 2008). The electron transport rate (ETR) was calculated as ETR = Φ_PSII × PAR × 0.84 × 0.5, where the absorbed photon energy (PAR) is assumed to be equally distributed between PSI and PSII and 0.84 is the assumed light absorbance of the leaf.

Leaf chlorophyll content was determined according to Lichtenthaler and Buschmann (2001). One Hundred Fifty milligrams fresh leaf was grinded using liquid nitrogen and extracted in 80% acetone overnight at −20°C. Absorbance at 470 nm (A₄₇₀), 645 nm (A₆₄₅), and 663 nm (A₆₆₃) was measured with a spectrophotometer (Infinite 200, Tecan Group Ltd., Switzerland) and the pigment contents were calculated from the following equations: Chl a = 12.25 × A₆₆₃−2.79 × A₆₄₅; Chl b = 21.50 × A₆₄₅−5.10 × A₆₆₃, and Carotenoids = (1000 × A₄₇₀−1.82 × Chl a−85.02 × Chl b)/198.

The second fully expanded leaf area counting from the apex was measured using a leaf area meter (Li-Cor 3000, Li-Cor, USA) this in four replicates. Four plants per treatment and cultivar were used for the biomass measurements. After aerial fresh weight (FW) determination plants were oven-dried at 85°C for 3 days until a constant mass was reached to determine dry weight (DW).

Data are presented as means ± SE. Data were analyzed for light quality for each species by one-way analysis of variance (ANOVA), after verifying homoscedasticity by Levene's test. Tukey's HSD test was used to compare means at p < 0.05. Correlations between traits were tested using Pearson's correlation coefficients. A regression testing K_leaf as function of leaf thickness and stomatal conductance was performed. All statistical analyses were conducted using SPSS Statistics 22 (IBM Software, Chicago, USA).

In C. australis, total aboveground fresh weight was the greatest under W, followed by B and RB and significantly decreased under R, similar the dry weight was greatest under W and declined under R (Figure 2). Biomass (both FW and DW) of F. benjamina and S. speciosa were significantly lower under R, while no significant difference between the other light qualities were found.

The three species had very different leaf morphologies (Figure 2, Table 2). C. australis and F. benjamina had relative small leaves, while S. speciosa developed large leaves. B enhanced the leaf area of F. benjamina followed by RB and W while it significantly decreased under R. B tended to increase the individual leaf area in both C. australis and S. speciosa though this was not significant (P = 0.070 and 0.183, respectively).

Leaf thickness in C. australis was highest under W followed by RB and B while the thinnest leaves were found under R (Table 2, Figure 3). As C. australis is a monocot, the leaf anatomy is isobilateral and the mesophyll is hardly differentiated into palisade and spongy parenchyma cells. Therefore, only the adaxial and abaxial epidermal thickness was measured which contribute, respectively, 6.1 ± 0.23 and 6.7 ± 0.26% of the total leaf thickness. Abaxial epidermis was not affected by light quality while the thinnest adaxial epidermis was found under R while B and RB had the thickest epidermal cells.

Leaf thickness in F. benjamina was greatest under B, lower under RB and W while it was significantly thinner under R (Table 2). F. benjamina has evergreen glossy leaves and the adaxial and abaxial epidermis contribute, respectively, 21.8 ± 1.0 and 10.6 ± 0.5% to the leaf thickness. Especially the adaxial epidermis is strongly reduced under R followed by W. The effect on the abaxial epidermis is not as strong though also here the thinnest cell layers are under R and W. The leaf thickness difference is strongly influenced by the mesophyll. In absolute value the palisade parenchyma is highest under B although it represents only 15.5% of the total leaf thickness while the palisade layer is, respectively, 26% under RB and 24% under W. B also strongly enhances the spongy parenchyma while it is not affected by the other light qualities. In S. speciosa, leaf thickness was not affected by the different light qualities (Table 2). S. speciosa has velvety hairy leaves and the adaxial and abaxial epidermis contribute, respectively, 10.2 ± 0.5 and 6.5 ± 0.4% to the leaf thickness. Adaxial epidermal thickness was found thinnest under R while it tended to be thicker under B though not significant differing from RB and W. Abaxial epidermis was thickest under B. Palisade parenchyma thickness was found lower under R and W and significantly greater under B and RB while no effect were found for the spongy parenchyma.

Leaf thickness correlated with Φ_PSII in C. australis (r = 0.855) but this correlation was weaker in F. benjamina (r = 0.622) while thickness of the palisade parenchyma correlated moderately with Φ_PSII in S. speciosa (r = 0.674).

Light quality tended to influence the leaf hydraulic conductance of the selected ornamentals though effects were not significant (Figure 4). In C. australis K_leaf was lowest under B and slightly increased under R, RB, and W. In F. benjamina and S. speciosa K_leaf was lowest under R and highest under B. On average K_leaf was highest in C. australis, followed by F. benjamina and quite low in S. speciosa.

Correlation study between K_leaf and other leaf characteristics showed positive correlations with leaf thickness and stomatal conductance in F. benjamina and S. speciosa (Figure 6). However, for the monocot C. australis, a negative trend with stomatal conductance was found and no correlation with leaf thickness.

The effects of light quality on the stomatal characteristics are given in Table 3. The aperture length was not affected by light quality, this for the three species. An increase of aperture area was found in C. australis under B, while no effects were found in F. benjamina and S. speciosa. The width/length ratio was not affected by light quality (data not shown). Total aperture area per unit leaf area was not affected by light quality though it tended to be lower under R for F. benjamina and S. speciosa. Stomatal index and density were significantly affected by the light quality treatments. In C. australis stomatal index decreased under R though density was not affected. C. australis also showed the highest stomatal density of the studied ornamentals, as it ranged between 274.75 N° mm⁻² under B up to 325.10 N° mm⁻² under R. Likewise a high density of epidermal cells per unit leaf area was present but B significantly reduced the number of epidermal cells (Table 3). In F. benjamina, both R and B gave the lowest stomatal index while the highest index was found under W; the stomatal density was lowest under R and highest under W. In S. speciosa both the highest stomatal density and index were found under B and W and the lowest under R.

The stomatal conductance of the ornamentals was differentially affected by the different light qualities (Figure 4). For C. australis, no effects were noted on the stomatal conductance with respect to increasing B. For both F. benjamina and S. speciosa stomatal conductance increased with increasing B when comparing R, RB, and B. However, multispectral W yielded the highest stomatal conductance in both species. A strong correlation of stomatal density (r = 0.979) with g_s and stomatal index (r = 0.995) with g_s was found in S. speciosa.

Effects of light quality on chlorophyll fluorescence parameters of the studied ornamentals are given in Table 4. The maximum quantum efficiency F_v/F_m, was influenced by the applied light quality and overall we saw a lower value of F_v/F_m for R (P = 0.003). For C. australis, the lowest value was observed under R, F_v/F_m increased under W and RB while B gave the highest F_v/F_m-value. For F. benjamina and S. speciosa F_v/F_m declined under R compared to the other spectral qualities.

Φ_PSII, qP and ETR showed a similar reaction to the light quality treatments. For both C. australis and S. speciosa the lowest values for Φ_PSII were observed under R. For F. benjamina, Φ_PSII was significant higher under B, while R and W gave lower values. For both C. australis and S. speciosa highest qP were found for RB and W while no effect of light quality was found for F. benjamina.

NPQ significantly increased under B followed by RB compared to W and R in C. australis, while for S. speciosa, it significantly increased under R and W followed by B compared with RB. However, for F. benjamina, no effect of light quality was found on NPQ (P = 0.117), though it tended to be higher under W.

The total pigment content was different between the species (Figure 5). In F. benjamina, the total chlorophyll content ranged from 1.102 to 1.338 mg g⁻¹, while it was 0.395 to 0.668 mg g⁻¹ and 0.395 to 0.668 mg g⁻¹ for S. speciosa and C. australis, respectively. The carotenoids were higher in C. australis (0.103–0.138 mg g⁻¹) and F. benjamina (0.100–0.190 mg g⁻¹) followed by S. speciosa (0.050–0.103 mg g⁻¹). Overall the total chlorophyll content was not significantly affected by the light quality (P = 0.468) though there were species differences (Figure 5). In C. australis the highest Chl a, Chl b, and Chl a/b was found under RB and the lowest content was found under R, while no significant effect on carotenoid content was present. In F. benjamina, no significant effects of light quality on chlorophyll and carotenoid content were observed. Blue light yielded the highest Chl a, Chl a/b, and carotenoid content in S. speciosa leaves followed by R. The lowest Chl a, Chl a/b, and carotenoid content were found for W, this treatment lead to a decrease of 55 and 51% for Chl a and carotenoids compared to B.

The authors declare that the research was conducted in the absence of any commercial or financial relationships that could be construed as a potential conflict of interest.