Accumulation of total phenolics and flavonoids in L. pumila were influenced by potassium levels (p ≤ 0.01; Table 1). Generally total phenolics content was highest in the leaves followed by stems and roots. As the plant receives high K levels (0 > 270 kg K/ha) the production of total phenolics and flavonoids was enhanced. The total phenolics content in leaf-180 kg K/ha, leaf-90 kg K/ha, leaf-0 kg K/ha, stem-270 kg K/ha, stem-180 kg K/ha, stem-90 kg K/ha, stem-0 kg K/ha, root-270 kg K/ha, root-180 kg K/ha, root-90 kg K/ha and root-0 kg K/ha were 10, 23, 35, 47, 52, 61, 65, 66, 68, 77 and 78%, respectively, were low compared to the leaf at 270 kg K/ha that registered 1.82 mg gallic acid g⁻¹ dry weight. Total flavonoids content followed the same trend as total phenolics where the highest total flavonoids was observed in leaf at 270 kg K/ha that registered 0.94 mg rutin g⁻¹ dry weight and the lowest was in the root at 0 kg K/ha that contained only 0.18 mg rutin g⁻¹ dry weight.

The increase in the production of total phenolics and flavonoids with increasing potassium levels in the study might be due to enhancement of total non-structural carbohydrate (TNC) [43,44]. This might due to potassium’s role of stimulating photosynthesis activity and increasing the translocation of carbohydrate to plant parts [45,46]. The increase in translocation indirectly enhanced the biosynthesis of total phenolics and flavonoids of L. pumila treated with high potassium fertilizer. From correlation Table 2 it was shown that TNC had a high significant positive correlation with total phenolics (R² = 0.985; p ≤ 0.05) and total flavonoids (R² = 0.812; p ≤ 0.05) that indicate the increase of production of total phenolics and flavonoids under high potassium levels might be due to increase availability of TNC [47,48]. The present findings was also in agreement with Wei et al.[8] that found the increase in total phenolics and flavonoids in Chrysanthemum morifolium leaves was due to increase in production of TNC under high K application. This implies that an increasing potassium supply can enhance production of TNC and simultaneously increase the production of plant secondary metabolites in L. pumila seedlings.

Ascorbic acid, also known as vitamin C, is one of the most abundant antioxidants in plant where the role of ascorbate is to protect plants against oxidative stress [49]. It is a powerful water soluble antioxidant and its established role is to prevent scurvy [50]. The profiling of ascorbic acid in L. pumila plants followed the same trend as the total phenolics and flavonoids, where the availability of ascorbic acid was found to be higher in the leaves followed by stems and lowest in roots (Table 1). The imposition of high K levels has resulted in significantly higher ascorbic acid contents in the leaves, stems and roots of L. pumila. By the end of week 15 after start of treatments, the ascorbic acid contents in the leaves of plants receiving 0, 90 and 180 kg K/ha were 0.030, 0.047 and 0.062 mg g⁻¹l-ascorbic acid fresh weight, respectively, compared to 0.083 mg g⁻¹l-ascorbic acid fresh weight achieved with 270 kg K/ha application. The same observation was made by Liaqat et al. [34] and Prasad and Spiers [51] when they observed ascorbic acid content in blackberries and Kiwi fruit was substantially enhanced with application of high amounts of K fertilizer. The increase in ascorbic acid content under high K application levels in L. pumila seedlings might possibly be attributed to high production of TNC under high K application. This is because TNC (d-glucose) is a precursor for ascorbic acid biosynthesis in plants, as more availability of TNC more ascorbic acid would be produced in the l-galactose pathways [52,53]. This fact was supported by positive relationship of ascorbic acid with TNC (R² = 0.765; p ≤ 0.05; Table 2). The present result indicated that accumulation of TNC in high fertilized K plant would increase the production of ascorbic acid in L. pumila.

Primary metabolites also followed the same patterns as total phenolics, flavonoids and vitamin C. The accumulation of soluble sugar, starch and TNC was influenced by K levels (p ≤ 0.05; Table 3). The highest accumulation of primary metabolites was also highest in leaves and lowest in the roots. The soluble sugar content in leaf-270 kg K/ha, leaf-180 kg K/ha, leaf-90 kg K/ha, leaf-0 kg K/ha, stem-270 kg K/ha, stem-180 kg K/ha, stem-90 kg K/ha, stem-0 kg K/ha, root-270 kg K/ha, root-180 kg K/ha, root-90 kg K/ha and root -0 kg K/ha were 79, 69, 60, 46, 41, 28, 26, 18 and 14%, respectively, higher compared to the root at 0 kg K/ha that only registered 29.61 mg g⁻¹ dry weight. The starch and TNC showed similar patterns as soluble sugar. Previous reports showed there was less accumulation of soluble sugar and increase in starch content under high K fertilization [29,40,41]. In the present results it was shown as the K increased the soluble sugar and starch content tended to accumulate. The differences in the carbohydrate accumulation in the study with the previous study might be due to the sink strength of the plant to K fertilization response [54]. Usually, plants with high sink strength would accumulate more starch and soluble sugar in their plant parts. In the present study, it was observed that there was more accumulation of starch than soluble sugar in L. pumila plants fertilized with high K without impairment of photosynthesis. Usually, plant with high sink strength would accumulate more starch and less soluble sugar with no impairment of photosynthesis under high potassium fertilization [29,55]. This indicate that L. pumila have high sink strength under high supplementation of K, however the response are different reported in other plant due to differences in the plant species, the plant sink strength and the sources of K that been used [26–28,56–58]. In the present study, the excess in TNC under high K fertilization would be used for growth and production of secondary metabolites [42]. The increase in production of TNC under excess K fertilization was also observed by Pettigrew [22] and Liu et al. [58] in cotton and Chrysanthemum morifolium. The increase in carbohydrate production under high K fertilization might be due to increase in starch synthetase (EC 2.4.1.21) activity that enhanced the production of carbohydrate. This is because starch synthetase activity is activated by K, thus with adequate K the levels of starch and soluble sugar accumulated [59–61]. From the correlations in Table 2, soluble sugar, starch and TNC were observed to have strong significant positive correlations with total phenolics and flavonoids. Carbohydrates are basic compounds required to produce phenolic compounds through the shikimic acid pathway where extra carbohydrates derived from glycolysis and the pentose phosphate pathway are converted into aromatic amino acids [62]. Previous studies by Shui et al.[62] showed that an increase in secondary metabolites was related to the balance between carbohydrate source and sink; the greater the source-sink ratio, the greater the production of secondary metabolites that might occur.

The PAL activity was influenced by potassium levels (p ≤ 0.05; Table 4). The PAL activity was found to be highest under 270 kg K/ha (37.28 nM transcinnamic mg⁻¹ protein⁻¹ hour⁻¹) followed by 180 kg K/ha (25.61 nM transcinnamic mg⁻¹ protein⁻¹ hour⁻¹), 90 kg K/ha (19.28 nM transcinnamic mg⁻¹ protein⁻¹ hour⁻¹) and the lowest at 0 kg K/ha that just recorded 8.24 nM transcinnamic mg⁻¹ protein⁻¹ hour⁻¹. The soluble protein content followed the same trend as PAL activity. As potassium fertilization increases from 90 > 180 > 270 kg K/ha soluble protein increased by 180%, 495% and 958%, respectively, compared to control (0 kg K/ha) that just recorded 1.17 mg/g fresh weight protein. The increase in production of secondary metabolites in the present work could be due to increase in PAL activities under high K fertilization. Correlation analysis (Table 2) showed that PAL had a significant positive relationship with total phenolics and flavonoids (R² = 0.945; R² = 0.987; p ≤ 0.05), which might indicate an upregulation of plant secondary metabolite production with increased PAL activity. This is basically due to the fact that PAL is a precursor to total phenolics and flavonoids biosynthesis [63]. The increase in PAL activity under high K fertilization was also observed by Wei et al. [8] in Chrysanthemum morifolium. Reasons can be higher PAL gene expression which was observed by Li et al.[63] in maize when exposed to high K levels or general enzyme activation by potassium. This also might be due the role of potassium in enzyme activation. The increase in total phenolics and flavonoid biosynthesis with increasing potassium fertilization might be due to the role of potassium as a cofactor in phenylalanine ammonia-lyase (PAL) enzyme activity. With more potassium available, the activity of PAL enzyme increases and more biosynthesis of secondary metabolites might occur [64,65]. This was observed by Devi et al. [66] where a high supply of potassium increased the PAL and antioxidant activity in Panax ginseng.

For this study, the net photosynthetic rate (A), stomatal conductance (g_s), intercellular CO₂ (C_i), dark respiration rates (R_d) and apparent quantum yield (ξ) were determined by a portable infrared photosynthesis system LI-6400 (LI-COR, Lincoln, NE, USA). In general, the photosynthesis rate (A) was found to be higher in 270 kg K/ha (12.72 μmol m⁻² s⁻¹), followed by 180 kg K/ha (8.45 μmol m⁻² s⁻¹), 90 kg K/ha (6.75 μmol m⁻² s⁻¹) and lowest in 0 kg K/ha that just recorded 3.45 μmol m⁻² s⁻¹ (Table 5). The same pattern was followed with stomatal conductance (g_s) and intercellular CO₂ (C_i). It was found that g_s and C_i have a significant positive correlation with net photosynthesis (R² = 0.987; R² = 0.908; p ≤ 0.01; Table 2) respectively, this indicate that the increase in photosynthesis rate under high K supplementation might be due increase in g_s that was stimulated with high K that simultaneously enhanced the C_i inside the leaves [67,68]. A previous study by Peoples and Coach [69] had shown that increased K fertilization can enhance the conductance in Medicago sativa and simultaneously enhance the uptake of CO₂ that increases C_i levels. The role of K in photosynthesis is complex. The activation of enzymes by K and its involvement in adenosine triphosphate (ATP) production is more important in regulating the rate of photosynthesis than the role of K in stomatal conductance [70]. This is because when plant are lacking in K, the rate of photosynthesis and the rate of ATP production are reduced, and all the process reliant on ATP are slowed down [71]. Conversely, plant respiration rate increases, which also contributes to slower growth and development. This phenomenon was shown in the present study where L. pumila fertilized with 270, 180 and 90 kg K/ha had lower respiration rates (2.14, 4.11, 6.23 μmol m⁻² s⁻¹), respectively, compared to control plants (0 kg K/ha) that registered the highest respiration rate (8.24 μmol m⁻² s⁻¹). Furthermore, the plant treated with high potassium fertilization have shown enhanced light harvesting efficiency by having a high apparent quantum yield. At 270 kg K/ha, the apparent quantum yield of L. pumila was 0.097 μmol m⁻² s⁻¹, compared to 180, 90 and 0 kg K/ha that recorded 0.078, 0.043 and 0.032 μmol m⁻² s⁻¹, respectively. According to Lam et al.[72] the increase in net photosynthesis with increase K might be due to increase in nitrate reductase activity that directly involved in photosynthetic enzyme regulation. From correlation Table 2 it were found that A have a significant positive correlationship with total phenolics (R² = 0.998; p ≤ 0.05) and total flavonoid (R² = 0.897 p ≤ 0.05) that indicate an increase in production of secondary metabolites under high K fertilization might be due to increase photosynthetic capacity of L. pumila that stimulated the shikimic acid pathway. [73]. This suggest the importance of K in regulation of leaf gas exchange and secondary metabolites in L. pumila. [74–76].

The leaf invertase activity was influenced by K levels applied (p ≤ 0.05; Figure 1). Leaf acid invertase and alkaline invertase increased with increasing K fertilization in an ascending order 0 < 90 < 180 < 270 kg K/ha. The highest acid and alkaline invertase activity was obtained in L. pumila fertilized with 270 kg K/ha (18.65 and 21.335 mg g⁻¹ glucose fresh weight hour⁻¹), respectively, compared to control (0 kg K/ha), that recorded 6.32 mg g⁻¹ glucose fresh weight hour⁻¹ for acid invertase and 7.89 mg g⁻¹ glucose fresh weight hour⁻¹ for alkaline invertase. Invertase can be considered a key enzyme in carbohydrate metabolism since it catalyzes the irreversible reaction that converts sucrose into glucose and fructose [77]. Potassium plays a critical role in phloem translocation mechanism. The increase in invertase activity showed enhanced phloem translocation mechanism under high K fertilization [78]. This is because an increase in invertase activity would increase the conversion of sucrose to glucose and fructose. Higher glucose levels with more potassium supply would produce more ATP in the citric acid cycle that will be used in phloem translocation of nitrates, minerals and amino acids [79]. According to Lei and Yan [77] the enhanced invertase activity correspond to enhanced malondialdehyde and nitrate reductase in Stevia rebaudiana. From correlation Table 2 it was found that acid and alkaline invertase activity have a significant positive correlation with total phenolics and flavonoids that indicated the increase in conversion of sucrose to glucose and fructose in high invertase activity might increase the availability of glucose that might enhanced production of plant secondary metabolites (total phenolics, flavonoid and ascorbic acid). The present finding showed the importance of K in enhancing the availability of glucose by promoting high invertase activity.

Activities of the antioxidant enzymes ascorbate peroxidase (APX), catalase (CAT) and superoxide dismutase (SOD) were significantly (p ≤ 0.05) affected by potassium fertilization (Table 6). The APX, CAT and SOD activities were found lowest at maximum potassium fertilization at 270 kg P/ha. These activities are an indication that enhanced potassium fertilization can reduce the oxidative stress to L. pumila seedlings. Under stress potassium plays an important role in the synthesis of protein by participating in polypeptide synthesis in ribosomes, since that process requires a high concentration of potassium [80]. It was reported by Tripathi et al.[81] that proteins such as thioredoxin, glutaredoxin and cyclophilin are known to facilitate the regeneration of the reduced (catalytically active) form of peroxyredoxin that plays an important role in reducing the ROS formation in plants under biotic and abiotic stress. The reduction in APX, CAT and SOD activities have shown to have a negative significant correlation with production of total phenolics and flavonoids that indicate the impairment of oxidative stress can enhance the secondary metabolites of L. pumila seedlings under high potassium fertilization. Generation of ROS, particularly H₂O₂, had been proposed to be part of the signaling cascades that lead to protection from stress. Induction of antioxidant enzymes was reported to be a general strategy adopted by plants to overcome oxidative stresses. The APX, CAT and SOD function as effective quenchers for ROS [82]. CAT plays an essential role in scavenging from H₂O₂ toxicity. The combined action of CAT and SOD converts the O²⁻ and H₂O₂ to water and molecular oxygen (O₂), thus prevent the cellular damage under unfavorable condition [83]. In the present study potassium plays a key role in reduction of ROS production by reducing activity of NAD(P)H oxidase and maintaining electron transport [84]. The present result indicated that if oxidative stress can be reduced by increasing potassium fertilization, the production of secondary metabolites in L. pumila can be enhanced. These result also implied that fertilization with high potassium can play important protective role in O²⁻ and H₂O₂ scavenging processes. In correlation Table 2 it is shown that APX, CAT and SOD have a significant negative correlation with total phenolics and flavonoids (p ≤ 0.05). This indicates that the antioxidant activity and secondary metabolite levels that indicate up-regulation of secondary metabolites might be occurring under low oxidative stress with high potassium fertilization in L. pumila benth seedlings.

The experiment was carried out in growth houses at Field 2, Faculty of Agriculture Greenhouse Complex, Universiti Putra Malaysia (longitude 101°44′ N and latitude 2°58′ S, 68 m above sea level) with a mean atmospheric pressure of 1.013 kPa. Three-month old L. pumila seedlings of var. alata, were left for a month to acclimatize in a nursery until ready for the treatments, then they were fertilized with four rates of potassium applied in the form of muriate of potash (MOP), viz. 0 kg K/ha (0.0 g per plant), 90 kg K/ha (0.25 g per plant), 180 kg K/ha (0.51 g per plant) and 270 kg K/ha (0.76 g per plant). The potassium was split into three fertilization phases, and each phase was about 33.3% of total potassium fertilizer. Every potassium treatment received urea (46% N; 0.72 g per plant) and Triple Super Phosphate, TSP (60% K; 0.51 g per plant) at standard rates of 180 kg /ha during the studies there were no indication of K deficiency in all the plant in 0 kg K/ha. The seedlings were planted in soilless medium containing coco-peat and well composted chicken manure in 5:1 (v/v) ratio in 25 cm diameter polyethylene bags. The medium properties are presented in Table 7. Day and night temperatures in the greenhouse were maintained at 27–30 °C and 18–21 °C, respectively, and relative humidity from 50% to 60%. All the seedlings were irrigated using overhead mist irrigation given four times a day or when necessary. Each irrigation session lasted for 7 min [85]. The experiment was based on a Randomized Complete Block Design (RCBD) with four replicates. The factor was four levels of potassium fertilization (0, 90, 180 and 270 kg K/ha). Each combination treatment consisted of 10 plants totaling a sum of 160 plants used in the experiment. Plants were harvested at 12 weeks after planting.

The method of extraction and quantification for total phenolics and flavonoids contents followed after Ibrahim et al. [86]. An amount of ground dried tissue samples (0.1 g) was extracted with 80% ethanol (10 mL) on an orbital shaker for 120 min at 50 °C. The mixture was subsequently filtered (Whatman™ No.1), and the filtrate was used for the quantification of total phenolics and total flavonoids. Folin-Ciocalteu reagent (diluted 10-fold) was used to determine the total phenolics content of the leaf samples. The sample extract (200 μL) was mixed with Folin-Ciocalteau reagent (1.5 mL) and allowed to stand at 22 °C for 5 min before adding NaNO₃ solution (1.5 mL, 60 g L⁻¹). After two hours at 22 °C, absorbance was measured at 725 nm. The results were expressed as mg g⁻¹ gallic acid equivalent (mg GAE g⁻¹ dry sample). For total flavonoids determination, a sample (1 mL) was mixed with NaNO₃ (0.3 mL) in a test tube covered with aluminium foil, and left for 5 min. Then 10% AlCl₃ (0.3 mL) was added followed by addition of 1 M NaOH (2 mL). Later, the absorbance was measured at 510 nm using a spectrophotometer with rutin as a standard (results expressed as mg g⁻¹ rutin dry sample).

The ascorbic acid content was measured using a modified method of Davis and Masten [87]. The fresh leaf samples (1 g) were extracted in 1% of phosphate-citrate buffer (2 mL, pH 3.5) using a chilled pestle and mortar. The homogenate was filtered. The filtrate was added to the 1 mL of 1.7 mM 2,6-dichloroindophenol (2,6-DCPIP) in a 3 mL cuvette. The absorbance at 520 nm was read within 10 min of mixing the reagents. The extraction buffer was used as a blank. l-Ascorbic acid was used as a standard. Ascorbic acid was recorded as mg g⁻¹l-ascorbic acid fresh leaves

Total soluble sugar was measured spectrophotometrically using the method of Ibrahim and Jaafar [88] Samples (0.5 g) were placed in 15 mL conical tubes, and distilled water added to make up the volume to 10 mL. The mixture was then vortexed and later incubated for 10 min. Anthrone reagent was prepared using anthrone (Sigma Aldrich, St. Louis, MO, USA, 0.1 g) that was dissolved in 95% sulphuric acid (Fisher Scientific, Chicago, IL, USA, 50 mL). Sucrose was used as a standard stock solution to prepare a standard curve for the quantification of sucrose in the sample. The mixed sample of ground dry sample and distilled water was centrifuged at a speed of 3400 rpm for 10 min and then filtered to get the supernatant. A sample (4 mL) was mixed with anthrone reagent (8 mL) and then placed in a water-bath set at 100 °C for 5 min before the sample was measured at an absorbance of 620 nm using a spectrophotometer model UV160U (Shimadzu Scientific, Kyoto, Japan). The total soluble sugar in the sample was expressed as mg sucrose g⁻¹ dry sample.

Starch content was determined spectrophotometrically using a method described by Thayumanavam and Sadasivam [89]. In this method, dry sample (about 0.5 g) was homogenized in hot 80% ethanol to remove the sugar. The sample was then centrifuged at 5000 rpm for 5 min and the residue retained. After that, distilled water (5.0 mL) and 52% perchloric acid (6.5 mL) were added to the residue. Then the solution was centrifuged and the supernatant separated and then filtered with Whatman No. 5 filter paper. The processes were repeated until the supernatant was made up to 100 mL. A sample (100 μL) of the supernatant was added to distilled water until the volume became 1 mL in a test tube. After that, anthrone reagent (4 mL, prepared with 95% sulphuric acid) was added to the test tube. The mixed solution was placed in the water bath at 100 °C for eight min and then cooled to room temperature, and then the sample was read at absorbance of 630 nm to determine the sample starch content. Glucose was used as a standard and starch content was expressed as mg glucose equivalent g⁻¹ dry sample.

The total non-structural carbohydrate was calculated as the sum of total soluble sugar and starch content [90].

Phenylalanine-ammonia-lyase (PAL) activity was measured using the method described by Martinez and Lafuante [91]. The enzyme activity was determined by measuring spectrophotometrically the production of trans-cinnamic acid from l-phenylalanine. Enzyme extract (10 μL) was incubated at 40 °C with 12.1 mM l-phenylalanine (90 μL, Sigma) that were prepared in 50 mM Tris-HCl, (pH 8.5). After 15 min of reaction, trans-cinnamic acid yield was estimated by measuring increase in the absorbance at 290 nm. Standard curve was prepared by using a trans-cinnamic acid standard (Sigma) and the PAL activity was expressed as nM trans-cinnamic acidμg protein⁻¹ h⁻¹.

Protein content was determined using the method of Ibrahim and Jaafar [92]. In this method, fresh leaf samples (about 2 g) were cut into pieces using scissors and ground in mortar with 0.05 M Tris buffer (1 mL, pH 8.5) and powdered with liquid nitrogen. The homogenate was then centrifuged at 9000 rpm for 10 min and then stored under refrigeration at 4 °C for 24 h. After the extraction, supernatant from the samples (about 100 μL) was added to Bradford reagent (3 mL, Sigma, prepared using 10 mL of the reagent diluted with 50 mL distilled water) and then incubated for 5 min before being measured at 595 nm with the spectrophotometer. In this method bovine serum (Sigma) was used as a standard to produce calibration curve between actual protein content and spectrophotometer readings. The protein was expressed as mg g⁻¹ protein fresh weight.

The measurement was obtained from a closed infra-red gas analyzer LICOR 6400 Portable Photosynthesis System (IRGA, Licor. Inc. Nebraska, NE, USA). Prior to use, the instrument was warmed for 30 min and calibrated with the ZERO IRGA mode. Two steps are required in the calibration process: first, the initial zeroing process for the built-in flow meter; and second, zeroing process for the infra-red gas analyzer. The measurements used optimal conditions set of 400 μmol mol⁻¹ CO₂ 30 °C cuvette temperature, 60% relative humidity with air flow rate set at 500 cm³ min⁻¹, and modified cuvette condition of 800 μmol m⁻² s⁻¹ photosynthetically photon flux density (PPFD). The measurements of gas exchange were carried out between 09:00 and 11:00 a.m. using fully expanded young leaves numbered three and four from plant apex to record net photosynthesis rate (A). The operation was automatic and the data were stored in the LI-6400 console and analyzed by “Photosyn Assistant” software (Version 3, Lincoln Inc., Columbus, OH, USA). Several precautions were taken to avoid errors during measurements. Leaf surfaces were cleaned and dried using tissue paper before enclosed in the leaf cuvette [93]. The light response curve was produced followed procedures from Ibrahim and Jaafar [94] to generate the apparent quantum yield and dark respiration rate.

Invertases were assayed according to Schaffer et al.[95] with some modifications. Leaf tissue (2.0 g fresh weight) was extracted in buffer (10 mL) containing 50 mM HEPES-NaOH (pH 7.5), 5 mM MgCl₂, 1 mM Na₂EDTA, and 0.05% Triton X-100. The extract was passed through Microcloth and centrifuged at 17,300× g for 20 min at 4 °C. The supernatant was concentrated by the addition of (NH₄)₂SO₄ to 80% saturation. After centrifugation for 15 min at 20,000 rpm the precipitate was resuspended in 2 mL of extraction buffer and dialyzed overnight against 5 mM HEPES-NaOH (pH 7.5), 5 mM MgCl₂, and 1 mM Na₂EDTA. The invertase assay mixture contained 200 μL dialyzed extract, 200 mM sucrose, and 100 mM citrate buffer (pH 5.0) for acid invertase and 100 mM HEPES-NaOH (pH 7.5) for alkaline invertase. The reaction mixtures were incubated for 45 min at 37 °C and the glucose and fructose released were determined with dinitrosalicylic acid.

Data were analyzed using analysis of variance by SAS version 17. Mean separation test between treatments was performed using Duncan multiple range test and standard error of differences between means was calculated with the assumption that data were normally distributed and equally replicated. Correlation analysis was analyzed using SPSS version 13 using Pearson correlation methods [99–103].