2.1 Survey of lamp use for off‐season blooming control of dragon fruits

A questionnaire was designed aiming at collecting relevant information on dragon fruit farming and techniques for flowering stimulation from farmers in Vietnam. Some typical categories include type of lamp, irradiation process, irradiation duration, and bulb density per hectare.

A total of 100 respondents were selected from the farmer lists prepared by the local authorities in three provinces, including Binh Thuan (50 households), Tien Giang (25 households), and Tay Ninh (25 households) according to probabilistic randomized sampling procedure. We also collected lamp samples that were being used for floral stimulation from local farmers for further analysis.

2.2 Lamp spectrum analysis

A PG100N Handheld Spectral PAR Meter (UPRtek, Miaoli, Taiwan) connected to a computer was used to measure emission spectra and determine photosynthetically active photon flux density (PPFD), photon flux density in the red region (PFD R, 600–700 nm), and photon flux density in the far‐red region (PFD FR, 700–780 nm) of lamps.

The testing lamp was positioned 30 cm in front of the instrument sensor in the dark. The procedure was carried out at Laboratory of Agricultural Biotechnology, Nguyen Tat Thanh University, Vietnam.

2.3 Field experiments

Based on the survey results and spectrum analysis, we designed new CFL bulbs with spectra focused in the red and far‐red regions. The light bulbs were manufactured by Rang Dong Light Source & Vacuum Flask Joint Stock Company (Vietnam) and were screened before being used in field experiment.

In the field experiment, three experimental CFL bulbs (20 W) selected from the spectra analysis results were used as treatments (denoted from 1 to 3) and one commercial incandescent lamp (60 W) acted as the control bulb (denoted as control).

No lighting and domestic CFL lighting were not included as controls due to zero or very low productivity resulted from the adoption of the two options. In addition, Yamada et al. (2008) also maintained that the use of domestic CFL lighting gave longer budding time and worse flowering efficiency.

The subject plants were mature white‐flesh pitaya (Hylocereus undatus) and red‐flesh pitaya (Hylocereus polyrhizus) that were normally bearing fruits with the age ranging from 5 to 8 years. The experiment was carried out in household fields located in three provinces with largest dragon fruit production in Vietnam, namely Binh Thuan, Tien Giang, and Tay Ninh (Table 1).

In each province, one field was selected to perform the experiment. Of the three fields, the field that was located in Binh Thuan province and specialized in cultivating white‐flesh pitaya was experimentally employed for three consecutive harvesting seasons. Meanwhile, the other two fields solely focused on cultivation of red‐flesh pitaya and were treated with experimental light bulbs only for two seasons.

White‐flesh dragon fruit plants were irradiated for 17 consecutive nights with each night involving around 11.5 hr of continuous irradiation starting from 18:30 p.m. to 6:00 a.m. For red‐flesh plants, artificial lighting commenced for 14 consecutive nights and the plants were continuously irradiated for 8 hr starting from 21:00 p.m. to 5:00 a.m. each night.

These timings and durations were selected based on current cultivation practices for commercial fruit production during off‐season periods. After stopping irradiation for 7–10 days, pitaya plants began to develop floral buds. Plants blossomed after 25–28 days since artificial lighting had ended.

Fruits fully developed and were harvested after 55–60 days following lighting stoppage. Experiments in each field were designed according to Randomized Completely Block Design with four replications for each treatment.

Each replication consisted of 16 plants. Of which, only 9 plants were monitored, and the remainder were neglected due to light interference between different lamps (Scheme 1). Watering, fertilization, use of chemicals, and tending procedure followed the guideline recommended by MARD (Ministry of Agriculture & Rural Development of Vietnam, 2006).

2.4 Statistical analysis

Excel software was used in producing descriptive statistical analysis. The experiments were calculated according to the method for field survey (Gomez & Gomez, 1984). Average monitoring parameters of each replication between experimental treatments were treated by the method of analysis of variance (ANOVA), then compared with Duncan’s test at confidence level p ≤ .05 by using SAS software 9.1.

3.1 Current status on bulb use for controlling off‐season flowering of dragon fruit

Preliminary interviewing results on some farmers and daytime measurement in the three provinces that have largest commercial dragon fruit cultivation area in Vietnam (Binh Thuan, Tien Giang, and Tay Ninh province) confirmed that, from September to March of the following year, dragon fruit plants cultivated in these areas were unable to bloom under natural conditions.

In fact, this is also the period in which the daytime is shorter than 12 hr. As a result, a majority of farming households adopted artificial irradiation as a measure to stimulate flowering of dragon fruit crops.

Prior to 2012, off‐season floral stimulation of dragon fruit often involved incandescent bulbs with the capacity of either 100 W, 75 W, or 60 W to supplement lighting during nighttime.

In the years 2013–2014, there was a shift in the use of bulbs for dragon fruit flowering induction toward incandescent bulbs with lower capacity (40 W) and compact fluorescent bulbs (20 W) to save electricity.

To gain further insights into farmers’ behavior in dragon fruit farming, a typical sample of 100 farming households located in the three regions was surveyed for their use of light bulbs in off‐season harvests. Table 2 showed the summarized survey results under some main indicators.

On the whole, it was showed that all surveyed household relied on artificial lighting to control blossoming of both white and red‐flesh dragon fruits. Incandescent light bulbs with power capacity of 40 W and 60 W were the dominantly used lamps for the floral stimulation while the households that used CFL bulbs (20 W) only accounted for around 10% of the total surveyed respondents.

Moreover, no other types of bulbs were recorded for dragon fruit irradiation purpose. Measures aimed to improve energy efficiency of irradiation, such as intermittent lighting were not adopted (Saradhuldhat et al., 2009).

There is a discrepancy in bulb use between Tay Ninh and other two provinces. To be specific, the percentage of households using CFL for dragon fruit irradiation in Tay Ninh was 32%, which was higher than that recorded in both Tien Giang and Binh Thuan (4%). Tay Ninh is also the province in which many households adopted the combinational irradiation with both incandescent and CFL light bulbs.

In addition, there were 5 out of 25 surveyed households in Tay Ninh reporting the use of incandescent bulbs with lower capacity of 40 W. Higher adoption of CFL in Tay Ninh could be explained by the recently emerged inclination to commercially grow dragon fruit in the province, as opposed to the other two regions where cultivation traditions have already been established; thus, the local farmers are hesitant to endorse new, improved growing techniques.

Another notable feature is that growers in Binh Thuan province started their artificial lighting earlier than those in Tay Ninh and Tien Giang. Specifically, the latest time in Binh Thuan in which irradiation started was 19:00 p.m. meanwhile 100% of farmers in Tien Giang and 92% of surveyed Tay Ninh growers commenced their lighting 2 hr later, at 21:00 p.m.

Two households who were able to begin early irradiation in Tay Ninh also reported the use of CFL bulbs, which are less energy consuming than incandescent counterparts, in their fields. This is reportedly due to delayed provision of electricity source reserved for agricultural activities and possibly reflects the inadequacy between grid capacity and production readiness of the electricity supplier and the recent, rising energy needs for dragon fruit cultivation in these areas.

Regarding lamp density, most growers allocated around 1,000–1,200 bulbs per hectare of cultivation. Higher distribution densities were only implemented in the harvest season from January to March where dragon fruits face difficulties in blooming.

Farmers also reported that, during this time period, white‐flesh dragon fruit plants were less susceptible to light‐induced floral stimulation than red‐flesh plants and in some cases in Binh Thuan, a denser distribution of more than 2,000 bulbs per hectare was required. With respect to lamp use specific to dragon fruit variety, there was no indication suggesting light bulb preference against certain variety.

In Binh Thuan, 45 out of 50 surveyed growers specializing in white‐flesh variety mainly utilized incandescent lamps for floral stimulation. Of which, incandescent bulbs with capacity of 60 W and 40 W accounted for nearly 90% and 6%, respectively. In two areas specialized in cultivating red‐flesh dragon fruit, the percentages are also similar in which only 4% of households adopted CFL lamps.

Overall, from the survey results, it is indicated that efficient control of off‐season blossoming is an important concern in commercial dragon fruit farming and that possible improvements should be called for in efforts to conserve energy for such processes. Currently, all dragon fruit farmers adopted artificial irradiation as the main measure for controlling off‐season flowering and ruled out other stimulation methods such as chemical intervention and foliar fertilization.

Moreover, the irradiation process was carried out based on farmers’ experience, thus lacking considerations on standardizing critical aspects including bulb type, height of bulb hanging, bulb density, irradiation duration, and implementation of intermittent lighting.

3.2 Spectrum analysis of currently used lamps for controlling off‐season flowering of dragon fruit

We collected three types of lamps commonly used by dragon fruit farmers (60 W incandescent, 40 W incandescent lamp, and 20 W domestic CFL lamp) and measured their emission spectra. The results are shown as in Figure 1.

Examination of the spectrum of the three lamps shows that two incandescent lamps exhibited great intensity in the far‐red and red region of their emission spectra, which coincides with the absorption spectrum of two main forms of phytochrome, namely P₆₆₀ and P₇₃₀ (Figure 2). Both of them are mutually transformable under appropriate irradiation and dark conditions and the ratio between the two determines blossoming state of the plant (Taiz & Zeiger, 1991).

The consistency between intensities of far‐red and red regions of the incandescent lamp emission spectra and maximum absorption wavelength in P₆₆₀ and P₇₃₀ absorption spectra also sides with the widespread use of incandescent lamps with the capacity ranging from 60 to 100 W for dragon fruit flowering control in Taiwan and in Thailand (Saradhuldhat et al., 2009; Yen & Chang, 1997).

However, the spectral regions that are not consistent with absorption spectra of phytochromes, including yellow, green, blue, and dark red areas, of incandescent lamps also exhibited high intensities. Most notably, the dark red region with the wavelength >730 nm, which is close to infrared light, displayed the greatest intensity compared with other regions, indicating possibly redundant heat and energy waste when the bulbs are used for floral stimulation.

The spectrum of the 20 W domestic CFL lamp (Figure 1c) displayed high intensity in the green and yellow light, which is engineered to suit the conventional lighting for human eye. The CFL lamp also has fair intensity in the far‐red and red region but is comparatively lower than that in the absorption spectra. To further highlight the difference in red and far‐red intensities between the two types of lamps, we measured the photon density of the two emission regions. The results are summarized in Table 3.

In terms of PPFD, the 60 W incandescent type exhibited the highest value, at 44.5 µmol m⁻² s⁻¹, which is twofold higher than that of 40 W counterpart and of the 20 W CFL. PPFD of the 40 W incandescent and 20 W CFL bulb fell within a narrow range from 21.4 to 22.4 µmol m⁻² s⁻¹, suggesting that they have similar photosynthetic effects. However, analysis results of photon density in the red and far‐red regions revealed that two incandescent lamps had different capacities in terms of controlling phytochrome transformation. To be specific, the 60 W incandescent lamp gave the highest total PFD (71.3 µmol m⁻² s⁻¹), followed by the 40 W incandescent lamp (38.0 µmol m⁻² s⁻¹). The 20 W CFL lamps usually have total PFD at the lowest value (9.69 µmol m⁻² s⁻¹).

Clearly, currently used lamps all had emission spectra focused on red and far‐red regions. Of which, two incandescent lamps showed that the density of the far‐red and red regions were proportional to the bulb capacity and the CFL lamp usually have much lower red and far‐red photon density than incandescent lamps.

As a result, the use of domestic CFL lamps might induce weaker floral stimulation effect on dragon fruit cactus than incandescent counterparts, even at similar power capacities.

This is in part illustrated by the previous survey results where only 9% of surveyed farmers with CFL irradiation reported the productivity of higher than 10 kg per plant, as opposed to 92% of growers using incandescent lamps reported the same yield. Current results suggest possible improvements in stimulation efficiency could be achieved by manipulating spectral intensity of CFL lamps toward increased far‐red and far‐red photon density.

3.3 Proposed CFL lamps and field experiment results on their effects on some growth indicators of dragon fruits

We obtained various improved CFL specimen lamps from Rang Dong Light Source & Vacuum Flask Joint Stock Company, one of leading manufacturers of light bulbs in Vietnam.

All bulbs had the power capacity of 20 W and were specifically designed with increased intensity in red and far‐red regions and reduced green and yellow intensity in their emission spectra.

After measuring their spectra, we selected three CFL bulbs with highest total PFD and used them in the following field experiment. The spectra and analysis results are shown as in Figure 3 and Table 4.

Three improved CFL bulbs were used in field experiments conducted in three main fields located in three typical provinces with commercial dragon fruit farming activities. The experiments spanned across multiple harvest seasons of the year from 2013–2014 and were conducted on the two varieties of dragon fruit. Incandescent lamp with the capacity of 60 W was used as the control. The results are presented in Table 5.

Flowering indicators of dragon fruits in the experiment 2 are considerably lower than those in the experiment 1. This is possibly due to the lower temperature in this harvesting season (24.4–24.6°C) compared with the last season (26.9–27.5°C) and is consistent with the experiments of Kim et al. (2009) and of Pin and Nilsson (2012) conducted on facultative long‐day plants.

However, in current experiment, CFL bulbs displayed better floral stimulation than incandescent bulbs. To be specific, fruit yields of plants of plants in the treatment 2 and 3 (4.1 and 3.8, respectively) are statistically equal and higher than yields of control and treatment 1 (1.3 and 1.5, respectively).

The experiment 3 was carried out at the same location and under climatic conditions similar to the experiment 1. The experiment results again confirmed the stimulus efficiencies of newly designed light bulbs in the experiment 1 in which plants treated with the incandescent bulbs and CFL bulb 2 showed higher number of floral stems, buds, and fruits per plant compared with those of other plants.

Strikingly, plants illuminated with the control and treatment 2 displayed exceptionally large number of bubs per plant (60.4 and 59.4, respectively), approximately 46% higher than plants treated with other treatments and were therefore subjected to bud trimming (to around 25–30 buds/plant) to ensure productivity. Ultimately, trimmed plants showed high fruit yield and were among best white‐flesh dragon fruits plants in terms of yield per plant throughout the three examined harvesting seasons.

Some interesting observations could be made from experiment conducted on red‐flesh dragon fruit plants in Tien Giang province (experiments 4 and 5). Firstly, dragon fruit plants treated with the incandescent light (control) displayed higher number of floral stems, buds, and in turn fruits per plant than those treated with CFL bulbs and even higher than those indicators of the control group in the experiment 3.

This suggests that the influence of lamp type on the floral stimulation efficiency is more profound in red‐flesh dragon fruit than in white‐flesh plants and that the development from bud to flowers of the red‐flesh variety is expedited compared with the white‐flesh counterpart under incandescent irradiation.

To date, the mechanism explaining for lower supplemented irradiation required for flowering of the red‐flesh dragon fruit has not been clearly elaborated. Nevertheless, current findings on variety‐specific response to artificial irradiation of dragon fruit are in agreement with previous experimental studies asserting that off‐season flowering of red‐flesh dragon fruit could be more easily stimulated than of white‐flesh variety (Be et al., 2014; Yen & Chang, 1997).

Secondly, the CFL treatment 2 performed equally well in comparison with the incandescent light in terms of plant productivity and flowering induction capability in the third harvesting season (experiment 5).

To be specific, the number of buds, fruits, and productivity between control and treatment 2 plants displayed no statistical differences. In addition, unlike experiment 4, all plants in this experiment did not require trimming and virtually all buds of dragon fruit plants developed into fruits. This result may carry implication that controlling the number of buds is essential in the light‐induced stimulation process for off‐season fruiting.

Similar to experiment 2, experiment 6 was conducted under colder growing condition (24.9–25.9°C), resulting in comparatively poor growth and productivity. The treatment 2 gave plants with higher number of buds than other groups. In some plants in the control group, no budding was observed.

Despite that, productivities and bud numbers of the control and treatment groups in this experiment are still higher than those in the experiment 2, confirming that red‐flesh dragon fruit is more responsive to floral stimulation by artificial lighting. Both thermal and light sensitivity of the flowering process in red‐flesh dragon fruit plant have been confirmed by a previous study where a temperature of at least 15°C was recommended for night lighting for the red‐flesh cultivar (Jiang & Yang, 2015).

Results of experiment 7 generally confirmed the findings of the experiment 5. Both experiments were carried out on the same cultivar and at the same time period. Control incandescent bulb and treatment 2 exerted better floral induction effect on the plant than other lights.

To summarize, the floral stimulation performance of the CFL light bulb of treatment 2 has been shown to be either equal or better than that of the incandescent light bulb through results of field experiments. In six out of seven experiments, treatments 1 and 3 pale in comparison with treatment 2 when it comes to bud‐stimulating capability and productivity.

One exception was experiment 4 where treatments 1 and 3 showed comparable floral stimulation capability to that of the treatment 2 and the control bulb was able to result in white‐flesh dragon fruits with marginally better yield per plant than the treatment 2 (15.1 vs. 12.9).

Given that the tested CFL lamps have power capacity almost threefold lower than the control, adoption of improved CFL lamps is recommended in energy‐scarce areas with intensive dragon fruit farming activities.

In fact, treatment 2 has been commercialized under the code name of CFL‐20 W NN R for flowering stimulation of agricultural crops and ornamental plants in various provinces including Binh Thuan, Long An, Tien Giang, Tay Ninh, Lam Dong, and Ha Noi city since 2016. The product earned good acceptance from farmers due to its affordability and short payback period.