2024 Volume 93 Issue 2 Pages 101-113
Three rose varieties, ‘Meivildo’, ‘Meikatana’, and ‘Korcut0122’ were grown using an “arching” method for three years. The relationship between temperature and light intensity, and the yield and quality of cut flowers, were analyzed. Flowering flushes were observed 18 times in ‘Meivildo’ and ‘Meikatana’ and 16 times in ‘Korcut0122’ during the experimental period. In three varieties, significant negative linear regressions between the interval of flowering flush (growth period, GP) and the mean temperature per GP were observed. Significant positive correlations and linear regression were observed between the number of cut flowers per plant and the mean temperature in ‘Meivildo’ and ‘Meikatana’, and the total light integral per GP (TLI) in ‘Korcut0122’. In three varieties, significant positive correlations and linear regressions were observed between the total cut flower weight per plant, and the mean daily light integral per GP (DLI) and TLI. Highly significant positive correlations and linear regressions were observed between the daily gain in flower weight (DGW) of cut flowers per plant (the total cut flower weight divided by the number of days of GP) and the mean DLI in all three varieties. In ‘Meivildo’ and ‘Meikatana’, cut flower weight per stem had significant negative correlations and linear regression with mean temperature and positive ones with TLI, while ‘Korcut0122’ showed no significant correlation with the three environmental factors. Significant positive correlations and linear regressions existed between the specific cut flower weight (the cut flower weight per stem length) and TLI in all three varieties, and negative ones existed between the specific cut flower weight and mean temperature in ‘Meivildo’ and ‘Meikatana’. The relationship between the results in this experiment and previous reports on the relation between the environment and cut flower yield quality are discussed. Finally, we present the significance of the method used in this experiment for 1) prediction of flowering and shipping of cut flowers, 2) evaluation of differences in characteristics among the rose varieties, and 3) contribution to the development of a growth model.
In current cut flower production, there is a need to provide accurate harvest and shipping information in response to customer demand. To meet this requirement, it is necessary to develop “data-based techniques” to predict harvest date, number, and quality of cut flowers. These “data-based techniques” are approaches that have been developed in recent years to optimize cultivation management and operations based on analysis of monitored environmental data. Two contrasting views are considered necessary for this: a “microscopic” perspective, which focuses on the behavior of plant tissues or individual plants, and a “macroscopic” perspective, which summarizes the overall plant community.
The rose plant is a woody, perennial plant and is one of the major cut flowers in the horticulture market. In the commercial production of cut roses, cultivation and harvest are continued for several years after planting, while other herbaceous plants are harvested one time only in a cropping season. Interestingly, it is assumed that the quality and yield are affected by seasonal environmental changes in other plants, and a few studies reported long-term quantitative analysis of these effects on rose plants. Plaut et al. (1979) in a 6-month study of four rose varieties and Khosh-Khui and George (1977) in a 1-year study of the variety ‘Baccara’ investigated the effects of light intensity and duration on growth and reproductive characteristics. Cockshull (1975), using six months of cultivation data, also found that rose flower production follows seasonal changes in light intensity. Plaut and Zieslin (1977) and Plaut et al. (1979) noted that the seasonal variation in rose flower quality may be due to temperature, light, or both, and may also be influenced by the moisture status of the plant.
Much of the analysis on the influence of environmental factors on rose cut flower production has been conducted at high-latitudes in Europe and North America in the 20th century (Zieslin and Mor, 1990). The low temperature and low-light intensity features in these regions are very different from those in Japan (Khosh-Khui and George, 1977). A large number of studies on supplemental lighting, especially in Europe (Zieslin and Mor, 1990), are related to the low-light intensity in this region, especially in winter. Global rose production has now shifted to low-latitude highlands with minor annual environmental changes, such as in India, Ecuador, Colombia, and Kenya (International Association of Horticultural Producers, 2019). Japanese rose production is still strongly affected by seasonal environmental variability as opposed to conditions in high-latitude and low-latitude regions.
This study was conducted to demonstrate the possibility of quantitatively understanding the relationship between environmental conditions, yield, and quality of cut roses at the plant community level. Here, harvested cut flowers of three varieties with different characteristics, cultured for a 3-year period in an intermediate area of Japan, were investigated. Further, the relationship between previous reports on the environment and cut flower yield quality is discussed, and finally, the significance and application of this analysis method are suggested.
Three rose varieties ‘Meivildo’ (trade name in Japan: ‘Yves Piaget’), ‘Meikatana’ (‘Samurai 08’), and ‘Korcut0122’ (‘Brillante’) were used as experimental varieties. Two plastic greenhouses (length 10 m × width 5 m × height 2.7 m) at NARO (Fujimoto, Tsukuba, Japan) were used as the growing sites. Four growing benches (length 1,840 mm × width 940 mm × height 700 mm) were arranged longitudinally, and three rows of benches were arranged in the transverse direction. On July 5, 2017, young plants grafted on Rosa odorata as rootstock were obtained from a supplier and were planted in 21 cm-diameter plastic pots filled with growing medium (Royal Culture Soil; Tachikawa Heiwa Noen, Tochigi, Japan). The pots were arranged in the longitudinal direction of the growing benches, with 32 plants of each variety allocated to a row of the pots (11 cm between pot centers). The arrangement of varieties and individual plants in the two greenhouse chambers was identical. From the date of planting to September 27, 2017, one layer of white cheesecloth curtain with a shading rate of 20% was installed in all areas to avoid excessive light intensity.
For the light intensity experimental plots, each of two greenhouses were divided in half to create a high-light and a low-light zone with different degrees of shading. From September 27, 2017, the shading curtain was removed from the high-light zone, while the 20% shading curtain remained in place in the low-light zone. From June 1 to September 25, 2018; from May 30 to September 23, 2019; and from May 29, 2020 to the end of the experiment (August 20, 2020), one layer of the curtain with 20% shading was applied to the high-light zone and two layers to the low-light zone. For the other periods, no shading curtain was installed in the high-light zone and one layer of the curtain was installed in the low-light zone.
A humidification device (Cool Pescon; H. IKEUCHI & Co., Ltd., Osaka, Japan) was installed in one of the two greenhouses. Humidification began on October 9, 2018, after installation and was set to spray at 1-minute intervals when the relative humidity fell below 70%.
The nutrient solution for irrigation was made from precipitation B with OAT House No. 1 and No. 2 fertilizer (OAT Agrio Co., Ltd., Tokyo, Japan). The EC was controlled at 1.0 for the high-temperature season and 1.5 for the low temperature season, and the pH was adjusted to 6.0.
Heating temperature in the two greenhouses was set at 15°C until December 27, 2019 and 10°C after December 28, 2019 (Fig. 1).
Temperature, light intensity, and CO2 concentration in daytime during the experiment.
The temperature, light intensity, relative humidity, and CO2 concentration in the houses were measured using copper-constantan thermocouples, quantum sensors (IKS-27; Koito Industry Co., Yokohama, Japan), humidity sensors (Model RHT-2; Sensatec Co. Ltd., Kyoto, Japan), and CO2 transmitters (GMT220; Vaisala, Helsinki, Finland). These sensors were placed in the middle row of the benches, in the center of each light zone, and about 30 cm above the canopy of bent shoots. The temperature sensors were placed inside a ventilating pipe. The data from these sensors were recorded by a data logger (XL124; Yokogawa Electric Co., Tokyo, Japan).
Shoot management, cut flower harvest, and collection of cut flower dataThe rose plants were cultured by a simple “arching” method. The first shoot bending for “photosynthetic branches community” was performed on August 24, 2017 for ‘Meikatana’ and ‘Korcut0122’, and for ‘Meivildo’ on August 31, 2017. Thereafter, shoots emerging from the base of the plant (“crown”) were bent successively. Harvest of cut flowers began on November 10, 2017, and continued until August 20, 2020. However, weak branches that were considered to have no commercial value as cut flowers were sequentially bent and used to maintain the photosynthetic branch community.
Harvest was performed by cutting the flowering shoots at the site of emergence from the crown. Stem length (from the cut end to the bottom of the flower), maximum leaf length attached to stem, and weight per cut flower (including the flower organs) were measured at harvest. After harvest on March 8, 2019, six flowering stages of each cut flower were recorded for 0 (hard buds) to 5 (full flowering) and values for stem length and weight per cut flower were corrected by flowering stage (details of the method are omitted). From July 23 to August 10, 2018, and from July 29 to August 17, 2019, harvest was suspended with removal of the developing shoots due to extremely high temperatures in the greenhouse chambers (Fig. 1) and extreme deterioration of flowering shoot growth.
Determination of flowering and growth periodsThe number of cut flowers was counted for each variety at ten or eleven days from the early, middle, and late period of the month to determine the duration of individual flush. The average harvest date of cut flowers per flush was defined as the “peak flowering date”, and then the growth period (GP) was defined as the time between the adjacent peak flowering dates.
Parameters for environmental factors, and yield and quality of cut flowersEnvironmental parameters, daily mean temperature, the difference in daily mean day and night temperature (DIF), mean daily total light integral (DLI), and total light integral (TLI) per each GP were calculated (Table 1). In the present analysis, the temperature was measured every 10 min and defined as the daily average temperature. It has been reported that in rose cut flower production, productivity and cut flower quality depends on the mean temperature, and fluctuations in temperature have no effect on the average temperature (Buwalda et al., 2000).
Mutual correlations among the three environmental factors.
Parameters of cut flower yield were calculated as the number of cut flowers per plant, total cut flower weight per plant, and daily gain in cut flower weight (DGW: total cut flower weight per plant/number of days of GP) for each GP. As parameters of cut flower quality, weight per cut flower, stem length, and specific cut flower weight (weight per cut flower/stem length) were adopted.
Correlation and regression analysis of the parametersCriteria were established to discuss the relationship between environmental parameters and cut flower parameters. First, a single correlation analysis was performed between each environmental parameter, and each cut flower yield and quality parameter. Next, the regression line was calculated by single linear regression analysis, based on the assumption that a causal relationship between the parameters existed when the correlation was significant.
Y = a × X + b
where Y is the parameter of yield or quality of the cut flower, X is the environmental parameter, and a and b are constants. Significance was determined based on a P value < 0.05 level, and a P < 0.01 level was considered highly significant. The first GPs, when the starting point of growth cannot be defined, and the GPs after the harvest break period in summer (the 6th and 13th GPs of ‘Meivildo’ and ‘Meikatana’, and the 7th and 12th GPs of ‘Korcut0122’) were excluded from the analysis.
The novelty of this report lies in the fact that it was a long-term study conducted over a period of three years under actual growing conditions at a grower’s site to evaluate the effects of temperature and light intensity on yield and quality in terms of seasonal variations. Zieslin et al. (1973) stated that the productivity of rose plants depends on the variability of various yield components such as the number of lateral buds, the position of mature flower buds, the abortion of flower buds, and the formation of renewal buds, which are influenced by the variety, the presence or absence of grafted rootstock, and seasonal factors. In this report, some of the correlations or linear regressions between the four environmental parameters and the eight cut flower parameters were significant, while others were not. The DIF was not significantly correlated with any cut flower parameters, except for DGW in ‘Meikatana’.
It has been reported that low growing temperature and low-light intensity (or without supplemental lighting) or shading can cause flower abortion and lower yields (Carpenter and Anderson, 1972; Cockshull, 1975; Khosh-Khui and George, 1977; Zieslin and Halevy, 1975a, b), but in the present experiment, flower abortion was hardly observed. This may be because Tsukuba City in Japan, where the experiment was conducted, has a higher light intensity even in winter than high-latitude regions such as Europe.
Environmental conditionsFigure 1 shows the environmental conditions in the growth rooms during the experiment. Large seasonal variations in temperature and solar radiation were noted in the greenhouse chambers. The average light intensity in the high-light zone was about 1.6 times higher than that in the low-light zone. The daily mean CO2 concentration in the rooms during the daytime (6:00–18:00) was 300–400 ppm, with a minimum value of 100–300 ppm, and the daily minimum value was low during the winter season when the room was closed. Correlations among environmental factors are shown in Table 1. There was a significant correlation between the mean temperature and mean DLI for each growing season (correlation coefficient of 0.3121), and a high degree of correlation between the mean DLI and TLI (correlation coefficient of 0.8759). On the other hand, the correlation between the mean temperature and TLI was not significant. The mean DIF showed a highly significant correlation with mean DLI (correlation coefficient of 0.5664) and TLI (correlation coefficient of 0.5403), but was not significantly correlated with mean temperature. In the following discussion, no “confounding” is assumed for these parameters, but we carefully considered those that are highly interrelated.
Yield and quality parameters for each variety and their distributionIn this experiment, 3,724 cut flowers of ‘Meivildo’, 4,471 of ‘Meikatana’, and 4,568 of ‘Korcut0122’ were harvested and studied. Figure 2 shows the median and distribution of yield and cut flower quality parameters, as well as the coefficient of variation (standard deviation/mean), for each variety throughout the experimental period. The total number and total cut flower weight per GP in the yield parameters were similar in ‘Meikatana’ and ‘Korcut0122’, and lower in ‘Meivildo’. Stem length and cut flower weight per stem were the highest in ‘Meikatana’, followed by ‘Korcut0122’, and the lowest in ‘Meivildo’. ‘Meikatana’ had the heaviest specific cut flower weight, followed by ‘Korcut0122’, and ‘Meivildo’ had the lightest. Leaf length was the longest in ‘Meikatana’, followed by ‘Meivildo’, and the shortest in ‘Korcut0122’.
Distribution of yield and quality parameters. MV: ‘Meivildo’, MK: ‘Meikatana’, and KC: ‘Korcut0122’. The box-and-whisker plots represent medians and quartiles. The bars represent maximum and minimum values (excluding outliers). The numbers indicate coefficients of variation. The letters indicate mean separation by Tukey’s multiple test (P < 0.05 level).
The coefficient of variation of all three varieties, except for leaf length, was greater than 0.2, and the coefficient of variation for cut flower weight per stem was particularly high at around 0.4. Here, it may be emphasized that in such cases it is difficult to present a clear-cut analytical result. Although we have not been able to find any data that discusses the characteristics of this type of data for cut roses, we assume that the occurrence of dominance-recession relationships among shoots or individual plants during the long growing season in cut roses is a major cause of these events. However, potted miniature roses differ from those in cut flower production in this respect (Yu et al., 2006), as they have a short growing season and are easy to cultivate and each plant can be managed appropriately. The difficulty in the study of cut roses is that we need to discuss them within these limitations.
Effect of humidityIn this experiment, it was not possible to establish a clear difference in humidity between the humidified and non-humidified plants, and no clear trends in yield or quality differences between the two sections were obtained (data omitted). Humidity control using the humidification device did not work well because the experimental greenhouse was a single building with a small area, and the effects of temperature and light intensity were too large to mask the effect of humidity. Therefore, data from the humidified and non-humidified treatments were included in the subsequent analysis. Plaut et al. (1979) reported that the effect of evaporative cooling reduced yield, but improved cut flower quality in summer.
Flowering flush, GP, and temperatureThe interval of flowering flush is important in production as follows: 1) it determines the time of shipping to market and, 2) it determines the yield. For the latter, Zieslin et al. (1973) noted that determining the number of growth cycles per year or season affects flower production. The annual yield of rose cut flowers is the product of two factors, i.e., the flowering cycle and the number (or weight) of cut flowers per flowering cycle.
Flowering flushes were observed 18 times in ‘Meivildo’ and ‘Meikatana’, and 16 times in ‘Korcut0122’ during the experimental period (Fig. 3). High negative correlations with coefficients of −0.8730 (‘Meivildo’), −0.6474 (‘Meikatana’), and −0.9065 (‘Korcut0122’) were observed between days of GP and the mean temperature in all three varieties (Fig. 4).
The flowering flushes (closed circles) and their corresponding growth periods (GP) (bars). The growth periods after the harvest break period in summer (black bars) were excluded from the correlation and regression analysis.
Relationships between the length of the growth period and mean temperature. ** indicates significance at the P < 0.01 level. The arrows in ‘Meikatana’ are explained in the text.
Significant linear regressions between GP and temperature were observed in all three varieties (Fig. 4). The coefficients of determination were high in ‘Meivildo’ (0.7621) and ‘Korcut0122’ (0.8218). On the other hand, the coefficient of determination of ‘Meikatana’ was lower at 0.4192, which may be attributed to the values of the 15th and 17th growth periods indicated by the arrows in Figure 4B. These were long periods of two years after the start of cultivation, reflecting the fact that the peak of flowering was obscured at this time, making it difficult to identify a definite GP. The slopes of the regression lines (Fig. 4) in ‘Meivildo’ and ‘Meikatana’ were close (−3.162 and −2.919, respectively), whereas ‘Korcut0122’ had a larger slope (−5.576), indicating that the effect of the peak-to-peak interval on temperature was more pronounced in ‘Korcut0122’.
There are many reports on the influence of temperature on the initiation of flowering shoot activity, and flower bud formation and development. Higher temperatures accelerated the rate of budding and subsequent development (Buwalda et al., 2000; Horridge and Cockshull, 1974; Kim and Lee, 2002; Marcelis-van Acker, 1995; Moe and Kristoffersen, 1969). However, in our preliminary analysis, comparing the correlations between mean daily temperature, minimum daily temperature or maximum daily temperature and GPs, the correlation with mean daily temperature was the highest.
The results for light intensity differ among previous reports. Long days under low-light (10 μmol·m−2·s−1 PPFD) affected the number of days to flowering (Carpenter and Anderson, 1972; Tsujita, 1982). Moe (1972) reported that long days with white fluorescent lamps (relative high-light intensity of 10,000 lx) suppressed bud break, but shortened the days to flowering. The effect of reducing the number of days to flowering was correlated with the total amount of irradiance (Armitage and Tsujita, 1979). A previous study reported only a mild effect of irradiation (Wiseley and Lindstrom, 1972). The effect of supplemental lighting on shortening the time to complete flowering of rose flowers interacted with temperature and was more pronounced at higher temperatures than at lower temperatures (Moe and Kristoffersen, 1969). However, as also noted by Moe (1972), there was a correlation between mean temperature and mean DLI under normal growing conditions (Table 1), making it difficult to separate these two factors. Therefore, although we cannot rule out the influence of light intensity, we conclude that the temperature determines the flowering cycle due to the markedly high correlation coefficients for mean temperature.
Cut flower yield—number of cut flowers per plant and GPYield can be considered in two ways: number-based or weight-based. As shown in Figure 2, ‘Meivildo’ was smaller than the other two varieties in both number and weight, and ‘Meikatana’ and ‘Korcut0122’ were comparable in number, while ‘Meikatana’ was the higher yielding variety in terms of weight.
Between the number of cut flowers per plant and mean temperature, a high (correlation coefficient of 0.7255 in ‘Meivildo’; Fig. 5A) or a moderate (0.5267 in ‘Meikatana’; Fig. 5B) positive correlation were observed, except for ‘Korcut0122’ (Fig. 5C). On the other hand, only ‘Korcut0122’ showed a moderate, but significant, positive correlation (correlation coefficient of 0.5282) with the TLI (Fig. 5F).
Relationships between the number of cut flowers per plant with each growth period (GP) and mean temperature (A, B, C) and total light integral (TLI) (D, E, F). White symbols are those grown in the high-light zones, black symbols are those grown in the low-light zones, and gray symbols include both. ** and * indicate significance at the P < 0.01 and P < 0.05 level, respectively. NS indicates no significance at the P ≥ 0.05 level.
Although the temperature has been shown to affect the number of cut flowers (Toudou et al., 2012; Zieslin et al., 1978), there has been little discussion on whether this is because it accelerates the flowering cycle or increases the number of flowers per flowering cycle. The results of this experiment showed that the number of cut flowers tended to increase at higher temperatures in ‘Meivildo’ and ‘Meikatana’ (Fig. 5A, B), but further quantitative experiments in more strictly controlled environments are needed.
Numerous reports have been published on the effect of light intensity on the number of cut flowers. Supplemental lighting had the effect of increasing production, but its main effect was to stimulate axillary bud growth (Carpenter and Anderson, 1972; Carpenter and Rodriguez, 1971; Cockshull, 1975; Khosh-Khui and George, 1977). In an interesting report, Khayat and Zieslin (1982) showed that exposure to light was necessary for the emergence of basal shoots.
Cut flower yield—cut flower weight per GP and DGWUnderstandably, weight-based yield, i.e., total cut flower weight per GP, was positively correlated with the environmental parameters related to light intensity. Moderate positive correlations (correlation coefficient of 0.6710 in ‘Meivildo’, 0.5194 in ‘Meikatana’, and 0.5820 in ‘Korcut0122’) were observed between the mean DLI and the weight of cut flowers per plant for all three varieties (Fig. 6A, B, C). Moderately positive correlations were also observed between the weight of cut flowers per plant and the TLI in ‘Meivildo’ (correlation coefficient of 0.5789; Fig. 6D) and ‘Meikatana’ (0.3716; Fig. 6E), and a high positive correlation of 0.7994 was observed in ‘Korcut0122’ (Fig. 6F).
Relationships between the total cut flower weight per plant with each growth period and mean daily light integral (DLI) (A, B, C) and total light integral (TLI) (D, E, F). White symbols are those grown in the high-light zones, black symbols are those grown in the low-light zones. ** and * indicate significance at the P < 0.01 and P < 0.05 level, respectively.
There were significant positive correlations between DGW and mean DLI, (Fig. 7G, H, I) and TLI (Fig. 7J, K, L) in the three varieties. Significant correlations between DGW and mean temperature were observed in ‘Meivildo’ (correlation coefficient of 0.6464) and ‘Meikatana’ (0.4275), but were not observed in ‘Korcut0122’ (Fig. 7A, B, C). The correlation coefficients between DGW and mean DLI (‘Meivildo’: 0.6951, ‘Meikatana’: 0.6682, ‘Korcut0122’: 0.7766) were the highest among the three environmental parameters in all varieties (Fig. 7G, H, I). The results of this experiment are consistent with the results of Kim and Lee (2002, 2008), that is, the efficiency of dry matter accumulation increases in rose plants that are given sufficient light. The DGW was used as an indicator of the productivity of the plant, showing the fresh weight of cut flowers per day. These results were expected in terms of material production due to photosynthesis.
Relationships between the daily gain of cut flower weight with each growth period and mean temperature (A, B, C), mean difference of daily mean day and night temperature (DIF) (D, E, F), mean daily light integral (DLI) (G,H, I) and total light integral (TLI) (J, K, L). White symbols are those grown in the high-light zones, black symbols are those grown in the low-light zones, and gray symbols include both. ** and * indicate significance at the P < 0.01 and P < 0.05 level, respectively. NS indicates no significance at the P ≥ 0.05 level.
The positive correlation with DGW in ‘Meikatana’ was the only significant effect of mean DIF on growth parameters in this experiment (Fig. 7E). We previously observed that low-night temperature cultivation increased dry matter accumulation in lilies (Inamoto et al., 2016). In roses, Dieleman et al. (2007) showed that day-night temperature differences affect the dry matter weight and starch content of cut flowers, and Zieslin et al. (1987) showed that low-night temperature settings cause yield loss, but the trend varies among varieties. Mortensen and Moe (1992) showed that low diurnal and high night temperatures resulted in shorter cut flower lengths in growing experiments at different diurnal and night temperatures. However, Vogelezang et al. (2000) showed that day-night temperature differences only have a small effect on cut flower quality. Compared to their experiment in which plants were grown under artificially controlled temperatures, we grew plants under natural day-night temperatures; the variation was not so stable, but did not have large day-night temperature differences. More detailed studies on day-night temperature differences and dry matter accumulation are anticipated in the near future.
There are many reports on photosynthesis in roses, and the light compensation point has been evaluated to be about 30–70 μmol·m−2·s−1 PPFD, depending on the variety and season (Jiao et al., 1988; Lieth and Pasian, 1990). The light saturation was evaluated to be 600–700 μmol·m−2·s−1 PPFD (Lieth and Pasian, 1990; Pasian and Lieth, 1989) based on individual leaf-based measurements. However, Righetti et al. (2007) pointed out that, in general, community photosynthesis does not coincide with individual leaf photosynthesis. Our evaluation based on community-based measurements showed no saturation even at 1,000 μmol·m−2·s−1 PPFD (unpublished data) higher than those individual leaf-based reports. Kim and Lieth (2001) also found the photosynthetic saturation of whole rose plants was 900 μmol·m−2·s−1 PPFD.
Positive correlations were also observed between temperature and DGW in ‘Meivildo’ and ‘Meikatana’ (Fig. 7A, B). In the measured and mathematical model of Lieth and Pasian (1990), the temperature at which maximum photosynthesis was obtained in the individual rose leaves was as high as 30°C. Bozarth et al. (1982) showed that the change in the photosynthetic rate of roses was not so pronounced in the range of 20–30°C. Ushio et al. (2008) reported that when roses were grown at low temperatures (20°C/15°C day/night), their photosynthetic rate was initially lower than that of roses grown at high temperatures (30°C/25°C), but that this rate eventually reversed. In contrast with these reports, it is assumed that there was no inhibition of photosynthesis at the higher end of the temperature range encountered by the plants in this experiment because the correlations between temperature and DGW were positive.
Cut flower quality—stem lengthThere was little seasonal variation in the stem length, i.e., the absolute correlation coefficients between temperature, mean DLI, and TLI and stem length in all three varieties were very low (less than 0.3) and not significant. Previous result on the effects of environmental conditions on stem length were mixed, indicating the instability of environmental influences. De Vries et al. (1982) reported that in seedlings, stem length at flowering is shortened by increased temperature. Dieleman and Meinen (2007) found that the day-night temperature difference also affects stem length, with greater temperature differences resulting in shorter stem length. However, in this experiment, no significant correlation was found between the mean DIF and stem length, and this may be due to the smaller day-night temperature difference as compared to the experiment conducted by Dieleman and Meinen (2007). Interestingly, Mortensen and Moe (1995) concluded that stem length cannot be controlled by setting temperature differences in miniature roses. Concerning light, no change in cut flower length (Armitage and Tsujita, 1979) or a slight decrease (Tsujita, 1982; White and Richter, 1973) have been observed with relatively intense supplemental lighting.
Cut flower quality—cut flower weight per stemAs for the weight of individual cut flowers, the results differed from that of the stem length, i.e., seasonal variations were observed in the weight of cut flowers per stem. Moderate (P < 0.05) negative correlations were observed between cut flower per stem and mean temperature in ‘Meivildo’ (correlation coefficient of −0.4025; Fig. 8A) and ‘Meikatana’ (−0.4573; Fig. 8B). Between the cut flower weight per stem and the TLI, a moderate positive correlation in ‘Meivildo’ (correlation coefficient of 0.4061; Fig. 8D) and a weak one in ‘Meikatana’ (0.3635; Fig. 8E) were observed. In ‘Korcut0122’, there was no significant correlation between cut flower weight per stem and any of the other environmental parameters at the P < 0.05 level (Fig. 8C, F).
Relationships between the cut flower weight per stem with each growth period and mean temperature (A, B, C) and total light integral (TLI) (D, E, F). White symbols are those grown in the high-light zones, black symbols are those grown in the low-light zones, and gray symbols include both. * indicates significance at the P < 0.05. NS indicates no significance at the P ≥ 0.05 level.
Regarding light, Armitage and Tsujita (1979) and Tsujita (1982) reported that supplemental lighting had little effect on the fresh weight of rose cut flowers. More interestingly, Carpenter and Rodriguez (1971) and Moe (1972) reported that the number of rose cut flowers increased with supplemental lighting, but it was accompanied by an increase in the number or proportion of cut flowers with shorter stems. With reference to these reports, in our experiment, the decrease in cut flower weight per stem in response to an increase in temperature in ‘Meivildo’ and ‘Meikatana’ (Fig. 8A, B) was related to the distribution of dry matter (photosynthetic product) to individual cut flower stems, as with the increase in the number of cut flowers (Fig. 5A, B). Another idea is to balance the time from bud emergence to flowering with the rate of dry matter translocation, which is an issue to be considered in the future using the “growth analysis method”.
Cut flower quality—specific cut flower weightIn all three varieties, with TLI, specific cut flower weight had significant positive correlations (correlation coefficient in ‘Meivildo’ was 0.4181, 0.3706 in ‘Meikatana’, and 0.4075 in ‘Korcut0122’), and linear regressions (Fig. 9D, E, F). In addition, moderate negative correlations and linear regressions were observed between mean temperature and specific cut flower weight in ‘Meivildo’ (correlation coefficient of −0.4307; Fig. 9A) and ‘Meikatana’ (correlation coefficient of −0.4582; Fig. 9B). Specific cut flower weight is considered an indicator that reflects the thickness and fullness of cut flower stems. Buwalda et al. (2000) reported that higher growing temperatures reduced the specific cut flower weight regardless of the light intensity. On the other hand, Tsujita (1982) showed that supplemental lighting increased the specific cut flower weight. The results supported the empirical reference that higher growing temperatures reduce the fullness of cut flowers, while higher light levels increase it.
Relationships between the specific cut flower weight with each growth period and mean temperature (A, B, C) and total light integral (TLI) (D, E, F). White symbols are those grown in the high-light zones, black symbols are those grown in the low-light zones, and gray symbols include both. * indicates significance at the P < 0.05. NS indicates no significance at the P ≥ 0.05 level.
Moderate significant (P < 0.05) negative correlations were observed between leaf length and mean temperature in ‘Meikatana’ (correlation coefficient of −0.4463; Fig. 10B) and ‘Korcut0122’ (−0.3954; Fig. 10C), but not in ‘Meivildo’ (−0.2125; Fig. 10A). No significant correlation was observed between leaf length and mean DLI, TLI, and mean DIF. For rose plants, Shin et al. (2001) reported that a high growing temperature resulted in a small leaf area. Photosynthesis of individual plants is the product of the photosynthetic rate per unit leaf area and leaf area of individual plants. Therefore, it is possible that smaller leaves under high temperatures affect the rate of photosynthesis and reduce productivity. However, the slope of the regression lines between temperature and leaf length (i.e., the degree of influence) was small (Fig. 10B, C), while the positive correlations between temperature and daily increase in cut flower weight in Figure 7A, B, and C suggest that such an effect is unlikely to be significant.
Relationships between the maximum leaf length per stem with each growth period and mean temperature. * indicates significance at the P < 0.05. NS indicates no significance at the P ≥ 0.05 level.
The significance of the method used in this experiment is summarized by the following three points: First, it allows the prediction of flowering and shipping. The highly significant linear regressions between the peak of flowering flush and growing temperature suggests that it is possible to predict or control the next flowering peak to a considerable degree. It is difficult to predict the precise harvest date and the number of cut flowers under such highly variable conditions; however, there is technological potential to combine new technologies such as video analysis to overcome these weak points. Moreover, as shown in Figure 2, although there is a large variation in the quality of cut roses, the statistical distribution data could be used to predict the quality of the product for shipping.
The second point is the importance of the evaluation of differences in characteristics among the rose varieties. It is likely that there are differences in response to environmental conditions among cultivated rose varieties, even if not as extreme as among wild species (Ueda et al., 2000). Although it is ideal to evaluate variety characteristics by conducting simultaneous trials under strictly controlled conditions, it is difficult to make quantitative comparisons of the characteristics of many varieties simultaneously. However, in this experiment, the collection and accumulation of data on harvest and quality during the actual cultivation phase was sufficient for evaluation. In fact, in this experiment, ‘Korcut0122’ was found to have different properties compared to the other two varieties as follows: 1) longer intervals between flowering flushes (Fig. 3). 2) a steeper slope of regression lines of the temperature versus interval of flowering flush (GP) than the other two varieties (Fig. 4). 3) the number of cut flowers was significantly affected by light intensity (Fig. 5F), not temperature (Fig. 5C), and 4) the regression slopes of the parameters of the weight-based yield component with an increase in light intensity were steeper (Figs. 6 and 7). These results may indicate that ‘Korcut0122’ is suitable for cultivation under higher temperatures and light intensity.
The third point is the contribution to the development of a growth model by providing supporting data. We are also developing a system dynamics model to predict the yield, quality, and flowering yield for “arching” roses using temperature and light intensity as input parameters. We require a basic dataset on the actual relationship between the environment, yield, and quality, that will clarify the “behavior” of rose plants in response to changes in the environmental conditions. The analysis in this paper uses statistical and static methods; this illustrated the relationship between environmental conditions, and yield and quality of roses clearly and visibly. On the other hand, it is limited in its explanation of the causal relationships and mechanisms. The environment factors are considered to have direct effects on material production by photosynthesis and an indirect effect through a time factor (i.e., number of days to flowering) for material accumulation; in such a case, a dynamic simulation model is useful. Previously, Inamoto et al. (2001a, b) demonstrated the effectiveness of dynamic simulation models in such cases in a report using tulips.
As mentioned in the introduction, Japan has a large seasonal variation in temperature unlike the major rose production areas, and this trend has recently been accelerated by the warming of the climate in summer and a reduction in heating temperatures to save energy in winter. In addition, we have reported in a previous paper that long-term observations have shown that seasonal changes in environmental factors significantly affect the stomatal aperture, transpiration characteristics, and flower longevity of ‘Asami Red’ rose cut flowers (In et al., 2007, 2009). Therefore, combined analysis of yield, visual quality, and flower longevity is needed to provide accurate supply and quality information to markets and consumers.
