Prediction of Differences in the Blooming Dates between Japanese Apricot ‘Nanko’ and Pollinizer Cultivars Using Development Rate Models

Abstract

The selection of appropriate pollinizer cultivars is essential for stable fruit production of Japanese apricot because this species has many self-incompatible cultivars. In this study, the chilling responses of flower buds during endodormancy and the heat responses of flower buds during ecodormancy were quantified as development rates (DVRs) in ‘Kotsubu-nanko’, ‘Hakuo’, and ‘NK14’, three pollinizer cultivars of the leading cultivar ‘Nanko’ in Wakayama Prefecture. Approximating functions of DVRs for both endodormancy release and ecodormancy release were obtained on the basis of chilling and heat exposure tests on young trees with various combinations of temperatures and periods. These analyses demonstrated that temperatures over 15°C were only effective for endodormancy release of ‘Hakuo’ buds. Compared with the buds of ‘Kotsubu-nanko’ and ‘Hakuo’, those of ‘NK14’ were less sensitive to 15–20°C during the ecodormancy stage. Blooming date prediction models for these cultivars were constructed using the DVR values. Then, the applicability of the three cultivars as pollinizers for ‘Nanko’ was evaluated based on whether their predicted blooming times overlapped with that of ‘Nanko’. The models were optimized by adjusting the threshold of blooming percentages and initial points of heat accumulation during ecodormancy release to achieve the smallest differences between predicted and observed blooming dates (root mean squared error = 3.72–5.90). Simulations under different temperatures revealed the most suitable pollinizer cultivar for ‘Nanko’ going forward. Our predictions indicate that, in the simulated warmer conditions, the blooming date of ‘Kotsubu-nanko’ will be delayed 10 or more days compared with that of ‘Nanko’, and larger differences between the blooming dates of ‘Hakuo’ and ‘Nanko’ will occur under lower and higher temperatures. The blooming period of ‘NK14’ was predicted to remain stably synchronized with that of ‘Nanko’ under a range of simulated warmer and colder temperature conditions. These results showed that ‘NK14’ is an appropriate pollinizer for stable fruit production of ‘Nanko’ in the future, and highlight the importance of blooming time assessments based on the temperature responses of flower buds.

Introduction

Japanese apricot (Prunus mume Sieb. et Zucc.) is one of the most popular trees in Japan because it produces ornamental flowers as well as fruit (Mega et al., 1988). It typically blooms from early February to mid-March in a cultivar- and region-dependent manner, and it is the earliest to bloom among temperate deciduous tree species. Because most cultivars of Japanese apricot, including ‘Nanko’, a primary cultivar for fruit in Japan, are self-incompatible and rely on flower-visiting insects for pollination, the weather conditions at blooming time can affect fruit bearing and crop yields (Hasebe, 1980).

To stabilize fruit production, appropriate pollinizer cultivars must be selected and interplanted within orchards (Watanabe, 1984a). For successful fertilization, pollinizer cultivars should have several traits, such as male fertility, cross-compatibility, and a blooming period that overlaps with that of the main cultivar. In Wakayama Prefecture, the largest production region for Japanese apricot in Japan, ‘Kotsubu-nanko’ and ‘Hakuo’ (a small-fruit cultivar; locally referred to as ‘Koume’) are frequently planted as pollinizer cultivars of ‘Nanko’ (Yaegaki, 2013). Additionally, the self-compatible cultivar ‘NK14’ (a progeny of the cross between ‘Nanko’ and ‘Kensaki’) has recently been bred and introduced into ‘Nanko’ orchards for both fruit production and as a pollinizer (Negoro et al., 2009). The cross-compatibility of the three cultivars with ‘Nanko’ has been confirmed in conventional cultivation by local producers. However, the differences in their blooming periods may be larger in the future as a result of climate change, and the blooming period, especially that of ‘Kotsubu-nanko’, has already been observed to greatly differ from that of ‘Nanko’ depending on the season.

The selection of pollinizer cultivars with appropriate blooming periods has relied on observations of their flowering in the past. However, it is important to forecast the blooming time in the future because it is likely to change under altered climate conditions that will occur with climate change. The genotype-specific blooming time is determined by the chilling requirement during endodormancy (dormancy phase in which internal factors cause bud outgrowth inhibition), and the heat requirement during ecodormancy (dormancy phase in which external factors such as unfavorable environmental factors cause bud outgrowth inhibition) (Faust et al., 1997). To estimate the blooming time of deciduous ornamental trees such as Prunus yedoensis, Rhododendron kaempferi, Wisteria floribunda, and Hydrangea macrophylla, the number of days transformed to standard temperature (DTS) method has been used (Aono and Omoto, 1990, 1992). Although modified DTS models that include endodormancy completion dates have been developed to improve the precision of this method, the applicability of DTS varies depending on regions and years for deciduous woody plants, including fruit tree species (Aono and Moriya, 2003; Aono and Sato, 1996; Ono and Konno, 1999). Focusing on dormancy release, three chilling quantification models that convert the ambient temperature during the endodormancy phase of buds into chilling values have been proposed; chill hour (CH), chill unit (CU), and chill portion (CP) (Fishman et al., 1987a, b; Richardson et al., 1974; Weinberger, 1950). Compared with the CH and CU models, the CP model was found to provide the most accurate prediction of the endodormancy release date in high-chill cultivars of Japanese apricot in Nanjing, China (Gao et al., 2012). However, the suitability of the CP model for predicting the endodormancy release date in other production regions and cultivars is unclear.

The development rate (DVR) model has been used in previous studies to predict the blooming dates of temperate deciduous fruit tree species (Sugiura and Honjo, 1997; Sugiura et al., 2010). In the DVR model, the amount of dormancy release per hour at given temperatures is calculated and the development index (DVI) is defined as the cumulative value of DVR. This model assumes that dormancy release is complete when DVI = 1 (Horie and Nakagawa, 1990). DVR values can be determined by dormancy release tests under various combinations of temperatures and periods, as well as regression analysis using observed climate and phenotype data (Takezawa and Tamura, 1991). The latter method can be performed without experiments using plants, but requires long-term records of temperatures and blooming or dormancy-release dates. We previously reported the possibility of blooming date prediction based on DVRs calculated from experimental data for young trees of the Japanese apricot ‘Nanko’, for which the dormancy stage was divided into endodormancy and ecodormancy phases (Kitamura et al., 2017). Because our DVR model reflects the physiological responses of trees to temperature, we speculated that it could be used to predict blooming dates under climate change conditions.

In this study, we constructed blooming date prediction models for three pollinizer cultivars introduced into Wakayama Prefecture, ‘Kotsubu-nanko’, ‘Hakuo’, and ‘NK14’. We calculated the DVR values based on chilling and heating tests using young potted trees, and constructed prediction models combining continuous accumulation models of two DVR models; DVR for chilling requirement fulfillment for endodormancy release and for heat requirement fulfillment for ecodormancy release. The suitability of these three cultivars as pollinizers, now and in the future, was evaluated by comparing their predicted blooming dates with that of ‘Nanko’ under a range of temperature conditions.

Materials and Methods

Plant materials

Three-year-old trees of Japanese apricot ‘Kotsubu-nanko’, ‘Hakuo’, and ‘NK14’ grown in 25-L pots at the Japanese Apricot Laboratory (Minabe, Hidaka, Wakayama, Japan; 33.49°N, 135.21°E) were used in this study. All trees were pruned to maintain one upright trunk and one-year-old shoots bearing flower buds, and were kept in a greenhouse with the temperature maintained above 15°C to avoid exposure to low temperatures until use.

Modeling chilling responses during endodormancy release

The surveys were conducted from the 2014–15 to 2018–19 seasons. Trees were manually defoliated in early November and exposed to various combinations of chilling temperatures (2°C to 20°C) and periods (240 h to 960 h) in darkened phytotrons (Table S1). One tree per treatment and per season was used, and the number of flower buds per tree ranged from 35 to 292, 92 to 430, and 47 to 179 in ‘Kotsubu-nanko’, ‘Hakuo’ and ‘NK14’, respectively. The values from the combinations surveyed over two seasons were averaged. After chilling exposure, the trees were sequentially transferred to a greenhouse with the temperature maintained above 15°C. The blooming percentage of each tree was recorded every 2 or 3 days for 2 months. The relationships between final blooming percentages during the forcing test and the period of chilling exposure at each temperature were regressed using the following logistic function:

  

$$f\left(d_c\right)=100/\left\{1+\mathit{exp}\left[{\mathit{-\alpha }}_1\left(d_c-{\alpha }_2\right)\right]\right\}$$

where d_c, f(d_c), α₁, and α₂ indicate the chilling exposure period (h), estimated blooming percentage, relative increasing rate of blooming (an indicator of the increasing rate in blooming percentages in the logistic function), and chilling exposure period at the inflection point (i.e., the chilling period required for 50% endodormancy release), respectively. The parameters α₁ and α₂ were determined by the least-squares method. The DVR values during the endodormancy phase (DVR_endo) for each tested temperature were calculated as the reciprocals of chilling exposure periods expected to induce 80% blooming by the function above. The DVR_endo for the whole range of temperatures was approximated using cubic or quadratic functions. The degrees of approximation functions were determined based on the following criterion: the curves had single peaks within the range, exhibited positive values in DVR_endo, and included tested temperatures.

Modeling heat responses during ecodormancy release

The surveys were conducted in the 2020–21 season. Defoliated trees were preliminarily exposed to 5°C for 500 hours from early November in darkened phytotrons to fulfill the chilling requirement for endodormancy release. After chilling exposure, the trees were transferred to phytotrons set at 5°C, 10°C, 15°C, or 20°C with a 12-h light/12-h dark photoperiod. Three trees were used for each temperature treatment. The total number of flower buds per temperature treatment ranged from 300 to 1080, 158 to 1334, and 117 to 516 in ‘Kotsubu-nanko’, ‘Hakuo’ and ‘NK14’, respectively. Blooming percentages were recorded every 1 or 2 days, and relationships between heating periods and blooming percentages were regressed using the following function:

  

$$f\left(d_h\right)=100/\left\{1+\mathit{exp}\left[{\mathit{-\beta }}_1\left(d_h-{\beta }_2\right)\right]\right\}$$

where d_h, f(d_h), β₁, and β₂ indicate the heating period (h), estimated blooming percentage, relative increasing rate, and heating period at the inflection point (i.e., the period of heating required for a blooming rate of 50%), respectively. The parameters β₁ and β₂ were determined as described above. The DVR values during the ecodormancy phase (DVR_eco) were calculated as the reciprocals of the estimated heating period corresponding to the predetermined blooming percentages. The DVR_eco values for other temperatures were determined by interpolating between experimental data points. Those for < 5°C and > 20°C were extrapolated using the linear functions for 5–10°C and 15–20°C, respectively.

Optimization of blooming models

To verify the precision of the blooming models, predicted blooming dates were compared with observed blooming dates at the Japanese Apricot Laboratory from the 2014–15 to 2020–21 seasons (only from the 2016–17 to 2020–21 seasons for ‘NK14’ due to the lack of reliable blooming date data before the 2015–16 season). The blooming dates (approximately 20% blooming) of representative adult trees (> 15 years old) of ‘Kotsubu-nanko’, ‘Hakuo’, and ‘NK14’ within the field were recorded. Daily temperature data were obtained using meteorological equipment (E-734-00; YDK Technologies Co., Ltd., Tokyo, Japan). Blooming dates were predicted using the prediction program for ‘Nanko’ developed previously (Kitamura et al., 2020). In this model, the hourly temperatures were approximated by a sine function using the daily maximum temperature and minimum temperature on the following day. DVR_endo starts to accumulate from November 1^st, and the cumulative value of DVR_endo is referred to as DVI_endo. DVR_eco starts to accumulate from a predetermined DVI_endo value (0.5 in the ‘Nanko’ model), and the predicted blooming date is defined as the date when DVI_eco = 1.0 for the first time. In the predictions for the pollinizer cultivars, the coefficients of regression function for both DVR_endo and DVR_eco were replaced with those for each cultivar determined above. When DVR_endo and DVR_eco calculated using these functions were negative values, the DVRs were regarded as zero.

The following two values were verified to optimize the models for each cultivar. First, the definition of blooming was adjusted within the range of 10% to 50% at 10 percentage intervals to calculate the heating period required for blooming during ecodormancy. Second, DVI_endo values as the initial point of DVR_eco accumulation were adjusted from 0.1 to 1.0 at 0.05 intervals. For each combination of these two values, the root mean squared error (RMSE) was calculated using the following equation:

  

$$\mathit{RMSE}=\sqrt{\frac{1}{n}\sum {\left(D_{\mathit{pred}}-D_{\mathit{obs}}\right)}^2}$$

where n, D_pred, and D_obs indicate the number of survey seasons, predicted blooming date, and observed blooming date, respectively.

Simulation of blooming dates under warmer and cooler temperatures

Daily maximum and minimum temperatures recorded at the Japanese Apricot Laboratory from 2005–06 to 2020–21 were used. Mean values of both maximum and minimum temperatures during the 16 seasons were separately calculated. Additionally, modified temperature data consisting of the mean values described above ±2.0°C with 0.1°C intervals were prepared. Predicted blooming dates of ‘Kotsubu-nanko’, ‘Hakuo’, ‘NK14’, and ‘Nanko’ (Kitamura et al., 2020) were obtained using the temperature dataset and prediction programs optimized for each cultivar as described above.

Results and Discussion

Modeling the temperature responses of three pollinizer cultivars

The blooming percentages of the ‘Kotsubu-nanko’, ‘Hakuo’, and ‘NK14’ trees under forcing conditions after exposure to chilling at various temperatures for various periods are shown in Table 1. More than 80% of flower buds bloomed after exposure for up to 960 h at 2°C to 12°C in all cultivars. After 960 h of exposure to 15°C, the blooming percentage of ‘Hakuo’ reached 77.8%, while those of ‘Kotsubu-nanko’ (26.5%) and ‘NK14’ (1.3%) were extremely low. For 2 to 12°C in ‘Kotsubu-nanko’ and ‘NK14’, and 2 to 15°C in ‘Hakuo’, where high percentages of blooming were induced, the parameters of regression curves for the relationships between chilling periods and blooming percentages were determined (Table 1). The regression curves generally fitted well, although there were some discrepancies with experimental values of blooming percentages, especially when the chilling periods were long and temperatures were high in ‘Kotsubu-nanko’ and ‘Hakuo’ (Fig. S1). The DVR_endo values were calculated as the reciprocals of the estimated chilling periods required to induce 80% blooming (Table 1), because clear differences in regression curves among temperatures were observed at this level. The DVR_endo values at temperatures of 15°C in ‘Kotsubu-nanko’ and ‘NK14’, and 20°C in ‘Hakuo’ were regarded as zero because of their extremely weak effects on endodormancy release (Table 1). Including these data indicating DVR_endo = 0 as experimental values similar to other temperature data, approximation curves for the relationships between temperatures and DVR_endo in the three cultivars were created (Fig. 1). All three curves had a local maximum near 6°C (5.69°C, 6.26°C, and 5.88°C in ‘Kotsubu-nanko’, ‘Hakuo’, and ‘NK14’, respectively) as the most effective temperature for endodormancy release. The DVR_endo values under high temperatures were higher in ‘Hakuo’ than in the other two cultivars and the previous data for ‘Nanko’ (Kitamura et al., 2017), indicating that ‘Hakuo’ had a broader range of temperature sensitivity and a low chilling requirement for endodormancy release under ambient conditions.

Table 1

Blooming percentages under forcing conditions after chilling exposure and parameters and coefficients of determination of regression curves in ‘Kotsubu-nanko’, ‘Hakuo’, and ‘NK14’.

Fig. 1

Regression curves of relationships between chilling temperatures and DVR values during the endodormancy phase in ‘Kotsubu-nanko’, ‘Hakuo’, and ‘NK14’. DVRs were calculated as reciprocals of chilling periods required for endodormancy release in 80% of flower buds. DVR_endo = 7.840e–07 t³—4.545e–05 t² + 4.409e–04 t + 1.001e–03 for ‘Kotsubu-nanko’, 4.540e–07 t³—2.575e–05 t² + 2.691e–04 t + 1.300e–03 for ‘Hakuo’, and −2.453e–05 t² + 2.884e–04 t + 1.322e–03 for ‘NK14’, where t indicates temperature.

All trees exposed to heat after 500 h of chilling treatment at 5°C achieved almost 100% blooming, indicating that this chilling treatment was sufficient for endodormancy release in the tested cultivars. The parameters of regression curves for the blooming percentages in the three cultivars during heating were determined (Table 2). The regression curves between heating periods and blooming percentages in all tested cultivars and temperatures were also well-fitted to experimental values (Fig. S2). To optimize the blooming models, the reciprocals of the heating periods required for 10%, 20%, 30%, 40%, and 50% blooming were calculated using these regression functions (Fig. 2, all heating periods and coefficients are shown in Table S2). Similar patterns of DVR_eco values showing continuously steep rises at higher temperatures were observed in ‘Kotsubu-nanko’ and ‘Hakuo’. In contrast, the DVR_eco of ‘NK14’ had a lower rate of increase above 15°C, indicative of relatively slow ecodormancy release at higher temperatures. This trend in ‘NK14’ was also observed in ‘Nanko’ (Kitamura et al., 2017). Due to global warming, it is anticipated that there will be an increase in the likelihood of encountering relatively high-temperature zones of 15°C or higher in winter. The similar heat responses of ‘Nanko’ and ‘NK14’ suggest that they would show similar changes in blooming time under the conditions of climate change during the ecodormancy period, typically February and March.

Table 2

Parameters and coefficients of determination of regression curves for relationships between blooming percentages and heating periods after endodormancy release.

Fig. 2

Relationships between DVR values and temperature in the ecodormancy phase in ‘Kotsubu-nanko’, ‘Hakuo’, and ‘NK14’. DVRs were calculated as reciprocals of heating periods required for 10% to 50% blooming as the definition of ecodormancy release.

Verification of blooming models

To optimize the blooming models, the RMSE values that were calculated with various blooming percentages to indicate completion of ecodormancy release (BP), and DVI_endo values as initial points of DVR_eco accumulation (IP), were evaluated. BP means the definitions of DVI_eco = 1 in the DVR_eco calculation using young potted trees (Fig. 2; Table S2). The minimum RMSEs were detected where the two parameters were set at BP: 50% and IP: 0.45 for ‘Kotsubu-nanko’ (RMSE = 3.72), BP: 30% and IP: 0.65 for ‘Hakuo’ (RMSE = 5.90), and BP: 20% and IP: 0.5 for ‘NK14’ (RMSE = 5.44) (Fig. 3). The optimized BP and IP were different, cultivar-dependent values. We propose the reason why these values differ across cultivars. Blooming time in Japanese apricot is affected by physiological factors such as tree vigor and shoot length (Taniguchi, 2006; Watanabe, 1984b). While the young trees used in this study uniformly consisted of a single trunk and bearing shoots of similar length, adult Japanese apricot trees have various types of shoots and complex architecture that differs among cultivars. Additionally, the structures of the potted trees used in this study were variable among tested cultivars. For example, because the trees of ‘Kotsubu-nanko’ had a lot of short shoots, the blooming rates were relatively high compared to the adult trees (approximately 50% blooming in potted trees corresponded to 20% blooming in the adult trees in the field). In contrast, those of ‘Hakuo’ and ‘NK14’ had a larger proportion of middle to long shoots. The ‘Nanko’ model, in which young potted trees mainly produced short to middle shoots, was optimized at 40% BP (Kitamura et al., 2017). Therefore, the variations in optimum BP among cultivars may reflect differences in blooming rates between young potted trees and adult trees.

Fig. 3

RSME values for each combination of blooming definitions and DVI_endo as initial points of DVR_eco accumulation in ‘Kotsubu-nanko’, ‘Hakuo’, and ‘NK14’. RMSEs were calculated using temperature data from the 2014–15 to 2020–21 seasons for ‘Kotsubu-nanko’ and ‘Hakuo’, and from the 2016–17 to 2020–21 seasons for ‘NK14’.

In our models, the large differences between observed and predicted blooming dates, especially in the case of delayed blooming in all tested cultivars; at most 7-, 9-, and 11-day differences for ‘Kotsubu-nanko’, ‘Hakuo’, and ‘NK14’, respectively (Fig. 4). Predictions of blooming dates based on DVR values calculated by statistical analyses have been reported for other deciduous fruit trees such as the Japanese pear ‘Kosui’, peach ‘Akatsuki’, apple ‘Fuji’, and grape ‘Delaware’ (Adachi et al., 2018; Kamimori et al., 2020; Oya, 2006). Highly accurate predictions with RMSE values of approximately 2 were obtained in those studies. Compared to other deciduous fruit trees, Japanese apricot generally blooms over a long period (for approximately 1 month at most under sunny conditions), even in one individual tree, and blooming at a certain level continues for a relatively longer time. This lower resolution of blooming time observations may result in higher RMSE values. Additionally, the blooming time of Japanese apricot showed high annual variations. The standard deviations (SDs) of blooming dates observed during the survey years were 9.53, 7.57, and 9.93 days for ‘Kotsubu-nanko’ (Jan. 28^th to Feb. 26^th), ‘Hakuo’ (Jan. 31^st to Feb. 19^th), and ‘NK14’ (Jan. 24^th to Feb. 16^th), respectively. In previous studies, the applicability of prediction models has been judged by comparisons between SDs over the years and RMSEs (Adachi et al., 2018; Kamimori et al., 2020; Takezawa et al., 1989). Our models’ RMSEs were lower than the SDs of blooming dates. Considering the large differences between the earliest and latest blooming date (19–29 days, Fig. 4), these prediction results with the range of RMSEs can be also utilized for precise blooming time forecasting and to prepare pollination management, such as the pasturing of honey bees in production orchards, at a suitable time. Therefore, we propose that the constructed models in this study are suitable for practical use for Japanese apricot blooming.

Fig. 4

Relationships between observed blooming date and that predicted using optimized models for ‘Kotsubu-nanko’, ‘Hakuo’, and ‘NK14’ from 2014–15 to 2020–21 (from 2016–17 to 2020–21 for ‘NK14’). Blooming dates are expressed as number of days from January 1^st.

Suitability of ‘Kotsubu-nanko’, ‘Hakuo’, and ‘NK14’ as pollinizer cultivars for ‘Nanko’ under changing climate conditions

The blooming dates for ‘Nanko’, ‘Kotsubu-nanko’, ‘Hakuo’, and ‘NK14’ predicted using optimized models with mean temperatures (MT) during the past 16 years at the Japanese Apricot Laboratory, Wakayama, Japan were February 14^th, February 25^th, February 13^th, and February 15^th, respectively (Fig. 5). Regarding pollinizer cultivars for ‘Nanko’, the predicted blooming date of ‘NK14’ was closest to that of ‘Nanko’: only −1 to +2 days at the range of −2°C to +2°C of MT, indicating that ‘NK14’ is the most suitable and sustainable pollinizer for ‘Nanko’ among the cultivars tested in this study. The simulations suggested that ‘Hakuo’ will bloom later than ‘Nanko’ (at most +5 days) under colder temperatures than MT, whereas ‘Hakuo’ will bloom earlier than ‘Nanko’ (at most −7 days) under higher temperatures than MT. The results suggest that the blooming time of ‘Hakuo’, one of the most popular pollinizer cultivars for ‘Nanko’, will show a larger difference from that of ‘Nanko’ under extreme climate conditions, especially under the higher temperatures accompanying global warming. In the simulations, the blooming dates of ‘Kotsubu-nanko’ showed constant +10 to +14 day differences from those of ‘Nanko’ in the tested range of temperatures. Although a single Japanese apricot tree can continue to bloom for a month, there is a risk of decreased pollination rates between ‘Kotsubu-nanko’ and ‘Nanko’ because of early petal drop during rainfall and windstorms. Considering this risk, the interplanting of multiple pollinizer cultivars is recommended.

Fig. 5

Predicted blooming dates under a range of temperature conditions. 0°C on the horizontal axis corresponds to the mean temperature in the 2005–06 to 2020–21 seasons recorded at the Japanese Apricot Laboratory, Wakayama, Japan. Blooming dates are expressed as number of days from January 1^st.

The potential of various cultivars as pollinizers has been well-studied in terms of genotypic compatibility based on S-haplotype variation in Prunus fruit tree species such as sweet cherry, Japanese plum, and Japanese apricot (Amemiya et al., 2006; Guerra et al., 2020; Hayashi et al., 2003; Takemura et al., 2023; Yaegaki et al., 2001). Artificial pollination methods are not popular in Japanese apricot cultivation. Despite the importance of overlapping blooming periods between the main cultivar and pollinizer cultivars, the pollinizers have been selected based only on past observations of blooming time. This study is the first to demonstrate the suitability of pollinizers using blooming models that include future predictions. Previous studies have predicted changes in suitable production areas for fruit tree crops under global warming scenarios (Sugiura et al., 2012), and noted that insufficiently low temperatures could hinder the fulfillment of chilling requirements (Campoy et al., 2011; Luedeling et al., 2011). The symptoms of a flowering disorder arising from deficient endodormancy completion have already been detected in Japanese pear in southwestern areas of Japan (Tominaga et al., 2022). Our results suggest that there will be substantial differences in blooming times between some conventional Japanese apricot pollinizer cultivars in Wakayama Prefecture and ‘Nanko’ under milder winter conditions. Therefore, not only the current blooming time, but also the temperature responses of flower buds should be considered in the evaluation and breeding of pollinizers. The construction of DVR models for dormancy release requires a number of trees to be tested and temperature-controlled rooms. Simple and precise methods to quantify temperature responses during dormancy stages need to be developed so that the potential blooming time of newly bred cultivars that lack long-term observation data can be predicted accurately.

In conclusion, chilling responses during endodormancy and heat responses during ecodormancy in flower buds of three major pollinizer cultivars of the Japanese apricot ‘Nanko’ were experimentally quantified as DVR values. Models to predict blooming dates were constructed by sequentially combining the accumulation of the two DVRs and minimizing the errors based on observed data. Predictions of blooming dates in the future revealed the suitability of the current pollinizers under various temperature conditions. These findings will contribute to the sustainable fruit production of Japanese apricot.

Acknowledgements

We are grateful to researchers and technicians at the Japanese Apricot Laboratory for maintaining the numerous plant materials and recording blooming dates in the experimental orchard. We thank Jennifer Smith, PhD, from Edanz (https://jp.edanz.com/ac) for editing a draft of this manuscript.

Literature Cited

•  Adachi,  Y.,  H.  Shimura,  M.  Nukada,  Y.  Saito,  A.  Kuwana and  M.  Abe. 2018. Verification of the developmental rate (DVR) model of fruit tree on global warming. Bull. Fukushima Agric. Technol. Cent. 9: 11–21 (In Japanese with English abstract).
•  Amemiya,  H.,  T.  Tezuka and  T.  Sakurai. 2006. Genotyping of S-RNase gene in sweet cherry cultivars by PCR and crossing. Bull. Yamanashi Fruit Tree Exp. Stn. 11: 29–34 (In Japanese with English summary).
•  Aono,  Y. and  C.  Moriya. 2003. A generalized model to estimate flowering for cherry tree (Prunus yedoensis) considering both processes of endodormancy completion and development. J. Agric. Meteorol. 59: 165–177 (In Japanese with English abstract).
•  Aono,  Y. and  Y.  Omoto. 1990. A simplified method for estimation of blooming date for the cherry employing DTS. J. Agric. Meteorol. 46: 147–151 (In Japanese with English abstract).
•  Aono,  Y. and  Y.  Omoto. 1992. Estimation of blooming date of Rhododendron kaempferi, Wisteria floribunda and Hydrangea macrophylla. J. Agric. Meteorol. 47: 233–240 (In Japanese with English abstract).
•  Aono,  Y. and  K.  Sato. 1996. Estimation of flowering date for Japanese apricot tree (Prunus mume) in western Japan considering both processes of rest break and development. J. Agric. Meteorol. 52: 125–134 (In Japanese with English abstract).
•  Campoy,  J. A.,  D.  Ruiz and  J.  Egea. 2011. Dormancy in temperate fruit trees in a global warming context: A review. Sci. Hortic. 130: 357–372.
•  Faust,  M.,  A.  Erez,  L. J.  Rowland,  S. Y.  Wang and  H. A.  Norman. 1997. Bud dormancy in perennial fruit trees: Physiological basis for dormancy induction, maintenance, and release. HortScience 32: 623–629.
•  Fishman,  S.,  A.  Erez and  G. A.  Couvillon. 1987a. The temperature dependence of dormancy breaking in plants: Mathematical analysis of a two-step model involving a cooperative transition. J. Theor. Biol. 124: 473–483.
•  Fishman,  S.,  A.  Erez and  G. A.  Couvillon. 1987b. The temperature dependence of dormancy breaking in plants: Computer simulation of processes studied under controlled temperature. J. Theor. Biol. 126: 309–321.
•  Gao,  Z.,  W.  Zhuang,  L.  Wang,  J.  Shao,  X.  Luo,  B.  Cai and  Z.  Zhang. 2012. Evaluation of chilling and heat requirements in Japanese apricot with three models. HortScience 47: 1826–1831.
•  Guerra,  M. E.,  B. I.  Guerrero,  C.  Casadomet and  J.  Rodrigo. 2020. Self-(in)compatibility, S-RNase allele identification, and selection of pollinizers in new Japanese plum-type cultivars. Sci. Hortic. 261: 109022. DOI: 10.1016/j.scienta.2019.109022.
• Hasebe, H. 1980. Ume no hinsyu to saibai (In Japanese). Nobunkyo, Tokyo.
•  Hayashi,  K.,  K.  Negoro,  K.  Iwamoto,  M.  Hosohira and  H.  Sugai. 2003. Typing of S-RNase gene in Japanese apricot (prunus mume) by PCR. Bull. Wakayama Res. Cent. Agric. Forest. Fish. 5: 67–73 (In Japanese).
•  Horie,  T. and  H.  Nakagawa. 1990. Modelling and prediction of developmental process in rice. Japan. J. Crop Sci. 59: 687–695 (In Japanese and English abstract).
•  Kamimori,  M.,  Y.  Miwa,  T.  Isobe and  A.  Hosomi. 2020. Models for predicting dates of budding and full bloom of ‘Delaware’ grape based on the daily average temperature. Hort. Res. (Japan) 19: 175–181 (In Japanese with English abstract).
•  Kitamura,  Y.,  K.  Numaguchi and  Y.  Naka. 2020. Application of a simple blooming time prediction program using development rate models for temperature requirements of Japanese apricot ‘Nanko’ flower buds. Bull. Wakayama Pref. Exp. Stn. Agric. Forest. Fish. 8: 79–84 (In Japanese).
•  Kitamura,  Y.,  H.  Yamane,  A.  Yukimori,  H.  Shimo,  K.  Numaguchi and  R.  Tao. 2017. Blooming date predictions based on Japanese apricot ‘Nanko’ flower bud responses to temperatures during dormancy. HortScience 52: 366–370.
•  Luedeling,  E.,  E. H.  Girvetz,  M. A.  Semenov and  P. H.  Brown. 2011. Climate change affects winter chill for temperate fruit and nut trees. PLoS ONE 6: e20155. DOI: 10.1371/journal.pone.0020155.
• Mega, K., E. Tomita, S. Kitamura, S. Saito and S. Mizukami. 1988. Ume. p. 289–300 (In Japanese). In: T. Aoba et al. (eds.). The grand dictionary of horticulture. Shogakukan, Tokyo.
•  Negoro,  K.,  K.  Hayashi and  K.  Iwamoto. 2009. Breeding of self-compatible new Japanese apricot cultivar ‘NK14’. Hort. Res. (Japan) 8 (Suppl. 1): 311 (In Japanese).
•  Ono,  S. and  T.  Konno. 1999. Estimation of flowering date and temperature characteristics of fruit trees by DTS method. Japan. Agric. Res. Q. 33: 105–108.
•  Oya,  Y. 2006. Prediction method for anthesis of Japanese pears based on weather habit reactions. Bull. Tochigi Agric. Exp. Stn. 58: 7–16 (In Japanese with English summary).
•  Richardson,  E. A.,  S. D.  Seely and  D. R.  Walker. 1974. A model for estimating the completion of rest for ‘Redhaven’ and ‘Elberta’ peach trees. HortScience 9: 331–332.
•  Sugiura,  T. and  H.  Honjo. 1997. The effects of temperature on endodormancy completion in Japanese pear (Pyrus pyrifolia Nakai) and modeling the relationship. J. Agric. Meteorol. 53: 285–290 (In Japanese with English abstract).
•  Sugiura,  T.,  D.  Sakamoto,  T.  Asakura and  H.  Sugiura. 2010. The relationship between temperature and effect on endodormancy completion in the flower bud of ‘Hakuho’ peach. J. Agric. Meteorol. 66: 173–179 (In Japanese with English abstract).
•  Sugiura,  T.,  H.  Sumida,  S.  Yokoyama and  H.  Ono. 2012. Overview of recent effects of global warming on agricultural production in Japan. Japan. Agric. Res. Q. 46: 7–13.
•  Takemura,  Y.,  K.  Tochimoto,  M.  Kitamura,  H.  Moroto,  M.  Sakata,  T.  Yoshida,  T.  Inamoto,  S.  Takazawa,  A.  Okamoto,  N.  Machida,  Y.  Nakano,  K.  Hirai and  F.  Tamura. 2023. Potential of Myrobalan plum as a new pollinizer for Japanese plum cultivars. Hort. J. 92: 22–29.
•  Takezawa,  K. and  Y.  Tamura. 1991. Use of smoothing splines to estimate rates of development. Agric. Forest Meteorol. 57: 129–145.
•  Takezawa,  K.,  Y.  Tamura and  Y.  Ono. 1989. A nonparametric method to estimate developmental index. J. Agric. Meteorol. 45: 151–154 (In Japanese with English abstract).
• Taniguchi, M. 2006. Ume no sagyo benricho. Nobunkyo, Tokyo, Japan.
•  Tominaga,  A.,  A.  Ito,  T.  Sugiura and  H.  Yamane. 2022. How is global warming affecting fruit tree blooming? “Flowering (dormancy) disorder” in Japanese pear (Pyrus pyrifolia) as a case study. Front. Plant Sci. 12: 787638. DOI: 10.3389/fpls.2021.787638.
•  Watanabe,  S. 1984a. Ume no seisan antei to hinsitsu kaizen. Nougyo Gijutsu 39: 363–368 (In Japanese).
• Watanabe, S. 1984b. Kaikaki no souban to sonogono seiiku. p. 3–6 (In Japanese). In: Ume. Kihon gijutsu hen. Nogyo gijutsu taikei. Nobunkyo, Tokyo.
•  Weinberger,  J. H. 1950. Chilling requirements of peach varieties. Proc. Amer. Soc. Hort. Sci. 56: 122–128.
•  Yaegaki,  H. 2013. Production, marketing, processing and breeding objective of Japanese apricot. Bull. NARO Inst. Fruit Tree Sci. 16: 1–12 (In Japanese with English abstract).
•  Yaegaki,  H.,  T.  Shimada,  T.  Moriguchi,  H.  Hayama,  T.  Haji and  M.  Yamaguchi. 2001. Molecular characterization of S-RNase genes and S-genotypes in the Japanese apricot (Prunus mume Sieb. et Zucc.). Sex. Plant Reprod. 13: 251–257.