Chilling and Heat Requirements of Temperate Stone Fruit Trees ( Prunus sp.)

: Stone fruit trees of genus Prunus , like other temperate woody species, need to accumulate a cultivar-speciﬁc amount of chilling during endodormancy, and of heat during ecodormancy to ﬂower properly in spring. Knowing the requirements of a cultivar can be critical in determining if it can be adapted to a particular area. Growers can use this information to anticipate the future performance of their orchards and the adaptation of new cultivars to their region. In this work, the available information on chilling- and heat-requirements of almond, apricot, plum, peach, and sweet cherry cultivars is reviewed. We pay special attention to the method used for the determination of breaking dormancy, the method used to quantify chilling and heat temperatures, and the place where experiments were conducted. The results reveal di ﬀ erent gaps in the information available, both in the lack of information of cultivars with unknown requirements and in the methodologies used. The main emerging challenges are the standardization of the conditions of each methodology and the search for biological markers for dormancy. These will help to deal with the growing number of new cultivars and the reduction of winter cold in many areas due to global warming.


Introduction
Temperate stone fruits belong to the genus Prunus in the Rosaceae and produce a fruit called drupe, whose seed is covered by the woody endocarp which in turn is covered by the endocarp. In most cultivated Prunus species, the edible part of the fruit is the endocarp, which includes the fleshy pulp (mesocarp) and skin (exocarp) such as apricot (P. armeniaca L.), European plum (P. domestica L.), Japanese apricot (P. mume Siebold and Zucc.), Japanese plum (P. salicina Lindl.), peach (P. persica L. Batsch), sour cherry (P. cerasus L.) and sweet cherry (P. avium L.) [1]. On the other hand, in almond (P. dulcis (Mill.) D.A. Webb), the edible part of the fruit is the seed. The annual global stone fruit production reached in 2017 more than 47 million t in 7.3 million ha [2]. The most cultivated species are peach (P. persica L. Batsch) (24.6 million t in 1.5 million ha), plum (including European and Japanese plum) (11.7 million t in 2.6 million ha), apricot (4.2 million t in 0.5 million ha), sweet cherry [30]. The model is based on a possible biological process in which a thermally unstable precursor would lead to the accumulation of a factor in the buds. This process would follow the Arrhenius law that fits the mathematical relationship between temperature and the rate of a chemical reaction.
Once dormancy is predicted allowing the quantification of chilling, it is also needed to quantified warm temperatures after dormancy for flowering to occur (Figure 1b.2). The modelization of warm temperatures was early developed in agriculture to predict the different phenological stages of crops [41]. The combination of a chilling model with a heat model to predict flowering was firstly described with the combination of the Utah model with the Growing Degree Hours (GDH) quantification [25], and then this combination was also applied with the other chilling models. A GDH is defined as one hour at 1 °C above the base temperature (4.5 °C), this linearly progresses until the upper limit (25° C) [13]. One of the main drawbacks of using these models is the necessity of hourly temperature data records, whose availability is limited. Thus, equivalent models have been developed based on maximum and minimum temperatures [42].  The experimental determination of dormancy consists of evaluating when the buds recover the capacity to grow (Figure 1a.1). This is usually performed by transferring shoots into a growth chamber sequentially during winter, thus after different chilling exposures. Shoots remain a certain period in the warm conditions, and then bud growth is evaluated. This approach has been widely used from Agronomy 2020, 10, 409 4 of 32 early [24] to recent studies [31] that determine the chilling requirements of the cultivars. Furthermore, these experiments serve as a base for physiological studies on dormancy [6].
The statistical approach estimates the date of chilling fulfillment based on a long series of phenological observations (flowering dates) and relating them with the previous temperature records (Figure 1a.2). Tabuenca et al. established a statistical methodology by calculating the correlation coefficients between the maximum, minimum and mean temperatures of certain time periods and flowering dates in apple, apricot, cherry, peach, pear, plum [32], and almonds [33]. Then, Alonso et al. determined the temperature requirements correlations between the flowering dates of almond cultivars and daily minimum, mean and maximum temperatures calculated as the mean of the surrounding 5, 10, 15 . . . until 30 days, with a set of data from 7 years. The endo-dormancy to eco-dormancy transition was considered to be when the significant correlation coefficients change from being mainly positive to be mainly negative [26]. Ashcroft et al. firstly estimated chilling and heat accumulation of peaches [29] based on when the chilling and heat accumulation presented the least squared residuals methods [25]. A new approach has been recently developed based on the statistical analysis of long-term phenological records and temperature series. The application of partial least squares (PLS) regression leads to the estimation of the agroclimatic requirements. PLS regression is especially applicable when the number of independent variables (daily temperatures, 365 data per year) substantially exceeds the number of dependent variables (one flowering date per cultivar and year). The results of these analyses include the model coefficients and variable-importance-in-the-projection. Significant positive model coefficients correspond with the chilling accumulation, endo-dormancy, while negative coefficients correspond with the heat accumulation, eco-dormancy [34]. It was initially applied in sweet cherry [34] and later in other fruit trees as almond [35,36], pistachio [19], apricot [27] or apple [36].

Temperature Based Models for Phenology Prediction
Three main models are currently used in agriculture to quantify chilling over the dormancy period [37]. They were developed in peach: the Chilling Hours model [28], the Utah model [29], and the Dynamic model [30] (Figure 1b.1). The Chilling Hours model was developed in the early fifties of the 20 th century, and it has been widely used up to now due to its simplicity and easy comprehension and calculation. This model establishes that a Chilling Hour (CH) corresponds to an hour at temperature between 0 and 7.2 • C (45 • F), since this range of temperatures is considered to affect dormancy completion. While temperatures below 0 • C are assumed not contributing due to at such low temperatures biological processes were considered slowed or not occurring, temperatures over 7.2 • C (45 F) were considered not low enough to affect dormancy completion [28].
The Utah model bases on the quantification of Chilling Units (CU) and establishes different ranges of temperatures with a different contribution to dormancy completion. A chilling unit corresponds to one hour under temperatures between 2.5-9.1 • C, a range that is considered the most effective temperatures on dormancy completion. Other ranges of temperatures are considered to have half (1.5-2.4 • C and 9.2-12.4 • C), null (<1.4 • C and 12.5-15.9 • C) or negative (>16 • C) contribution to dormancy [29].
The Dynamic model, dated back from the 1980s [30], is based on a series of experiments that evaluated the effect of different series of temperatures on dormancy release [38][39][40]. This model proposed the accumulation of an intermediate product promoted by cold temperatures that can be reversed by warm temperatures (first step). Once this intermediate product has reached a certain level, the chill portions are permanently fixed and are considered not affected by warm temperatures [30]. The model is based on a possible biological process in which a thermally unstable precursor would lead to the accumulation of a factor in the buds. This process would follow the Arrhenius law that fits the mathematical relationship between temperature and the rate of a chemical reaction.
Once dormancy is predicted allowing the quantification of chilling, it is also needed to quantified warm temperatures after dormancy for flowering to occur (Figure 1b.2). The modelization of warm temperatures was early developed in agriculture to predict the different phenological stages of crops [41]. The combination of a chilling model with a heat model to predict flowering was firstly described with the combination of the Utah model with the Growing Degree Hours (GDH) quantification [25], and then this combination was also applied with the other chilling models. A GDH is defined as one hour at 1 • C above the base temperature (4.5 • C), this linearly progresses until the upper limit (25 • C) [13]. One of the main drawbacks of using these models is the necessity of hourly temperature data records, whose availability is limited. Thus, equivalent models have been developed based on maximum and minimum temperatures [42].
In this species, the most data (96 out of 106 cultivars) were calculated with statistical approaches, which contrast with the other Prunus sp. reported in this work. Almond data were obtained according to three different statistical methodologies [26,33,34]. The initial phenological data set also differed between works: the PLS analysis was performed over the date of flowering initiation (BBCH phenological stage 61, 10% flowers open) during 30 years [34,35], while the other approaches based on the dates of full bloom (BBCH phenological stage 65, 50% flowers open) over 7 [26] and 4-10 years [33].
A comparison between experimental (E) [43] and statistical (S) [26] approaches reveals similar results of chilling requirements and heat requirements for 'Ferragnès' (558 and 444 CU, 7309 and 8051 GDH), 'Marcona' (435 and 428 CU, 6681 and 6603 GDH) and 'Ramillete' (326 and 444 CU, 6538 and 5947 GDH) in Spain. Unfortunately, it is not possible to make more comparisons due to the different models used to quantify chilling and the different cultivars used in each study.

European and Japanese Apricot (P. armeniaca and P. mume)
European apricot is one of the most economically important fruit crops in temperate regions worldwide [44]. It is mainly produced in the Mediterranean area and the Middle East, being the higher producers Turkey, Uzbekistan, Italy, Algeria, and Iran [2]. A total of 15 works have experimentally evaluated the chilling requirements of 68 apricot cultivars all around the world (Iran, Italy, Serbia, South Africa, Spain, and the USA) ( Table 2). The range of chilling requirements is between 274 CU in 'Palsteyn' [45] to 1450-1600 CU in 'Orangered' [46]. This crop is cultivated mainly in Mediterranean regions and it has traditionally been considered that most cultivars had low chilling requirements. However, some traditional cultivars showed high chilling requirements as 'Búlida' (1048 CU), 'Canino' (806 CU), 'Currot' (642 CU) or 'Moniqui' (1139 CU) [42,43] (Table 2).
In the last decades, an important renewal is taking place due to sharka, a disease caused by the Plum Pox Virus (PPV). High chilling PPV-resistant cultivars from North America, such as 'Goldrich' (950-108 CU / 65-59 CP), 'Harcot' (920-1665 CP), 'Orangered' (568-1481 CH / 902-1600CU / 55-69 CP), and 'Stark Early Orange' (1411 CU / 79 CP) ( Table 2), have been used as parentals in different breeding programs with the aim of introducing a source of the resistance to the disease. The release of a high number of new cultivars is resulting in a lack of information about the chilling and heat requirements of the majority of the new commercial cultivars [44,47].
Japanese apricot originated in China and has been widely cultivated for about 3000 years in Asian countries as China, Japan, and Korea. However, this crop is hardly known in other countries probably due to its poor adaptation to other areas of different climatic conditions, since it requires warmer and more humid conditions than European apricot [44].
Some cultivars with a wide range of chilling requirements, such as 'Nanko', a high-chilling cultivar from Japan, and 'Ellching', a low-chilling cultivar from the subtropical region in Taiwan, have been used in studies on dormancy physiology [54] and genetic regulation [55][56][57][58][59].

Peach (P. persica)
Peach is the stone fruit crop with higher economic importance. It has been confined traditionally to latitudes between 30 • and 50 • North and South [68], but in the last years, there is an increasing interest to expand it to warmer areas, including tropical and subtropical regions [68][69][70]. In recent decades, intense breeding has led to the release of an enormous number of cultivars of different types of fruit, including pubescent (peaches) or glabrous skin (nectarines), round or flat shape, white or yellow flesh, and freestone o clingstone [70] (Table 4). Several peach cultivars have been used to develop models in dormancy studies, both in experimental approaches to determine the date of breaking of endodormancy [23,38,71] and in models to quantify chilling and forcing temperatures [13,28,29,72]. The DORMANCY-ASSOCIATED MAD-BOX (DAM) genes that regulate dormancy were first reported in an 'evergreen' peach mutant [73,74].

European and Japanese plum (P. domestica and P. salicina).
World production of plums increased by almost 20% in the last 10 years (from 9.5 million tons in 2007 to 12 million tons in 2017) [2]. These data include European plums, Japanese plums and hybrids between different Prunus sp. In spite of the economic importance of this crop, temperature requirements are little studied, with data available for only nine cultivars of European plum (Table 5) [91] and 16 cultivars of Japanese plum (Table 6) from two studies performed in Spain [31,91]. The experimental procedure used was slightly different between studies, with variation in the temperature of the growing chamber and in the growth evaluation procedure.
Sour cherry requirements have been poorly studied. There is only data of cultivar Montmorency, which showed 954 CU and 6130 GDH [42] (Table 7).

Concluding Remarks and Perspectives
This study compiles the temperature requirements of a total of 530 cultivars of eight Prunus ssp. Most of the data correspond to peach (204 cultivars), almond (106 cultivars), Japanese apricot (77 cultivars), European apricot (68 cultivars), and sweet cherry (49 cultivars) since there is little information available for European plum (nine cultivars), Japanese plum (16 cultivars), and sour cherry (1 cultivar) ( Table 8). These data represent a very small percentage of the existing commercial cultivars and, in addition, 84 out of 530 cultivars came from studies published more than 25 years ago (Table 8). Therefore, temperature requirements are only available for very current growing cultivars. To serve as a reference, more than 1500 stone fruit cultivars were registered in the European Union in the last 25 years, including 946 for peach, 320 for apricot, 132 for Japanese plum, and 130 for sweet cherry [99]. Two main methodologies, statistical and experimental, have been used to obtain the temperature requirements reported here. Chilling requirements were experimentally determined for all the cultivars of European and Japanese apricot and European and Japanese plum. In contrast, most of the data available for cultivars of almond (153 statistical data vs. 56 experimental data) and peach (173 statistical data vs. 42 experimental data) were statistically determined (Table 8). bud growth is performed after different periods in the growth chamber (a week, 10 days, 20 days), and it is based on analyzing vegetative [103] or flower buds [95]. Furthermore, several criteria are used to determine dormancy overcome. These can consist of significant increases in fresh [23] or dry weight [60], and/or phenological changes in bud phenology [24], which can result in an underestimation of the chilling requirements [98]. All these variations make this methodology easily adaptable to the characteristics of each fruit species and regional variations, but the results obtained under specific conditions should be taken with caution when applied to other regions or climates.
The statistical determination of dormancy is mainly based on two approaches. On one side, the correlation of winter temperatures with the flowering dates, which has been used to establish the chilling requirements of almond [26,33], apricot, peach, plum and sweet cherry [33] cultivars in Spain. On the other side, PLS analysis has been recently developed in a sweet cherry cultivar in Germany [34], and was subsequently applied to other temperate fruit crops as almond in Tunisia [35] and Spain [36], and apricot in China [27,104]. This methodology has reported interesting results on predicting phenology under future scenarios of global warming [19,104]. However, statistical analyses present low applicability on new cultivars released from breeding programs, since they are based on a long series of flowering dates records (more than 20 years).
Once the endo-and eco-dormancy periods have been established, specific temperature models are used to calculate the duration of each phase. Although these models were specially designed to quantify chilling or warm temperatures, they present numerous drawbacks. The Chilling Hours model is easy to understand and calculate and is commonly used by growers who often know the accumulated CH in their location, despite the lack of information about the requirements of their cultivars. This model does not fit perfectly with the behavior of trees, especially in mild and warm areas [105,106]. The Utah model attempts to be more accurate by weighting the temperature ranges. However, the fact that it establishes negative values for warm temperatures hampers its applicability in mild winter conditions. The different criteria for establishing the starting point to quantify temperatures, either an established date for dormancy starting (e.g., November 1 st ) [35] or the date with the maximum negative value [29], had resulted in high differences even for the same cultivar as occurs in almond [26,35,43] or European apricot [45,48,50], making the comparison between studies difficult. The Dynamic model has been proposed as the best model available but also presents limitations on fitting the plant responses to chill [37]. It was designed as a process-based model, however, the physiological process behind is still unknown [30,40,72]. Finally, the Growing Degree Hours model allows quantifying forcing temperatures in a wide range of biological processes, such as phenological stages of annual crops or even insect growth [107]. The results of GDH quantification are highly variable between species, especially at different locations [34]). This could be due to both the model and the interaction between chill and heat accumulation [108][109][110].
In spite of both the methodologies to determine dormancy periods and the temperature models have numerous pitfalls, the available data of temperature requirements are useful to predict the adaptability of a particular cultivar to a certain area. However, very few of the cultivars currently grown have known temperature requirements. This means that, in most cases, the flowering period is the unique information available to assess the adaptation of a cultivar. Flowering periods are usually related to a reference cultivar (e.g., 'Burlat' in sweet cherry) [5] and successfully used for spring frost risk assessment and for pollination purposes, to predict flowering overlap between pollinating and pollinated cultivars. However, assessing the possible adaptation to an area based on relative flowering dates has many limitations, and temperature requirements offer a more reliable approach [111]. However, this review shows that information is not available for the most important cultivars nowadays, and, when available, they are usually imprecise estimations based only on flowering dates. Furthermore, temperature requirements are scarcely evaluated in most breeding programs, which could lead to an increasing lack of information in the coming years.
In conclusion, numerous improvements may be needed to obtain an accurate determination of the temperature requirements of stone fruit cultivars. The standardization of the experimental conditions would allow obtaining more robust and comparable data. However, the increasing number of new cultivars in most Prunus species and the expected reduction of winter chill due to global warming emphasize the necessity of a proper biological marker for dormancy. This would allow the analysis of samples collected directly from the field, without depending on external factors such as forcing conditions in the experimental approach or the availability of a large set of phenological data. Recent reports have revealed several processes as promising candidates for dormancy markers such as the expression of the DORMANCY-ASSOCIATED MAD-BOX (DAM) genes in peach [112], starch accumulation within the ovary primordia cell in sweet cherry [113], anther meiosis in apricot [114,115], and hormone regulation in sweet cherry [116]. Establishing the relationship between temperature records and a biological dormancy marker would lead to a process-based model that would allow direct determination of dormancy and a more accurate estimation of the temperature requirements of particular cultivars. Funding: research was funded by Instituto Nacional de Investigación y Tecnología Agraria y Alimentaria (RFP2015-00015-00, RTA2017-00003-00); Gobierno de Aragón-European Social Fund, European Union (Grupo Consolidado A12_17R).