Dormancy predictions for fruit trees

In perennial fruit trees, activity usually slows down in the cooler months each year, with some species from temperate and cool sub-tropical regions going through a period when no matter how favourable environmental conditions are, flowers don’t bloom and new vegetative growth may be arrested. This phenomenon is known as dormancy and a certain amount of cultivar (cv)-specific cool weather chilling is required for it to be relieved. Many species native to tropical regions where there is very little change in temperature throughout the year are less affected in this way, and any pauses in growth are largely driven by other variables, often low precipitation. Temperate and cool sub-tropical species more commonly experience cold weather slowing every year, many being deciduous. Dormancy provides a means of protecting delicate leaf and flowering meristematic bud tissues from low temperature damage, especially if there are frosts. In an effort to extend the range where cool climate species can be grown, many low chill cvs have been bred or selected, and some of these are on the border line of what’s possible to grow in sth west WA’s warm climate.

There are 3 main types of dormancy. The one that most of us would be aware of when wishing to grow cool climate fruit trees around Perth is called endodormancy (ED). This is driven by internal plant mechanisms (genetics, biochemistry, physiology, hormones etc) that control the period of chilling that must be satisfied before growth can resume. As enthusiasts, we need to be confident that our area will provide this level of chilling for a species or cv we wish to grow, otherwise there may be no flowering or it may be so delayed or erratic that fruit yield and quality is compromised. Despite many decades of research given its importance, we still don’t have any reliable biomarkers that indicate when chill has been satisfied. So we have to rely on proxies (models) to help us with our predictions. None of the models provides a functional understanding of plant physiology. The most common one, developed in the 1930-40s for peaches, is called the Chill Model (CH). This is where a low temperature is specified, usually 7.2oC, and the number of hours in the range 0-7.2 throughout cooler months is counted. Below zero is typically excluded because research has shown that freezing temperatures usually don’t contribute. Species/ cvs may then be quoted as needing eg 300 or 1500 hours to flower. In the Perth coastal metro region we usually get ≤ 300 chill hours each year, so our plants should preferably be under this. In addition to predicting whether and when bloom may occur with warmer weather, the model is sometimes also used to predict the recommencement of vegetative growth in deciduous species. There are two other types of dormancy called ecodormancy (ecoD) and paradormancy, phenomena that we may be aware of in other aspects of horticulture but usually don’t associate with fulfilling ED. EcoD is the phase after ED when a certain amount of heat must be accumulated before buds will spring into action, as this re-start doesn’t happen immediately. Around Perth we usually don’t have to worry too much about satisfying this as we have an abundance of heat in spring and summer. But ecoD (often called forcing) can influence bloom time, especially if chill hours have not been fully met. Paradormancy is where latent buds may be inhibited by other plant organs, eg when lateral buds are inhibited by apical dominance, or tip pruning removes apical tissues producing inhibitory hormones that affect proximal buds, or tying down branches laterally changes hormonal regulation.

The CH model for ED has been used globally for so long now that it’ll probably continue to be used for years to come by other than serious commercial growers and the research community as it’s relatively simple to understand, measure and use. But we should recognise it’s only a very rough guide, and unless your cv is well under the quoted CH your harvest may be compromised. For enthusiasts where livelihood and income is not involved, a reduced harvest may be acceptable. However commercial growers have to set a much higher bar as they need a reliably good yield every year and have to time general horticultural management, possible chemical flowering treatment (eg hydrogen cyanamide), contract labour/equipment, and judge favourable market conditions. So they not only have to know that flowering will always occur, but when, how complete it is, and over what period. Even enthusiasts may be interested in this timing if they’re trying to stage early, mid- or late season cvs to eg minimise pest and disease problems or spread their harvests. If your cv requires cross pollination by another cv (eg plums) then we need to be confident that flowering will overlap.

Some of the reasons for anomalies with the CH model that you may have experienced are the following:

The first might seem to be a self-sufficient explanation for any possible problems as the reliability of industry CH data is not ideal. A single temperature threshold does not capture all the factors driving ED relief (as discussed below), and CH figures vary for different regions subject to different climatic environments. Plus, CH can vary markedly from year to year. In a 2013 Mediterranean study of low chill pistachios (latitude 34N, for comparison Perth is 32S) daily temperatures were collected for 12 successive years. Pistachios are normally a high chill species (900-1500 CH) but the cv selected in this study was reportedly low chill, requiring only 350. Over the 12 years studied, the lowest recorded CH was 72 and the highest 346, 70% below and 46% above the 12 year mean resp. In low CH years with less than what is needed, any or all of the following can occur – flower bud drop, underdevelopment of the pistil, floral and leaf bud bursting delay, insufficient carbohydrate reserves, poor fruit set and low quality, irregular flowering, smaller terminal shoot extension, abnormal leaves etc. In the low chill year, nut yield for orchard trees was 0.1kg/tree compared to the 12 year average of 5.0. Despite precipitation for that year being 50% above the mean, it could not make up for inadequate chill. If a limited-year industry study included the 72 CH year in their data it will inevitably be an underestimate; this can lead to you selecting an inappropriate cv and you might then incorrectly assign underperformance to all sorts of other factors. Agronomists suggest use of Safe Chill Hours, defined as the CH that can be expected in 90% of years, and this level should be the maximum for fruit trees selected to be grown in a region. Another impacting factor is gradual warming due to climate change that is slowly decreasing CHs globally.

The Bureau of Meteorology provides data on daily temperatures for major cities and regions but usually only in the form of average, min and max, not CH in the range 0 – 7.2oC. Methods have therefore been developed to calculate approximate CHs from such data, but these will be error-prone depending on the assumptions made. It’s sometimes possible to obtain actual hourly temperature recordings from Agriculture Dept stations or research studies, but the sites where they’ve been measured may behave differently to your local area because of proximity to the coast or other bodies of water, altitude, winds, crests and valleys etc. Then on a smaller scale on your own property there might be micro-climate variations according to where individual plants are located – facing walls, presence of wind breaks, tree shading etc.

It’s also important to know the extent of flowering used in arriving at quoted CH requirements. We all know that flowering in trees doesn’t occur exactly in unison on the same day; it can be tightly spaced over a few weeks for some cvs or in others it may last several months. Quoted values that you’re basing your decisions on may refer to when 10% flowering has occurred and in others it might be 50 or 90%. This is typically not made clear; if 10% was used, the quoted CH will be smaller than what may be more important for you, especially if the flowering period is prolonged.

There are 2 main ways to record and then publish CHs – under controlled conditions in laboratory settings or with data collected under field conditions. The former can be controlled exactly and could be thought superior, but the latter, although more problematic, variable and time consuming to gather, is more relevant because it incorporates the many factors influencing trees and flowering in real life settings.

The development of ED usually begins with the onset of shorter days (photoperiod) and cooler weather each year, and may take all of autumn and some of winter to develop fully. This can use up CHs that would otherwise be accumulated to break dormancy, so the plant will start later than normal and perhaps leave insufficient time to break ED before warm spring weather commences. Conversely, full ED may be achieved before winter, so ecoD heat hours will commence early. A key question therefore is when should you start counting? The two extreme approaches are sequential (ED complete before EcoD starts) and parallel (where they can overlap). Traditionally, the start of winter or when the first hour of recorded effective chill has been used, but newer research is suggesting ecoD can start before ED has been completed, at least for some cvs and regions.

The 7.2oC threshold can never be a magical switch with biological processes, where 7 is CH effective but 7.4 is not; it’s only a rough yardstick. Since the model was developed, it has been found that temperatures above 7.2 contribute to chill relief and there’s a gradual tapering off of the effectiveness for ever higher temperatures. This is the basis of the Utah Model (UM), developed in the 1970s for peaches, that assigns varying positive weights to temperature ranges and negative weights for high temperatures, as follows:

< 1.4oC0
1.5 – 2.40.5
2.5 – 9.11
9.2 – 12.40.5
12.5 – 15.90
16 – 18-0.5
Utah Model (UM)

After weighting of the data, the summed totals are called Chill Units and they’re specific for each cv; they’re really specific also for the location given variation in field conditions between sites. Like CH, this model is similarly used widely but it performs poorly in warm climates like ours. Also, the negative weightings can sometimes result in negative totals, yet flowers still bloom – an obvious problem for the model. This is addressed in the positive UM developed in sth Africa for warm sub-tropical climates, where the negative weights in the UM for temperatures of 16 and above are not applied.

The 3rd main chill model is called the Dynamic Model (DM), developed in Israel in the 1980-90s for peaches and other warm climate fruit. The model is much more complex and realistic than the other two and employs exponential functions to weight temperatures and their temporal relationships. Chilling is considered to take place in 2 steps, the first being a thermally-labile precursor that only becomes permanent after a certain chill has been accumulated without interruption by a warm spell which cancels the precursor. The permanent chill segments are summed and called Chill Portions; as with other models, they are specific for each cv. The Model is capable of bypassing arbitrary assignment of the start date for each of the 2 phases, statistically indicating whether there is any overlap in ED and ecoD, and it can estimate the three sequential phases of ED (development, full and relief). Most studies comparing the 3 principal models and their variants conclude the DM is superior, predicting bloom date to within a few days, particularly for the warm climate fruits we’re interested in growing in sth west WA. In addition, an authoritative 2021 study on apples and pears using 60 years of climate records and bloom data found that it can also be the best model for these cool climate fruits at 50°N. Similarly to the 2013 low chill pistachio study discussed above, the DM likewise outperformed other models for pistachio in Australia, and it was also best in a study of 63 nectarine and 118 peach cvs over 7 years in Argentina (32°S). These comparative findings for the 3 models have been replicated in many countries and cvs, including Israel, Sth Africa, Chile, Spain, Tunisia, France, California and Australia.

To achieve such good predictions, all 3 models and their variants have to include a model for the heat units required after ED has been relieved as bud growth does not follow immediately afterwards. There has been far less work on this than for ED, but if predicted bloom date is desired beyond expecting whether a cv will flower or not, it has to be considered. The most common ecoD model, with some variations, counts Growing Degree Hours (GDH); trigonometric weights are applied to warm temperature hours using thresholds for the base effective level (commonly around 4.4oC), the optimum and a high temperature cut-off. Unlike ED, ecoD relief is subject to heat but also needs external factors such as nutrition, sunlight, moisture etc to be favourable. The resulting GDHs for each cv are usually much larger than a simple count of relevant hourly temperatures because of weighting. A program combining the DM and GDH models is available free on the internet.

In summary, these proxy models are only approximate stand-ins for ED and ecoD till we have reliable biomarkers. For enthusiast growers in the south west of WA, the CH model will likely continue to be used, but we should realise it’s a pretty rough guide – it takes no account of ecoD requirements and has several other limitations. When considering ED requirements for your cv selections, it’s wise to err on the underside for this factor to ensure yield quantity and quality is as good as it can be and you don’t have variable performance from one year to the next. But don’t overdo it; in a particular cool year, bloom may start too early if ecoD is also satisfied following a brief period of higher temperatures, only to be killed off when temperatures return back to the remaining normal winter levels.