Olive dormancy

Savvas Pastopoulos, Agronomist MSc, Pastopoulos Agronomics, Neos Mylotopos, Pella, Greece

The olive keeps its leaves all winter, and that alone changes what the word dormancy means. In a deciduous tree the winter is a clean pause, with a beginning and an end that are visible to the eye: the tree drops its leaves, closes, counts chilling and opens again in the spring. In the olive there is no such visible pause, the canopy is still working in January, and in recent years even the existence of a true dormancy in the olive has been questioned, as against a period during which the cold simply does not allow it to grow. The text that follows examines four questions in turn: what exactly stops shoot growth in the autumn, what chilling does to the buds that will carry flowers, where the scientific disagreement over the nature of that state currently stands, and what all of this means for a Greek olive grove.

1. What stops shoot growth in the autumn: temperature and not day length

In most woody species of the temperate regions the signal that stops shoot growth at the end of the summer is the shortening day and not the falling temperature. The plant, in other words, reads a calendar rather than a thermometer, and the advantage of a calendar is that it cannot be fooled: day length is the same every year for a given location, whereas temperature varies. Apple, pear and a few other genera are a well known exception and stop with temperature. For the olive the question had not been answered by measurement, and it was answered in 2020 by a three-part study carried out in southern Spain, at Cordoba and at Espiel [1].

In the first part, young trees of the cultivar Arbequina were divided among five treatments, designed so that the effect of temperature could be separated from that of day length. What was measured was the rate at which new pairs of leaves appeared, which is the simplest indicator of whether the shoot is still extending. The result was unambiguous. Growth stopped only where the temperature fell, while the plants held above 18 °C continued to produce leaves normally even when day length was short [1]. Photoperiod, in other words, exerts no control at all over growth cessation in the olive, and any attempt to predict the behaviour of the tree from the calendar has no basis from the outset.

The two models that predict the date of growth cessation

Two simple models for the date on which shoot extension stops were derived from the same data. Both work with hourly temperatures and with a single counter, which starts at zero in early autumn and increases for as long as the cold lasts. When the counter reaches the value required, growth stops. The logic is the same as that of winter chilling hours, with the difference that something else is being measured here, and in another season.

The first model is the simplest that can be written. It counts only the hours during which the temperature lies below a threshold, adds one unit for each such hour, and never subtracts anything. Fitting it to the data of the study gave a threshold of 14.4 °C and a requirement of 160 hours, with a standard error of 3.1 days [1]. The 160 hours correspond to roughly six full days of total time spent below the threshold, but in the field they are not accumulated in six days. They are spread over several weeks, because on autumn days only part of the twenty-four hours, usually the night and the first hours of the morning, lies below 14.4 °C. In a region where the autumn gives eight such hours per night, the requirement is met in about twenty days, whereas in a region with four hours per night twice as long is needed.

The second model adds a second term, and that term is the interesting part. Every hour with a temperature above the threshold subtracts 0.32 of a unit from what has already accumulated. The fit gave a threshold of 14.5 °C and the same 160 hours, with a standard error of 2.5 days [1]. That coefficient is the quantitative expression of the reversibility of the process: since the reciprocal of 0.32 is 3.1, about three hours of warmth are needed to cancel one hour of chilling. A warm spell in mid autumn can therefore turn the counter appreciably back and postpone growth cessation by days. It is precisely this term that improved the accuracy relative to the first model, and the fact that the improvement exists is in itself an indication that the cancellation is a real phenomenon and not a mathematical convenience.

The models were tested against independent data, that is against measurements that had not been used in the fitting, which is the only way to show whether a model predicts or merely describes. For Arbequina the second model predicted the date of growth cessation with a standard error of 3.16 days and a mean absolute error of 2.67 days. When the other four cultivars of the study were included in the test, the corresponding errors rose to 4.79 and 4.00 days [1]. That accuracy applies to the specific location and the specific cultivars, and it is retained as an indication of the order of magnitude and not as a guarantee for a Greek olive grove.

The limits the authors themselves state

Two limitations of the model are stated within the paper itself, and it matters that they are read together with the figures. The first is that the requirement of 160 hours is not a chilling requirement for flowering but a requirement for growth cessation, and the two processes are entirely separate. The second is that the calibration was carried out on young potted trees, one to three years old, so the transfer of the figures to mature trees of another cultivar and another climate is not a given [1]. A mature tree has reserves, a deep root system and a large thermal mass of wood, and none of these exists in a potted seedling.

Two further details from the same work have direct practical value. The first is that the base temperature for leaf appearance was found to be 11.1 °C, that is appreciably lower than the 14.4 °C threshold that stops growth. That difference of about three degrees means that the tree does not stop because it has reached the point at which it can no longer grow, but stops while it still can, and the cessation is therefore an active response and not passive exhaustion. The second is that the five cultivars tested, among them the three largest in the world by planted area, differed very little in the date of cessation [1]. Location mattered far more than cultivar, with differences of one to three weeks between the two sites of the study, and that means the question of when growth stops is answered mainly by the climate of the region.

Sources: 1

2. The action of chilling on the buds that will carry inflorescences

Growth cessation is only one of the two questions, and in economic terms the less important one. The second, which decides the following year's crop, is what happens inside the axillary buds, that is those formed in the axils of the leaves of the current season's growth and from which the inflorescences of the following spring will emerge.

The sequence of events, as described by the Cordoba group after decades of field observation and of measurements on cuttings, has four stages [2]. First, the axillary buds are formed over a long period within the growing season rather than all at once, and they pass from undifferentiated to potentially floral unless the presence of fruit on the tree prevents it. Second, throughout that period they are held closed by the leaves and the fruit themselves, through a mechanism called paradormancy which has nothing to do with cold. Third, when autumn temperatures fall, that state becomes general and now includes every bud, irrespective of its position. Fourth, what releases them from then on is the accumulation of chilling.

The finding that runs against intuition is that the morphological differentiation of the inflorescence begins after the winter rest and not before it. In the bud sections examined, on 9 January only a few buds had met their chilling requirement and a zone of cell activity was only beginning to be visible. On 7 February, when all of them had met their requirement, the inflorescence axis was already formed [2]. In the middle of winter, in other words, a bud that will carry flowers and a bud that will carry a shoot cannot be told apart even under the microscope, and any estimate of flowering made by examining buds in December or January rests on something that does not yet exist.

Which temperatures count, and why the two thresholds are not interchangeable

For the flower buds, chilling is not measured against the same threshold that stops shoot growth. Measurements on cuttings showed that 12.5 °C is an efficient accumulation temperature, and indeed allows slow budburst once enough chilling has accumulated, whereas at 30 °C natural accumulation falls to zero [2]. The International Olive Council gives the same range in more detail: chilling accumulates from 7 to 15 °C, with an optimum at 12.5 °C, above 16 °C it does not accumulate at all, and still higher temperatures cancel chilling that has already accumulated [4].

Confusing the two thresholds is the commonest error in this subject, which is why the distinction is worth stating explicitly. These are two different processes, measured in different ways, which happen to occur in the same tree in the same season.

One practical observation follows from the table. The threshold for growth cessation is higher than the optimum for chilling accumulation, so there is a band of temperatures, roughly from 15 to 16 °C, in which growth has stopped but no chilling useful for flowering is being accumulated. A mild winter that moves continuously within that band presents the grower with the picture of a tree resting normally, while the process on which flowering depends does not advance.

The synchronisation hypothesis

The explanation the authors offer for the role of chilling is that it acts as a synchroniser. The buds were formed over a period of months, they are of different ages and in different positions along the shoot, and if each one started as soon as it was ready, flowering would be spread over a long period. The winter rest imposed by chilling prevents all of them from starting early and separately, and once the requirement is met it allows them to start together [2]. Pollination in the olive is wind borne and requires simultaneous flowering within the grove, so synchronisation is not a detail but a precondition of fruit set.

The same hypothesis also explains the converse phenomenon, which is the one most easily observed in the field. When the winter is warm and chilling is insufficient, budburst, flowering and subsequently ripening come out unsynchronised and drawn out, with stages of different ages on the same tree [4]. That picture is often attributed to a peculiarity of the cultivar or to an error in fertilisation, whereas on the interpretation above it is the expected outcome of a winter that did not supply enough chilling.

Sources: 1 2 4

3. Endodormancy, paradormancy and ecodormancy: where the disagreement lies

The term dormancy is used for three different things, and confusing them is not an academic matter, because each one predicts different behaviour of the tree during the winter. The three states are distinguished by the question of what holds the bud closed and what is needed for it to open.

In 2024 a position paper was published in the Journal of Experimental Botany arguing that the olive does not pass through true endodormancy [3]. There are two arguments, and both rest on the test column of the table. The first is that when buds or cuttings are moved to favourable conditions the proportion that bursts is very high even in the middle of winter, which does not happen in a plant with an internal lock. The second is that removal of the leaves combined with warming gives one hundred per cent vegetative budburst. If paradormancy is lifted by defoliation and ecodormancy by warming, then nothing is left to hold the bud back, and therefore there is no third mechanism [3].

The same work also restates the role of chilling, and this is the more substantial consequence of its position. Cold, on this account, does not break any lock but induces the expression of the gene OeFT, which is the flowering gene. The evidence invoked has three strands: that chilling induces OeFT, that a warm treatment of buds during the winter abolishes flowering entirely, and that transgenic plants overexpressing OeFT flowered out of season [3]. The conclusion is that chilling is not needed as such for the bud to start, but for the activation of the gene that decides what kind of organ it will produce. That interpretation agrees with an observation reported by both of the other works, that buds which burst before or early in the winter produce shoots and not inflorescences.

Why the distinction matters in practice

The difference among the three states is not theoretical, because they predict different behaviour during a warm spell in winter. If the olive is in ecodormancy, that is held back only by the cold, then a warm period is enough for growth to resume. The study shows this by measurement rather than by estimate: plants moved into a glasshouse above 18 °C on eight different dates through the winter all started, irrespective of how much chilling they had received up to that moment, and usually within one to two weeks [1]. There is therefore no internal safeguard keeping the tree closed until March, as there is in a deciduous species with deep endodormancy.

The consequence for management is that a warm spell in January or February does not pass without cost, and the cost in fact arrives by two routes at once. On the one hand growth may start and then find itself exposed to the next frost, without having had time to acquire tolerance. On the other hand, and with greater significance for the crop, high temperatures cancel chilling that had already accumulated [2] [4], so the tree ends up behind where it stood before the warm spell, and the time remaining until spring for the requirement to be met is shorter.

Sources: 1 2 3 4

4. What all of this means in a Greek olive grove

All the figures of the preceding sections were measured on Arbequina, in pots, at Cordoba and at Espiel in southern Spain, on young trees one to three years old [1], or on cuttings from the same region [2]. What transfers from them is the mechanism, that is which factor controls what, because the physiology of the species does not change with geography. What does not transfer are the thresholds and the chilling requirements, because these are calibrated on another cultivar and another climate. The Greek question is therefore a different one and is answered by Greek sources: which Greek cultivars require how much chilling, and in which regions that is secured.

Greek cultivars do not all have the same requirement

In the teaching material of the Agricultural University of Athens the Greek cultivars are divided into two groups according to their chilling requirement, and the same distinction is repeated in technical writing within the sector [5] [6].

The values are given as an upper threshold below which chilling counts, not as a requirement in hours. The second source reports 12 °C and the first 10 °C for the same group, and that discrepancy shows that these are rules of thumb rather than calibrated thresholds.

The practical consequence of the distinction concerns the siting of new plantings. Chondrolia Chalkidikis and Konservolia Amfissis require prolonged winter chilling and perform poorly in regions with a mild winter, such as the coastal zones of Crete [6]. Koroneiki and the other cultivars of the first group crop normally in regions where the winter mean stays above 10 °C [6]. Choosing a cultivar for a new planting is therefore, from this point of view, also a decision about whether the particular site supplies enough chilling for the particular cultivar, and an error here is not corrected afterwards by cultural practice.

As to the total requirement, the Greek teaching material gives a range of 10 to 13 °C for 1,500 to 2,000 hours [5]. A technical article from the sector reports at least ten weeks below 16 °C, and refers to earlier research in which trees receiving 1,200 hours below 7.2 °C flowered and cropped normally, while trees that received no hours at all below that threshold formed no flowers [6]. The two sets of values are not in competition, because they are measured against different thresholds, and that is precisely why chilling figures are never compared without stating the model from which they came.

What the measured Greek data show

The only extensive Greek series combining phenology with chilling accumulation is from Crete. It covers roughly 1,200 flowering observations of Koroneiki from 1984 to 2018, at four locations spread from 45 to 624 metres in altitude, and the calculation was made with the Dynamic model and chill portions [7].

Accumulation ranged from 50.6 chill portions at Gazi, at 45 metres, to 97.3 portions at Agia Varvara, at 624 metres, with Dafnes and Kastelli, at 315 and 340 metres, giving intermediate values [7]. The difference between the coastal and the upland location is close to twice the accumulation within the same island, and over a distance of a few tens of kilometres. The size of that difference shows that altitude outweighs geographical distance by a wide margin, and that a chilling estimate based on the prefecture or on the island says almost nothing about the particular holding.

The projections of the same work show a reduction in chilling of 12.0 to 28.3 per cent for the period 2021 to 2060 and of 22.7 to 70.9 per cent for the period 2061 to 2100, depending on the location and the emissions scenario [7]. Under the most adverse scenario and for the distant future, accumulation at Gazi falls practically to zero in certain years, and the authors warn explicitly that such low accumulation may not be sufficient for floral induction in Koroneiki [7]. That observation concerns the coastal zones and not the crop as a whole, but the coastal zones are also those with the largest planted areas.

What changes in management

Four points follow directly from the above and concern decisions taken on the holding.

• Siting Chondrolia and Konservolia is a chilling decision. In a region with a mild winter the planting will underperform every year, and that underperformance is corrected neither by fertilisation nor by irrigation nor by pruning, because the cause lies in the climate of the site.
• A warm winter costs twice. It cancels chilling that had already accumulated, and at the same time it may start premature growth, because in the olive there is no strong internal lock to prevent it.
• Unsynchronised flowering is a symptom and not a quirk of the season. When inflorescences of different ages coexist on the same tree, the most likely cause is insufficient chilling, and that observation is worth recording by year and by block, so that a data series exists for the particular holding.
• Biennial bearing begins the previous summer. The presence of fruit prevents the conversion of axillary buds into flower buds [2], so whatever concerns this season's crop load also decides the flowering potential of the next, long before winter enters the equation.

The points that remain open

The research leaves three gaps that are worth stating, because a gap that is declared is a research question, whereas a gap filled with a foreign figure becomes an error that is repeated.

• The growth cessation model, with its threshold of 14.4 °C and its requirement of 160 hours, has not been calibrated on a Greek cultivar nor on a mature tree. The figures serve as an order of magnitude and as a description of the mechanism, not as a calendar for a Greek olive grove.
• A chilling requirement expressed in portions or in hours, per Greek cultivar, has not been published. The Greek data give upper temperature thresholds per group of cultivars and one accumulation series for Koroneiki in Crete, without closing the circle between the two.
• A series comparable to that of Crete does not exist for northern Greece, where Chondrolia Chalkidikis is grown, that is the cultivar with the highest chilling requirement. That is the point at which a local series of flowering observations would have the greatest value.

For the way chilling is measured, and for the reason chilling hours and chill portions are not the same quantity, there is a separate text: Chilling hours: the three models and when each one tells the truth. Accumulation for five regions of northern Greece is calculated live on the page Chilling hours and chill portions.

Sources: 1 2 5 6 7

5. Sources

The numbers in the text refer to the list below.

• [1] López-Bernal, Á., García-Tejera, O., Testi, L., Orgaz, F. and Villalobos, F.J. (2020). Studying and modelling winter dormancy in olive trees. Agricultural and Forest Meteorology 280:107776. Source for the role of temperature rather than photoperiod, for the growth cessation models with thresholds of 14.4 and 14.5 °C and a requirement of 160 hours, for the subtraction coefficient of 0.32, for the validation errors of 2.67 and 4.00 days, for the base temperature of 11.1 °C, for reversibility within one to two weeks, for the small difference among cultivars and for the frost tolerance argument. The study was carried out at Cordoba and Espiel in southern Spain, on potted trees one to three years old. Read in full.
• [2] Rallo, L., Ramos, A., Rubio-Valdés, G. and Rapoport, H.F. Integrated Overview of Olive Reproductive Bud Dormancy and Biennial Bearing. Acta Horticulturae, International Society for Horticultural Science. Source for the sequence of events in the axillary buds, for the role of the leaves and the fruit in paradormancy, for the differentiation of the inflorescence after the winter rest, for the sections of 9 January and 7 February, for 12.5 °C as an efficient temperature, for the loss of accumulation at 30 °C and for the synchronisation hypothesis. The measurements were made on cuttings in the region of Cordoba. Read in full.
• [3] Arias-Sibillotte, M., Considine, M.J. and Signorelli, S. (2024). Reinterpreting olive bud dormancy. Journal of Experimental Botany 75(19):6017-6021. Source for the position that the olive does not pass through true endodormancy, for the distinction among endodormancy, paradormancy and ecodormancy, for one hundred per cent vegetative budburst after defoliation and warming, and for the role of chilling through the gene OeFT. A position paper and not new research, with no experimental data of its own. Read in full.
• [4] International Olive Council. World Olive Catalogue, Module 2, Chilling requirements. Source for the accumulation range of 7 to 15 °C, for the optimum of 12.5 °C, for the absence of accumulation above 16 °C and cancellation at higher temperatures, and for unsynchronised budburst and flowering when chilling is insufficient. An international source, with no reference to a particular country.
• [5] Roussos, P. Annual growth cycle of the olive, bud differentiation. Teaching material, Agricultural University of Athens, in Greek. Source for the requirement of 10 to 13 °C over 1,500 to 2,000 hours and for the division of the Greek cultivars into low and high chilling requirement. A Greek source.
• [6] Bartzokas, N. The role of chilling in olive fruiting. SEDIK, in Greek. Source for the Greek upper thresholds of 16 and 12 °C by cultivar, for the ten weeks below 16 °C, for the 1,200 hours below 7.2 °C and for the poor performance of the high requirement cultivars in the coastal zones of Crete. A technical article from the sector, not peer reviewed, and used only for the Greek context.
• [7] Grillakis, M.G., Kapetanakis, E.G. and Goumenaki, E. (2022). Climate change implications for olive flowering in Crete, Greece: projections based on historical data. Climatic Change 175:7. Source for the roughly 1,200 flowering observations of Koroneiki from 1984 to 2018, for the chill portions of 50.6 at Gazi and 97.3 at Agia Varvara, for the projected reductions of 12.0 to 28.3 and 22.7 to 70.9 per cent, and for the warning that accumulation may not be sufficient for Koroneiki. A Greek source, with data from Crete. Abstract and extracts of the full text were read.