Main

Olive oil embodies the essence of Mediterranean identity and is a fundamental pillar of both the region’s diet and economy1,2,3,4. In 2019, 3.314 million tons of olive oil were produced worldwide, 98% of which derived from the Mediterranean region5,6. Globally, olive is produced by 41 countries, representing 10.58 million hectares and upon which >6.7 million families currently depend for their livelihoods6,7. The economic importance of olive oil is attested by significant increases in global consumption (94% growth between 1990 and 2019, currently 3.234 million tons) and growth in exports ($5794 million in 2019)5. However, olive oil production is increasingly being affected by current climate change8,9,10,11,12. Recurrent warm winters, which (i) delay the fulfilment of chilling requirements, (ii) damage floral bud tissues, or (iii) reduce or entirely inhibit flowering13, along with strong heat waves (that scorch stamens) and low rainfall, have caused a sharp decline in Mediterranean yields. Spain, the world’s leading producer, has been particularly affected6. This situation worsened during the crop years 2022-2023 and 2023-202414. Unfavourable weather conditions in Western Europe significantly affected production, and some areas had to cope with exceptional climatic hazards that affected flowering and yields6. The major olive-growing regions of southern Europe (i.e., Spain, Italy, Portugal, and Greece) are all increasingly affected by climate change, with a notable drop in yields since 2012–201315,16,17.

The centre of gravity for olive oil production shifted towards the eastern Mediterranean during the 2022–2023 crop year but, since 2023, climate pressures in several countries, especially in Turkey and the Levant, have also caused a sharp decline in production5,14. In the 2023–2024 crop year, this area experienced an average decline of 44 ± 18% in olive oil production compared to the previous year5.

These data underline the importance of annual-seasonal weather conditions and long-term climate dynamics as modulators of olive yields, olive oil production and the oil-related economy. Climate has long been one of the major pressures put on olive production18,19,20, alongside the impact of pathogens21,22,23,24.

Among climate factors, precipitation stands out as a key determinant25,26 as olive production is highly dependent on water availability. Winter rainfall replenishes soil moisture, while the amount of precipitation during the olive phenological stages (from March to November to early December)27,28,29,30 is crucial for sustaining optimal crop yields. Olive, while adapted to dry environments, usually shows a decrease in photosynthesis that limits growth and yields under conditions of water stress10,31. Previous studies have highlighted that the projected reduction in precipitation amount across Mediterranean regions32 will intensify water stress, affecting numerous species, including olive33,34.

A second factor that, coupled with water stress, influences photosynthesis is solar activity/light intensity35. Plant productivity directly depends on the photosynthetic capacity of leaves, the dominant photosynthetic organs. Reduced irradiation (e.g. solar irradiance refers to the amount of power per unit area received from the Sun by the Earth’s system), simulated through various shade techniques, has been shown to significantly diminish fruit yield per olive tree36,37. Studies, considering the compensation between the number and size of fruits, have further demonstrated that dry fruit weight and oil concentration (and consequently yield) decrease when daily photosynthetically active radiation (e.g. spectral range of solar radiation from 400 to 700 nm that photosynthetic organisms are able to use in the process of photosynthesis) received falls below 40%38,39. Limited light availability particularly affects floral bud induction, fruit set and growth, as well as oil accumulation in the fruit40.

Finally, other factors, such as temperature and CO2 concentration, strongly influence photosynthetic activity during the olive phenological stages. Previous research has shown that, on a millennial scale, temperature range is a key factor affecting yields41, primarily due to its impact on photosynthesis42. Additionally, variation in CO2 concentration has been identified as a critical parameter for efficient photosynthetic activity43. In this study, we focused on solar activity and precipitation, as both factors are crucial for photosynthesis and, therefore, for maintaining sustainable olive cultivation. Solar variability is especially important due to its key influence on Earth’s climate, affecting both temperature and precipitation44.

Although numerous studies have documented the effects of light intensity variations and declining precipitation on olive trees during the instrumental period10,11,39,45, long-term trends over centennial-to-millennial timescales remain largely unknown. The extent to which these key environmental factors have influenced olive flowering and pollen production over millennia is still an open question. This issue is nevertheless key because, in the near future, the decline in solar irradiance due to the next Grand Solar Minimum46,47 and the reduction in precipitation due to anthropogenic climate change48,49 will profoundly alter olive tree activity and, therefore, olive production.

Here, we examined long-term trends in pollen production, a proxy for flowering and fruiting, in relation to key factors influencing photosynthesis, namely solar irradiance and precipitation. We focused on olive trees in the eastern Mediterranean, a region currently grappling with the impacts of climate change, economic instability and insecurity linked to recurring conflicts. At a time of food transition and recognition of the health benefits of a Mediterranean diet, ongoing climate change calls into question the medium to long-term sustainability of Mediterranean food resources, particularly olive harvests and olive oil production. Will decreasing solar activity and intensification of water stress become critical factors in shaping the future of table olive and olive oil production?

Results

A solar-driven periodicity in olive pollen production

The olive pollen data serving as a matrix for this study derives from eight different locations across the eastern Mediterranean (including Turkey, Cyprus, Syria, Lebanon and Israel; Fig. 1). For anemophilous species like the olive tree, pollen production is a key parameter related to flowering, fertilisation and fruit production50,51. The reconstructed pollen influx for the last 8000 years (Fig. 2a), a direct proxy for pollen production, reveals an oscillating signal. Notably, higher scores were recorded during three distinct periods: 6630–5930 ± 80 BP, 4120–2920 ± 80 BP and 2620–570 ± 80 BP. A sinusoidal model applied to the pollen influx data (Fig. 2a), supported by a wavelet analysis (Fig. 2b), identified a dominant 2350 ± 80-year cycle, suggesting that pollen production and consequently fruit production (due to better pollination efficiency) have experienced long-term periodic variations. The identified periodicity was further validated through a spectral analysis (Lomb periodogram algorithm; Supplementary Fig. 1a) and a REDFIT analysis (Supplementary Fig. 1b). Both analyses confirmed the presence of the same 2350 ± 80-year cycle, indicating that the mathematical signal related to pollen production over the past 8000 years is robust. The 2350 ± 80-year cycle, therefore, emerges as a key periodicity, as it closely parallels the Bray-Hallstatt solar cycle of 2300 ± 200 years52,53.

Fig. 1: Geographical location of the study sites.
figure 1

A separate colour has been attributed to each country.

Full size image
Fig. 2: A solar cycle in olive pollen production.
figure 2

a Olive pollen influx (expressed in pollen grains per cm−2 per year−1) is drawn on a linear age scale (expressed in kiloyears Before Present) and shown with the standard deviation. The long-term trends are depicted as 200 years moving average and as a sinusoidal regression (with the 1σ error). The periods of low/high pollen production are highlighted by different colours. b Wavelet transform (scalogram) for the olive pollen influx is detailed. The cone of influence is denoted as a grey line, and the significance level (P = 0.05) is as white lines. The main periodicity is noted on the graph.

Full size image

Comparison of olive pollen production and solar activity

The solar irradiance datasets were extracted from three different sources: (i) total solar irradiance (ΔTSI)54, (ii) sunspot number55 and (iii) the cosmogenic radionuclide 10Be56. A sinusoidal model, coupled with a wavelet analysis, was applied to each time series (Fig. 3a–c). The Bray-Hallstatt solar cycle, clearly identified in solar proxies, is centred on cycles of 2300–2330 years. When the periodic pollen influx signal is compared and contrasted with the different proxies for solar activity, a synchronous evolution of the different signals emerges, with a stronger correlation observed with the cosmogenic radionuclide 10Be. The pollen influx cycle (2350 ± 80-year) is chronologically correlated with the 2300-year cycle found in the 10Be signal (Fig. 3a), with a cross-correlation coefficient R2 of 0.91 at Lag0 (Pvalue < 10−3). A slight delay in pollen influx is recorded with the 2330-year cycle detected in the ΔTSI signal (Fig. 3b; cross-correlation coefficient R2 of 0.52 at Lag0 with a Pvalue < 10−3), and with the 2300-year cycle identified in the sunspot number signal (Fig. 3c; cross-correlation coefficient R2 of 0.51 at Lag0 with a Pvalue < 10−3). However, the alignment of the wavelet analyses (Supplementary Fig. 2a, b) shows the importance of the Bray–Hallstatt solar cycle in the ensemble of solar and pollen signals.

Fig. 3: Olive pollen production in relation to various proxies of solar activity.
figure 3

The periodicity of each proxy related to solar activity [a cosmogenic radionuclide 10Be, b ΔTSI and c sunspot number] is shown as wavelet transforms (scalogram) with the main periodicity noted on the graph. The long-term cycle of each proxy related to solar activity, as previously defined, was extracted and compared with the olive periodic signal [a pollen influx versus cosmogenic radionuclide 10Be, b pollen influx versus ΔTSI, and c pollen influx versus sunspot number]. The periodic olive signal, corresponding to the pollen influx curve, is the same for graphs (a–c).

Full size image

Focussing on the cosmogenic radionuclide 10Be, the violin boxplots, based on periods of low or high olive pollen influx, reveal a link between sunlight intensity and pollen production (Fig. 4a). The highest pollen influx was recorded during the period 4120–2920 ± 80 BP (increase of 42% in pollen production), consistent with a 55% rise in sunlight (Fig. 4a). Inversely, the 570–50 ± 80 BP period is marked by a significant decrease in pollen production (−57%) that occurred during a drop in sunlight (−44%). The link between sunlight and pollen production lies in photosynthesis, which is the foundation for tree growth, flowering, pollen production and ultimately fruit production. Tree growth primarily depends on photosynthesis, but continued growth also requires water, nutrients and CO257. Although other factors have played a role (e.g. changes in human activities with variations in land use and in the type of agriculture developed, different types of soil, different expositions, and different cultivars used), sunlight and water remain two key parameters in olive fruiting.

Fig. 4: Olive pollen production in relation to 10Be and September–November insolation.
figure 4

a Violin boxplot showing the distribution and probability density of each time series (pollen production influx and cosmogenic radionuclide 10Be). The periods of low/high pollen production are highlighted by different colours. Increases and decreases are expressed as percentages on the graph. b September–November insolation dataset (expressed in Wm−2), compared with the olive tree signal shown with a 200-year moving average and a sinusoidal regression.

Full size image

Olive pollen production and September–November insolation

Another key factor is the amount of insolation (e.g. amount of solar radiation received in the Earth’s atmosphere or at the Earth’s surface) during the morphological differentiation of flower buds (March–April) and the ripening phase (from September to November, sometimes early December). Olive production is directly influenced by changes in radiation throughout the different growing periods (from bud break to pit hardening), which can affect both fruit yields and oil quality58. March–April insolation59 is key as pollen production is directly dependent on the microsporogenesis that takes place in the flower buds during formation. Pollen develops during the four to six weeks prior to flowering (April–May according to temperature)28,29. The influx data show that more efficient pollen production occurs in tandem with increased insolation (from 7500 ± 80 to 100 ± 80 BP; Supplementary Fig. 3).

Focusing on sustainable food production, the photosynthetically active radiation is also key during the oil synthesis phase as variation in insolation can impact final fruit dry weight and oil concentration38. Comparing and contrasting pollen influx data with September–November insolation59 over the last 8000 years (Fig. 4b), our data show that higher pollen production also coincides with periods of increased insolation (from 4200 ± 80 to 700 ± 80 BP). Although these two processes are chronologically separated (pollen production typically occurs in early May but can begin in April, while ripening and early harvest occur in September to November), the first led to greater pollination and fruit setting. Combined with optimal insolation, this should result in optimal fruit ripening and a better harvest. The overall process favoured olive production and the olive oil-based economy.

Rainfall during the phenological stages of the olive tree

Focusing on water, particularly on water stress that affects photosynthesis, impacting bud development, flowering and later on fruit growth and yield10,31,60, we reconstructed the amount of precipitation during the different olive phenological stages. The pollen production is linearly correlated (R2 of 0.61—Pvalue < 10−3) with precipitation from March to November (Fig. 5a). When a polynomial model is applied (R2 of 0.67—Pvalue < 10−3), it shows that although pollen production increases starting at 330 ± 35 mm, it begins to decline after reaching 370 ± 35 mm. Considering the lowest and highest values, the 295-mm threshold (330–35 mm) and the upper 405-mm threshold (370 + 35 mm) establish potential limits for optimal olive production. Based on 325 modern olive growing areas distributed throughout the Mediterranean basin (Supplementary Fig. 4), we calculated a present-day average precipitation of 370 ± 50 mm from March to November, with a lower limit of 328 mm (25th percentile, with a lowest score of 265 mm) and an upper limit of 410 mm (75th percentile, with a highest score of 458 mm) for efficient fruit production. Both past and present data underline the key role of precipitation amount during the various phenological stages of the olive tree (Fig. 5b), highlighting consistent lower and upper thresholds for viable and sustainable olive production. This suggests that near-future water stress forecasts for the eastern Mediterranean will severely impact olive trees and olive oil production.

Fig. 5: Precipitation during the olive phenological stages.
figure 5

a Link between olive tree influx and March-November precipitation (expressed in mm) during the last 8000 years. The long-term trend is shown as a linear regression and a polynomial model (order 4). Boxplots show the evolution of olive tree influx every 10 mm. b The long-term periodicity of precipitation compared with the periodic olive signal. The periods of low/high pollen production are highlighted by different colours.

Full size image

Discussion

Cultural and historical influence

Historical events and cultural periods have influenced olive cultivation over the last 8000 years, especially in the eastern Mediterranean, the cradle of olive growing2,61. Olive cultivation, based on the domesticated form, is considered to have begun during the Chalcolithic period (also termed the Copper Age), when the development of techniques for oil extraction and the presence of olives was recorded at several sites in the southern Levant62. Radiocarbon data from the Jordan Valley give a maximum age of 6400–6270 BP for olive pits, and 6300–6000 BP for oil extraction, consistent with a Chalcolithic emergence of olive cultivation63,64. Following the domestication of olive trees during the Chalcolithic, fruit products were grown and widely traded during the Bronze Age (5300–3200 BP). During this period, olives featured among the main eastern Mediterranean orchard crops, and were valued primarily as a source of oil for eating and cooking65. Olive oil could be stored for long periods and became a valued trade commodity for regional exchange62. During the Iron Age, olive stones, as well as olive presses and olive oil installations were discovered at several sites in the southern Levant, suggesting a strong local presence of olive trees66,67. However, the acme of olive cultivation was recorded later, from the Persian–Achaemenid to the Byzantine eras in the eastern Mediterranean (approximately 2500–1310 BP). Moreover, the collision and fusion of the Greek and Roman notions of the economy with those of the newly won lands of the eastern Mediterranean generated important innovations with a remarkable variety of agricultural installations for oil extraction, which were in use concurrently in the Levant68. Industrial-scale oil extraction was a response to the high demand for olive oil in the Greek and Roman empires, and the Levant became a centre for olive cultivation69. Roman oleiculture was different from that recorded during the Bronze Age. Cultivation seems not to have been restricted to an area close to the Mediterranean coast but was widely extended using irrigation70. After the Byzantine era, the olive tree remained one of the most mentioned fruit trees during the Mameluke and Ottoman periods, but its economic importance moved towards the western Mediterranean.

All of these cultural periods influenced the olive signal obtained in pollen records because pollen production depends on the number of orchards, the size of the orchards and the density of trees per orchard as well as the technology used. However, when all olive pollen from Turkey to Israel is compiled into a single signal, the resulting proxy incorporates both local and regional events. The resulting signal comes from a broad geographical entity, which integrates different cultures and different periods and, above all, reflects the major trends related to the evolution of olive growing in the eastern Mediterranean facing environmental pressures.

Pollen has already been used as an important tool to reconstruct the economic history of Central Europe and Greece or even the historical ecology of the plague71,72,73. The investigation of long-term trends in olive pollen production through the lens of variability in two major climatic factors, namely solar activity and precipitation, brings new insights to understand the potential sustainability of olive yields, olive oil production and the oil-related economy in the near future.

Photosynthesis

Understanding how photosynthesis will evolve in the near future is critical74, as improved photosynthesis could contribute to greater food security in the coming decades amid a growing global population57. When photosynthetic capacity is affected, it can adversely impact plant and tree growth, bud differentiation, fruit set and overall yields. Focusing on the projected long-term Grand Solar Minimum46, solar irradiance is expected to decrease by 0.22%, resulting in a 1 °C decrease in surface temperature during the Grand Solar Minimum 1 (2020–2053 calendar years)75. This scenario is similar to the Maunder Minimum episode of the Little Ice Age (1645–1710 calendar years)76. It has previously been suggested that photosynthetically active solar radiation contains wavelengths between 400 and 700 nm (visible light)77. The maximum decrease in the ΔTSI during the Maunder Minimum occurred in the ultra-violet region, not in the visible range, which is essential for photosynthesis78. Over the last 8000 years, the observed lag between pollen production and ΔTSI seems to indicate that variations in solar energy have had an indirect and delayed impact on the photosynthetic activity of olive trees. Since ΔTSI is measured from space, it can be extended using ground-based observations such as sunspot numbers, which may explain why a similar delay is observed between olive pollen production and sunspot numbers. As a result, no direct correlation can be drawn between olive production and variations in solar energy (ΔTSI and sunspot number).

Inversely, the amplitude of global geomagnetic activity (measured with the geomagnetic aa index) reflects changes in the solar wind and the heliospheric magnetic field, which are responsible for the modulation of galactic cosmic rays (GCR)79 and, consequently, the cosmogenic radionuclide 10Be. The wavelet coherency connections between olive pollen production and the cosmogenic radionuclide 10Be during the last 8000 years seem to indicate that the pollen production rate was probably affected by GCR through a change in the cloud cover and, therefore, the radiation balance (e.g. based on the incoming and outgoing components of radiation on the Earth system). It has been previously demonstrated that GCR affects the cloud cover in the lower atmosphere80 through the ion-induced nucleation81 and/or aerosol charging mechanism82, and, as a consequence, changing the radiation balance78. The long-term link between olive pollen production and the cosmogenic radionuclide 10Be suggests that the radiation balance should be considered as a key and primary factor for the forthcoming Grand Solar Minimum to estimate how photosynthetic activity and, therefore, food production will be affected, particularly in terms of yields. This is especially important as insolation during March–April (morphological differentiation of flower buds) and September to November (ripening phase) is declining59.

The projected reduction in precipitation and intensification of heat waves in the Mediterranean region83,84 will act in concert with the radiation balance as photosynthesis relies on adequate solar radiation, moderate temperatures and sufficient water—including atmospheric humidity and relative humidity—throughout the different olive phenological stages. Based on past and present data, it has been determined that precipitation between the lower threshold of 290 mm and the upper threshold of 410 mm is necessary for efficient fruit development. The growing season in the Mediterranean dry regions is primarily limited by a lack of precipitation and a high degree of evapotranspiration, which recently reached critical levels30. In the 21st century, temperature increases and consequent higher evapotranspiration will occur during periods when water availability will decrease, leading to heat stress and severe droughts85. Drought has been shown, through complex mechanisms10, to lead to a decrease in photosynthesis, resulting in a substantial reduction in the economic yield of cultivated plants, which poses a major threat to food security86. Heat stress will moreover adversely affect the carbon efficiency of olives, leading to a loss in productivity42. Olive trees can tolerate low soil water availability by developing physiological adaptations to cope with water stress. However, these adaptation strategies are limited and permanent damage has been observed. Olive leaves can permanently lose up to half of their photosynthetic capacity during drought without irrigation87.

With climate change and the approaching Grand Solar Minimum, stresses on the olive sector are set to intensify, particularly under rainfed conditions, potentially threatening its long-term economic viability. This could lead to the decline of traditional olive groves in the near future, with devastating consequences, both environmental (e.g. soil erosion)10 and socio-economic (i.e. reduction in income and employment).

Sustainability

Current and future adaptation to climate change and solar variability is a major challenge facing the agricultural sector worldwide. Olive is a high-value crop currently cultivated in 28 countries worldwide and constitutes a key economic sector for the whole Mediterranean Basin88,89. Recent advances in molecular techniques, such as genetic fingerprinting for identifying cultivars adapted to drought or lower light intensity, and the evaluation of ecosystem services to assess the impact of increasing environmental pressures, biodiversity and ecophysiological parameters are crucial for preserving the olive tree. The olive tree continues to hold substantial values in terms of landscape, nutrition, cultural identity and its historical connection to human societies90. The adoption of appropriate technology for each olive orchard is necessary to enhance the production and the profit of the grower. Incorporating conservation practices, such as using cover crops in olive orchards, is an effective for addressing increasing drought. This approach is particularly beneficial for reducing soil erosion, especially on sloped terrains91. This technique, associated with mulching, ensures a permanent or semi-permanent protective cover over the soil surface, not only controlling soil erosion but also increasing organic matter content92. Soil amendment, including biochar93, olive mill wastes94 and natural zeolites95, also benefits olive orchards by controlling soil degradation, promoting soil remediation, enhancing soil–air–water relations, increasing fertility and nutrient availability, and improving drainage and aggregation96. From an agronomic perspective, one of the keys is olive breeding, favouring essential traits in cultivars including early bearing, high productivity and oil content, disease tolerance and suitability for mechanical harvesting97. Based on these criteria, several new varieties have already been developed in the eastern Mediterranean, such as “Askal”, “Barnea”, “Kadesh” and “Maalot” in Israel98.

For the Mediterranean region, future climate scenarios forecast a 15–20% decline in yields in plantations without irrigation99, with irrigation requirements expected to increase by 18.5%, suggesting that rainfed olive cultivation could become unsustainable100. Among the water-saving measures, deficit irrigation stands out as a viable, short-term and sustainable strategy. Deficit irrigation reduces water consumption by providing irrigation volumes lower than the total evapotranspiration requirements of the olive crop, thus achieving considerable water savings without compromising production101.

One of the keys concerning water deficit and lower light intensity is pruning management. Pruning should focus on balancing vegetative growth and fruiting, managing canopy density, or adapting the tree to a particular harvesting method, but also on improving light distribution within the canopy (photosynthesis-oriented pruning), increasing aeration of the foliage and promoting the development of bearing shoots97,102. In olive orchards, tree canopies should not cause mutual shading in order to maximise light penetration36,37,38,39.

Methods

Cores and chronology

Our pollen database comprises 805 samples from eight fossil records from similar sedimentary deposits (mainly alluvial deposits) with close biostratigraphical resolution facilitating data comparison and integration. All the sites, except for two located farther inland (Fig. 1), are situated near the sea and close to human settlements. The radiocarbon chronologies of the individual cores are detailed in the original publications. The original chronologies (age-depth models) have been preserved in this study. The associated dating error has been averaged to ±80 years based on all the age models.

Olive tree signal

The olive tree data used as a matrix for this study originate from eight different locations across the eastern Mediterranean. These locations include: Bereket basin in Turkey (0–3500 BP)103, Hala Sultan Tekke/Larnaca Salt Lake in Cyprus (500–5800 BP)104, Tell Tweini (2400–4200 BP)105, Jableh (0–1100 BP)106 and Wadi Jarrah (0–1600 BP)107 in Syria, Tyre in Lebanon (1300–7750 BP)20, Tel Dan (1800–4200 BP)108 and Tel Akko (50–6000 BP)109 in Israel (Fig. 1). Although based on eight sites, only one sequence (Tyre) covers the period 6000-8000 BP. Because this sequence shows no significant difference from the entire dataset for the period 1300–6000 BP, as indicated by the standard deviation, we concluded that it effectively represents the evolution of olive pollen production during the period 6000-8000 BP.

The raw olive-pollen data were first transformed into Pollen Accumulation Rates (PAR) in order to derive a local olive tree signal for each locality (expressed as influx—pollen grains per cm−2 per yr−1). For each pollen sample, we calculated the total pollen sum. We subsequently calculated the pollen concentration (grains per cm−3) and an age-depth model to convert olive tree pollen concentrations into PAR. A regular interpolation (10 years) was then performed on each site’s olive signal to facilitate a chronological comparison between the different areas. The different PAR values were averaged to generate a single olive signal (expressed as influx—pollen grains per cm−2 per yr−1) that encompasses all the studied locations and covers approximately the last 8000 years. The associated standard deviation has been averaged to ±60 grains per cm−2 per yr−1 based on all the sequences (Fig. 2a).

While the different eastern Mediterranean areas used here to generate the olive signal host wild and domestic olive subspecies/varieties that have always coexisted, the wild form is indistinguishable from the domesticated subspecies/varieties using pollen analysis110. The same is true for distinguishing between different domesticated subspecies/varieties using pollen analysis. It has also been previously shown that pollen productivity depends on genotypes111,112 and on different cultivars113. However, despite these limitations, environmental pressures influence pollen production in the same way, whether it is a wild or domestic form (although different thresholds may exist). Pollen production will be affected by unfavourable environmental conditions and, conversely, favoured when conditions are more optimal.

Our approach, based on PAR, which can provide direct estimates of past plant abundance114, has some limitations, notably the time interval from flowering to harvest, which is long and during which other climatic and environmental factors may also have played a key role.

Solar irradiance

The solar irradiance datasets were derived from three different sources. The total solar irradiance (ΔTSI), expressed in Wm−2, is from Steinhilber et al.54. The reconstructed 10-year averaged sunspot numbers were extracted from Solanski et al.55. The solar activity Φ [MeV] derives from a record of the cosmogenic radionuclide 10Be published by Vonmoos et al.56.

Olive tree versus solar irradiance

The long-term trend in the olive tree signal was first tested using a moving average function (200-yr; Fig. 2a). The periodicity was calculated using a sinusoidal model (Fig. 2a). The associated 1σ error derives from the standard deviation of the influx. The periodicity of the olive tree signal (Fig. 2b) was further investigated using a wavelet analysis (wavelet transform) based on Morlet as the basis function. A wavelet analysis allows the decomposition of a time series into time-frequency space. It is thus possible to determine both the dominant modes of variability and the way in which these modes vary over time. The scalogram is displayed with a log2 scale and periods against a linear age scale. The cone of influence marks the increasing importance of the edge effects. The significance level (Pvalue = 0.05) is plotted as a contour (chi-squared test). The main periodicity is noted on the graph. A spectral analysis (Lomb periodogram algorithm) was also calculated to analyse the periodicity in terms of frequency and power value. The 0.01 and 0.05 significance levels (white noise lines) are shown as dashed lines (Supplementary Fig. 1a). For time series, obtaining an accurate estimate of their red noise spectrum is challenging. This can be corrected by fitting a first-order autoregressive (AR1) process directly to the time series. REDFIT analysis allows us to test whether the peaks in the spectrum of a time series are significant compared to the red noise background of an AR1 process. A REDFIT analysis (oversampling: 1, segments: 1, window: rectangular) was then performed to reduce the red noise, using the theoretical AR1 red noise model and the chi-squared 90–95% levels (parametric approximation; Supplementary Fig. 1b).

The periodicity of each proxy related to solar activity (10Be, ΔTSI and sunspot number) was also calculated using both sinusoidal model and wavelet analyses (Fig. 3a–c). The long-term cycle of each time series, previously defined by the sinusoidal model, was extracted and compared with the olive periodic signal (Fig. 3a–c). All wavelets were then aligned, and the periodicities related to long-term cycles were highlighted by dotted lines (Supplementary Fig. 2a, b).

Cross-correlations were performed to ascertain the best temporal match between long-term pollen production versus each proxy related to solar activity. The correlation coefficient (R2) is provided, with a focus on the Lag0 value (with the associated Pvalue), and the significant thresholds are set at +0.50 and −0.50.

Violin boxplots were further used to visualise the distribution of each time-series and their probability density (Fig. 4a). Violin boxplots were used to compare and contrast the periods of low or high pollen production with solar activity.

Insolation

The March-April and September-November insolation datasets (expressed in Wm−2) are derived from the astronomical calculation of insolation quantities received by the Earth published by Laskar et al.59. The March–April insolation covers the period corresponding to the morphological differentiation of flower buds. The September–November insolation corresponds to the ripening and harvest season. The March–April and September–November insolation signals were then visually compared with the olive tree signal (Fig. 4b and Supplementary Fig. 3).

Precipitation

The March to November precipitation dataset originates from a composite climate reconstruction published by Kaniewski et al.41. Precipitation in the eastern Mediterranean was reconstructed using pollen assemblages. Initially, drawing on expertise in ecology, botany and palynology, fossil pollen taxa, typically identified to the genus level, are matched to their most likely source plant species. These species and their associated climate ranges are stored in a database. The assigned plant species and their corresponding climate values are then used to calculate a weighted median value and standard deviations for the target climate variable in each fossil sample. The method employed115 incorporates a “leave-one-out calculation” to address temporal fluctuations in pollen taxa within the fossil record. For each fossil sample, this process involves systematically excluding one taxon and calculating a weighted median using the remaining taxa, with pollen percentages serving as weights. This procedure is repeated for every taxon within a given fossil level, resulting in multiple weighted median calculations. The final output is the median value derived from all these iterations. Additionally, the standard deviations used correspond to the median value obtained from the standard deviations generated during these iterations41,115.

The link between olive tree influx and March to November precipitation for the last 8000 years was tested, ranking the precipitation values in ascending order and retaining the associated influx scores. The long-term trend was calculated using a linear model (Pvalue < 10−3) and a polynomial model (order 4, Pvalue < 10−3) with a Pvalue based on an F test - two-tailed, with no adjustment. We then added boxplots to probe the evolution of olive tree influx every 10 mm (Fig. 5a). The periodicity of precipitation was calculated using a sinusoidal model. The long-term cycle defined by the sinusoidal model was extracted and compared with the olive periodic signal (Fig. 5b). The precipitation dataset from the 325 modern olive growing areas (Supplementary Fig. 4) was derived from the data gathered by Kaniewski et al.20.

Reporting summary

Further information on research design is available in the Nature Portfolio Reporting Summary linked to this article.