Although there remain substantial unknowns regarding its exact mechanisms and scale, there is practically no longer any doubt about the reality of climate change, nor about its origin at least partly anthropogenic [1]. This complex phenomenon began slowly during the Industrial Revolution in the mid-19th century, accelerated worldwide from the 1990s onwards, and is expected to further escalate in the future, making it one of the greatest challenges facing humanity in the 21st century [2]. Its main symptoms are rising average temperatures, stronger and longer heat waves, as well as changes in precipitation patterns, with fluctuations both upward and downward in their frequency, abundance, duration, and spatial extent depending on geographical area, but also with an increase in the intensity of different extreme weather events such as thunderstorms, torrential rains, and episodes of severe drought [3]. The decade from 2011 to 2020 was the hottest on record, and the trend has continued to strengthen in the early 2020s, with global surface temperatures breaking records year after year. Under such conditions, the most pessimistic scenarios predict an average global warming of approximately 5°C by the end of the 21st century compared to the reference period of 1850–1900.

Since Wilkinson’s landmark paper in 1989 [4], a growing number of peer-reviewed research papers have emphasized that the combined effects of elevated temperatures and increased carbon dioxide levels stimulate plant growth and reproduction, while disrupting the life cycle of vegetation with usually earlier, prolonged, and more robust flowering seasons [5–7]. To date, apart from a few Korean studies, the majority of publications and those based on the most reliable methodologies have focused on Europe and North America. Their interest is not limited to naturalistic aspects, but also pertains equally to the health sector. Indeed, for trees, grasses, and weeds that release allergenic pollen, changes in the characteristics of pollination, both in its abundance and timing, may already have, and will continue to have, significant consequences on the prevalence and severity of allergies [8, 9]. Two other potential impacts of climate change have been described as likely to either worsen or reduce allergies, depending on the context. One relates to an increase or, in other cases, a decrease in the allergen content of pollen grains [10]; however, unlike pollen concentrations, there is still no routine detection and quantification of the allergens in pollen grains, nor of the allergen content of the air. The other refers to the frequent shift towards higher latitudes and elevations in the spatial distribution of at least some wind-pollinated species, whether they are wild or cultivated [11]. For instance, simulations have shown that, by 2085, changes in temperature and precipitation could lead to modifications in vegetation types across 31 to 42% of Europe [12], with the obvious implication that when the range of an allergenic species expands or contracts, the risk of allergy is profoundly altered.

Without claiming to be exhaustive, which would far exceed the scope of an article, this work will provide an objective overview and a quantitative synthesis of the available literature on the evolution of the pollen season intensity and timing of plants with a higher allergenic potential. The focus will primarily be on the last few decades, but attention will also be given to projections concerning the most foreseeable future, roughly until the years 2040–2060.

The three bibliographical databases Medline^®, PubMed^®, and Google Scholar^® were searched from inception until August 15, 2025, to identify studies published in peer-reviewed journals, conference proceedings, or books that have investigated the impact of ongoing and future climate change on allergenic pollen. The following search string was used, with some modification as necessary (e.g., ~ instead of * in Google Scholar^®): (climat* change OR warming OR carbon dioxide OR trend) AND (pollen OR flower* OR sensitization). Additional articles or book chapters were also retained when they were cited in the previously selected publications.

After removing duplicates, the full texts of all studies deemed potentially relevant based on their titles were assessed to determine their eligibility. Four inclusion criteria were used for this purpose:

(1)

Studies were required to be available online in the English language. A subsequent query of the databases led to the inclusion of five more references published in French and one in Spanish, because they provided important data for which, to our knowledge, there is no equivalent in any English-language publication. All of them contained an abstract in English.

Although other potential indicators of the impact of climate change on pollination can be found in the literature, such as the end date of the pollen season or the date and intensity of the annual peak, only the three most studied parameters have been retained here, for which the terminology recommended by European and international aerobiology authorities [15] has been adopted. These three parameters are (1) the pollen integral (PIn), which, expressed in number of grains per cubic meter of air, is obtained by summing the daily pollen counts for the entire year or the main pollen season; (2) the pollen season start date (PSS, calculated based on the number of days elapsed since January 1st); and (3) the pollen season duration (PSD, corresponding to the number of days that separate the beginning and end of the season).

For each eligible study, the following data were extracted and compiled, taxon by taxon, in large tables: reference in brackets and year of publication, location of the study and number of monitoring stations involved, study period, investigation method, trend identification technique, temporal trends observed successively for PIn, for PSS, and for PSD. Long-term trends were presented in a standardized form (increase or decrease in the number of grains or the number of days per decade), and their statistical significance was noted as calculated by the authors (p < 0.05, p < 0.01, or p < 0.001). When boxes are left empty in the tables, it means that the parameter in question has not been studied in the article analyzed, without the authors specifying the reasons; this applies especially to the statistical significance of the results presented. Simple linear regression was by far the most used statistical method for trend analysis in the reviewed publications. It has been quite often supplemented by the non-parametric Mann-Kendall test with the Theil-Sen slope estimator for detecting the presence of monotonic upward or downward temporal trends, as these tools are considered the most robust for heterogeneous time series data. Spearman’s rank correlation analysis was also regularly performed to determine possible relationships between the meteorological data recorded before and during the main pollen season and the aerobiological variables described for each taxon. Bayesian statistics have sometimes also been used to highlight discontinuities within the series and quantify the direction and speed of changes. However, many other statistical methods have been applied (e.g., reduced major axis linear regression, seasonal-trend decomposition procedure based on LOESS technique, canonical correlation analysis, etc.), so that it was not possible to differentiate the trends based on the method used, which varies too much from one study to another.

In the temperate regions of the Northern Hemisphere, pollen from the Betulaceae family is almost always both the most abundant and the primary cause of respiratory allergies that occur in late winter or early spring. However, given that this pollen can be easily identified at the genus level under optical microscopy, only three studies [16–18], which are moreover incomplete, have focused on the family level. They report (Table 1 and Figure 1) a pollen production that is strengthening over time (up to +1,400 grains/m³ per decade in Emilia-Romagna [16]) and beginning increasingly earlier, although these trends are not predominantly significant and no clear conclusions can be drawn about the PSD.

However, there is strong evidence that pollination of all trees in this family has not always evolved in the same way over recent decades.

The Cupressaceae family includes a significant number of species that are cultivated as ornamental plants in urban environments; other species occur naturally. Although pollen allergy from this family was considered a rarity until around 1975, especially in Europe, it is now recognized as an indisputable clinical entity in many regions, such as the Mediterranean basin, Texas, or New Mexico [87]. However, one should be cautious about the fact that the various studies analyzed here are not strictly comparable. Indeed, the pollen grains from Cupressaceae are difficult to differentiate under the light microscope at the species or genus level, and also difficult to distinguish from those of the Taxaceae or Taxodiaceae because of their close morphological similarity. In routine pollen counts, all these taxa are usually grouped under the name Cupressaceae-Taxaceae but, while some authors do use this designation, others refer to a specific genus, Cupressus (cypress), for instance, [20] or Juniperus (juniper) [27], without specifying how it has been identified, or remain vague concerning the content they ascribe to the term Cupressaceae.

Despite the diversity of species classified as Cupressaceae, most of the pollen collected in the aerobiological traps comes from one or another variety of cypress: other ornamental species are less abundant, and their periodical pruning eliminates many of their flowers, while wild species pollinate with less intensity, and usually grow far away from sampling sites. Furthermore, the general trends (Table 2 and Figure 1) are sufficiently strong to leave virtually no room for doubt. Thus, the PIn is oriented upwards in 83.3% of the 24 publications dealing with this point [20, 26, 27, 36, 37, 47, 51, 56, 62, 74, 81, 86, 88–95], often with a highly significant progression exceeding 2,000 grains/m³ per decade in Basel [37], as well as in Milan [88] and in Oklahoma [92], or even 5,000 in Southern Spain [91] and 8,500 in Granada [74]. It is sometimes noted that the trend has accelerated after 2001 [88] or after 2012 [89]. While three studies concluded that there was no clear trend [57, 78, 96], only one reported a significant local decline in Málaga [86].

Slightly fewer, reduced to 17, are the publications that have focused on the pollen calendar of Cupressaceae. It is true that the successive flowering of the different species means that this taxon is present in the atmosphere for an extended period (generally from Autumn to Spring, and even into Summer for certain junipers), to the point that it becomes difficult to delineate a main pollen release season. But here again a broad consensus emerges, with 76.4% of studies in favor of an increasing precocity of the onset of pollination [20, 26, 37, 47, 51, 57, 62, 73, 86, 92–95], the record coming from Neuchâtel with an advance of 12.9 days every ten years [26]. Only two papers report a diametrically opposite trend; the first [74] mentions an average delay of 1.6 days per decade in the Iberian Peninsula from 1994 to 2023, while the other [97] largely aligns with this observation regarding the specific case of Badajoz between 1993 and 2013.

Twelve publications address the PSD of the Cupressaceae. Four of these do not highlight any trend [26, 57, 81, 92], while the remaining eight [20, 37, 51, 62, 90, 92, 94, 95] indicate a lengthening of between 3.25 and 17 days per decade, with the record coming from Madrid.

Finally, two studies have focused on simulating the effects of various climate warming scenarios on the start date and, incidentally, on the PSS and, partly, on the PSD of Cupressaceae. The first one [73], referred to the juniper in the Western part of the Netherlands and without providing much explanation about the method used, anticipates a very sharp slowdown in the advance of the onset of pollination which, after being 6.25 days every ten years between the 1970s and the 1990s, would decrease to 0.8 days per decade in the 21st century, somewhat abusively regarded as a homogeneous whole. The second study [98] relied on a phenological model capable of simulating the development of the male flower of Cupressus sempervirens (Mediterranean cypress) as a function of the average daily temperature from October 1st to the end of the flowering period. The results indicate for the Florence region, in Central Italy, that the pollen season would become increasingly earlier (from mid-February today to early January in the middle of the century and to mid-December by the end of the century). The duration of this season, on the contrary, would remain virtually unchanged.

Typically, within the Oleaceae family, only Olea (olive tree) pollen is considered a significant aeroallergen. However, there is at least one other species, namely Fraxinus (ash), which can constitute a significant source of airborne allergens, especially since cross-reactivity between various species in this family is now a proven fact. The two publications dealing with the Oleaceae family (Table 3 and Figure 1), without delving into the genus level [18, 51], thus also including Ligustrum (privet), conclude that pollen seasons are gradually becoming more productive and starting earlier. However, they do not allow for a determination regarding the duration of these seasons, which are sometimes shortened (in seven out of ten studied European sites) and sometimes lengthened (in the Veneto region, Northern Italy).

Beyond inevitable nuances, the studies dedicated separately to Olea and Fraxinus agree with these trends.

The Poaceae or Gramineae family (grasses), found worldwide, occurs in all terrestrial biomes. It is probably its near-ubiquity that makes it the most common source of pollen allergies worldwide, affecting around 80% of allergy sufferers [112]. Within this family, pollen grain determination at the genus level, and even more so at the species level, proves to be very difficult, if not impossible, using conventional aerobiological methods. It would therefore be unrealistic to claim to assess the partial contribution of each species to the airborne pollen spectrum, which constitutes a significant drawback given that not all species have the same allergenic potential or sensitivity to climate change.

The prevailing trend in the annual PIn (Table 4 and Figure 1), documented in 41.2% of publications [16, 20, 22, 25, 27, 28, 30, 37, 40, 41, 47, 52, 56, 61, 74, 85, 92, 113–116], is a decline, without a marked difference between the North and South of Europe, although the pollen spectrum of grasses differs notably between these two geographical areas. Most often, the decline remains fairly moderate and not statistically significant (averaging about –119 grains/m³ per decade across twelve stations in the Iberian Peninsula [74]), but there are notable exceptions (–1,588 grains/m³ every ten years in Badajoz [74], –1,620 in Perugia, Central Italy [116]; consequently, in the latter case, the annual amount of pollen collected in recent years has decreased by half compared to the end of the 20th century). The opposite trend is slightly less frequent, as an increase in concentrations was observed in only 35.3% of studies [24, 31, 36, 38, 44, 45, 51, 55, 62, 63, 81, 90, 91, 93, 94, 100, 117, 118], but it is sometimes very pronounced, reaching over +15,500 grains/m³ in ten years in Córdoba, albeit based on a relatively short series [91]. No trend was identified in 23.5% of publications [19, 23, 26, 57, 59, 82, 95, 96, 109, 119–121]. It should also be noted that in several studies, the authors report contradictory trends depending on the site considered [18, 42, 50, 122] and/or reversals in trends from one decade to the next [22, 34, 43, 122]. Thus, since the late 1980s, the annual pollen count for grasses has approximately doubled in North-Eastern and Southern Germany, while it has sharply declined (–2,180 grains/m³ per decade) in the North-West of the country [42]. Similarly, at the East Midland locations of Derby and Leicester in the United Kingdom, where a 52-year series could be utilized, PIn initially decreased from the late 1970’s to the early 1990’s, before entering a phase of vigorous growth [43].

The PSS brings us back to a pattern close to that described for trees. Indeed, 80.5% of publications addressing the subject conclude that pollination is occurring increasingly earlier [18–20, 24, 26, 28, 35, 40, 42, 44, 45, 47, 51, 57–59, 61–63, 73, 74, 83, 90, 92–95, 114–116, 118, 119, 123], while only 7.3% report a postponed onset [16, 96, 113]. There are still very few differences between Northern Europe and Southern Europe. Once again, the trend may be positive, negative, or neutral depending on the site, whether in France [22], in the United Kingdom [122], throughout Western Europe [124] or in the United States [55], and some trend reversals over time can be detected, for instance in Central England where grass shows a U-shaped trend with the earliest start of season near 1990 [43]. Lastly, and notwithstanding the reservations already expressed about this work, the sole article that attempted to look ahead predicted for the Netherlands a potential continuation of the trend toward increasing earliness, at an average rate of 1.1 days per decade until the 2090s [73].

Because the numerous species of the family bloom at different times, the grass pollen season almost always lasts longer than that of other taxa, particularly in Southern Europe. It is therefore more difficult to propose a simple scheme to characterize the evolution of the duration of this season. In fact, 37.9% of cases show an extension [28, 42, 44, 47, 51, 62, 92, 94, 118, 119, 123], 27.6% show a decrease in length [37, 57, 59, 83, 95, 96, 113, 115] and 34.5% show unchanged duration [16, 20, 26, 61, 81, 90, 100, 114, 116, 120]. Furthermore, it is not uncommon for trends to change in magnitude, and sometimes even in direction [18, 22, 34, 55] over short distances. But this time, a significant difference emerges between Northern Europe (where the pollen season is prolonged in 58.3% of studies) and Southern Europe (where it is prolonged in only 29.4%). Given that the start dates of this season are relatively stable, it can be inferred that the end dates are often earlier or remain unchanged in the South, likely due to the combined effect of higher temperatures and greater water stress.

At least in Europe, the Urticaceae family includes two main genera, Urtica (nettle) and Parietaria (pellitory). Both are widely cosmopolitan, but in practice, only Parietaria plays a significant role in triggering allergic diseases. Regrettably, genus differentiation based on pollen morphology remains difficult, even though the two genera have very slight differences in size (only U. membranacea is easily identified). In aerobiological databases, the name Urticaceae is therefore systematically used to group together many species. Nevertheless, pellitory (at least P. officinalis) is more abundant in the South than in the North of Europe, whereas the opposite is true for nettle. This has led us, albeit as a makeshift solution, to consider only the studies conducted in the Mediterranean region, thereby increasing our certainty that most allergenic pollen is labelled under Urticaceae.

Hence, unlike grasses, a slight predominance of the tendency toward increased pollen concentrations (Table 5 and Figure 1) is observed for the Urticaceae, present in 42.9% of cases [20, 81, 82, 93, 100, 121] compared to 35.7% for the opposite trend [36, 91, 94, 96, 125] and 21.4% for no trend [51, 57, 59]. A special mention is warranted for Córdoba, where PIn recorded a considerable drop estimated at 41,500 grains/m³ over ten years [91], broadly corroborated by another study [36].

Pollen calendars are marked, in more than three-quarters of cases, by an increasingly early onset of flowering of the Urticaceae [20, 51, 57, 59, 74, 83, 90, 93, 94, 100], with a progression reaching up to one month every ten years in Western Liguria [20]. In contrast, Vigo, located at the Northwestern tip of Spain, which shares the same latitude as the Mediterranean but has a significantly different climate, has recorded an average delay of around thirty days per decade [125]. A very early pollination of Urticaceae, which are mainly believed to be Parietaria, is regularly associated with an extension of the pollen season [20, 51, 59, 83, 100], with the increase ranging from less than 1 to more than 31 days per ten-year period. However, there are also sites where this duration remains unchanged [57, 81, 96], and others where it tends to decrease [94, 125]. The cases of Malaga and Vigo still appear to be anomalous, with pollen seasons becoming both later and shorter [125]. In any case, such findings challenge the widely held belief, especially in Italy, that the pellitory season has extended to cover the entire year [126].

The Asteraceae family, which includes 12,000 species classified into 780 genera, approximately, is one of the largest groups of flowering plants with members across all continents, but particularly in Northern temperate latitudes. Most are entomogamous and are unlikely to cause anything other than contact or proximity allergies. However, there are at least two genera, Artemisia (mugwort) and Ambrosia (ragweed), which are wind-pollinated and are prone to causing respiratory allergies in late summer and autumn. These are indeed short-day plants with a generative growth phase induced after the day length decreases and the summer temperatures have reached their peak.

Only a single publication [18] is dedicated to the evolution of pollination in the Asteraceae family without differentiation. It covers about ten European sites and observes, from 1990 to 2009, both a decrease in airborne pollen concentrations (in 8 of the studied locations), an earlier onset of the seasons (in 6 of these sites), and a lengthening of the exposure period to one or another pollen of this family (everywhere except Prague); however, it is not possible to determine the respective contributions of mugwort and ragweed to these trends.

The synthesis of available data indicates a dominant trend toward earlier and more abundant pollen seasons, particularly for winter-spring flowering trees such as birch, cypress, and olive. In contrast, trends for late-pollinating herbaceous species such as grasses or mugwort are less consistent and often region-specific. These variations are taxon-, site-, and period-dependent, with some species even showing opposing trends within the same botanical family, illustrating the complex interactions between biological adaptation and climatic variability. Moreover, it cannot be denied that this literature review has its limitations, due to three factors: the heterogeneity of the methods used, both for the characterization of pollen seasons and for the statistical demonstration of trends; the heterogeneity of the series studied, both in terms of their date and their duration; and the overrepresentation of certain regions, particularly the Mediterranean area. The conclusions stated must therefore always be approached with great caution, and are confined to the most general trends. While the current influence of climate change on pollen production and phenology is well established, the magnitude of its future impact remains uncertain, and the diversity of methodologies and study durations limits the comparability of available data. Nevertheless, most projections support a continued, though possibly attenuated, increase in pollen intensity and season advancement, suggesting partial ecological adaptation.

Despite frequent divergent findings from one sampling location to another, the timing of pollen seasons is primarily dependent on temperature changes measured one to five or six months, or even eight to ten months, before the start of flowering [61, 79, 118]. The example of the Betulaceae is particularly illustrative in this regard. While flower and anther maturation in hazel, alder, and birch tends to be delayed after mild autumns and early winters, with longer vernalization periods [77], it occurs earlier as temperatures rise from mid-winter onwards, with shorter forcing periods [17, 40]. A similar pattern has been observed for mugwort, whose the start of pollination depends closely on the temperature recorded as early as February, while it only flowers in the middle of summer [40]. It should be noted that the start of the pollen season, all species combined, is also influenced by factors affecting short-term pollen release, such as rainfall or lack of wind, and may not coincide exactly with the maturation of the flowers. With regard to the quantities of pollen released, it was noted that higher mean temperature in the previous summer was often significantly associated with a higher seasonal integral for tree pollen, while higher mean temperature in the previous winter was associated with a higher seasonal integral for grass pollen [43]. Changes in rainfall patterns due to global warming are typically less correlated with pollen characteristics than temperature changes are, due largely to the complexity of rainfall distribution over space and time. However, water availability during the period preceding pollination most often promotes the development of grasses and weeds, and therefore their pollen production [40], while a large amount or prolonged precipitation during the flowering period can significantly reduce the amount of pollen in the air. Finally, and this applies to all selected species, there is experimental evidence to suggest that increasing CO₂ concentration in the atmosphere stimulates photosynthesis and growth rates, reduces the time required to reach the minimal critical size for floral induction, increases the number of flowers, and enhances pollen production [35, 59, 79, 113]. Nevertheless, it is not always easy to disentangle the direct impact of carbon dioxide (fertilizing effect) from its indirect impact (which involves global warming).

Furthermore, while the trends described for the intensity and timing of pollen seasons are largely undisputed regarding changes in pollen exposure, it would be a serious mistake to attribute all or even most of this to climate change. Many other explanatory factors exist, which operate in very variable proportions. Changes in local or regional land use and land cover play a key role in this regard [91, 93], as do increasing urbanization (and the increasing greening of cities where pollen traps are located), air pollution (which boosts pollen production, at least in certain species), and biological invasions, which are themselves partly, but only partly, linked to climate change.

The evolution of allergic sensitization, however, has been less frequently investigated, particularly in relation to its connection with climate change [20, 23, 88, 95, 100, 121, 138–140]. In Western Liguria, analysis of over 25,000 skin prick tests (1981–2007) revealed increased sensitization to Parietaria, birch, olive, and cypress, species with rising airborne pollen concentrations, while grass sensitization remained stable in line with unchanged Poaceae pollen loads [20]. Other studies in the same region found no consistent link between pollen trends and sensitization frequencies. In Central Germany (1998–2017), sensitization to birch, hazel, alder, and grasses increased markedly despite stable pollen counts [23]. Overall, available evidence suggests that pollen exposure and sensitization prevalence do not always evolve in parallel, reflecting the multifactorial nature of allergic disease development. A better understanding of sensitization profiles and their temporal evolution is therefore needed to clarify how environmental and climatic drivers shape allergic susceptibility across populations.

Beyond these gradual shifts, extreme weather events such as thunderstorm asthma represent an emerging manifestation of climate-related risk [141]. During such events, pollen grains can rupture into respirable allergenic particles, triggering acute asthma attacks even in individuals with mild or previously unrecognized allergic disease. The growing frequency of these phenomena underscores the need for integrated surveillance that combines aerobiology, meteorology, and clinical reporting, as well as for targeted public health preparedness.

In the years and decades to come, the anticipation of a higher burden of allergic diseases will inevitably influence both clinical practice and public health planning [142]. Allergen immunotherapy (AIT), the only treatment capable of modifying the natural course of allergic disease, will remain central to managing climate-related changes in allergic respiratory conditions. Earlier and more intense pollen exposure may require advancing the initiation of AIT or adjusting treatment doses and duration. But there are many other areas where the medical profession, as a reliable source of knowledge, can contribute to the implementation of effective primary and secondary prevention strategies. Among other examples, physicians have a key role to play in ensuring that, in urban and peri-urban environments, ornamental plantings contain an increasingly smaller proportion of species that are not only highly allergenic but also highly sensitive to global change. Ultimately, strengthening longitudinal pollen monitoring, harmonizing methodologies, and fostering collaboration between aerobiologists, clinicians, environmental professionals, urban planners, and policymakers will be essential to improving forecasting, prevention, and personalized treatments in a rapidly changing environment.