Pollen Production of Quercus in the North-Western Iberian Peninsula and Airborne Pollen Concentration Trends during the Last 27 Years

Natural forests are considered a reservoir of great biological diversity constituting one of the most important ecosystems in Europe. Quercus study is essential to assess ecological conservation of forests, and also of economic importance for different industries. In addition, oak pollen can cause high sensitization rates of respiratory allergies in pollen-allergy sufferers. This study sought to know the pollen production of six oak species in the transitional area between the Eurosiberian and Mediterranean Bioclimatic Regions, and to assess the impact of climate change on airborne oak pollen concentrations. The study was conducted in Ourense (NW Spain) over the 1993–2019 period. A Lanzoni VPPS 2000 volumetric trap monitored airborne pollen. A pollen production study was carried out in ten trees randomly selected in several Quercus forest around the Ourense city. Oak pollen represented around 14% of annual total pollen registered in the atmosphere of Ourense, showing an increasing trend during the last decade. Pollen production of the six studied oak species follow the proportions 1:1:2:5:90:276 for Q. ilex, Q. faginea, Q. rubra, Q. suber, Q. pyrenaica, and Q. robur respectively. We detected a significant trend to the increase of the annual maximum temperature, whereas a decrease of the maximum and mean temperatures during three previous months to oak flowering. This could be related with the detected trend to a delay of the oak Main Pollen Season onset of 0.47 days per year. We also found significant trends to an increase of the annual pollen integral of 7.9% pollen grains per year, and the pollen peak concentration of 7.5% pollen grains per year. Quercus airborne pollen monitoring as well as the knowledge of the reproductive behavior of the main oak species, bring us an important support tool offering a promising bio-indicator to detect ecological variations induced by climate change.


Background
Oak is the dominant tree in many of Europe's deciduous forests [1]. Natural oak forests are considered a reservoir of great biological diversity, constituting one of the most important ecosystems in Europe. In Northwestern Spain, the presence of the Quercus tree species in forests is an indicator of

Impact on Human Health
In addition, oak pollen can induce respiratory allergies in many countries [4]. Although it is traditionally assumed that pollen from European Quercus species has low allergenicity, some researchers noted the potential responsibility of oak for major cases of allergies in areas with abundant Quercus vegetation [21,22]. Previous studies demonstrated that oaks produce a high amount of pollen grains per flower (between 1.9 × 10 4 and 1.3 × 10 6 pollen grains) depending on the species [23,24]. Furthermore, symptoms appear frequently over long periods due to flowering overlaps of the different species [4] and their related reactivity with pollen from Alnus, Betula, Castanea, Olea and Poaceae [25]. This fact increases their allergenic potential due cross-reaction processes and the so-called "priming Forests 2020, 11, 702 3 of 18 effect" [2,7,26]. In Northwest Spain, the sensitization rate to oak pollen among pollen-allergy sufferers, assessed by skin-prick tests, varies between 4% to 12% [27,28]. This high sensitization rate indicates the importance of research in oak pollen allergy [29]. Information about pollen production and airborne dispersion will allow us to enhance the prevention of clinical pollinosis symptomatology and to reduce medicine consumption.

Goal of the Study
This study plays an important role since it analyzes an abundant taxon in the sampling area whose pollen is considered to be responsible for major cases of allergies. Since pollen masses can increase the productivity of an ecosystem, both terrestrial and aquatic, we studied the pollen production of the most important six oak species in the transitional area between the Eurosiberian and Mediterranean Bioclimatic Regions in Northwest Spain. We also assessed the impact of climate change on the seasonal dynamics of airborne oak pollen.

Location and Classification of the Study Area
This study was conducted in the Ourense province, located in a depression at 139 m above sea level in North-western Spain (42 • 20 N; 7 • 52 W) in a transition zone between the Eurosiberian and Mediterranean Bioclimatic Regions ( Figure 1). The climate is oceanic with strong Mediterranean influence. According to the Worldwide Bioclimatic Classification System, Ourense presents a very warm temperature and has low humidity, with an annual mean temperature of 14 • C and annual rainfall around 772 mm [30]. This high sensitization rate indicates the importance of research in oak pollen allergy [29]. Information about pollen production and airborne dispersion will allow us to enhance the prevention of clinical pollinosis symptomatology and to reduce medicine consumption.

Goal of the Study
This study plays an important role since it analyzes an abundant taxon in the sampling area whose pollen is considered to be responsible for major cases of allergies. Since pollen masses can increase the productivity of an ecosystem, both terrestrial and aquatic, we studied the pollen production of the most important six oak species in the transitional area between the Eurosiberian and Mediterranean Bioclimatic Regions in Northwest Spain. We also assessed the impact of climate change on the seasonal dynamics of airborne oak pollen.

Location and Classification of the Study Area
This study was conducted in the Ourense province, located in a depression at 139 m above sea level in North-western Spain (42°20′ N; 7°52′ W) in a transition zone between the Eurosiberian and Mediterranean Bioclimatic Regions ( Figure 1). The climate is oceanic with strong Mediterranean influence. According to the Worldwide Bioclimatic Classification System, Ourense presents a very warm temperature and has low humidity, with an annual mean temperature of 14 °C and annual rainfall around 772 mm [30].

Pollen Production Study
A pollen production study was carried out in several oak forests around the city of Ourense. The selected populations belonged to the main species represented in North-western Spain: Quercus robur L., Quercus pyrenaica Wild, Quercus ilex subsp. ballota, Quercus suber L., Quercus faginea Lam. and Q. rubra L. The most prominent forests belongs to Q. robur with 61,904 ha and Q. pyrenaica with 544,943 ha [31]. In each populations, ten trees were randomly selected to estimate pollen grains per anther, anthers per flower, flowers per catkin, pollen grains per catkin, catkins per tree and pollen grains per tree. The methods developed by Cruden [32] and Hidalgo [33] were used to estimate pollen production per anther in three catkins per tree. Three flowers were selected on each catkin, and three anthers of each flower were sampled. Therefore, pollen production was assessed with 27 anthers per tree (3 catkins × 3 flowers/catkin ×3 anthers/flower). Anthers from fresh flowers close to the anthesis were collected, placed in the bottom of a test tube and 1 mL of distilled water colored with 1% basic fuchsine was added. Anthers were crushed and the suspension was shaken to achieve the highest homogeneity. Then we deposited 10 microliters of the suspension in a slide and counted all pollen grains at 400× magnification with an optical microscope. We repeated the pollen count three times for each dilution, considering the average value of the three replicates for each sample. To obtain the number of anthers per flower, we sampled three flowers of three catkins from each tree. To estimate the number of flowers per catkin we counted the total number of flowers in ten catkins of each tree in the studied populations. To estimate pollen production per tree we assessed the number of catkins on three branches of each tree, considering the average of catkins per branch and the number of branches per tree.

Airborne Pollen Study
Airborne pollen has been continuously monitored over a period of 27 years (from 1993 to 2019) by means of a 7-day VPPS 2000 volumetric pollen trap (LANZONI, Bologna, Italy) [34], located on the roof of the Sciences Faculty in Ourense (approximately 20 m above ground level). The device has a vane tail (orientating the sampler according to the wind direction) and a built-in vacuum pump that has a suction flow rate of 10 L/min. By passing through a narrow intake orifice (2 × 14 mm), the sampled air impacts a clock-driven drum rotating at an angular velocity of 2 mm/h, taking 7 days to perform a complete turn. The drum is covered with a Melinex tape coated with an adhesive solution where airborne pollen is retained. After exposure, this tape is cut into segments 48 mm long, representing 1 day of sampling, that are then mounted on a microscopic slide with a glycerogelatine-basic fuchsin mounting media and a cover glass. Pollen counts were conducted following the methodology proposed by the Spanish Aerobiological Network (REA) [35] based on four equidistant longitudinal transects along the slides. Pollen data were expressed as pollen grains per cubic meter of air (pollen grains/m 3 ). The Main Pollen Season (MPS) was determined using the Andersen [36] method, which defines MPS as the period from the accumulated sum of daily mean pollen concentrations that reach 2.5% of the annual total to the day when 97.5% of the annual total is achieved. The main characteristics of the airborne oak MPS were estimated during the studied period: start date, end date, length in days, annual pollen integral, and the annual pollen peak and its date.

Meteorological Data and Statistical Analysis
Meteorological data were obtained from the Galician Institute for Meteorology and Oceanography METEOGALICIA "Ourense" station, placed at 300 m of the pollen sampler. Measured parameters were temperature ( • C), relative humidity (%), rainfall (mm) and sun hours (h). With the aim to ascertain the degree of association between the studied reproductive variables of the different sampled oak species, a non-parametric Spearman's correlation test was carried out (considering p values of p < 0.1 and p < 0.05). The analysis was conducted considering the single species and all species as a whole with the objective to detect possible reproductive strategies. In addition, a linear regression analysis was conducted to estimate increase or decrease trends of the aerobiological and meteorological variables during three different periods (1993-2019; 1993-2004 and 2005-2019). Statistical analysis was performed by means the STATISTICA 7 10.0 program (Dell, Newton, MA-Boston, USA).

Pollen Production Study
The pollen production study of the six oak species analyzed in different forests around the city of Ourense showed wide variations between species (Table 1 and Figure 2). Table 1. Average, minimum (Min), and maximum (Max) values, standard deviation (S.D.) and the relative standard variation in percentage (%RSD) of the pollen per anther, anthers per flower, flowers per catkin, pollen per catkin, catkins per tree and pollen per tree of the studied Quercus species (robur, pyrenaica, rubra, faginea, ilex and suber) in the North-western Spain.  The average pollen amount per anther ranged between 3400 pollen grains registered for Q. robur to 6282 recorded in Q. ilex. The lowest relative standard deviation between trees of the same species was estimated for Q. suber and the highest for Q. ilex. The number of anthers per flower ranged between 5 and 8, with the highest quantity registered for Q. robur and Q pyrenaica and the lowest for Q. rubra and Q. suber. The standard deviations between the sampled trees of a given species were the lowest among the six analyzed flowering parameters. Regarding the quantity of flowers per catkin the highest number was estimated for Q rubra, whilst the lowest was estimated for Q. ilex. The number of pollen grains produced per catkin fluctuated from 355,237 pollen grains obtained for Q. suber to 716,385 pollen grains registered for Q. rubra. The average pollen amount per anther ranged between 3400 pollen grains registered for Q. robur to 6282 recorded in Q. ilex. The lowest relative standard deviation between trees of the same species was estimated for Q. suber and the highest for Q. ilex. The number of anthers per flower ranged between 5 and 8, with the highest quantity registered for Q. robur and Q. pyrenaica and the lowest for Q. rubra and Q. suber. The standard deviations between the sampled trees of a given species were the lowest among the six analyzed flowering parameters. Regarding the quantity of flowers per catkin the highest number was estimated for Q. rubra, whilst the lowest was estimated for Q. ilex. The number of pollen grains produced per catkin fluctuated from 355,237 pollen grains obtained for Q. suber to 716,385 pollen grains registered for Q. rubra. The relative standard deviation between trees of the same species ranged from 23.3% to 42.3%. The number of catkins per tree showed important variations as a consequence of the size of mature individual trees in each oak specie. The highest catkin production was estimated for Q. robur with 667,231 catkins, whereas the lowest was detected for Q. ilex with only 3770. The number of catkins per tree was the parameter with the largest variations of the relative standard deviation. Finally, the highest number of pollen grains per tree was estimated for Q. robur with an average of 436 billion pollen grains per tree and the lowest was observed for Q. ilex with only 1.6 billion pollen grains produced. Important differences regarding the relative standard deviation were also registered in the case of pollen produced per tree.

Catkins/ Tree
Spearman's correlation analysis was applied to ascertain the degree of association between the evaluated flowering parameters, considering data of the six oak species altogether, and several significant correlations were found (Table 2). Negative, significant correlations were observed between pollen grains per anther with number of anthers per flower as well as number of flowers per catkin. A positive, significant correlation was obtained between pollen grains per anther and pollen grains per catkin. The number of anthers formed per flower positively correlated with the number of catkins per tree and pollen production per tree. The number of flowers per catkin positively correlated with pollen grains per catkin. Finally, a positive correlation was found between the number of catkins per tree and the final pollen production per tree.

Airborne Pollen Study
Oak airborne pollen was present in the atmosphere of North-western Spain during spring, mainly in March, April and May ( Figure 3). The average seasonal behavior of oak pollen in the atmosphere is shown in Figure 2. During the 27 years of study, the MPS started on March 26th and ended on June 7th (Table 3). The relative standard deviation between trees of the same species ranged from 23.3% to 42.3%. The number of catkins per tree showed important variations as a consequence of the size of mature individual trees in each oak specie. The highest catkin production was estimated for Q. robur with 667,231 catkins, whereas the lowest was detected for Q. ilex with only 3770. The number of catkins per tree was the parameter with the largest variations of the relative standard deviation. Finally, the highest number of pollen grains per tree was estimated for Q. robur with an average of 436 billion pollen grains per tree and the lowest was observed for Q. ilex with only 1.6 billion pollen grains produced. Important differences regarding the relative standard deviation were also registered in the case of pollen produced per tree.
Spearman´s correlation analysis was applied to ascertain the degree of association between the evaluated flowering parameters, considering data of the six oak species altogether, and several significant correlations were found (Table 2). Negative, significant correlations were observed between pollen grains per anther with number of anthers per flower as well as number of flowers per catkin. A positive, significant correlation was obtained between pollen grains per anther and pollen grains per catkin. The number of anthers formed per flower positively correlated with the number of catkins per tree and pollen production per tree. The number of flowers per catkin positively correlated with pollen grains per catkin. Finally, a positive correlation was found between the number of catkins per tree and the final pollen production per tree

Airborne Pollen Study
Oak airborne pollen was present in the atmosphere of North-western Spain during spring, mainly in March, April and May (Figure 3). The average seasonal behavior of oak pollen in the atmosphere is shown in Figure 2. During the 27 years of study, the MPS started on March 26th and ended on June 7th (Table 3).  (Table 3).   (Table 3). Table 3. Characteristics of the airborne oak MPS during the 27 years of study: start date, end date, length, annual pollen integral (pollen grains), daily pollen peak (pollen/m 3 ) and its date. Average value (Mean), minimum value (Min), maximum value (Max), standard deviation (SD) and relative standard variation in percentage (%RSD) of sampled trees for each species were also shown. Oak pollen represented, on average, around 14% of the annual total pollen registered in the atmosphere. The percentage varied substantially from year to year, but the highest rates were observed during the last decade ( Figure 4). Oak pollen represented, on average, around 14% of the annual total pollen registered in the atmosphere. The percentage varied substantially from year to year, but the highest rates were observed during the last decade ( Figure 4).  The annual pollen integral was also highly variable, ranging between 18,269 pollen grains registered in 2017 to 1296 observed in 1998 (Table 3). Peak pollen concentration was generally observed in April, with the latest record on 16 April 2014, with 1250 pollen/m 3 .

Date
The intensity and temporal trends of the oak MPS features and the main meteorological parameters considering the entire year and the three months prior to flowering (January, February, and March), were calculated by means of simple linear regressions (Table 4).  With the purpose to achieve a deeper understanding of the climate change effects on oak flowering, trends were calculated using the entire 27 years of data (1993-2019) as well as considering separately the first 13 years (1993)(1994)(1995)(1996)(1997)(1998)(1999)(2000)(2001)(2002)(2003)(2004)  The total rainfall recorded during January, February and March seemed to strongly affect the total oak pollen integral ( Figure 6). Regarding the two periods considered separately, significant trends were registered only for the annual pollen integral and the pollen peak during the last 15 years. No statistically significant trends were observed for the exploratory analysis of meteorological variables over the years, with the exception of the increase in maximum temperatures in the entire data set (Table 4). Considering meteorological data of the three previous months to oak flowering, significant trends of maximum temperature decrease (−0.058 • C per year) and mean temperature decrease (−0.071 • C per year) were observed.

Discussion
The main characteristics of wind as a pollination vector are its lack of specificity and its random dispersal mechanism. Therefore, anemophilous trees stimulated the production of high pollen amounts to facilitate fertilization processes as a response to the wind-limited efficiency. Successful flowering depends not only on abiotic parameters, such as meteorological factors, and tree biotic

Discussion
The main characteristics of wind as a pollination vector are its lack of specificity and its random dispersal mechanism. Therefore, anemophilous trees stimulated the production of high pollen amounts to facilitate fertilization processes as a response to the wind-limited efficiency. Successful flowering depends not only on abiotic parameters, such as meteorological factors, and tree biotic features that are genetically regulated [37] such as the number of pollen grains per anther, anthers per flower, flowers per catkin, and catkins per tree [3,38].
Our study showed the amount of pollen produced by individual members of six oak species, which is approximately summarized in the proportions 1:1:2:5:90:276 for Q. ilex, Q. faginea, Q. rubra, Q. suber, Q. pyrenaica and Q. robur respectively. Previous studies have rarely focused on pollen production in the two more prominent species (Q. robur and Q. pyrenaica) represented in oak forests of North-western Spain; they only examined their pollen viability characteristics [39]. Some research has been conducted in order to measure the pollen production of Q. ilex [11,24], Q. suber and Q. faginea [11] in the Mediterranean basin. In the case of pollen per anther and pollen per flower production, previous studies verified lower values of pollen production than that registered in our research for Q. ilex, Q. suber and Q. faginea [11,24]. The amount of pollen produced per catkin or per tree was similar to that obtained in North-western Spain for Q. ilex, whilst it was higher in the cases of Q. suber and Q. faginea [11,24]. The number of anthers formed per flower or catkin registered in the study area, at the limit of the Q. ilex, Q. suber and Q. faginea distribution, was lower than that observed for trees of the same species located in their optimal Mediterranean bioclimatic area [11,24]. These high values were similar to the observed for the North Spain oak species, Q. robur and Q. pyrenaica, located in their optimal bioclimatic conditions [11,24].
Several statistically significant associations have been registered between the evaluated flowering parameters, with higher correlation coefficients when data of all species were considered altogether. Our study showed that higher pollen production per anther is associated with a lower quantity of anthers per flower and flowers per catkin, which is in accordance with the results pointed out by Tormo [24]. The reproductive strategy of the Quercus genus consists of a higher production of pollen per anther in tree species with smaller size, mainly in the case of perennial species, in order to ensure fertilization. It was pointed out that pollen production per anther appears to be genetically fixed [37]. Another detected pollination approach was that a higher number of anthers per flower coincided with a higher number of pollen grains per tree. In our study, the number of anthers per flower was higher in deciduous species, such as Q. robur, Q. pyrenaica and Q. rubra, which presented the highest number of flowers per catkin. On the contrary, studies conducted by Tormo [24] noted that a higher number of anthers or flowers per catkin, corresponded to the production of a lower number of catkins by trees. In our study, statistical correlation revealed that species with lower pollen per anther production presented a higher number of flowers on each catkin, and species with a higher production of pollen per catkin presented a higher number of flowers in the catkin. Our data also showed a highly positive correlation between the number of catkins per tree and total production of pollen per tree. Studies conducted by Tormo [24] noted the same behavior, as teller trees increase their production of anthers, flowers, catkins and pollen. Oak species showed a tendency to compensate their sexual characteristics (such as pollen grains per anther, or inflorescences per tree) by increasing some of them or reducing others [11]. Nonetheless, in the study area Q. robur and Q. pyrenaica were the species with the largest pollen production as a consequence of their height and tree-crown diameter. In fact, pollen production depends on the production rate of pollen grains for individual plants, which in turn depends on the genotype, age, plant size, phenology, regional climatic and edaphic factors, diseases, and so forth [12,[40][41][42].
Information about the total pollen production per oak species, supported by knowledge on the coverage of various species in a given area, is useful to estimate the number of pollen grains in the atmosphere [11,24,43]. The curve of daily mean values of airborne oak pollen in the study area was characterized by several concentration peaks corresponding to successive flowerings of different species [2,44]. Quercus robur was the first species to flower in Northwest Spain; therefore, onset of the pollen season refers to this species. Q. robur pollination was closely followed by Q. faginea, Q. pyrenaica and Q. ilex flowerings [44]. The last species to flower was Q. suber, whose pollination period did not overlap with any flowering period of other oak species. Therefore, the pollination period is very long as a result of the abundance of late-flowering oak species such as Q. ilex and Q. suber [7].
Regarding annual totals, oak pollen in the atmosphere of the study area has shown an increasing trend during recent years. This tendency could denote the restoration of oak natural forests as a sign of ecological maturity in the overall vegetation distribution [2,4,44]. Previous studies in the study area pointed out a continuous growing trend of autochthonous oak forests during the end the 20th century and in the first decade of the 21st century mainly resulting from changes in land use through reforestation or from natural regeneration of abandoned agricultural land [31]. Other factors, such as, forest fires or agricultural practices, could also have caused for this change [42,45,46]. In contrast, oak natural forests have been almost completely replaced by non-irrigated olive groves, reducing their representation in the atmosphere in some Mediterranean areas [12].
Pollen rains from anemophilous plants induce rapid pulses of nutrients into the food web of ecosystems [16]. Several decomposers, such as fungi, bacteria, protozoans, and various groups of invertebrates, use pollen pulses as an easily available, digestible, and nutritious food, facilitating the entry of limiting nutrients within and across ecosystem nutrient cycling [16]. Forests may produce pollen in masses reaching 100 to 1000 kg/ha [13] increasing the productivity of the ecosystem by the addition of non-C limiting elements, such as P, K, S or Mg, even at a distance of thousands of kilometres between ecosystems [47]. Thus, estimation of the pollen mass from the main wind-related pollination forest in the study area should be included when calculating the annual biomass input in ecosystems. Scarce studies in scientific literature have considered the production of oak pollen in mass units, pointing out the mean mass of a single oak pollen grain as 0.011 µg as average [48]. This value allows us to calculate the mass of pollen produced in our area of study where the most prominent forests belongs to Q. robur, Q. pyrenaica and Q. ilex species. Considering the registered pollen production per tree, the natural forested cover of the three species [31] (as the other oak species occupied low surfaces due to their ornamental use), and the number of tress per ha (ranging from 516 for Q. robur to 696 for Q. pyrenaica [31]), the mass of pollen produced by the forests was estimated. An amount of 157 × 10 6 kg/ha was produced by Q. robur, 121 × 10 6 kg/ha in the case of Q. pyrenaica and only 0.00009 × 10 6 kg/ha for Q. ilex pollen mass introduced in the ecosystem food webs. This information would be valuable in the contexts of nutrient cycling and food web functions in forests. In addition, pollen from wind-pollinated trees, including various oak species, is a very important source of "high-quality" larval food for both solitary wild bees and managed honey bees [49]. Bees and larvae may be limited by the availability of P, Na, Mn, Mg, K, Fe, Ca, Zn and Cu [50], which are essential nutrients present in airborne pollen grains. The large mass of pollen produced by oak forests in the study area could help satisfy the demand for a nutritionally balanced diet in growing bees [51].
Increases in oak pollen concentrations in the atmosphere could also have a negative impact, as a consequence of the related development of sensitization in allergic people. Some researchers considered this pollen type responsible for major cases of allergies in areas with abundant oak vegetation [21,25]. The maximum concentration values were generally attained between the last week of March and the first two weeks of April, coinciding with the flowering of the most abundant species in the area, Q. robur and Q. pyrenaica [44]. Both species were the main contributors to the oak pollen curve, although Q. ilex was responsible for 10% of airborne pollen in years with high pollen production and favorable weather conditions [3]. Peak concentrations were produced slightly later in certain Mediterranean areas as a consequence of cork oak abundance [7].
Due to the high representation of oak trees in Northwestern Spain, the annual pollen integral was very high. Previous studies detected statistically significant differences between the oak annual pollen integral in the Eurosiberian region and those of the rest of the Iberian Peninsula [7]. Similar values were detected in the Northeast [52], while higher values were noted in the central part, where up to 66,000 pollen grains were registered in some years [53], and lower values were noted in the South [4,54]. Pollen production and dispersion is strongly modulated by a complex combination of biotic and environmental factors [12]. Annual pollen integral differences between years could be explained by abiotic factors, such as meteorological conditions, or biotic factors such as biennial behavior rhythms due to an alternation in the mobilization of nutrient reserves towards the tree's vegetative growth or towards the reproductive structures [1,45,55]. The obtained pollen data from 2011-2013, 2014-2016 and 2017-2019 could suggest this behavior, represented by a three-year cycle with one year of high pollen concentration followed by two years with lower values. Fluctuations in pollen production may account for biannual, triannual, or four-year cycles, a phenomenon that has been widely reported for oak pollen [4,56]. Moreover, the influence of other biotic factors may be very important, particularly during rainy years, since the insect Neuropterus triggers the development of gall on catkins [57]. A generation of asexual Neuropterus emerges from these galls and will deposit spherical galls in young leaves and catkins. In a short time, a new sexual generation with males and females will emerge and form lenticular galls [44]. In our study, years with the lowest pollen values mainly coincided with high accumulation of rainfall during the previous months to flowering. This could also hinder accurate pollination predictions, as under this condition the catkins are moistened, which impedes the formation of pollen grains. This has a higher effect in early flowering oak trees, mainly Q. robur, than in late flowering species such as Q. pyrenaica [44]. In the rainiest years, galls of the insect can cause half the catkins to develop compared to those that would be formed in a normal year.
During recent years, temperature trends have displayed regional differences with a non-uniform influence on ecological processes [10,58,59]. These trends have much more influence in winter and early spring flowering trees, since in these seasons the temperature registered during the last years has varied widely [4,60]. Our study detected a statistically, significant increasing trend in yearly maximum temperature, whereas a decrease in the maximum and mean temperatures during the three previous months to oak flowering was observed. The main reported changes related to the flowering phenology of plants have been the earlier onset [59,61], a longer pollen season [58,62], and an increase in airborne pollen concentrations [21,63,64]. Previous studies have been conducted in different Spanish localities to ascertain the atmospheric oak MPS behavior and to predict possible variations induced by climate change [4,7,12,18]. Our study showed a statistically significant in delay of MPS onset (0.47 days per year), possibly as a consequence of the decrease in maximum and mean temperatures during the three months preceding oak flowering, mainly during the 2005-2019 period. On the contrary, an earlier onset in some trees, including oaks, has been reported in different European areas [65], including Spain, associated with increased temperatures in the pre-flowering periods [1].
In the present study, a statistically, significant increasing trend was also observed for the annual pollen integral (a 7.9% of pollen grains per year) and for the peak pollen concentration (a 7.5% of pollen grains per year). These changes were also mainly produced during the 2005-2019 period. Increases in oak pollen counts have been observed for many Spanish cities, mainly in inland locations [1]. Some authors argued that the increase in atmospheric CO 2 concentration could stimulate pollen production by plants [66,67]. As a consequence of the rapid response to environmental changes, some authors have proposed oak pollen levels as the best bio-indicators to detect ecological variations induced by climate change [11,12].
Finally, it is important to highlight that oak flowering out of the pollination period was often noticed during the last ten years in North-western Spain. Evidences was found through field phenological observations and the detection of elevated amounts of oak pollen grains in the atmosphere during autumn months. The same behavior was noted in South Spain, where populations of Q. rotundifolia and Q. suber bloom twice a year [4].

Conclusions
Pollen produced by the six studied oak species follow the proportions 1:1:2:5:90:276 for Q. ilex, Q. faginea, Q. rubra, Q. suber, Q. pyrenaica and Q. robur respectively. Among the evaluated flowering parameters, the number of anthers per flower showed the lowest variation between trees from each oak species, what could reflect the genetic regulation that drives this parameter.
Pollen per anther was negatively correlated with the number of anthers per flower and the number of flowers per catkin. This fact shows the reproductive strategy of oak, since a higher production of pollen per anther compensates the lower number of anthers and flowers formed in order to ensure appropriate fertilization. The number of anthers per flower or catkin per tree registered in the bioclimatic transition region of the study area, at the limit of the Q. ilex, Q. suber and Q. faginea distribution, reflected a condition of boundary closeness, with a lower anther formation than that noted for trees of the same species located in their optimal Mediterranean bioclimatic area. This was also observed for Q. robur and Q. pyrenaica, which presented higher anther formation rates when located in their optimal bioclimatic conditions.
Oak pollen represented around 14% of the total annual airborne pollen, although it varied substantially from year to year, and recorded an increasing trend during the last decade. This could denote the restoration of oak natural forests as signal of ecological maturity in vegetation distribution, but it can have a negative impact on oak pollen-related allergies, since this genus is considered responsible for major cases of allergies in areas with abundant oak vegetation.
Additionally, the possible use of oak pollen levels as a bio-indicator to detect ecological variations induced by climate change is remarkable due to its rapid response to environmental stimulus. Considering the 1993-2019 data set, we observed a significant delay in the MPS onset of 0.47 days per year, which could be related to the decreasing trend observed in maximum and mean temperature during the three previous months to flowering (January, February, and March). Moreover, we also found a significant increasing trend of a 7.9% of pollen grains per year in the annual pollen integral, and a 7.5% in the case of the peak pollen concentration per year, which was noticeable during the last fifteen years in the study period (2005-2019 data set).