The effects of climate change on the flowering phenology of alder trees in southwestern Europe

Global warming impacts plant phenology and the effect of climate change will be more intensely experienced at the edges of a plant's distribution. This work focuses on Iberian alder's climatic range (Alnus lusitanica Vít, Douda & Mandák). The Iberian Peninsula constitutes the Southwestern edge of the global chorological distribution of European black alder (Alnus glutinosa (L.) Gaertn. s.l.), and some of the warmest and driest conditions for the alder population are located in the center of Spain. The critical temperature-relevant periods that regulate the reproductive phenology of alder were analyzed using a statistical-based method for modeling chilling and forcing accumulation periods in temperate trees. Our results reveal that autumn chilling was the most important thermal accumulation period for alder in a Mediterranean climate while forcing requirements are satisfied in a short period of time. Autumn temperatures were significantly correlated with the timing of flowering, and chill units during this season directly influence start-dates of alder flowering. A positive trend was observed in pollen seasons' timing, meaning a slight delay of alder flowering in central Spain. It coincided with autumn warming during the period 2004-2018. If this warming trend continues, our results predict a delay in the start-date of flowering by around 3-days for every degree increase in maximum autumn temperatures according to the most optimistic emission scenarios.


Introduction
Climate change has multiple implications for biodiversity (Garcia et al., 2014), as well as many other impacts on interconnected environmental domains (e.g., air, ecosystems, food, health, water, etc.) (Cramer et al., 2018). The potential effects of climate change on plants are very diverse affecting individuals, species, and whole ecosystems (Fernández-González et al., 2005). Increasing CO 2 concentrations and global temperatures may increase net primary productivity and extend the growing seasons (Donmez et al., 2016). In some waterlimited ecosystems, the effect of aridity has the opposite effect (Osborne et al., 2000). Vegetation in areas with increasingly warmer and drier conditions will suffer changes in competitive interactions because plant species' drought tolerance determines plant survival to intense summer droughts and extreme heat events (García-Madrid et al., 2016;Peñuelas et al., 2001). Furthermore, global warming affects the timing of the main stages of plant life cycles from a phenological perspective Rojo and Pérez-Badia, 2014).
Climate change in the Mediterranean basin has been revealed to be more dramatic than average worldwide changes during the 20 th century (Guiot & Cramer, 2016). In the central Iberian Peninsula (Castilla-La Mancha region) annual mean temperatures have increased about 1.5 ºC during the last three decades (Gómez-Cantero et al., 2018). However, warming is not homogeneous throughout the year, and summer and autumn are the seasons with the greatest increase in temperature concerning the mean values for previous climatological periods. The irregular distribution of precipitation has also marked recent decades. Most future climate scenarios forecast a decrease in rainfall and a generalized increase in the Mediterranean region's temperatures (Spanish Meteorological Agency, 2020).
Thermal requirements strongly regulate vegetative and reproductive plant development in deciduous forest tree species (Peaucelle et al., 2019;Picornell et al., 2019). The mid-latitudes climate is characterized by successive favorable and non-favorable periods related to a marked seasonality. In these climatic conditions, deciduous trees' dormant state is a physiological phenomenon related to frost tolerance and the protection of vegetative and ARTICLES reproductive tissues (Dewald & Steiner, 1986). During the dormancy period, the plant must achieve a certain level of chilling accumulation characterized by low temperatures. On the other hand, once chilling requirements have been reached, plants need to fulfill heat requirements (high temperatures) to initiate the plant development during the so-called forcing period (Luedeling et al., 2013). Warming negatively impacts chilling accumulation, and considerable reductions in chilling units have been documented across the Mediterranean region . It is especially worrying for Eurasian species like black alder, whose Mediterranean populations are located at the warmer edge of their distribution.
In Europe, the distribution of black alder (Alnus glutinosa (L.) Gaertn. s.l.) ranges from Scandinavia to the Mediterranean Basin along a wide ecological gradient characterized by a broad range of annual mean temperature and annual precipitation (Houston-Durrant et al., 2016) (Figure 1). Alder trees also cover parts of Asia, and this species was introduced in North America and Australia in the late 19th century (Kamocki et al., 2018). The plant is also present in central and southern parts of the Iberian Peninsula (and Northern Africa), where alder populations are considered to be at the edge of the European distribution of A. glutinosa s.l. (Pérez Latorre et al., 2011;Salazar et al., 2001), i.e., the most vulnerable alder populations to global environmental change (Lepais et al., 2013). Recent molecular and karyological studies recommend the disaggregation of alders of the Iberian Peninsula (except in the Pyrenees and Cantabrian Mountains) into a new species named Alnus lusitanica Vít, Douda & Mandák (Vít et al., 2017) (Figure 1).
Alder forests on the Iberian Peninsula are floodplain forests associated with periodically flooded riparian zones (Fernández-González et al., 2012). The alder forests have been classified separately at the level of alliance in the syntaxonomical classification system according to their biogeographical distribution (Rivas-Martínez et al., 2011): Alnion incanae for Eurosiberian and Osmundo-Alnion for Mediterranean forests. Recent classification studies confirm the strict Mediterranean character of the Osmundo-Alnion alliance . Alder forests are widely distributed in the North and western parts of the Iberian Peninsula (Douda et al., 2016) and are characterized by their ecological importance, provision of ecosystem services, and high conservation priority .
From the legal protection point, alder forests are included in the list of habitat types in the European Habitat Directive (Annex 1, Directive 92/43/EEC). In central Spain, this type of riparian habitat has been included in the Regional Catalogue of Special Protected Habitats of Castilla-La Mancha (Nature Protection Law 9/1999 of Castilla-La Mancha) (Martín-Herrero et al., 2003). Furthermore, Alnus glutinosa s.l. is also included in the Regional Catalogue of Protected Species of Castilla-La Mancha (Decree 33/1998).
This work focuses on the environmental ranges occupied by Iberian alder in the center of the Iberian Peninsula that, together with those located in southern areas of the Iberian Peninsula and Northern Africa, constitutes the edge of the global chorological distribution of European black alder. We hypothesize that alder populations located in the rear edge of the global distribution will be more affected by global warming than central populations, as supported by the Centre-Periphery Hypothesis (Pironon et al., 2017;Vilà-Cabrera et al., 2019), therefore significant changes will occur in the flowering phenology of alder trees in the central Iberian Peninsula. The main aims of this study are to: 1) Analyse the critical temperature-relevant periods that strongly regulate the reproductive phenology of alder trees in central Spain; 2) Study the current phenological changes occurring as a consequence of warming climate during recent years; 3) Predict potential changes to alder phenology if warming trends follow Representative Concentration Pathways proposed in the IPCC Fifth Assessment Report (Pachauri & Meyer, 2014).

Study area
In this study, we considered the alder forests of the Special Areas of Conservation (SAC) (Natura 2000 Network) located in the Northwest of the Castilla-La  (Figure 2). Figure 2. Location of the study area. Green points represent alder forests and red polygons, the study area belonging to the four Special Areas of Conservation (SACs) from the Network Natura 2000 in central Spain.
Source of the base map: GoogleMaps .
The climate is characterized by two periods with the highest precipitation (autumn and spring) and a summer drought typical of Mediterranean climates. The annual precipitation in Montes de Toledo ranges from 500 to 1000 mm, decreasing to less than 400 mm in the bottom of the Tagus basin, and the annual mean temperature ranges from 10 to 14 ºC, reaching up to 16 ºC toward the Tagus valley (Ninyerola et al., 2005). The annual precipitation of the northern SACs ranges from 500 mm in the lower valley of Alberche river to 600-1200 mm in the Sierra de San Vicente, and the mean temperature ranges from 11 to 17 ºC (Ninyerola et al., 2005) (Figure 3A-B). Bioclimatically, the study area is characterized by a Mediterranean Pluviseasonal-Oceanic Bioclimate. Biogeographically, it is framed into the Mediterranean Iberian Occidental Province (Lusitan-Extremadurean Subprovince, Toledan-Taganean Sector) (Rivas-Martínez et al., 2011).

Meteorological and climatic data
Different meteorological and climatic sources were used during this study. In the first step, climatic monthly data in the past (period 1951-1999) were used to characterise the climatic range where alder forests grow. For this purpose, we used a high-resolution climatic map as the digital climate atlas of the Iberian Peninsula (Ninyerola et al., 2005), which has a grid cell size of 200 x 200 m.
Daily meteorological data (maximum, minimum, and average temperatures) were used during the period 2004-2018 for modelling of the alder phenology. In this case, meteorological datasets were provided by the Spanish Meteorological Agency in Spain (www. aemet.es). Finally, future projections of the models required applying the phenological models to the future. The last version of the WorldClim project was used (Fick & Hijmans, 2017). These climatic datasets were downscaled to 1x1 km of spatial resolution and a monthly temporal resolution.

Mapping of alder forests
The alder forests studied were mapped in detail at a 1:5000 scale. Vegetation mapping was carried out in vector format using QGIS 3.0 to delineate the polygons based on the high-resolution digital aerial orthophotographs provided by the Spanish National Aerial Orthophoto Program (PNOA) supported by the National Geographical Institute (www.ign.es). Mapping was completed with field campaigns to check the vegetation content of the polygons, including the presence of alder forests. The vector map has been rasterized using a grid cell size of 200 x 200 m. Then, the environmental range covered by the alder forests in central Spain was characterized with respect to the distribution of the Iberian alder (Alnus lusitanica). The occurrences of the alder tree (A. glutinosa s.l.) have been downloaded from the Global Biodiversity Information Facility (GBIF) (www.gbif.org), only clipping the distribution area proposed by Vít et al. (2017) for A. lusitanica in the Iberian Peninsula.

Distribution of alder forests
Alder forests are well represented in the Alberche and Tiétar rivers, both tributaries of the right margin of the Tagus river in North (Figure 2). Alder trees grow sparsely in the Estena river and several streams in the Fresnedoso river's birth, all these areas located to the foothills of the Montes de Toledo. The number of occurrences (pixels of 200 x 200 m) of alder forests in the Alberche and Tiétar rivers was 138 points. In contrast, in the Estena and Fresnedoso rivers (Montes de Toledo) only 22 occurrences were identified ( Figure 2). Alder forests in Montes de Toledo ranged from 550 to 950 m asl, and these forests grow in narrow water streams. On the other hand, alder trees in the Alberche and Tiétar rivers were studied in the middle of the rivers at an altitude of 300 to 500 m asl. These riparian habitats grow along wider rivers with higher flow regimes.
Alder trees in the study area of central Spain grow in a climatic range characterized by annual mean temperature ranging from 15 to 16 ºC, and annual precipitation of 500-1100 mm in Sierra de San Vicente (Alberche and Tiétar rivers). Conversely, alder trees in Montes de Toledo (Estena and Fresnedoso rivers) grow in areas with an annual mean temperature of 13-15 ºC and annual precipitation 400-800 mm. Figure 3C compares the local climatic range in the study area with the much broader environmental space occupied by A. lusitanica in the Iberian Peninsula. Figure 3. Annual mean temperature (ºC) and annual precipitation (mm) in the study area; A-B, climatic range of A. lusitanica (according to Vít et al., 2017) in the Iberian Peninsula and occupied range in the study area (green points); C, source of the climatological data (Ninyerola et al., 2005).

Phenological data
Phenological monitoring of alder forests located in the study area was based on an indirect methodology derived from airborne pollen sampling during the pollination period (hereafter flowering period). This procedure has been widely used in plant phenological studies from an ecological point of view (Dąbrowska & Kaszewski, 2012;Picornell et al., 2019;Rojo et al., 2018). Pollen seasons were defined using the 95% method, i.e., the pollen season's start-date occurred when 2.5% of annual pollen was collected, and the enddate of the pollen season occurred when 97.5% of the season's catch was reached. The peak-date was the day when the maximum pollen concentration was recorded every year. Phenological calculations were carried out using the AeRobiology R package, a specific tool for conducting aerobiological tasks .
Pollen monitoring was carried out in the city of Toledo during a 15-year period (2004)(2005)(2006)(2007)(2008)(2009)(2010)(2011)(2012)(2013)(2014)(2015)(2016)(2017)(2018). The historical pollen time-series come from the dataset of the Aerobiological Network of Castilla-La Mancha of the University of Castilla-La Mancha (www.polencastillalamancha.com). In order to determine whether the long pollen time-series recorded in Toledo could be used in this study, the curve of airborne alder pollen recorded in Toledo was compared through linear correlation to the pollen curve recorded in Talavera de la Reina (a site situated close to the alder forests) during the period 2008-2013. A high correlation (r = 0.71, p < 0.001) indicates that although the pollen intensity could be different among stations, the pollen sources are the same and both stations could be used to calculate the timing of the pollen season (Skjøth et al., 2013). Trends of the phenological parameters (start, peak, and end-dates) were therefore calculated for the period 2004-2018 in Toledo using linear regression (slope) and Mann-Kendall test (tau).

Modeling
Start-dates of alder pollen seasons were correlated with monthly and seasonal temperatures recorded in the months before the beginning of the flowering period (from September of the previous year to January of the current year). Thermal accumulation periods were also defined in order to estimate chilling and heat requirements for alder flowering. For this purpose, we followed a statistical-based method proposed by Luedeling et al. (2013) to identify the critical temperature-relevant periods that influence the floral phenology of alder, and is widely used for other deciduous woody plants (Benmoussa et al., 2017;Martínez-Lüscher et al., 2016). Daily meteorological data (maximum, minimum and average temperatures) were used.
The detailed procedure to define the accumulation periods in this work is performed as follows: startdates of alder flowering (response variable) were related to daily mean temperatures for each day from September of the previous year to the commencement of flowering (predictor variables). A moving average was applied to the daily temperature to ensure that the thermal periods were clearly recognizable (Luedeling & Gassner, 2012). The influence of temperature for each day on the timing of flowering was obtained from the results of Partial Least Square (PLS) regression, and the most relevant days were highlighted following the criterion of the Variable Importance in Projection (VIP) scores (Luedeling et al., 2013). In this case, the days reaching a VIP score of 0.8 were considered relevant days for alder's physiological thermal accumulation. Another important result of the PLS regression is the sign of the coefficients retrieved for the daily temperatures. Positive coefficients were related to chilling accumulation periods since the temperatures were positively correlated with the flowering dates. On the other hand, negative coefficients were related to heat accumulation periods (forcing period) since higher temperatures during this period were associated to an advance of the flowering dates. Potential thermal accumulation periods were estimated using the continuous periods of the PLS analysis results showing positive (chilling period) or negative (forcing period) coefficients.
The Dynamic Chilling Model was used to calculate chilling effectiveness units during the chilling accumulation period previously estimated (Fishman et al., 1987). This model assumes that chilling accumulation results from a first intermediate chilling product derived from low temperatures that can be converted into permanent chill units at low and moderate temperatures (Darbyshire et al., 2011). The equations of this model and the calculations of the thermal accumulation periods have been implemented for R software in the 'chillR' package (Luedeling et al., 2013).

Future projections
The results of the Partial Least Square regression analyses were used to construct a model relating the timing of flowering to temperatures recorded in the thermal accumulation periods previously estimated. The model was spatially applied to all points where alder forests grow in the study area using the current climatological data for 1970-2000 released in the last version of the WorldClim project (Fick & Hijmans, 2017). Finally, the phenological model was projected to the future for the periods 2050 (average for 2041-2060) and 2070 (average for 2061-2080) for several General Circulation Models (GCMs) used in the IPCC Fifth Assessment Report (Hijmans et al., 2005;Pachauri & Meyer, 2014), taking the potential radiative forcing as Representative Concentration Pathways (RCPs) 4.5 and 8.5 W/m 2 (van Vuuren et al., 2011).

Flowering phenology of alder tree
In general, alder pollen seasons in central Spain occurred from mid-January (13th January ± 5 days of standard deviation) to late March (27th March ± 22 days). Therefore, start-dates of alder pollination occurred in a limited period from the 2 nd to 21 st January (2004-2018). Peak-dates usually occurred soon after the beginning of the pollination (30th January ± 9 days, Figure 4A). Clear positive trends were observed for the start, peak, and end-date of the alder pollination period (not significant), meaning there has been a slight delay in alder flowering in central Spain during the last 15 years. These results showed an averaged delay of 2.5 and 8 days every 10 years, respectively, to the start and peak-date of the alder pollen season ( Figure 4B).

Temperature-relevant periods for the initiation of flowering
The results showed a positive relationship between temperature during September-November (autumn) and the start-date of flowering in the next year (Table 1). The results coincide with those obtained from the PLS analysis reporting that the daily temperatures observed during the months September, October and November were positively correlated with flowering timing. In addition, these temperatures were important enough to be included as significant variables in the model of the start-date of the flowering (according to the VIP scores) ( Figure 5A).  Figure 5. Identification of the chilling (green) and heat (red) accumulation periods following the modeling of the timing of flowering based on the PLS regression using A, daily mean temperatures as predictors, and B, the relationship between the timing of flowering and chilling units calculated during the autumn of the previous year (September-November) using the Dynamic model.
Positive coefficients of the PLS regression mean that during this period (September-November), the temperature has a direct relationship with the onset of the flowering period in alder trees (chilling accumulation period). The calculations derived from the Dynamic Chilling Model demonstrated that a lower accumulation of low temperatures (chilling process) was related to the delay of the timing of flowering of alder ( Figure 5B).
The modeling results showed around a 3-day delay in the onset of the alder flowering for every degree increase in maximum temperatures during the autumn (mean for the three months September-October-November). Logarithmic models showed a stronger influence on the start of flowering than minimum temperatures ( Figure 6A,B).

Future scenarios of flowering phenology
The model generated for the timing of alder flowering showed that all predictions result in a delay in alder pollen seasons under future climatic scenarios (Figure 7). The most optimistic scenario corresponds to INMCM4 General Circulation Models (GCM) with RCP 4.5, which forecasts an increase of about 2ºC in maximum autumn temperatures in 2070 (Figure 8). In the most pessimistic emission scenario GFDL−CM3 (RCP 8.5), autumn warming in the study area would be 4.5 and 7.5 ºC for minimum and maximum temperatures, respectively. Therefore, the average range of the phenological delay is predicted to be 6-27 days in the timing of flowering for 2070 compared to current conditions (about a 3-day delay per degree of increase in maximum autumn temperatures), considering both GCMs INMCM4 (RCP 4.5) and GFDL−CM3 (RCP 8.5), respectively ( Figure  8). When the logarithmic model is applied, the most extreme phenological scenario (RCP 8.5) is slightly modulated for the 2070 period to a maximum delay of 25.5 days (Figure 8).

Discussion
The distribution of alder forests is limited by precipitation and temperature in Southern Europe, with aridity being one of the most important limiting factors for alder trees (Hemery et al., 2010). However, local precipitation rates in the areas where alder occurs is not particularly relevant since alder forest is a river flow dependent-vegetation and not directly rainfall dependent-vegetation. The head of the Alberche river is located in the Central System, a more humid area (annual precipitation 800−1000 mm) of about 1700 m asl. The periodic snow cover in the higher reaches of Alberche causes greater flow stability and high water availability throughout the year, which favors the development of the alder forests (Fernández- González et al., 2012). This is the reason for the wide ecological gradient and the fact that alder can grow even in dry parts of central and southern Spain (Pérez Latorre et al., 2011;Salazar et al., 2001). However, increasing aridity in upper river basins, as a consequence of climate change, would potentially trigger the reduction of alder tree distribution in the Mediterranean region (Attorre et al., 2011). While the precipitation rate and aridity index are the main limiting factors for the distribution of alder, the temperature is the main factor influencing the seasonal life cycle of the black alder as well as other deciduous trees Linkosalo et al., 2017;Picornell et al., 2019). The most important finding from our phenological modeling of alder flowering was the strong positive relationship between the timing of flowering and temperatures from the previous autumn. Moreover, a stronger effect was observed for maximum temperature during the chilling accumulation period of autumn than those retrieved by the minimum temperature. Therefore, the maximum temperature was more relevant for alder phenology than the minimum temperature as previously reported in Mediterranean plants (Gordo & Sanz, 2010).
A positive slope was observed for the timing of pollen seasons, meaning a slight delay of the entire pollen season of alder in central Spain during recent years. The warming of the autumn period could be an explanation for this phenological delay of alder trees. A significant trend was observed for monthly maximum October temperatures for the period 2004-2018 (slope = 0.24; R 2 = 0.32; p < 0.05), i.e. autumn temperatures have become warmer during recent years. Similar phenological trends were found in other Mediterranean areas of Southern Europe, such as Italy and the Iberian Peninsula, where a slight trend towards later start-dates of alder pollen seasons has been documented (Jato et al., 2013;Novara et al., 2016;Rodríguez-Rajo et al., 2011). On the other hand, these results contrasted with those from Central and Northwestern Europe (United Kingdom, Austria, Finland, Norway, Sweden, Iceland) (Emberlin et al., 2006;Jäger et al., 1996;Lind et al., 2016;Smith et al., 2014), all of which reported negative trends, i.e., alder flowering earlier over time. Our results contrasted with trends observed for other springflowering trees in the Mediterranean region that generally show an advance of the flowering period as a consequence of an increase of temperature during the forcing accumulation periods (García-Mozo et al., 2010;Gordo & Sanz, 2010).
Previous studies have noticed the importance of chilling accumulation for alder and other winterflowering trees in the last autumn. The chilling phase has been identified as the most important thermal accumulation period for alder in the Mediterranean climate while forcing requirements are satisfied in a short period of time in alder trees of Southern Europe (González-Parrado et al., 2006;Novara et al., 2016). It supports the results of our study, but such behavior with respect to the thermal requirements may be different in other climatic conditions. Previous research based on thermal requirements conducted on alder in a temperate climate reported different findings with an increase in the importance of the heat requirements during the forcing period to alder budbreak (Emberlin et al., 2006;Malkiewicz et al., 2016).
Continentality and Thermicity indexes may play an important key role in the ecophysiology of alder trees associated with thermal requirements. Rodríguez-Rajo et al. (2006) documented significantly different chilling and forcing accumulations when very different bioclimatic areas, in terms of continentality and altitude, were compared in the boundary between Mediterranean and Eurosiberian biogeographical areas. Colder areas required greater chilling and forcing requirements since the trees need to be maintained in the dormant rest until the environmental conditions are suitable for plant development (Campoy et al., 2011;Lang, 1987). The flowering period of alder sequentially occurs following a continentality gradient from Southern Europe, Central and Northern Europe to Western Asia (Biedermann et al., 2019;Dąbrowska & Kaszewski, 2012;Jato et al., 2013;Kasprzyk, 2003;McVean, 1953;Turchina, 2019) Bioclimatic triggers for changes in thermal accumulation periods and the phenological behaviour of alder trees in different bioclimatic areas are interesting topics for investigation (Dewald & Steiner, 1986). We hypothesize that the most critical temperature-relevant periods for alder in Mediterranean areas are the chilling phases that the tree must fulfill for bud break, and the achievements of chilling requirements may be threatened in the context of global warming (Chuine, 2010;Luedeling et al., 2011;Ma et al., 2018). On the contrary, in colder climates achieving chilling requirements during the autumn and winter would not be a limitation for the plant, forcing accumulation to become the most important thermal requirement . Further large-scale studies should be carried out that compare the reproductive phenology, and thermal requirements of black alder in different climate ranges using a unified modeling method. Also, considering the recent disaggregation of species in Europe and North Africa (Vít et al., 2017), the ecophysiology of the different species should be studied and compared in the future.
All considered General Circulation Models (GCMs) predicted an increase in autumn temperatures in central Spain. From a phenological point of view, according to our results, alder flowering will be delayed by about 3-days for every degree increase in maximum autumn temperatures, but the response was not linear. We observed that a logarithmic pattern could reliably be used to relate the timing of flowering and environmental variables, and a nonlinear phenological trend was consistent with the data from the sampling period (Jochner et al., 2016;Morin et al., 2010). Adaptability to climate change, with declining rates of phenological change of the reproductive cycles, has been exhibited by several plants during recent decades (De Kort et al., 2016). Vegetative plant development has shown significant phenological changes during the last decades due to global warming (Morin et al., 2009). This fact may cause interference also in the reproductive cycle because both phases could overlap. In the same way as the flowering, alder trees showed advanced spring leaf unfolding to warming trends, although in this case, deciduous trees showed a nonlinear response (Fu et al., 2015).
Phenological changes have important consequences of planting survival. A reduction of the distribution area of alder may result from the limitation of the phenological plasticity of populations in the face of extreme environmental changes (Attorre et al., 2011;Turchina, 2019). Plant phenology is associated with the local adaptability of plant species to environmental conditions. Usually, flowering traits are genetically conserved by the populations, even those located in geographically distinct communities but occupying the same ecological niche (Chuine, 2010;De Kort et al., 2016;Thuiller et al., 2004). Moreover, phenology has relevance to the reproductive success of plants and the timing of flowering affects several phases of reproduction such as the pollination period, fruit set, fruit maturation and the quality of plant offspring, but other positive or negative coevolutionary processes are also very important such as those based on the phenological synchrony with herbivores, or pollinators and seed dispersers in other plants (Levin, 2006).
Other important changes in Mediterranean riparian vegetation would be those associated with the deterioration of habitat quality (Salinas et al., 2000). Such impacts would be magnified in marginal populations (Hampe & Petit, 2005). Compositional changes in vegetation are also found at the edge of the chorological areas (Rodríguez- Rojo & Fernández-González, 2014). Hydrogeological characteristics of Mediterranean river courses are expected to be altered by climate change, whose negative effects over riparian vegetation will be brought about by human management of river flows (Lytle et al., 2017;Palmer et al., 2008). In addition, climatic stresses may make riparian trees more susceptible to disease and speed up the decline of alder forests. Oomycete pathogens like Phytophthora alni Brasier and S.A. Kirk have already been associated with high mortality rates of alder trees in riparian ecosystems in Europe (Aguayo et al., 2014;Haque & Casero, 2012).
Riparian vegetation such as alder forests plays a key role in protecting riverbeds from an ecological point of view and providing important ecosystem services, especially in dry climates (González et al., 2017). Furthermore, marginal populations of alder located at the rear edge of alder distribution have a great biogeographical value. They have served as genetic sources to expand the temperate species over Europe after the Last Glacial Maximum of the Quaternary Era (Douda et al., 2014). High conservation priority is given to alder trees' Mediterranean populations because of their ecological and phylogenetic characteristics Lepais et al., 2013).

Conclusions
Advances in the timing of flowering of forest tree species due to increasing temperatures during the forcing period have been documented in recent decades. However, in some species whose chilling period is becoming more and more relevant, the phenological response may be different as in the case of alder in the Iberian Peninsula center. Alder is a winter-flowering species whose onset of the flowering period normally occurs during January, and so the chilling accumulation period is crucial for achieving suitable reproductive development. It is, in part, a consequence of being part of the rear edge of the ecological gradient of the species, which makes the population in central Spain extremely vulnerable to climatic warming. If the current trends of phenological traits continue over time, the flowering period of Iberian alder (Alnus lusitanica) will experience a considerable delay of around 3 days per degree of increase in maximum autumn temperatures according to the most optimistic emission scenarios. Other potential changes, caused by increasing aridity or greater irregularity in precipitation projected for the Mediterranean Basin, are associated with a reduction in the potential ecological niche of alder and the deterioration of the structure and floristic composition of Iberian alder forests.