Premature losses of leaf area in response to drought and insect herbivory through a leaf lifespan gradient

Sonia Mediavilla · Montserrat Martínez-Ortega · Santiago Andrés · Javier Bobo · Alfonso Escudero

Abstract Implications of the differences in leaf life span are still subject to debate in the field of ecophysiology.Since leaf traits associated with these differences may be decisive for determining the distribution of tree species, this topic is particularly relevant in the context of climate change.This study analyzes the effects of the differences in leaf life span on premature losses of leaf area owing to insect herbivory and to abiotic stress.Loss of leaf area may be an important determinant of total leaf carbon assimilation.Seven Mediterranean tree species, distributed on four sites with different climates were studied.The species exhibited strong differences in leaf life span and in leaf traits, especially leaf mass per unit area.Premature leaf area losses were estimated in response to insect herbivory and summer drought over two years.The results revealed that, despite having older leaf cohorts with more damage, species with longer leaf duration had lower area lost to herbivores and less damage due to accelerated senescence during the summer drought.With respect to the predicted increase in water stress, deciduous species are at a disadvantage due to their high premature loss of leaf area and thus loss of photosynthetic capacity.

Keywords Accelerated senescence · Insect herbivory · Leaf life span · Leaf structural reinforcement · Summer drought

Introduction

Forests occupy about 30% of the earth’s surface, represent about 45% of the world’s terrestrial carbon reserves and provide basic environmental services, all of which makes their conservation a priority (Choat et al.2018; Jonsson et al.2019).However, in recent decades there have been significant episodes of mortality in forest stands around the world (Buras et al.2018; Hartmann et al.2018) that are no doubt associated with increased temperatures and intensified drought events accompanying new climatic conditions (Trenberth et al.2014; Senf et al.2020).An increase in the frequency and severity of drought is one of the main projections of climate models for the future (IPCC 2014).This is of particular concern for regions with a Mediterranean climate, where the availability of soil water represents the most significant environmental constraint (Flexas et al.2014).The new conditions envisaged make them one of the most vulnerable areas to climate change (Pe?uelas et al.2018).The fate of many forest ecosystems, including Mediterranean ecosystems, will depend on how different tree species respond to new conditions of water limitation.It is therefore essential to provide information on this topic to be incorporated into the management and conservation of these ecosystems.

The death of the individual tree or the early loss of a part of the canopy are the most visible symptoms of the effects of drought.However, under stressful conditions, different leaf symptoms may also appear (necrotic spots, discolorations, disappearance of parts of the leaf surface) before leaves reach their maximum life span (Günthardt-Goerg and Vollenweider 2007).This implies that part of the photosynthetic machinery remains in the crown for a shorter time than the mean duration of the individual leaves (Mediavilla and Escudero 2003a).This is one aspect on the effects of drought that has not been studied sufficiently, despite its implications on the capacity of carbon sequestration by different species and therefore on their competitive capacity under new expected climatic conditions.

Plant resistance to drought reflects two contrasting physiological strategies: drought avoidance (isohydric plants) and drought tolerance (anisohydric plants) (Kozlowski et al.1991; Larcher 1995).Isohydric plants maintain a constant leaf water potential regardless of drought intensity, either by reducing stomatal conductance or by increasing water absorption (e.g., through deep root systems) (McDowell et al.2008; Miller et al.2010).In contrast, drought-tolerant species maintain constant gas exchange rates, allowing the water potential of leaves to decrease with decreasing water availability through features such as a conductive system that favors continuous water transport (Martínez-Vilalta et al.2004; West et al.2008).Both strategies allow the survival of leaves and the plant under drought conditions.However, prolonged reductions in gas exchange rates through stomatal closure expose isohydric species to carbon depletion, while extreme water potential gradients in anisohydric species result in cavitation, xylem embolism, and subsequent hydraulic failure (Sevanto et al.2014; Choat et al.2018).In Mediterranean-climate environments, high temperatures and low water availability are often accompanied by levels of photosynthetically active radiation that exceed the saturation value of photosynthesis during part of the year.This leads to photoinhibition phenomena which can generate necrosis in part of the photosynthetic tissues or the appearance of spots by degradation of photosynthetic pigments (Mafakheri et al.2010; Anjum et al.2011).Whether by one mechanism or another, if part of the photosynthetic area of the leaf is lost before the leaf dies, these premature losses will inevitably affect the carbon balance of the plant (Mediavilla and Escudero 2003a; Mediavilla et al.2018a).Therefore, their quantification in different species may be essential to evaluate their response to new climatic conditions and to provide relevant information to correct and improve predictions about forest carbon sequestration capacity.

In this study, we analyze the premature loss of leaf area in seven tree species typical of the Iberian Peninsula.The species selected had a broad range of leaf life-span, which implies that this group includes a variety of leaf traits associated with differences in leaf longevity through the "leaf economics spectrum" (Wright et al.2004).Leaf life span is a fundamental trait because it reflects the amount of time the leaves have to provide carbohydrates to the plant and to compensate for the resources invested in the construction of leaf biomass.Leaves with longer longevity are structurally reinforced by a greater thickness, larger mass per unit area (LMA) and higher fiber concentrations, but in return, they invest less nitrogen in photosynthetic functions and achieve lower instantaneous CO2assimilation rates.Various studies have tried to clarify which of the two strategies, greater or lesser longevity, is more advantageous in terms of final leaf-level CO2assimilation (Mediavilla and Escudero 2003b; Kikuzawa and Lechowicz 2006; Falster et al.2011).Some studies have concluded that the coexistence of both strategies is possible because evergreen species compensate for a lower instantaneous production with a longer production time, so that the total leaf-level CO2assimilation is similar in both cases (Mediavilla and Escudero 2003b), while others give advantage to species with longer-lived leaves (Falster et al.2011).However, there are many variables that have not been considered to date such as differences in the premature loss of leaf area associated with differences in leaf life span.

In principle, since a longer duration implies structural reinforcing traits, we would expect that longer-lived leaves would be more protected against stress factors and would experience less premature losses of area.In addition, for a leaf designed to last and photosynthesize for several years, losing some of its photosynthetic area early in life implies larger costs in terms of assimilation loss than in the case of a leaf designed to last only a few months, so we could also expect that longer-lived leaves should exhibit more stress damage avoidance mechanisms.However, it is also true that a leaf with greater longevity would accumulate more damage over its lifetime as a result of longer exposure to stress events.Depending on which of these alternatives prevail, evergreen species could be harmed or benefited compared to deciduous species by the expected intensification of water stress, at least as far as premature losses of leaf area are concerned.Since the different levels of structural reinforcement that accompany the differences in leaf longevity have often been associated with different levels of herbivory (Carmona et al.2011; Mediavilla et al.2018b), also included in this study was the premature loss of leaf area by insects to take into account the implications of leaf longevity on total area losses, both by biotic and abiotic (drought) factors.

Our objectives were: (1) To quantify damages to the photosynthetic area caused by drought and by consumption by herbivorous insects in various species with different leaf life spans and in different geographical situations; (2) To identify those species most sensitive to drought, as related to premature leaf area losses in order to predict their responses to climate change; and, (3) To contribute to the understanding of the consequences of leaf strategy differences by introducing a variable, accelerated senescence, which has received scarce attention to date.Since we expected that the impact of accelerated senescence varies with leaf longevity, this may be crucial to further explain the coexistence of species with different leaf strategies.

Materials and methods

Species and study areas

The research was carried out on seven tree species, four hardwoods of the genusQuercus(Quercus fagineaLam.,Q.pyrenaicaWilld.,Q.suberL.,Q.ilexL.subspballota(Desf.) Samp.) and three conifers of the genusPinus(P.pinasterAiton,P.pineaL.andP.sylvestrisL.).These species are among the most widely represented in the forests of the Iberian Peninsula, allowing them to be found in various enclaves with different intensities of summer water stress.In addition, they represent different leaf habits and cover a wide range of leaf longevity values from approximately five and seven months in the two deciduous oaks,Q.pyrenaicaandQ.faginea, respectively, 15 months inQ.suber, 24 inQ.ilex, 35 inP.pinea, 49 inP.sylvestrisand 52 inP.pinaster(Mediavilla and Escudero 2003b).

The species were distributed on four sites in the Castilla-León region (central-western Spain).Since there are site differences in the intensity of stress factors (González-Zurdo et al.2016a, 2016b; Mediavilla et al.2018a), differences in leaf damage of the same species associated with between-sites differences in stress can be investigated.On each site, several species coexist which also allows for interspecific comparisons under similar water stress conditions.The coordinates, altitude and climatic characteristics of the four sites are shown in (Table 1).The soils correspond to dystric Cambisol in sites A and D, humic Cambisol in site C and dystric-humic Cambisol in site B (González-Zurdo et al.2016a).Mean climate values over the last five years were extracted from daily data provided by the AEMET (Spanish Agency of Meteorology) from the closest stations.According to the K?ppen classifciation (Peel et al.2007), the climate of A, B and C sites is Csb, ie a temperate climate with warm and dry summer.Site D has the highest levels of rainfall and a temperate climate without dry season and with warm summer (type Cfb).The between-site differences in temperature and rainfall were consistent in the two study years (Table 1).On all sites, the spring of 2018 was colder and more humid than that of 2019.The differences between years, however, were not significant in summer, except for plot D where rainfall levels were higher in summer 2018.In this year, the Emberger index (Emberger 1932) was also higher for all the plots with respect to 2019, but with little significant differences.

Table 1 Characteristics of the study sites

Sampling and analysis

In each site, five branches with leaves were sampled from the upper crown from each species.Sampling was done in mid-June once all the leaves of the current year had been fully formed, and in early October, to determine the increase in leaf area loss during summer.Sampling was conducted before and after the summers of 2018 and 2019.All selected trees were fully sun-exposed mature specimens with heights between 8 and 10 m and diameters (at 1.3 m) between 40 and 60 cm.For the evergreen species, leaves were separated into different age classes, placed in bags and transported to a refrigerated cabinet in the laboratory until processing.

All leaves were classified according to whether they showed symptoms of damage and according to the type of damage (either herbivory or partial senescence in the form of necrotic spots or chlorosis), and the number of leaves in each category was counted.There were no losses to herbivores for the three conifers.In leaves where damage was observed, five leaves of each type were selected (drought or herbivory damage) and scanned at high resolution (2300 × 3300 pixels).The percentage of damaged area was estimated from these images using the ImageJ program (http://rsb.info.nih.gov/ ij/; Abràmoff et al.2004), which allowed the measurement of both the whole leaf area and the parts with a different coloring due to necrosis or chlorosis.In cases where a part of the leaf surface had completely disappeared, the leaf contour was reconstructed to estimate the area lost.

In addition, from each individual and leaf age class on each site, five healthy leaves were taken at random from those collected in mid-June in the two study years.Leaf area was measured with an image analyzer (Delta-T Devices LTD, Cambridge, UK).The leaves were then dried at 70 °C to constant mass, and the dry mass determined with an analytical balance (Sauter AR70).Leaf mass/unit area (LMA) was obtained as the ratio of dry mass to leaf area.

Data analysis

After verifying that there were no differences in LMA between the two study years, an average LMA of healthy leaves was obtained for each species and age class on each site as an average corresponding to the 10 trees analyzed in each case (five/year).

For each tree, the percentage of leaf area lost or damaged per branch in leaves of each age class (% TAL in branchcohortn) was obtained as a product of the fraction of damaged leaves of that age class by the average area lost per leaf of the same age class.Thus, for each tree, a value of area loss per branch for leaves of each age class was obtained.As leaves age, they showed more damage from accelerated senescence since they had been subjected to stress for a longer time.However, due to gradual shedding, older leaf cohorts represented a smaller proportion of the total leaf population of each tree.Therefore, to determine the contribution of each age class to the total losses in each tree, total damage per branch of each age class (% TAL in branchcohortn) was multiplied by the percentage of leaves pertaining to that age class (% L in branchcohortn).Data on the contribution of each age class to the total leaf area were taken from Mediavilla et al.(2018a).Previous studies (González-Zurdo et al.2016a) have revealed the absence of between-site differences in leaf life span of a single species, which allowed us to assume a similar percentage of leaves from each cohort of the same species between the different sites.Finally, the total damage to each tree (% TD tree) was obtained as the sum of the damage recorded for the different leaf cohorts:

An average value of total damage by species, including the contribution of all cohorts, was obtained for each site, year and season as an average of the five trees analyzed and expressed as mean ± standard error.For the oak species, values were obtained separately for each type of damage (partial senescence by abiotic factors and by herbivory), as well as the total damage values.Given the absence of herbivory symptoms in pine needles, total damages corresponded exclusively to accelerated senescence due to drought or heat stress during the summer.

An analysis of variance was used to establish significant differences between years, sampling dates, species, age classes and sites, followed by a post-hoc Student-Newman-Keuls test.Normality was checked with the Kolmogorov-Smirnov test and homoscedasticity confirmed with the Levene test.Linear regression analysis was used to explore the relationships between leaf longevity and damage levels.Fractional data were previously subjected to the logit transformation [ln(p/(1-p))] (Warton and Hui 2011).All analyses were performed with the SPSS statistical package (SPSS Inc., Chicago, IL).

Results

Healthy leaf traits of different species

Within each site, LMA increased with the increase in leaf life span of each species (Table 2).OnlyP.sylvestrisshowed a significantly lower LMA than expected based on the average leaf life span.Among the evergreens, LMA also increased with leaf age, with the highest values corresponding to the oldest leaves (Table 2).No between-site differences were observed in the LMA of the deciduousQ.pyrenaica, but for the evergreens, all species present on more than one site tended to show higher LMA values on colder sites, particularly site D (Table 2).

Herbivory levels

As noted previously, after inspection of the needle samples, there were no symptoms of herbivore attacks on the pine species.Levels of consumption by insects on the oak species were similar over the two study years and for the two sampling dates (before and after summer) in spite of the climatic differences.There were also no differences in losses to herbivores between sites for the same species or betweenleaves of the different age classes for the two evergreen oaks (Table 3).In contrast, the differences among species were significant.The two deciduous species showed greater leaf area losses compared to the evergreens (Fig.1).Between the two evergreen oaks,Q.ilexhad 25% lower leaf area losses thanQ.suber(Fig.1).

Fig.1 Leaf area lost to herbivores (%) in the different species (mean + SE, n=12 for Q.pyrenaica and Q.ilex, n=4 for Q.faginea and Q.suber).Different letters indicate significant differences among species

Levels of accelerated senescence

There were no differences between the two study years in levels of premature senescence due to abiotic stress (Table 4).In addition, levels of abiotic damage were also similar between sites for a single species (Table 4).ForQ.ilexandP.pinaster, there was a tendency for more damage on warmer sites, but the differences were not significant (data not shown).

In evergreen leaves there were significant differences in damage between seasons, species and age classes (Table 4).However, the increase in damage in the autumn relative to spring decreased in magnitude as leaf age increased, implying that new leaves are more sensitive to drought and high summer temperatures (Table 5).For the youngest leaves, damage decreased with an increase in LMA and leaf life span.Two species, however, did not follow this pattern.P.pineashowed lower than expected damage levels in both spring and in autumn, whereasP.sylvestrishad greater losses than expected, based on its leaf longevity, especially in autumn (Table 5).

Total loss of leaf area (herbivory + accelerated senescence)

Among the fourQuercusspecies, herbivory losses per tree were markedly reduced as leaf longevity and LMA increased (Table 6).Losses per tree due to premature senescence were particularly high inP.sylvestrisand low inP.pinea, especially after the summer drought.For theother species, losses tended to be higher in the deciduous species, but similar in the two evergreenQuercusspecies and inP.pinaster.This last species had a larger proportion of older leaves that accumulated more damage (Table 6).As a result of both trends, increased leaf longevity contributed to reducing the total premature leaf area losses.Thus, both deciduous species had the highest level of total losses, with values exceeding 70% of those recorded in the species with the longest leaf life (P.pinaster), most of which corresponded to herbivory.With the exception ofP.pinea, differences in total losses between seasons were significant for all species, with higher values in the autumn than in the spring.However, these between-season differences were due solely to increases in area lost by accelerated senescence, since the losses to herbivores in the oak species were similar in both seasons.

Table 3 ANOVA results of differences in average herbivory levels per branch in Quercus species

Table 4 ANOVA results of differences in mean levels of accelerated senescence per branch

Table 5 Mean values (+ SE, n=2-8) of the percentage of leaf area damaged per branch for the different leaf types, obtained as an average of those recorded on the different sites where the same species appeared and in different study years (C=current year leaves)

Table 6 Mean percentages (SE in brackets, n=2-8) of total premature losses per tree recorded by the different species and times of the year; average of the data obtained for each species on the different sites and years of study; different letters indicate significant differences between species for P ＜ 0.05 (ANOVA)

The relationship between leaf life span and total damage was highly significant for both sampling periods (Fig.2a), with a clear decrease in total losses in species with the longest-lived leaves.However,P.pineaandP.sylvestrisdeviated from the pattern.By mid-June,P.pineaneedles showed less damage than expected (44% less), while the opposite occurred forP.sylvestris(45% more) (Fig.2a).These differences intensified after the summer drought, withP.pineashowing 56% less damage than expected andP.sylvestris66% more losses than expected (Fig.2c).Total area losses also decreased significantly with the increase in LMA in both spring and autumn (Fig.2b, d).Most data points were close to the fitted line, with the exception ofP.pinea, which again showed lower losses than expected for its LMA.

Fig.2 Losses of leaf area by herbivory and abiotic stress in function of a, c leaf life span and b, d leaf mass per unit area (LMA) of the different species

Discussion

The implications of the differences in leaf longevity continue to be questioned in ecophysiology (Mediavilla and Escudero 2003b; Kikuzawa and Lechowicz 2006; Falster et al.2011; Edwards et al.2014).One of the objectives of our study was to focus on an aspect rarely addressed, that of partial leaf area losses owing to biotic and abiotic factors.To this end, a group of species were evaluated that had a wide range of leaf longevities and therefore, had differences in the degree of leaf structural reinforcement as measured by LMA.A higher LMA has been interpreted as a trait that confers mechanical resistance against different environmental factors (drought, low winter temperatures, herbivory), and thus promotes a longer leaf life span (Niinemets et al.2009; González-Zurdo et al.2016a).Longer-lived leaves are exposed to more stress events and therefore accumulate the damages produced bythese events.However, if a stronger structural reinforcement confers adequate protection, this could compensate for the longer exposure to stress factors, which would allow maintaining most of the photosynthetic biomass until the end of the leaf life.This may have an important effect on total leaflevel CO2assimilation (Mediavilla et al.2018a).

Among the species in this study, there was a favorable effect of leaf longevity through the reduction of the loss of leaf area by herbivory because a longer duration and exposure time were not accompanied by cumulative increases in consumption.Thus, for the oak species, the percentages of area lost to herbivores were similar in the two sampling periods and also similar for the different age classes in the two evergreen oaks.Therefore, our results reveal, in agreement with Coley (1988) and Lawrence et al.(2003), that consumption by herbivores occurs mainly during the initial phase of leaf growth when leaves have larger nutrient concentrations and are less structurally reinforced.In addition, for the current-year leaves, there was a decrease in herbivory levels with increased leaf longevity.Some studies have found no relationship between herbivory and leaf life span (Kozlov et al.2015), others found positive relationships (Loranger et al.2012), while some researchers have reported higher levels of herbivory in species with intermediate leaf longevity (Zhang et al.2017).However, most studies have shown, as in this study, a negative relationship between leaf life span and consumption levels (Coley 1988; Silva et al.2015).Recent studies on herbivory attribute a predominant role to the degree of structural reinforcement of the leaves more than to the concentration of phenols or nitrogen (Agrawal and Weber 2015; Mediavilla et al.2018b).Therefore, the high LMA typical of longer-lived leaves allows significantly lower herbivory rates than those of deciduous species.

In view of the effects of the differences in structural reinforcement on herbivory, we expected that insect damage would be less for a single species on sites where its leaves have higher LMA.Of the species found on more than one site, there were no differences in LMA between sites forQ.pyrenaica.In contrast,Q.ilexhad larger LMA values on site D, possibly in response to low winter temperatures and the high number of days of frost on this site (González-Zurdo et al.2016a).However, there were no differences in herbivory between sites even forQ.ilex.It is possible that the more favorable conditions on site D, especially precipitation, during periods of high insect activity in the spring might have favored an increase in insect populations (Jamieson et al.2012) that would counteract the favorable effect of the higher LMA ofQ.ilexleaves.

The higher LMA associated with an increased leaf longevity apparently also contributed to reducing leaf area losses in response to abiotic factors in current year leaves.This was observed not only after the summer drought since in mid-June some leaf area loss was already detected owing to high temperatures and radiation levels typical of late spring under a Mediterranean climate.The favorable effect of larger LMA of longer-lived leaves was offset to some extent by the presence of older leaf cohorts that accumulated more damage as they experienced more stress events.However, although an important part of the crown of the evergreen species consisted of older leaves, most of the leaf biomass was contributed by 1-2 year-old leaves (Mediavilla et al.2018a), which showed less abiotic damage.These contrasting effects contributed to reduce interspecifci differences in damage by early senescence, although the losses remained slightly higher for the deciduous species.

Although a higher LMA seems to be an important characteristic for minimizing drought damage, in this study, for a species occupying different sites, we did not find higher LMA in the driest environments.However, it tended to be greater on the coldest sites, suggesting that this trait may mainly respond to differences in frost intensity (Ogaya and Pe?uelas 2007; González Zurdo et al.2016a).The lack of LMA response to rainfall levels translated to more damage after summer on the driest sites both forQ.ilexandP.pinaster, although the differences were not significant.

As a result of the effects of leaf life span on the losses by herbivory and early senescence, a significant negative relationship was observed between leaf longevity and total leaf area loss, which corroborates the results of a study by Mediavilla et al.(2018a).The two deciduous species had losses up to three times higher than the longer-lived species.Accordingly, interspecific differences in premature leaf area loss should be considered as another determinant of the consequences of differences in leaf life span.Sixty percent of the leaf losses in deciduous species was due to herbivory, which has greater consequences on assimilation losses.Consumption by insects occurred on young leaves resulting in the loss of photosynthetic area since the early stages of leaf life.The presence of older cohorts that accumulated damage over time contributed to increasing mean losses in longerlived leaves, but these losses were less in younger leaves.Since an increase in leaf age is accompanied by a decrease in photosynthetic capacity (Kitajima et al.2002; Zhang et al.2008), the impact of abiotic damage during the late leaf life stages should have less negative consequences for total leaflevel CO2assimilation.

However, in addition to the effects of LMA on premature losses of leaf area, other species-specific characteristics contribute to determine the losses by accelerated senescence in response to drought.P.pineaexhibited lower losses thanP.pinaster, in spite of its lower LMA, whileP.sylvestris had accelerated senescence losses higher than those of the deciduous species.

Quercusspecies are classified as anisohydric species because they maintain relatively constant gas exchange rates, allowing the water potential of the leaves to decrease with a decrease in water availability (Aranda et al.2014; Martínez-Sancho et al.2017).Several studies have shown that loss of hydraulic conductivity due to xylem cavitation reduces the plant’s ability to transport water to the leaves (Pockman and Sperry 2000; Kaack et al.2019), and that this is a key factor in determining drought resistance.Through stomatal regulation, plants control water loss and maintain water potential above the cavitation threshold, thus preserving the integrity of the vascular system (Oren et al.1999; Martinez-Vilalta et al.2003).However, the short life span of leaves ofQ.pyrenaicaandQ.fagineaforces these species to maximize gas exchange rates during the short time available for photosynthesis, resulting in high transpiration rates throughout the summer.Previous studies have shown that both species exhibit higher stomatal conductance and less sensitivity to increased edaphic and atmospheric water stress than evergreens such asQuercus ilexandQ.suber(Mediavilla and Escudero 2003c).This results in more pronounced decreases in leaf water potential that make the two deciduous species operate closer to the threshold of hydraulic failure, making them vulnerable to leaf damage by drought (Martínez-Vilalta et al.2003).The two evergreenQuercus, in contrast, show a more pronounced stomatal response to changes in water availability, allowing for less pronounced decreases in leaf water potential and helping to prevent hydraulic failure (Mediavilla and Escudero 2003c).Several studies confirm that, particularly in these Mediterranean evergreenQuercusspecies, there is a preferential allocation of biomass to the roots, thus increasing water availability and again reducing the risk of critical levels of water failure (Cano et al.2013; Barbeta et al.2015).Q.ilexis also one of the most cavitation-resistant species compared to otherQuercusspecies (Urli et al.2013), with a large part of its resources allocated to the maintenance and recovery of the xylem conductivity instead of growth (Castell et al.1994; Carnicer et al.2013).This characteristic allowsQ.ilexto experience damage levels slightly lower than would be expected based on its leaf life span and LMA.

Pine species, on the other hand, are isohydric species that, due to the structure of their xylem, are more sensitive to embolism than angiosperms, maintaining plant water status through strong stomatal control (Urli et al.2013; Hammond et al.2019).The combination of a high LMA and a watersaving strategy combines to explain the low leaf area losses in response to drought byPinus pinaster.P.sylvestrishad a lower LMA than expected based on its leaf life span, which could explain its higher levels of accelerated senescence after drought, in spite of occupying two sites with high water availability.However, it had more leaf senescence after summer than the two deciduousQuercusspecies, meaning that other factors contribute to explain these high losses.Several researchers have suggested that, among the different pine species,P.sylvestris, at the southernmost limit of its range as it is in the Iberian Peninsula, maintains a strict stomatal regulation to prevent the development of embolism (Poyatos et al.2013; Gea-Izquierdo et al.2014).This suggests that carbon absorption may also be compromised in this species under conditions of drought and thus the availability of carbohydrates to invest in growth and in structural strengthening of the leaves makes them more sensitive in subsequent years (Poyatos et al.2013; Gea-Izquierdo et al.2014; Rehschuh et al.2020).P.pinea, on the other hand, showed less loss than expected based on its LMA values even though it occupied the site with the highest temperatures and lowest levels of precipitation.In fact,P.pineahad the lowest levels of senescence after summer, perhaps because the proportion of older leaves with more accumulated abiotic damage was also lower than that of the other pines.This species exhibits a high tolerance to photoinhibition (Sperlich et al.2014), as well as an extensive and deep root system (reaching depths of 8-10 m, Caneva et al.2009), which confers an efficient buffer on the impact of drought.This allows the species to maintain less stomatal control, and thus higher photosynthetic rates and greater availability of resources, decreasing its sensitivity to drought.

Conclusion

Our results demonstrate the importance of including the analysis of premature leaf area loss in comparative studies of the responses of different species to stress factors, especially in the context of climate change.Loss of leaf area depends on the level of structural reinforcement that in turn varies with differences in leaf life span.The partial losses of photosynthetic area observed in this study contribute to determine the performance of species situated along the leaf economics spectrum.The high levels of accelerated leaf senescence inP.sylvestriscould exacerbate the unfavorable effect of carbon deficits due to the strong stomatal control that this species has in our latitudes.In contrast, the limited losses observed forP.pineacould increase the competitive advantage that this species seems to show over other coexisting species in environments under water stress.The intense premature losses of photosynthetic area in some species such as the two deciduous species, can also have important repercussions on their performance and thus on their competitive capacity under conditions of increasing drought and heat stress.

AcknowledgementsWe thank Teresa Malvar-Ferreras for her help in the analysis of the samples.