This is the accepted version of the article: Bogdziewicz, Michał; Fernández-Martínez, Marcos; Espelta Morral, Josep Maria; [et al.]. «Is forest fecundity resistant to drought? Results from an 18-yr rainfall- reduction experiment». New phytologist, Vol. 227, Issue 4 (August 2020), p. 1073-1080. DOI 10.1111/nph.16597 This version is avaible at https://ddd.uab.cat/record/224209 under the terms of the license 1 Is forest fecundity resistant to drought? Results from an 18-year rainfall-reduction experiment 2 3 Michał Bogdziewicz1,2, Marcos Fernández-Martínez3, Josep M. Espelta2, Romà Ogaya2, 4 Josep Penuelas2,4 5 6 1 Department of Systematic Zoology, Faculty of Biology, Adam Mickiewicz University, 7 Poznań, Poland 8 2 CREAF, Cerdanyola del, Vallès 08193, Catalonia, Spain 9 3 PLECO (Plants and Ecosystems), Department of Biology, University of Antwerp, 2610 10 Wilrijk, Belgium 11 4 CSIC, Global Ecology Unit, Cerdanyola del Vallès 08193, Catalonia, Spain 12 1 13 Summary 14 • Recruitment is a primary determinant of the long-term dynamics of plant populations 15 in changing environments. However, little is known about the effects of anthropogenic 16 environmental changes on reproductive ecology of trees. 17 • We evaluated the impact of experimentally induced 18 years of drought on 18 reproduction of three contrasting forest trees: Quercus ilex, Phillyrea latifolia, and 19 Arbutus unedo. 20 • Rainfall reduction did not decrease tree fecundity. Drought, however, affected the 21 allocation of resources in Q. ilex and A. unedo but not the more drought-tolerant P. 22 latifolia. Larger crop production by both Q. ilex and A. unedo was associated with a 23 stronger decrease in growth in the rainfall-reduction plots compared to the control 24 plots, suggesting that these species were able to maintain their fecundity by shifting 25 their allocation of resources away from growth. 26 • Our results indicated resistance to change in tree fecundity in Mediterranean-type 27 forest subjected to an average 15% decrease in the amount of soil moisture, suggesting 28 that these ecosystems may adapt to a progressive increase in arid conditions. 29 However, the species-specific reductions in growth may indirectly affect future 30 fecundity and ultimately shift community composition, even without immediate direct 31 effects of drought on tree fecundity. 32 33 Keywords: drought, fruit production, global change, mast seeding, rainfall-reduction, 34 reproduction, tree fecundity, 35 36 Introduction 37 Anthropogenic environmental changes are exerting increasing pressure on forests worldwide 38 (Gauthier et al. 2015, Seidl et al. 2017), and accumulating evidence indicates that climate 39 change is causing dramatic forest diebacks (Allen et al. 2010, Seidl et al. 2017, Lloret and 40 Kitzberger 2018). The critical question now concerns what ecosystems will follow from these 41 profound transformations. Few studies, however, have compared the impacts of 42 environmental change on the reproductive ecology of trees to other effects such as growth, 43 carbon sequestration, mortality, or phenology (Barbeta et al. 2013, Hacket-Pain et al. 2016, 44 Zohner et al. 2018, Luo et al. 2019). Ecosystem services, such as mitigating the risk of 45 avalanches, carbon storage, habitat availability, and value for the economy and recreation, can 2 46 suffer if reduced reproduction slows forest expansion or limits the recruitment of 47 merchantable tree species and seed producers that support wildlife (McShea 2000, Ostfeld 48 and Keesing 2000, Clark et al. 2007, 2019, Bogdziewicz et al. 2016). The volatility of seed 49 production and our poor understanding of the mechanisms that govern it are challenges for 50 anticipating change in forest reproduction (Bogdziewicz et al. 2020a). Reliable predictive 51 models are consequently not available, and the unpredictable recruitment of trees has become 52 a key obstacle to understanding forest change (Ibáñez et al. 2009, Zhu et al. 2012). 53 Tree reproduction is sensitive to climate change (Mckone et al. 1998, Pearse et al. 54 2014, Monks et al. 2016, Vacchiano et al. 2017). Observational studies of long-term trends in 55 fecundity report both increases and decreases in mean reproductive effort in many important 56 forest-forming species (Richardson et al. 2005, Mutke et al. 2005, Redmond et al. 2012, Allen 57 et al. 2014, Buechling et al. 2016, Bogdziewicz et al. 2020b). A wide array of statistical tools 58 used in these studies usually attributes these trends to global warming, but substantial 59 uncertainty remains, as causality remains unestablished. Experiments that simulate 60 environmental conditions projected by models of global change are thus useful for predicting 61 the impacts of environmental global change on the reproductive patterns of forest trees. Such 62 experiments usually report substantial effects. For example, Pinus taeda growing in an 63 atmosphere enriched in CO2 produced three-fold as many cones as trees growing under 64 natural conditions (LaDeau and Clark 2001). Excluding rain reduced the production of seed 65 biomass in Quercus ilex by thirty percent (Pérez-Ramos et al. 2010). Such experimental 66 studies are nonetheless almost inevitably limited in time. A meta-analysis of global-change 67 experiments have reported a dampening effect size of treatments (warming, nitrogen 68 fertilization, or drought) over time (Leuzinger et al. 2011). Monitoring experimental systems 69 as long as possible is thus desirable for assessing the long-term impacts of global change on 70 forest fecundity. 71 The drought experiment in the Prades Mountains in southern Catalonia runs since 72 1999, and is one of the longest running forest global-change experiments in the world (Wu et 73 al. 2011, Barbeta et al. 2013, Peñuelas et al. 2018). The experiment is being conducted in a 74 typical holm oak (Q. ilex) forest, where the oak is accompanied by other Mediterranean 75 woody species with more (Phillyrea latifolia) and less (Arbutus unedo) drought tolerance 76 (Peñuelas et al. 2018). Important demographic effects have already been observed, such as a 77 higher mortality of stems and reduced growth, especially in Q. ilex and A. unedo (Lloret et al. 78 2004, Ogaya and Peñuelas 2007b). The differences in the rates of growth and mortality 79 between drought and control plots recorded at the beginning of the experiment eventually 3 80 decreased after some years (Barbeta et al. 2013, Liu et al. 2015). Decreased competition with 81 high mortality after extreme droughts, and possible morphological and physiological 82 acclimation to drought during the study period, may buffer forests against drier conditions 83 (Barbeta et al. 2013, Liu et al. 2015, Peñuelas et al. 2018). 84 We evaluated the impact of experimentally induced drought on the fecundity of the 85 dominant forest trees at our experimental site: Q. ilex, P. latifolia, and A. unedo. All three 86 species mast at our sites, i.e. reproduced by the synchronous and temporally variable 87 production of seeds (Kelly 1994). The sensitivity of reproduction of mast-seeding species to 88 global change is predicted to be especially high, due to hypersensitivity of masting plants seed 89 production to variation in the weather (Mckone et al. 1998, Monks et al. 2016, Vacchiano et 90 al. 2017). In addition to the important trends in mean fecundity, changes in the strength of 91 masting (i.e. the interannual variability and synchrony of reproduction) are crucial for tree 92 fitness and forest regeneration, because masting is a life-history trade-off among missed 93 reproductive opportunities in low-seed years, increased pollination efficiency, and decreased 94 seed predation in mast years (Kelly 1994, Pearse et al. 2016, Bogdziewicz et al. 2020b). We 95 thus also tested the effects on coupling between plants and variation among years, in addition 96 to evaluating the effects of drought on mean fruit production. We predicted that drought 97 would reduce mean reproductive output in Q. ilex and A. unedo but not P. latifolia 98 (prediction 1) based on studies reporting that P. latifolia was much more drought-tolerant 99 than the other two species (Ogaya and Peñuelas 2007b, Barbeta et al. 2013, Peñuelas et al. 100 2018). We also predicted that the effect would dampen with time (prediction 2), paralleling 101 the diminishing effects of drought on growth and mortality (Barbeta et al. 2013). The theory 102 of mast seeding predicts that more frequent adverse weather would increase the interannual 103 variability of seed production and strengthen the synchrony of reproductive variation among 104 trees (Rees et al. 2002, Espelta et al. 2008, Bogdziewicz et al. 2018). We thus predicted an 105 increase in annual variability and synchrony in all three species on experimental plots relative 106 to the control, but likely less so in drought-resistant P. latifolia than in the other two species 107 (prediction 3). Finally, we expected that the drought experiment would induce variation in 108 the strength of the trade-off between growth and reproduction (prediction 4): the trade-off 109 would be stronger under stressful conditions (drought) (Martín et al. 2015, Berdanier and 110 Clark 2016, Hacket-Pain et al. 2017). 111 112 Materials and Methods 113 Rainfall-reduction experiment 4 114 We established the experimental site in 1999 on a 25% south-facing slope in the Prades Holm 115 oak forest in southern Catalonia (northeastern Spain) (41°21′N, 1°2′E; 930 m a.s.l.). As a 116 result of former coppicing the forest has a very dense multi-stem canopy layer (15 433 stems 117 ha-1) dominated by Q. ilex (5258 stems ha-1), P. latifolia (7675 stems ha-1), and A. unedo 118 (1100 stems ha-1), accompanied by other Mediterranean woody species that usually do not 119 reach the upper canopy (e.g. Erica arborea and Juniperus oxycedrus) and occasional isolated 120 deciduous trees (e.g. Sorbus torminalis and Acer monspessulanum). Holm oak forests in the 121 Prades Mountains grow throughout the altitudinal range (400–1200 m), presenting closed 122 canopies 3–10 m in height depending on site quality. This forest has been managed as a 123 coppice for centuries but has not been substantially disturbed for the last 70 years. 124 The site has a Mediterranean climate with a mean annual temperature of 12.4 °C and a 125 mean annual precipitation of 610 mm during the study period (see Results). Annual and 126 seasonal precipitation are irregularly distributed, with annual precipitation ranging from 355 127 to 1021 mm in the 19 years of this study. Spring and autumn are the wettest seasons, and 128 summer drought usually lasts three months, during which precipitation is ~10% of the annual 129 total and coincides with the highest temperatures. 130 The experimental system consisted of eight 150-m2 plots delimited at the same 131 altitude along the slope. We randomly selected half of the plots to receive the drought 132 treatment, and the other half had natural conditions. We partially excluded precipitation in the 133 drought treatment using PVC strips suspended 0.5–0.8 m above the soil and covering 134 approximately 30% of the plot surfaces, similarly as other drought experiments in 135 Mediterranean systems (Limousin et al. 2008). We dug a ditch 0.8 m in depth along the top 136 edge of the plots to intercept water runoff. 137 We installed an automatic meteorological station (Campbell Scientific Inc., Logan, 138 USA) between the plots to monitor temperature, photosynthetic active radiation, air humidity, 139 and precipitation, from which we obtained the Standardized Precipitation Evapotranspiration 140 Index (SPEI) as a measure of atmospheric hydric conditions. SPEI is calculated as the 141 difference between monthly precipitation and potential evapotranspiration (Beguería et al. 142 2014). High and low SPEI values therefore indicate wet and drought conditions, respectively. 143 We selected timescales of 3 (SPEI-3) and 6 (SPEI- 6) months, because they fit our annual 144 data on plant growth and population dynamics analyzed in earlier studies (Barbeta et al. 2013, 145 Liu et al. 2015). The SPEI values are provided for month and timescale of calculation (e.g. 146 SPEI_May3 refers to the water balance for March, April, and May of a given year). We also 147 measured soil moisture each month throughout the experiment by time-domain reflectometry 5 148 (Tektronix 1502C, Tektronix, Beaverton, USA), connecting the time-domain reflectometer to 149 the ends of three stainless-steel cylindrical rods, 25 cm long and fully driven into the soil, at 150 four randomly selected locations per plot. 151 We randomly distributed 20 circular baskets (27 cm in diameter with a 1.5-mm mesh) 152 on the ground in each of the eight plots and collected the fallen litter every two months from 153 1999 to 2017. Fruits were weighed after drying in an oven at 70 °C to constant weight. We 154 measured all living stems of all species with diameters >2 cm at a height of 50 cm each winter 155 since 1999. We then calculated stem basal area increments (BAIs) and began in winter 2009 156 to also measure the individuals with diameters <2 cm at the beginning of the study but which 157 then attained or exceeded this size. In total, we measured 735 stems in Q. ilex, 842 in P. 158 latifolia, 145 in A. unedo. Average (SD) number of stems per plot equaled 105 (47) in Q. ilex, 159 128 (90) in P. latifolia, and 20 (14) in A. unedo. 160 161 Model species 162 Holm oak (Quercus ilex L.) is a drought- tolerant tree widely distributed in the Mediterranean 163 basin. Phillyrea latifolia L. is a small tree associated with Q. ilex forests and more resistant to 164 drought and high temperatures than Q. ilex (Ogaya and Peñuelas, 2003; Peñuelas et al. 2018). 165 Arbutus unedo L. is another small tree typical of holm oak forests, less resistant to drought 166 than P. latifolia (Ogaya and Peñuelas, 2003; Peñuelas et al. 2018). The reproductive 167 phenologies of Q. ilex and P. latifolia are typical for Mediterranean species. Flowering takes 168 place in spring, fruit development in summer and fruit maturation and seed dispersal in 169 autumn (Ogaya and Peñuelas, 2004). In A. unedo, flower bud formation occurs in spring, but 170 flowering takes place in the following autumn, and fruit development continues over a 171 prolonged period until fruit matures in the autumn of the following year (Ogaya and Peñuelas, 172 2004). 173 174 Statistical analysis 175 We evaluated the impact of excluding rain on soil-water content by building a linear 176 mixed model (LMM), with soil moisture as the response and treatment (control vs drought) as 177 a fixed effect. Month and year were included as random intercepts. 178 We next evaluated the impact of excluding rain on fruit production (fruit dry mass per 179 plot) in the model species using LMMs that included log-transformed fruit mass as a 180 response, with the interaction between treatment and year as fixed effects (prediction 1 & 2). 181 The interaction was included to test for possible dampening effects of the drought treatment 6 182 on fruit dry mass production. We built a separate model for each species. Each model also 183 included plot as a random intercept and SPEI as a covariate. The specific month for each 184 SPEI was pre-identified for each species by fitting a partial least squares regression (PLS) of 185 fruit production vs all possible SPEI values. PLS is designed to analyze a large array of 186 related predictor variables, with insufficient sample sizes relative to the number of 187 independent variables (Carrascal et al. 2009). The number of plot-years for the analysis was 188 144 for Q. ilex and P. latifolia and 126 for A. unedo, which was absent in one of the plots. 189 We next evaluated the influence of the experimental drought on the interannual 190 variability and synchrony of fruit dry mass production (prediction 2). We estimated the 191 synchrony of fruit production for each treatment by calculating the mean pairwise cross- 192 correlation of fruit production over all plots of a treatment. The cross-correlations were 193 calculated using the mSynch function in the ncf package (Bjornstad and Falck 2001). We also 194 calculated measures of interannual variability for the treatments using the coefficient of 195 variation (CV) and a proportional variability index (PV) (Heath 2006). We used both indexes 196 because CV can be skewed by its mean-dependency, while PV is not (Fernández‐Martínez et 197 al. 2018). Yet, CV is widely used to measure inter-annual variation in seed production, 198 allowing among-studies comparisons. The corresponding 95% confidence intervals (CIs) for 199 the focal values were calculated by bootstrap resampling with 1000 replications. 200 We built LMMs that included annual stem diameter increment (BAI) as a response to 201 test whether excluding rain changes the trade-off between growth and fruit dry mass 202 production in our model species, with the interaction between log-transformed fruit mass and 203 treatment as a fixed effect (prediction 4). We built a separate model for each species. Each 204 model included plot and tree as random intercepts. We also included SPEI values as 205 covariates, which were similarly pre-selected for each species as in the models testing for the 206 effects of treatment on fruiting. The BAIs were standardized within trees before inclusion in 207 the models (i.e. we extracted the mean of each value and then divided it by the standard 208 error). In each model, we also included a matrix for a natural cubic spline (df = 5) of tree size 209 to account for growth-related trends in BAI, using the ns function from splines package. The 210 sample sizes for these models, i.e. per stem per year observations, were 11 288 for Q. ilex (735 211 stems measured), 15 301 for P. latifolia (842 stems), and 2242 for A. unedo (145 stems). 212 We fitted all models using the glmmTMB package (Brooks et al. 2017) in R. Models 213 were validated, including tests for homoscedasticity and normality of residuals and potential 214 outliers, using the DHARMA package (Hartig 2019). We also explored all models with 7 215 temporal autocorrelation structures (ar1) and retained or rejected them based on standard 216 Akaike information criteria. 217 218 Results 219 The drought treatment decreased the soil moisture by ~10–30% during the study period (z = 220 8.32, p < 0.001), with larger differences during periods of rain (Fig. 1). The drought treatment 221 decreased the amount of soil moisture by ~20% or more during these periods but by <10% 222 during dry periods. Median reduction equaled 13%, while mean 15%. 223 The drought treatment did not decrease fruit dry mass production in any of the species 224 (Fig. 2), contrary to our expectations (prediction 1). The interaction between treatment and 225 year was not significant for any of the species (p > 0.10) (prediction 2). The drought treatment 226 without the interaction term did not significantly affect fruit production by Q. ilex (z = -0.64, 227 p = 0.52), P. latifolia (z = 0.10, p = 0.92), or A. unedo (z = -1.60, p = 0.11). SPEI was 228 positively correlated with crop production for all three species (Q. ilex: SPEI_March3, z = 229 4.49, p > 0.001, P. latifolia: SPEI_March3, z = 5.29, p > 0.001, A. unedo: SPEI_December3, 230 z = 2.59, p = 0.01). 231 The variability of fruit production among years was high for all species, with positive 232 pairwise cross-correlations in fruiting among plots (mean cross-correlation >0.60 for all 233 species, Fig. 3). Both interannual variability, as measured by either PV or CV, and synchrony, 234 were unaffected by excluding rain for all three species (p > 0.05), contrary to our expectations 235 (prediction 3). 236 Excluding rain affected reproductive allocation for Q. ilex and A. unedo but not P. 237 latifolia (Fig. 4) (prediction 4). Q. ilex fruit production was positively correlated with growth 238 in the control plots (β [SE] = 0.06 [0.005]; z = 10.85, p < 0.001), but the slope of the 239 relationship (β [SE] = 0.04 [0.005]; z = 7.50, p < 0.001) was lower in the drought plots 240 (interaction term: z = -5.57, p < 0.001). Fruit production by A. unedo was similarly positively 241 correlated with growth in the control plots (β [SE] = 0.05 [0.009]; z = 5.86, p < 0.001), but the 242 slope of the relationship (β [SE] = 0.02 [0.01]; z = 2.06, p = 0.04) was lower in the drought 243 plots (interaction term: z = -2.15, p = 0.03). P. latifolia growth was positively correlated with 244 reproduction (β [SE] = 0.05 [0.003]; z = 14.78, p < 0.001), and the induced drought did not 245 affect this pattern (interaction term: z = -0.97, p = 0.33). Cubic splines of size were not 246 significant predictions of BAI in all three species (p < 0.10). 247 248 Discussion 8 249 The 18-year experimental rainfall reduction did not decrease tree fecundity, contrary to 250 predictions. The year-to-year variability and synchrony of reproduction was also unaffected 251 the levels of drought induced by our experiment. Drought, however, affected the allocation of 252 resources in Q. ilex and A. unedo but not the more drought-tolerant P. latifolia. Production of 253 larger crop size by both Q. ilex and A. unedo was associated with a stronger decrease in 254 growth in the rainfall-reduction plots compared to the control plots, suggesting that these 255 species were able to maintain their fecundity by shifting their allocation of resources away 256 from growth. 257 Theory predicts that tree reproduction will be sensitive to climate change, due to 258 strong correlations between seed production and annual variation in the weather (Pearse et al. 259 2016). Fruit production by the three species studied here was previously correlated with 260 drought severity (Ogaya and Peñuelas 2007a, Espelta et al. 2008, Bogdziewicz et al. 2017b), 261 but fecundity was resistant to the level of drought induced by this experiment. One possibility 262 is that water stress induced by natural drought also includes decrease in atmospheric water 263 availability. While our rainfall reduction treatment cannot affect that, both soil moisture and 264 atmospheric demand for water independently limit and affect vegetation productivity and 265 water use during periods of hydrologic stress (Novick et al. 2016). What is more, the 266 synchrony and interannual variability of reproduction were consequently also unaffected. 267 More frequent adverse weather, such as drought, should increase variability and synchrony 268 among years by decreasing reproduction in some years, thus strengthening the reinforcing 269 effects of stored resources on the synchrony of reproductive variation among trees (Rees et al. 270 2002, Espelta et al. 2008, Bogdziewicz et al. 2018, 2019). Such effect is, however, only 271 expected in the case of very intense drought episodes when reproductive failures happen more 272 frequently (Espelta et al. 2008), which was not the case in our forest. 273 Our results imply that the maintenance of fecundity under drought stress was possible 274 by decreasing growth in Q. ilex and A. unedo, indicated by changes in the trade-off between 275 growth and reproduction with and without stress. The slope of the positive relationship 276 between fruit production with growth was reduced by over 30% in Q. ilex, and 60% in A. 277 unedo in drought treatment compared to control. P. latifolia was in turn able to sustain both 278 growth and reproduction under induced drought. Stem growth in the drought treatment was 279 >60% lower for A. unedo, >17% lower for Q. ilex, and was unaffected for P. latifolia 280 compared to the control plots (Barbeta et al. 2013). Drought reduces transpiration by stomatal 281 closure in Q. ilex and A. unedo, which decreases the assimilation of carbon (Limousin et al. 282 2009, Ripullone et al. 2009). Drought also increases litterfall, likely due to xylem cavitation 9 283 that accelerates foliar senescence (Choat et al. 2012, Liu et al. 2015). Defoliation in turn 284 decreases the carbon content of plant tissues (Rosas et al. 2013), suggesting that drought 285 stress decreased resource availability in Q. ilex and A. unedo and forced the trees to partition 286 the limited resources to reproduction at the expense of growth, providing experimental 287 evidence for intraspecific variability and phenotypic plasticity in the cost of reproduction due 288 to habitat differences. Alternatively, the positive association between growth and fruit 289 production could follow from both growth and reproduction responding similarly to water 290 availability. If in the drought treatment decreases variability in water availability, it could 291 weaken the correlation between growth and fruiting. However, our data suggest that the 292 variance in soil moisture was similar in both treatments (Levene's Test, F = 0.44, p = 0.51; 293 Fig. 1d). Another important implication of these findings is that Q. ilex has avoided reduced 294 growth associated with reproduction throughout most of its range (Pérez-Ramos et al. 2010, 295 Fernández-Martínez et al. 2015), but our results imply that it may not continue do so in the 296 near future due to the progressive increase in drought frequency predicted by models of global 297 change. 298 Generally positive relationship between growth and reproduction in all three species, 299 indicate that favorable meteorological conditions could increase the accumulation of 300 resources and their subsequent allocation to both growth and reproduction in certain years 301 (Norton and Kelly 1988, Vergotti et al. 2019). Nonetheless, the maintenance of reproduction 302 at the expense of growth, together with the previously established link between drought, 303 reduced growth, and elevated mortality at our site (Ogaya and Peñuelas 2007b, Barbeta et al. 304 2013, Liu et al. 2015), well supports the theory of the cost of reproduction in plants, where 305 current reproductive allocation at the expense of growth is predicted to influence the 306 probability of future survival (Obeso 2002). The lack of direct effects of drought on fecundity 307 thus does not preclude indirect costs of fecundity from sustained lower growth rates, which 308 may influence future reproduction. 309 The results of this study indicated substantial resistance of tree fecundity in a Q. ilex 310 dominated forest subjected to an average 15% (median 13%) decrease in the amount of soil 311 moisture. The length of the study provides consistency to these results. Growing evidence 312 indicates that Q. ilex dominated forests are resistant to an increase in drought to some extent, 313 suggesting that these ecosystems may adapt to a progressive increase in arid conditions 314 (Peñuelas et al. 2018). Our study, however, comes with an important warning. The species- 315 specific reductions in growth and increased mortality (Ogaya and Peñuelas 2007b, Barbeta et 10 316 al. 2013) may indirectly affect future fecundity and ultimately shift community composition, 317 even without immediate direct effects of drought on tree fecundity. 318 319 Acknowledgements 320 This research was funded by the Spanish government project CGL2016-79835-P, the Catalan 321 government project SGR2014-274, and the European Research Council Synergy grant ERC- 322 2013-SyG-610028 IMBALANCE-P. MB was supported by (Polish) National Science Centre 323 grant no. 2018/28/U/NZ8/00003 (Uwertura) and 2017/24/C/NZ8/00151 (Sonatina). MFM is a 324 postdoctoral fellow of the Research Foundation-Flanders (FWO). 325 326 Author Contribution 327 All authors conceived the study, RO and JP collected data, MB and MFM ran the analysis, all 328 authors participated in the evaluation of the results, MB drafted the manuscript, and all 329 authors participated in the editing and approved the final version. 330 331 References 332 Allen, C. D., A. K. Macalady, H. Chenchouni, D. Bachelet, N. McDowell, M. Vennetier, T. 333 Kitzberger, A. Rigling, D. D. Breshears, E. H. (Ted) Hogg, P. Gonzalez, R. Fensham, 334 Z. Zhang, J. Castro, N. Demidova, J.-H. Lim, G. Allard, S. W. Running, A. Semerci, 335 and N. Cobb. 2010. A global overview of drought and heat-induced tree mortality 336 reveals emerging climate change risks for forests. Forest Ecology and Management 337 259:660–684. 338 Allen, R. B., J. M. Hurst, J. Portier, and S. J. Richardson. 2014. Elevation-dependent 339 responses of tree mast seeding to climate change over 45 years. Ecology and 340 Evolution 4:3525–3537. 341 Barbeta, A., R. Ogaya, and J. Peñuelas. 2013. Dampening effects of long-term experimental 342 drought on growth and mortality rates of a Holm oak forest. Global Change Biology 343 19:3133–3144. 344 Beguería, S., S. M. Vicente‐Serrano, F. Reig, and B. Latorre. 2014. Standardized precipitation 345 evapotranspiration index (SPEI) revisited: parameter fitting, evapotranspiration 346 models, tools, datasets and drought monitoring. International Journal of Climatology 347 34:3001–3023. 348 Berdanier, A. B., and J. S. Clark. 2016. Divergent reproductive allocation trade-offs with 349 canopy exposure across tree species in temperate forests. Ecosphere 7:e01313. 11 350 BjØrnstad, O. N., and W. Falck. 2001. Nonparametric spatial covariance functions: 351 Estimation and testing. Environmental and Ecological Statistics 8:53–70. 352 Bogdziewicz, M., D. Ascoli, A. Hacket‐Pain, W. D. Koenig, I. Pearse, M. Pesendorfer, A. 353 Satake, P. Thomas, G. Vacchiano, T. Wohlgemuth, and A. Tanentzap. (2020a). From 354 theory to experiments for testing the proximate mechanisms of mast seeding: an 355 agenda for an experimental ecology. Ecology Letters doi.org/10.1111/ele.13442 356 Bogdziewicz M., Kelly D., Thomas P., Lageard J., Hacket-Pain A. (2020) Climate warming 357 disrupts mast seeding and its fitness benefits in European beech. Nature Plants, 358 doi.org/10.1038/s41477-020-0592-8 359 Bogdziewicz, M., E. E. Crone, M. A. Steele, and R. Zwolak. 2017a. Effects of nitrogen 360 deposition on reproduction in a masting tree: benefits of higher seed production are 361 trumped by negative biotic interactions. Journal of Ecology 105:310–320. 362 Bogdziewicz, M., M. Fernández-Martínez, R. Bonal, J. Belmonte, and J. M. Espelta. 2017b. 363 The Moran effect and environmental vetoes: phenological synchrony and drought 364 drive seed production in a Mediterranean oak. Proceedings of the Royal Society B: 365 Biological Sciences 284:20171784. 366 Bogdziewicz, M., M. A. Steele, S. Marino, and E. E. Crone. 2018. Correlated seed failure as 367 an environmental veto to synchronize reproduction of masting plants. New Phytologist 368 219:98–108. 369 Bogdziewicz, M., R. Zwolak, and E. E. Crone. 2016. How do vertebrates respond to mast 370 seeding? Oikos 125:300–307. 371 Bogdziewicz, M., M. Żywiec, J. M. Espelta, M. Fernández-Martinez, R. Calama, M. Ledwoń, 372 E. McIntire, and E. E. Crone. 2019. Environmental Veto Synchronizes Mast Seeding 373 in Four Contrasting Tree Species. The American Naturalist 194:246–259. 374 Bonal, R., A. Muñoz, and J. M. Espelta. 2010. Mismatch between the timing of oviposition 375 and the seasonal optimum. The stochastic phenology of Mediterranean acorn weevils. 376 Ecological Entomology 35:270–278. 377 Brooks, M. E., K. Kristensen, K. J. van Benthem, A. Magnusson, C. W. Berg, A. Nielsen, H. 378 J. Skaug, M. Machler, and B. M. Bolker. 2017. glmmTMB balances speed and 379 flexibility among packages for zero-inflated generalized linear mixed modeling. The R 380 journal 9:378–400. 381 Buechling, A., P. H. Martin, C. D. Canham, W. D. Shepperd, and M. A. Battaglia. 2016. 382 Climate drivers of seed production in Picea engelmannii and response to warming 383 temperatures in the southern Rocky Mountains. Journal of Ecology 104:1051–1062. 12 384 Carrascal, L. M., I. Galván, and O. Gordo. 2009. Partial least squares regression as an 385 alternative to current regression methods used in ecology. Oikos 118:681–690. 386 Choat, B., S. Jansen, T. J. Brodribb, H. Cochard, S. Delzon, R. Bhaskar, S. J. Bucci, T. S. 387 Feild, S. M. Gleason, U. G. Hacke, A. L. Jacobsen, F. Lens, H. Maherali, J. Martínez- 388 Vilalta, S. Mayr, M. Mencuccini, P. J. Mitchell, A. Nardini, J. Pittermann, R. B. Pratt, 389 J. S. Sperry, M. Westoby, I. J. Wright, and A. E. Zanne. 2012. Global convergence in 390 the vulnerability of forests to drought. Nature 491:752–755. 391 Clark, C. J., J. R. Poulsen, D. J. Levey, and C. W. Osenberg. 2007. Are Plant Populations 392 Seed Limited? A Critique and Meta‐Analysis of Seed Addition Experiments. The 393 American Naturalist 170:128–142. 394 Clark, J. S., C. L. Nuñez, and B. Tomasek. 2019. Foodwebs based on unreliable foundations: 395 spatiotemporal masting merged with consumer movement, storage, and diet. 396 Ecological Monographs 89:e01381. 397 Espelta, J. M., H. Arias‐LeClaire, M. Fernández‐Martínez, E. Doblas‐Miranda, A. Muñoz, 398 and R. Bonal. 2017. Beyond predator satiation: Masting but also the effects of rainfall 399 stochasticity on weevils drive acorn predation. Ecosphere 8:e01836. 400 Espelta, J. M., P. Cortés, R. Molowny-Horas, B. Sánchez-Humanes, and J. Retana. 2008. 401 Masting mediated by summer drought reduces acorn predation in Mediterranean oak 402 forests. Ecology 89:805–817. 403 Fernández-Martínez, M., M. Garbulsky, J. Peñuelas, G. Peguero, and J. M. Espelta. 2015. 404 Temporal trends in the enhanced vegetation index and spring weather predict seed 405 production in Mediterranean oaks. Plant Ecology 216:1061–1072. 406 Fernández‐Martínez, M., Vicca, S., Janssens, I. A., Carnicer, J., Martín‐Vide, J., and 407 Peñuelas, J. (2018). The consecutive disparity index, D: a measure of temporal 408 variability in ecological studies. Ecosphere, 9(12), e02527. 409 Gauthier, S., P. Bernier, T. Kuuluvainen, A. Z. Shvidenko, and D. G. Schepaschenko. 2015. 410 Boreal forest health and global change. Science 349:819–822. 411 Hacket-Pain, A. J., L. Cavin, A. D. Friend, and A. S. Jump. 2016. Consistent limitation of 412 growth by high temperature and low precipitation from range core to southern edge of 413 European beech indicates widespread vulnerability to changing climate. European 414 Journal of Forest Research 135:897–909. 415 Hacket-Pain, A. J., J. G. A. Lageard, and P. A. Thomas. 2017. Drought and reproductive 416 effort interact to control growth of a temperate broadleaved tree species (Fagus 417 sylvatica). Tree Physiology 37:744–754. 13 418 Hartig, F. 2017. DHARMa: residual diagnostics for hierarchical (multi-level/mixed) 419 regression models. R package version 0.1 5. 420 Heath, J. P. 2006. Quantifying temporal variability in population abundances. Oikos 115:573– 421 581. 422 Ibáñez, I., J. S. Clark, and M. C. Dietze. 2009. Estimating colonization potential of migrant 423 tree species. Global Change Biology 15:1173–1188. 424 Kelly, D. 1994. The evolutionary ecology of mast seeding. Trends in Ecology & Evolution 425 9:465–470. 426 LaDeau, S. L., and J. S. Clark. 2001. Rising CO2 Levels and the Fecundity of Forest Trees. 427 Science 292:95–98. 428 Leuzinger, S., Y. Luo, C. Beier, W. Dieleman, S. Vicca, and C. Körner. 2011. Do global 429 change experiments overestimate impacts on terrestrial ecosystems? Trends in 430 Ecology & Evolution 26:236–241. 431 Limousin, J. M., Rambal, S., Ourcival, J. M., & Joffre, R. (2008). Modelling rainfall 432 interception in a Mediterranean Quercus ilex ecosystem: lesson from a throughfall 433 exclusion experiment. Journal of Hydrology, 357(1-2), 57-66. 434 Limousin, J. M., S. Rambal, J. M. Ourcival, A. Rocheteau, R. Joffre, and R. Rodriguez‐ 435 Cortina. 2009. Long-term transpiration change with rainfall decline in a Mediterranean 436 Quercus ilex forest. Global Change Biology 15:2163–2175. 437 Liu, D., R. Ogaya, A. Barbeta, X. Yang, and J. Peñuelas. 2015. Contrasting impacts of 438 continuous moderate drought and episodic severe droughts on the aboveground- 439 biomass increment and litterfall of three coexisting Mediterranean woody species. 440 Global Change Biology 21:4196–4209. 441 Lloret, F., and T. Kitzberger. 2018. Historical and event-based bioclimatic suitability predicts 442 regional forest vulnerability to compound effects of severe drought and bark beetle 443 infestation. Global Change Biology 24:1952–1964. 444 Lloret, F., J. Peñuelas, and R. Ogaya. 2004. Establishment of co-existing Mediterranean tree 445 species under a varying soil moisture regime. Journal of Vegetation Science 15:237– 446 244. 447 Luo, Y., E. J. B. McIntire, C. Boisvenue, P. P. Nikiema, and H. Y. H. Chen. (n.d.). Climatic 448 change only stimulated growth for trees under weak competition in central boreal 449 forests. Journal of Ecology n/a. 450 Martín, D., J. Vázquez-Piqué, F. S. Carevic, M. Fernández, and R. Alejano. 2015. Trade-off 451 between stem growth and acorn production in holm oak. Trees 29:825–834. 14 452 Mckone, M. J., D. Kelly, and W. G. Lee. 1998. Effect of climate change on mast-seeding 453 species: frequency of mass flowering and escape from specialist insect seed predators. 454 Global Change Biology 4:591–596. 455 McShea, W. J. 2000. The Influence of Acorn Crops on Annual Variation in Rodent and Bird 456 Populations. Ecology 81:228–238. 457 Monks, A., J. M. Monks, and A. J. Tanentzap. 2016. Resource limitation underlying multiple 458 masting models makes mast seeding sensitive to future climate change. New 459 Phytologist 210:419–430. 460 Mutke, S., J. Gordo, and L. Gil. 2005. Variability of Mediterranean Stone pine cone 461 production: Yield loss as response to climate change. Agricultural and Forest 462 Meteorology 132:263–272. 463 Norton, D. A., & Kelly, D. (1988). Mast seeding over 33 years by Dacrydium cupressinum 464 Lamb.(rimu)(Podocarpaceae) in New Zealand: the importance of economies of scale. 465 Functional ecology, 399-408. 466 Novick, K. A., Ficklin, D. L., Stoy, P. C., Williams, C. A., Bohrer, G., Oishi, A. C., Scott, R. 467 L. (2016). The increasing importance of atmospheric demand for ecosystem water and 468 carbon fluxes. Nature Climate Change, 6:1023-1027. 469 Obeso, J. R. 2002. The costs of reproduction in plants. New Phytologist 155:321–348. 470 Ogaya, R. and Peñuelas, J. 2003 Comparative field study of Quercus ilex and Phillyrea 471 latifolia: photosynthetic response to experimental drought conditions. Environmental 472 and Experimental Botany 50: 137–148. 473 Ogaya, R., and Peñuelas, J. (2004). Phenological patterns of Quercus ilex, Phillyrea latifolia, 474 and Arbutus unedo growing under a field experimental drought. Ecoscience, 11, 263- 475 270. 476 Ogaya, R., and J. Peñuelas. 2007a. Species-specific drought effects on flower and fruit 477 production in a Mediterranean holm oak forest. Forestry 80:351–357. 478 Ogaya, R., and J. Peñuelas. 2007b. Tree growth, mortality, and above-ground biomass 479 accumulation in a holm oak forest under a five-year experimental field drought. Plant 480 Ecology 189:291–299. 481 Ostfeld, R. S., and F. Keesing. 2000. Pulsed resources and community dynamics of 482 consumers in terrestrial ecosystems. Trends in Ecology & Evolution 15:232–237. 483 Pearse, I. S., W. D. Koenig, and D. Kelly. 2016. Mechanisms of mast seeding: resources, 484 weather, cues, and selection. New Phytologist 212:546–562. 15 485 Pearse, I. S., W. D. Koenig, and J. M. Knops. 2014. Cues versus proximate drivers: testing the 486 mechanism behind masting behavior. Oikos 123:179–184. 487 Peñuelas, J., J. Sardans, I. Filella, M. Estiarte, J. Llusià, R. Ogaya, J. Carnicer, M. Bartrons, 488 A. Rivas-Ubach, O. Grau, G. Peguero, O. Margalef, S. Pla-Rabés, C. Stefanescu, D. 489 Asensio, C. Preece, L. Liu, A. Verger, L. Rico, A. Barbeta, A. Achotegui-Castells, A. 490 Gargallo-Garriga, D. Sperlich, G. Farré-Armengol, M. Fernández-Martínez, D. Liu, C. 491 Zhang, I. Urbina, M. Camino, M. Vives, D. Nadal-Sala, S. Sabaté, C. Gracia, and J. 492 Terradas. 2018. Assessment of the impacts of climate change on Mediterranean 493 terrestrial ecosystems based on data from field experiments and long-term monitored 494 field gradients in Catalonia. Environmental and Experimental Botany 152:49–59. 495 Pérez-Ramos, I. M., J. M. Ourcival, J. M. Limousin, and S. Rambal. 2010. Mast seeding 496 under increasing drought: results from a long-term data set and from a rainfall 497 exclusion experiment. Ecology 91:3057–3068. 498 Pesendorfer, M., Bogdziewicz, M.,, Szymkowiak, J., Borowski, Z., Kantorowicz, W., Espelta, 499 J.M., Fernandez-Martinez, M., (2020). Investigating the relationship between climate, 500 stand age, and temporal trends in masting behavior of European forest trees. Global 501 Change Biology, DOI:10.1111/gcb.14945 502 Redmond, M. D., F. Forcella, and N. N. Barger. 2012. Declines in pinyon pine cone 503 production associated with regional warming. Ecosphere 3:art120. 504 Rees, M., D. Kelly, and O. N. Bjørnstad. 2002. Snow tussocks, chaos, and the evolution of 505 mast seeding. The American Naturalist 160:44–59. 506 Richardson, S. J., R. B. Allen, D. Whitehead, F. E. Carswell, W. A. Ruscoe, and K. H. Platt. 507 2005. Climate and Net Carbon Availability Determine Temporal Patterns of Seed 508 Production by Nothofagus. Ecology 86:972–981. 509 Ripullone, F., M. Borghetti, S. Raddi, E. Vicinelli, R. Baraldi, M. R. Guerrieri, A. Nolè, and 510 F. Magnani. 2009. Physiological and structural changes in response to altered 511 precipitation regimes in a Mediterranean macchia ecosystem. Trees 23:823. 512 Rosas, T., L. Galiano, R. Ogaya, J. Peñuelas, and J. Martínez-Vilalta. 2013. Dynamics of non- 513 structural carbohydrates in three Mediterranean woody species following long-term 514 experimental drought. Frontiers in Plant Science 4. 515 Seidl, R., D. Thom, M. Kautz, D. Martin-Benito, M. Peltoniemi, G. Vacchiano, J. Wild, D. 516 Ascoli, M. Petr, J. Honkaniemi, M. J. Lexer, V. Trotsiuk, P. Mairota, M. Svoboda, M. 517 Fabrika, T. A. Nagel, and C. P. O. Reyer. 2017. Forest disturbances under climate 518 change. Nature Climate Change 7:395–402. 16 519 Vacchiano, G., A. Hacket-Pain, M. Turco, R. Motta, J. Maringer, M. Conedera, I. Drobyshev, 520 and D. Ascoli. 2017. Spatial patterns and broad-scale weather cues of beech mast 521 seeding in Europe. New Phytologist 215:595–608. 522 Vergotti, M. J., Fernández-Martínez, M., Kefauver, S. C., Janssens, I. A., Peñuelas, J. (2019). 523 Weather and trade-offs between growth and reproduction regulate fruit production in 524 European forests. Agricultural and Forest Meteorology, 279, 107711. 525 Wu, Z., P. Dijkstra, G. W. Koch, J. Peñuelas, and B. A. Hungate. 2011. Responses of 526 terrestrial ecosystems to temperature and precipitation change: a meta-analysis of 527 experimental manipulation. Global Change Biology 17:927–942. 528 Zhu, K., C. W. Woodall, and J. S. Clark. 2012. Failure to migrate: lack of tree range 529 expansion in response to climate change. Global Change Biology 18:1042–1052. 530 Zohner, C. M., L. Mo, and S. S. Renner. 2018. Global warming reduces leaf-out and 531 flowering synchrony among individuals. eLife 7:e40214. 532 17 533 Figure 1. Abiotic conditions at the experimental site in the Prades Mountains in southern 534 Catalonia. a) SPEI-3, b) temperature, c) precipitation, and d) soil moisture. Data for soil 535 moisture was not collected in 2015 due to equipment malfunction. 536 18 537 Figure 2. Fruit production by a) Quercus ilex, b) Phillyrea latifolia, and c) Arbutus unedo in 538 the control and drought plots. The solid lines and shaded areas are annual means and 539 associated standard deviations, respectively. The number of observations of reproductive 540 events is 144 (plot-years) for Q. ilex and P. latifolia and 126 for A. unedo, which was absent 541 in one of the plots. Q. ilex and P. latifolia were observed at 8 plots, while A. unedo at 7. 542 543 544 545 546 547 548 549 19 550 Figure 3. Synchrony and interannual variability (PV) of fruit production by Quercus ilex, 551 Phillyrea latifolia, and Arbutus unedo in the control and drought plots. Synchrony was 552 measured by mean pairwise Pearson correlation among plots. CV not shown. 553 554 20 555 Figure 4. Scatterplots of standardized basal area increment (BAI) and fruit production for a) 556 Quercus ilex, b) Phillyrea latifolia, and c) Arbutus unedo in the control and drought plots. 557 The lines and shaded areas are the linear mixed model predictions and associated 95% 558 confidence intervals, respectively. Points and whiskers are plot-year means and associated 559 standard errors, respectively. Black represents the control, and yellow represents experimental 560 rainfall exclusion. The sample sizes for these models are 11 288 for Q. ilex (735 stems 561 measured), 15 301 for P. latifolia (842 stems), and 2242 for A. unedo (145 stems). 562 563 21