ORIGINAL RESEARCH published: 08 January 2020 doi: 10.3389/ffgc.2019.00089 Stem Mortality and Forest Dieback in a 20-Years Experimental Drought in a Mediterranean Holm Oak Forest Romà Ogaya 1,2*, Daijun Liu 1,2,3 , Adrià Barbeta 4 and Josep Peñuelas 1,2 1 Ecological and Forestry Applications Research Center (CREAF), Barcelona, Spain, 2 Global Ecology Unit, Ecological and Forestry Applications Research Center, Center for Advanced Studies of Blanes (CSIC), Universitat Autònoma de Barcelona, Barcelona, Spain, 3 School of Geography, Earth and Environmental Sciences, University of Birmingham Edgbaston, Birmingham, United Kingdom, 4 Department of Evolutionary Biology (BEECA), Ecology and Environmental Sciences, University of Barcelona, Barcelona, Spain Climatic warming is predicted to increase the intensity and duration of extreme weather such as droughts and heat waves. Climate change could therefore increase stem mortality and forest dieback in many ecosystems around the world, especially in semi-arid forests. We investigated the influence of climatic conditions and a throughfall displacement experimental treatment (15% decrease in the amount of soil moisture) on the intensity of stem mortality in a Mediterranean forest. We also investigated the use of remotely sensed data from MODIS as a tool to estimate the intensity of stem mortality and a possible strategy of forest management to mitigate this forest dieback Edited by: induced by climate change. Stem mortality was higher when mean annual temperature Charles Rhett Jackson, was higher and rainfall was lower, especially during spring and summer. Mortality was University of Georgia, Georgia higher and more affected by the drought treatment in the holm oak, Quercus ilex, than Reviewed by: Taehee Hwang, in co-occurring species of tall shrubs better adapted to drier conditions. Two spectral Indiana University Bloomington, MODIS indices (NDVI and EVI) were good predictors of stem mortality, but others (GPP, United States Randy Kolka, PsnNet, and NPP specifically calculated to predict forest health and development) were USDA Forest Service, United States not clearly correlated with stem mortality. Selective stem thinning strongly reduced stem *Correspondence: mortality (especially in Q. ilex) by buffering the effects of climate change on forest Romà Ogaya structure. Finally, the possible future substitution of the current dominant species of this r.ogaya@creaf.uab.cat forest, Q. ilex, by species of tall shrubs better adapted to drought is also discussed. Specialty section: Keywords: climate change, forest dieback, Mediterranean forest, MODIS, Quercus ilex, remote sensing, stem This article was submitted to mortality Forest Hydrology, a section of the journal Frontiers in Forests and Global Change INTRODUCTION Received: 09 October 2019 Higher air temperatures are projected globally for the coming decades, and climatic patterns Accepted: 09 December 2019 around the world are changing: the intensity and duration of extreme climate conditions, such Published: 08 January 2020 heat waves, floods, and droughts, are predicted to increase (Wang et al., 2012; IPCC, 2018). Higher Citation: rates of evapotranspiration in some semi-arid and Mediterranean areas would be induced by Ogaya R, Liu D, Barbeta A and these higher temperatures, with no any increase of precipitation (IPCC, 2018). General circulation Peñuelas J (2020) Stem Mortality and Forest Dieback in a 20-Years models thus project an average decrease in soil moisture of 15% in the next 50 years and a Experimental Drought in a return period of extreme droughts 10-fold shorter than in the twentieth century in Mediterranean Mediterranean Holm Oak Forest. regions (Bates et al., 2008). Forest ecosystems in these Mediterranean areas, seasonally exposed Front. For. Glob. Change 2:89. to water stress, may be particularly vulnerable to even slight increases in water deficits, which doi: 10.3389/ffgc.2019.00089 can induce reductions in tree growth (Barbeta et al., 2013; Liu et al., 2015, 2018) and crown Frontiers in Forests and Global Change | www.frontiersin.org 1 January 2020 | Volume 2 | Article 89 Ogaya et al. Stem Mortality Induced by Drought conditions (Carnicer et al., 2011; Galiano et al., 2012) and mean temperature and total rainfall from 1975 to 2018 averaged increases in tree mortality (Breshears et al., 2005; Allen et al., 11.9◦ C and 654 mm, respectively, with hot and dry summers and 2010; Peng et al., 2011; Williams et al., 2012). Precipitation most of the rainfall concentrated in spring and autumn. The anomalies have also been strongly correlated with the subsequent vegetation is a very dense multi-stem forest (16,616 stems ha−1 ), occurrence of hot extremes in most areas of the world (Mueller due to an ancient management to obtain charcoal, dominated and Seneviratne, 2012), and an increased frequency of heat by Q. ilex, Phillyrea latifolia L., and Arbutus unedo L., with waves is expected in Mediterranean regions, coinciding with an abundance of other evergreen species well-adapted to dry summer drought and higher evapotranspiration and thus a lower conditions such as Erica arborea L., Juniperus oxycedrus L., and availability of soil water (Fischer and Schar, 2010). Cistus albidus L. This forest was not perturbed during the last Several species of tall shrubs associated with holm oak 70 years, and the maximum heights of the dominant species are (Quercus ilex L.) forests have lower growth rates but are more about 6–10 m. resistant to drought conditions than the holm oak, the current dominant species in these forests (Peñuelas et al., 1998; Ogaya Meteorological Data and Peñuelas, 2003). Higher mortality rates and lower seed An automatic meteorological station installed at the study site has production and seedling survival of Q. ilex under future drier monitored temperature, photosynthetically active radiation, air conditions could drive a progressive substitution of dominant humidity, and precipitation since late 1998. We had previously oak species by the co-occurring species more resistant to low estimated temperature and rainfall from another meteorological water availability (Lloret et al., 2004; Ogaya and Peñuelas, 2007; station located in Poblet Monastery, 5.6 km northeast of our Liu et al., 2015, 2018). The selective thinning of stems is an study area at 510 m a.s.l. The Poblet station collected data from ancient practice of forest management widely used to enhance the late 1974 to July 2002, so we had climatic data from both growth and development of the remaining stems, increasing the meteorological stations from August 1998 to July 2002. We availability of resources per stem when stem density decreases. calculated linear correlations between temperature (R2 = 0.97) Selective thinning under a scenario of increasing drought could and rainfall (R2 = 0.75) for the data from these two stations therefore partly offset this water deficit by reducing competition during this period to estimate climatic data for the study site for water resources and thus contribute to the stabilization of since 1975. forest structure and function. Spectral indices from remotely sensed data are widely used to Throughfall Displacement and Stem evaluate the structure and functioning of terrestrial ecosystems Mortality (Peñuelas and Filella, 1998). The Normalized Difference Eleven 15 × 10 m plots were delimited at the study site at the Vegetation Index (NDVI) is the most widely used index, mainly same altitude along the slope; four were control plots, other four as an estimator of the fraction of the photosynthetically active were throughfall displacement exclusion (TDE) plots, where the radiation absorbed by the canopy (Tucker et al., 1985). The input of rainwater decreased the amount of soil moisture by 15% Enhanced Vegetation Index (EVI) was developed to reduce the using permanent plastic strips (Figure 1), and the other three noise produced by soil background and atmospheric aerosols and to reduce the saturation of the reflectance signal at increasing levels of green biomass, which is commonly associated with NDVI (Huete et al., 2002). The MODIS system, though, also provides several indices specifically calculated to estimate forest productivity [Gross Primary Production (GPP), Net Primary Production (NPP), and Net Photosynthesis (PsnNet)]. The main objectives of this study were: (1) to identify the relationships between stem mortality and climatic conditions, (2) to determine the effect of an experimental reduction of soil- water availability on stem mortality in the dominant species of the Mediterranean forest, (3) to test the utility of selectively thinning stems to mitigate the effects of the decrease in water availability, and (4) to evaluate the use of remotely sensed data from satellites as a tool for detecting and monitoring forest dieback and stem mortality. MATERIALS AND METHODS Site Description The study was conducted in the Prades Mountains, Catalonia, northeastern Spain (41◦ 21′ N, 1◦ 2′ E), on a 25◦ south-facing slope FIGURE 1 | Soil moisture in Control and TDE plots in the four annual seasons. at an altitude of 950 m a.s.l. The soil is a Dystric Cambisol over Error bars correspond to the standard error of the mean (n = 4). * P < 0.05. Paleozoic schist, ranging in depth from 35 to 100 cm. Annual Frontiers in Forests and Global Change | www.frontiersin.org 2 January 2020 | Volume 2 | Article 89 Ogaya et al. Stem Mortality Induced by Drought TABLE 1 | Basal area (BA) of the dominant species in each thinned plot before period, and t is the number of years of the study period. The and after thinning. control and drought treatments began in 1999 and ended at the Plot Species BA before thinning (m2 ha−1 ) BA after thinning (m2 ha−1 ) end of 2018, whereas M in the thinned plots was determined from 2012 (immediately after thinning) to 2015. 1 Q. ilex 43.57 38.41 P. latifolia 2.64 1.10 Remote Sensing A. unedo 10.06 5.80 The remotely sensed data for 2000–2012 were obtained from All species 56.27 45.31 MODIS sensors onboard the Terra and Aqua satellites. The Terra 2 Q. ilex 29.31 24.59 satellite orbits Earth from north to south in the morning, and P. latifolia 2.57 1.77 the Aqua satellite orbits from south to north in the afternoon. A. unedo 9.75 6.43 We used NDVI and EVI data obtained from the 16-days Terra All species 41.83 33.00 MOD13Q1 and 16-days Aqua MYD13Q1 products, combined 3 Q. ilex 42.14 33.86 in eight-day series based on a selected 250 × 250 m pixel P. latifolia 4.94 1.45 corresponding to our study site, and including the studied plots A. unedo 5.15 3.59 and the surrounding area. All species 53.71 39.53 We also used GPP and PsnNet data obtained from the 8- days Terra MOD17A2 product and NPP data obtained from the annual Terra MOD17A3 product. All these products provided an 1 × 1 km pixel for 2000–2010 centered in the studied area, and including the studied plots and the surrounding area. Mean annual values of the remote-sensing indices were calculated as an annual integration for each growing season covered by NDVI, EVI, GPP, and PsnNet, whereas NPP was calculated directly from annual data. Statistical Analyses We used the meteorological data for 1975–2018 to estimate the trends of changes in air temperature and rainfall throughout this period. The difference of soil moisture in control and TDE plots was tested using an ANOVA, where soil moisture was the dependent variable, and treatment the factor. General FIGURE 2 | Throughfall Displacement Exclusion system installed to decrease linear models were also conducted to determine the relationships soil moisture and water availability for the vegetation. between the rate of stem mortality and the changes in air temperature and rainfall during the study period. An analysis of variance (ANOVA) was performed, with mortality rate as a dependent variable, and species and treatment (control and plots were selectively thinned, where ca. 25% of the basal area of TDE plots) as independent factors. A similar ANOVA was then each species was removed (Table 1). In drought plots, six plastic performed to compare stem mortalities in the control and TDE strips were installed throughout plot slope, about 1 m above soil plots. We next calculated the difference in annual M for each surface and down of tree canopies. These plastic strips covered year and species and selected 5 extremely dry years and 5 wet also 2 m above plot margin, and 2 m down of plot margin, years. Another ANOVA was performed, with the difference in to avoid edge effect. Each plastic strip consisted of four rigid mortality rate as a dependent variable and species and treatment PVC tubes joined together by transparent PVC between them as independent factors. Other general linear models were used (Figure 2), these strips covered about 30% of plot surface and to determine the relationships of the rate of stem-mortality partially intercepted throughfall and removed it downslope the with mean annual NDVI, EVI, GPP, PsnNet, and NPP. We plots. Soil moisture was measured each annual season using a used linear regressions because they identified the relationships time domain reflectometer (Delta-T Devices, Cambridge, UK) in that fitted better with all variables studied. M was arcsin m0.5 four randomly selected sites per plot. The reflectometer measured transformed to satisfy the assumptions of a normal distribution. soil moisture in the first 15 cm depth. All linear models were performed using StatView (SAS Institute All living stems >2 cm in diameter at a height of 50 cm in Inc., Cary, USA). the plots were labeled and classified by species at the start of the experiment. All stems that had died during the year were RESULTS counted in each of the following winters to obtain the rates of mortality per year, plot, and species. The stem-mortality rate (M) Meteorological Data was calculated as (Sheil et al., 1995): M = 1–(1 – (No – Nt )/No )1/t , Mean annual temperature ranged from 10.7◦ C in 1984 to 13.1◦ C where No is the number of stems at the start of the study period in 2011, and total annual precipitation ranged from 355 mm (1999–2018), Nt is the number of stems at the end of the study in 2015 to 1021 mm in 2018 (Figure 3). Mean annual air Frontiers in Forests and Global Change | www.frontiersin.org 3 January 2020 | Volume 2 | Article 89 Ogaya et al. Stem Mortality Induced by Drought temperature during 1975–2018 increased by about 1.7◦ C (R2 Stem Mortality = 0.50). This increase was due to the strong increase of 2.7◦ C Annual M for all species together was positively correlated in spring and summer (R2 = 0.60), but autumn and winter with mean annual temperature (R2 = 0.17) and negatively with temperatures did not change. Annual precipitation, however, was spring and summer precipitation (R2 = 0.18) (Figure 4). Stem highly variable, with only a non-significant trend toward a slight mortality was higher in Q. ilex than P. latifolia and A. unedo decrease from 1975 to 2018, but precipitation during spring and throughout the study period, and M was significantly higher in summer decreased ca. 200 mm (R2 = 0.26). the TDE than in the control treatment for all species together FIGURE 3 | Changes in mean air temperatures and rainfall during the period of collection of meteorological data. Mean annual temperature and total annual rainfall [(A,B), respectively] and spring and summer temperature and rainfall [(C,D), respectively]. N.S. indicates no significant relationship. FIGURE 4 | Linear relationships between stem-mortality rates and mean annual temperatures (A) and rainfall during spring and summer (B). Frontiers in Forests and Global Change | www.frontiersin.org 4 January 2020 | Volume 2 | Article 89 Ogaya et al. Stem Mortality Induced by Drought FIGURE 5 | Average stem-mortality rates in the control and TDE plots for each of the three species studied. Error bars correspond to the standard error of the mean (n = 4). Different letters indicate significant differences (P < 0.05). (Figure 5). A comparison between M and the meteorological data indicated larger increases in stem mortality induced by the TDE treatment in hotter and drier years (Figure 6). The difference in M between the control and TDE plots was larger for Q. ilex than P. latifolia and A. unedo, and this difference for Q. ilex was larger in extremely dry than wet years (Figures 7, 8), even when M for this species was lower in the TDE than in the control plots in wet years. The difference in M for P. latifolia FIGURE 6 | Annual stem-mortality rates in the control and TDE plots for all and A. unedo between the control and TDE plots, however, was species studied (A) and annual meteorological data (mean air temperatures not significant in both extremely dry and wet years. Finally, and total rainfall) (B). Black dots represent temperature values, and gray bars represent rainfall values. Error bars correspond to the standard error of the selective thinning strongly decreased stem mortality only in mean (n = 4). *P < 0.05. Q. ilex, but not in P. latifolia or A. unedo (Figure 9). Stem mortality in the control plots was higher for Q. ilex than P. latifolia and A. unedo, and M in the thinned plots was lowest for Q. ilex. annual precipitation, but PsnNet and NPP were not significantly correlated with M or any climatic variable (Table 2). Remote Sensing DISCUSSION Mean annual NDVI (0.80) was highest in 2003, and mean annual EVI (0.38) was highest in 2008, when annual precipitation for Stem Drought-Induced Mortality 2000–2012 was highest (926 mm in 2003 and 837 mm in 2008). Drought-induced stem mortality is likely to increase in the Mean annual NDVI and EVI (0.74 and 0.33, respectively) were future. Climatic models forecast an increased frequency of lowest in late autumn 2011 and winter 2011–2012, coinciding drought together with higher air temperatures (Dai, 2013; Allen with the end of an extremely dry period. Mean annual NDVI et al., 2015). The progressive increase in air temperatures and EVI were positively correlated with annual precipitation, and accompanied by a slight decrease in rainfall, especially in spring negatively correlated with mean annual temperature and with (when plants flower and develop new leaves) and summer M (Table 2). (coinciding with the most critical period for plant development GPP and PnsNet had another annual pattern than NDVI and and survival), is exacerbating dry conditions and driving stem EVI, with low values in winter, peaking in spring, decreasing in mortality and a decline in tree growth in this holm oak forest summer, increasing again in autumn, and a final decrease in late (Ogaya and Peñuelas, 2007; Barbeta et al., 2013; Ogaya et al., autumn and the following winter. GPP was correlated only with 2016; Liu et al., 2018). The demand for atmospheric moisture Frontiers in Forests and Global Change | www.frontiersin.org 5 January 2020 | Volume 2 | Article 89 Ogaya et al. Stem Mortality Induced by Drought FIGURE 7 | Annual stem-mortality rates in the control and TDE plots for Q. ilex (A), P. latifolia (B), and A. unedo (C). Error bars correspond to the standard error of the mean (n = 4). *P < 0.05. FIGURE 8 | Average differences in stem-mortality rates between the control and TDE plots in dry and wet years for each of the three species studied. Error FIGURE 9 | Average stem-mortality rates in the control and thinned plots for bars correspond to the standard error of the mean (n = 4). *P < 0.05. each of the three species studied. Error bars correspond to the standard error of the mean (n = 4). *P < 0.05. has been described as the primary mechanism by which high temperature influences drought-sensitive tree populations (Liu contribute to the forest dieback as it was observed in the studied et al., 2013). This mechanism is illustrated by our finding site (Ogaya et al., 2016). As expected, mortality rates were higher that mean annual temperature was positively correlated with for Q. ilex than the co-occurring species of tall shrubs (Figure 3) stem-mortality rate, whereas spring and summer precipitation that are better adapted to drought (Peñuelas et al., 1998; Ogaya were negatively correlated with M. The deterioration of crown and Peñuelas, 2003). Stem mortality for Q. ilex also strongly condition has been associated with a reduction in levels of non- increased under low water availability in the TDE treatment or structural carbohydrates (Rosas et al., 2013), that could lead extremely dry years, which did not occur in the other species to a carbon starvation with a combination of hydraulic failure (Figures 4, 7). Most tree species globally are routinely subject to (Sevanto et al., 2014; McDowell and Allen, 2015), which may levels of physiological water stress near their limits of tolerance Frontiers in Forests and Global Change | www.frontiersin.org 6 January 2020 | Volume 2 | Article 89 Ogaya et al. Stem Mortality Induced by Drought TABLE 2 | Coefficients of determination (R2 ) of the linear relationships between (Garbulsky et al., 2013). Both indices, however, have also strongly stem-mortality rates and spectral indices. N.S. indicates no significant relationship. decreased under extremely dry conditions (Ogaya et al., 2016). Spectral index Stem-mortality rate Mean air temp Annual rainfall These indices are widely used to estimate vegetation activity, but they can also be used to estimate forest decay or stem mortality. NDVI 0.43 0.24 0.40 In contrast, annual GPP, PsnNet, and NPP in our study were EVI 0.35 0.17 0.56 poorly correlated with increased biomass or stem-mortality rates, GPP N.S. N.S. 0.29 and seasonal GPP, and PsnNet did not significantly decrease PsnNet N.S. N.S. N.S. when many trees died in 2011 (Ogaya et al., 2016). This result NPP N.S. N.S. N.S. was unexpected, because these indices are specifically calculated to directly estimate ecosystem productivity, but GPP, PsnNet, and NPP were not good predictors of either tree growth, or stem mortality in our forest. (Choat et al., 2012), and experimental evidence suggests high tree mortality induced by drought with relatively modest increases CONCLUSIONS in temperature (Adams et al., 2009; Duan et al., 2014). Higher stem mortality (Ogaya and Peñuelas, 2007; Barbeta et al., 2013; Hotter and drier conditions are driving an increase in Liu et al., 2015) and lower seedling survival (Lloret et al., 2004) forest dieback and stem mortality in many water-limited in Q. ilex are leading to a progressive substitution of the current ecosystems, such as Mediterranean forests, with a progressive dominant species in this forest by species of tall shrubs better substitution of the current dominant species by species of adapted to low water availability (Liu et al., 2018), which could tall shrubs better adapted to future climate. These effects be followed by rapid alterations in ecosystem composition and were followed by changes in forest structure and ecosystem structure and to the feedbacks between the biosphere and climate, services. Selective stem thinning was also identified as an as in many other forests around the world (Wang et al., 2012). excellent tool for managing this forest in a changing climate; Some ecosystem services such as the mitigation of climate change selective thinning not only buffered forest structure and function by CO2 uptake will also be seriously affected. against increasing drought, but also reduced the risk of forest fire. Selective Thinning Several studies have reported an increase in stem growth induced by selective thinning, increasing the availability of resources for DATA AVAILABILITY STATEMENT the remaining stems (Roberts and Harrington, 2008; Keyser, The datasets generated for this study are available on request to 2010; Ogaya et al., 2019). Higher rates of net photosynthesis and the corresponding author. stem growth in semi-arid habitats in response to thinning may be mediated by higher water availability and stomatal conductance (Moreno-Gutiérrez et al., 2011), whereas these responses were AUTHOR CONTRIBUTIONS observed in temperate forests only in unusually dry years (Herbst RO collected the data, performed the statistical analyses, et al., 2015). On the other hand, a decrease of tree mortality and drafted the manuscript. DL and AB collected the data induced by a selective thinning is also described in several and contributed to the statistical analyses. JP designed the forests around the world, especially for those that are water- experiment and contributed to the statistical analyses. All authors limited (Hood et al., 2016; Neumann et al., 2017; Restaino contributed to the interpretation of the results and the writing of et al., 2019). An increase in the rate of net photosynthesis the manuscript. can also contribute to mitigating climate change due to an increase in the uptake of atmospheric CO2 by forests. In addition to the benefits of selective thinning for forest conservation FUNDING and functioning (Wang et al., 2012), selective thinning has indirect benefits such as decreasing the risk of forest fires and This research was financially supported by the Spanish increasing human water supply, because a high stem density government grant CGL2016-79835-P, the European can favor the spread of forest fire due to the abundance of Research Council Synergy grant ERC-2013-SyG-2013- dead stems and branches (Ogaya and Peñuelas, 2007), and 610028 IMBALANCE-P, and the Catalan government highly dense vegetation consumes a lot of rainwater (Birot et al., grant SGR-2017-1005. 2011). ACKNOWLEDGMENTS Remote Sensing The NDVI and EVI indices were good predictors of forest We are grateful to DARP (Generalitat de Catalunya), Ester dieback and stem mortality. Good relationships between these Trullols, and Xavier Buqueras for their permission and help to indices and increased biomass have been previously observed conduct this research in the Poblet Holm Oak Forest. Frontiers in Forests and Global Change | www.frontiersin.org 7 January 2020 | Volume 2 | Article 89 Ogaya et al. Stem Mortality Induced by Drought REFERENCES Keyser, T. L. (2010). Thinning and site quality influence aboveground tree carbon stocks in yellow-poplar forests of the southern Appalachians. Can. J. For. Res. Adams, H. D., Guardiola-Claramonte, M., Barron-Gafford, G. A., Villegas, J. 40, 659–667. doi: 10.1139/X10-013 C., Breshears, D. D., Zou, C. B., et al. (2009). Temperature sensitivity of Liu, D., Ogaya, R., Barbeta, A., Yang, X., and Peñuelas, J. (2015). Contrasting drought-induced tree mortality portends increased regional die-off under impacts of continuous moderate drought and episodic severe droughts global-change-type drought. Proc. Natl. Acad. Sci. U. S. A. 106, 7063–7066. on the aboveground-biomass increment and litterfall of three coexisting doi: 10.1073/pnas.0901438106 Mediterranean Woody species. Glob. Cha. Biol. 21, 4196–4209. Allen, C. D., Breshears, D. D., and McDowell, N. G. (2015). On underestimation of doi: 10.1111/gcb.13029 global vulnerability to tree mortality and forest die-off from hotter drought in Liu, D., Ogaya, R., Barbeta, A., Yang, X., and Peñuelas, J. (2018). Long-term the Anthropocene. Ecosphere 6:129. doi: 10.1890/ES15-00203.1 experimental drought combined with natural extremes accelerate vegetation Allen, C. D., Macalady, A. K., Chenchouni, H., Bachelet, D., McDowell, N., shift in a Mediterranean holm oak forest. Env. Exp. Bot. 151, 1–11. Vennetier, M., et al. (2010). A global overview of drought and heat-induced tree doi: 10.1016/j.envexpbot.2018.02.008 mortality reveals emerging climate change risks for forests. For. Ecol. Manage. Liu, H., Williams, A. P., Allen, C. D., Guo, D., Wu, X., Anenkhonov, O. A., et al. 259, 660–684. doi: 10.1016/j.foreco.2009.09.001 (2013). Rapid warming accelerates tree growth decline in semi-arid forests of Barbeta, A., Ogaya, R., and Peñuelas, J. (2013). Dampening effects of long-term Inner Asia. Glob. Change Biol. 19, 2500–2510. doi: 10.1111/gcb.12217 experimental drought on growth and mortality rates of a Holm oak forest. Lloret, F., Peñuelas, J., and Ogaya, R. (2004). Establishment of co-existing Global Change Biol. 19, 3133–3144. doi: 10.1111/gcb.12269 Mediterranean tree species under a varying soil moisture regime. J. Veg. Sci. Bates, B. C., Kundzewicz, Z. W., Wu, S., and Palutikof, J. P. (2008). Climate 15, 237–244. doi: 10.1111/j.1654-1103.2004.tb02258.x Change and Water. Technical Paper of the Intergovernmental Panel on Climate McDowell, N. G., and Allen, C. D. (2015). Darcy’s law predicts widespread Change, IPCC Secretariat, Geneva, 210. forest mortality under climate warming. Nat. Clim. Change 5, 669–672. Birot, Y., Gracia, C., and Palahí, M. (2011). Water for Forests and People in the doi: 10.1038/nclimate2641 Mediterranean Region. Joensuu: European Forest Institute. Moreno-Gutiérrez, C., Barberà, G. G., Nicolás, E., de Luís, M., Castillo, V. M., Breshears, D. D., Cobb, N. S., Rich, P. M., Price, K. D., Allen, C. D., Balice, Martínez-Fernández, F., et al. (2011). Leaf δ18 O of remaining trees is affected by R. G., et al. (2005). Regional vegetation die-off in response to global- thinning intensity in a semiarid pine forest. Plant Cell Eviron. 34, 1009–1019. change-type drought. Proc. Natl. Acad. Sci. U. S. A. 102, 15144–15148. doi: 10.1111/j.1365-3040.2011.02300.x doi: 10.1073/pnas.0505734102 Mueller, B., and Seneviratne, S. (2012). Hot days induced by precipitation Carnicer, J., Coll, M., Ninyerola, M., Pons, X., Sanchez, G., and Peñuelas, J. (2011). deficits at the global scale. Proc. Natl. Acad. Sci. U. S. A. 109, 12398–12403. Widespread crown condition decline, food web disruption, and amplified tree doi: 10.1073/pnas.1204330109 mortality with increased climate change-type drought. Proc. Natl. Acad. Sci. U. Neumann, M., Mues, V., Moreno, A., Hasenauer, H., and Seidl, R. (2017). Climate S. A. 108, 1474–1478. doi: 10.1073/pnas.1010070108 variability drives recent tree mortality in Europe. Glob. Change Biol. 23: Choat, B., Jansen, S., Brodribb, T. J., Cochard, H., Delzon, S., Bhaskar, R., et al. 4788–4797. doi: 10.1111/gcb.13724 (2012). Global convergence in the vulnerability of forests to drought. Nature Ogaya, R., Barbeta, A., Basnou, C., and Peñuelas, J. (2016). Satellite data as 491, 752–755. doi: 10.1038/nature11688 indicators of tree biomass growth and forest dieback in a Mediterranean holm Dai, A. (2013). Increasing drought under global warming in observations and oak forest. Ann. For. Sci. 72, 135–144. doi: 10.1007/s13595-014-0408-y models. Nat. Clim. Change 3, 52–58. doi: 10.1038/nclimate1633 Ogaya, R., Escolà, A., Liu, D., Barbeta, A., and Peñuelas, J. (2019). Duan, H. L., Duursma, R. A., Huang, G. M., Smith, R. A., Choat, B., Effects of thinning in a water-limited holm oak forest. J. Sust. For. O’Grady, A. P., et al. (2014). Elevated CO2 does not ameliorate the negative doi: 10.1080/10549811.2019.1673179. [Epub ahead of print]. effects of elevated temperature on drought-induced mortality in Eucalyptus Ogaya, R., and Peñuelas, J. (2003). Comparative seasonal gas exchange and radiate seedlings. Plant Cell Environ. 37, 1598–1613. doi: 10.1111/pce. chlorophyll fluorescence of two dominant woody species in a Holm Oak forest. 12260 Flora 198, 132–141. doi: 10.1078/0367-2530-00085 Fischer, E. M., and Schar, C. (2010). Consistent geographical patterns of changes in Ogaya, R., and Peñuelas, J. (2007). Tree growth, mortality, and above-ground high-impact European heatwaves. Nat. Geosc. 3, 398–403. doi: 10.1038/ngeo866 biomass accumulation in a holm oak forest under a five-year experimental field Galiano, L., Martinez-Vilalta, J., Sabaté, S., and Lloret, F. (2012). Determinants of drought. Plant Ecol. 189, 291–299. doi: 10.1007/s11258-006-9184-6 drought effects on crown condition and their relationship with depletion of Peng, C., Ma, Z., and Lei, X. (2011). A drought-induced pervasive increase in carbon reserves in a Mediterranean holm oak forest. Tree Physiol. 32, 478–489. tree mortality across Canada’s boreal forests. Nat. Clim. Change. 1, 467–471. doi: 10.1093/treephys/tps025 doi: 10.1038/nclimate1293 Garbulsky, M., Peñuelas, J., Ogaya, R., and Filella, I. (2013). Leaf and stand- Peñuelas, J., and Filella, I. (1998). Visible and near-infrared reflectance techniques level carbon uptake of a Mediterranean forest estimated using the satellite- for diagnosing plant physiological status. Trends Plant Sci. 3, 151–156. derived reflectance indices EVI and PRI. Int. J. Remote Sens. 34, 1282–1296. doi: 10.1016/S1360-1385(98)01213-8 doi: 10.1080/01431161.2012.718457 Peñuelas, J., Filella, I., Llusià, J., Siscart, D., and Piñol, J. (1998). Comparative Herbst, M., Mund, M., Tamrakar, R., and Knohl, A. (2015). Differences in field study of spring and summer leaf gas Exchange and photobiology of the carbon uptake and water use between a managed and an unmanaged Mediterranean trees Quercus ilex and Phillyrea latifolia. J. Exp. Bot. 49, 229–238. beech forest in central Germany. For. Ecol. Manage. 355, 101–108. doi: 10.1093/jxb/49.319.229 doi: 10.1016/j.foreco.2015.05.034 Restaino, C., Young, D. J. N., Estes, B., Gross, S., Wuenschel, A., Meyer, M., Hood, S. M., Baker, S., and Sala, A. (2016). Fortifying the forest: thinning and et al. (2019). Forest structure and climate mediate drought-induced tree burning increase resistance to a bark beetle outbreak and promote forest mortality in forests of the Sierra Nevada, USA. Ecol. Appl. 2019:e01902. resilience. Ecol. Appl. 26: 1984–2000. doi: 10.1002/eap.1363 doi: 10.1002/eap.1902 Huete, A., Didan, K., Miura, T., Rodriguez, E. P., Gao, X., and Ferreira, Roberts, S. D., and Harrington, C. A. (2008). Individual tree growth response to L. G. (2002). Overview of the radiometric and biophysical performance variable-density thinning in coastal Pacific Northwest forest. For. Ecol. Manage. of the MODIS vegetation indices. Remote Sens. Environ. 83, 195–213. 255, 2771–2781. doi: 10.1016/j.foreco.2008.01.043 doi: 10.1016/S0034-4257(02)00096-2 Rosas, T., Galiano, L., Ogaya, R., Peñuelas, J., and Martínez-Vilalta, J. IPCC (2018). “IPCC 2018: summary of policymakers,” in Global Warming of 1.5◦ C. (2013). Dynamics of non-structural carbohydrates in three Mediterranean An IPCC Special Report on the Impacts of Global Warming of 1.5◦ C Above woody following long-term experimental drought. Front. Plant. Sci. 4:400. Pre-Industrial Levels and Related Global Greenhouse Gas Emission Pathways, doi: 10.3389/fpls.2013.00400 in the Context of Strengthening the Global Response to the Threat of Climate Sevanto, S., McDowell, N. A., Dickman, L. T., Pangle, R., and Pockman, W. T. Change, Sustainable Development, and Efforts to Eradicate Poverty, eds V. (2014). How do trees die? A test of the hydraulic failure and carbon starvation Masson-Delmotte, et al. (Geneva: World Meteorological Organization), 32. hypotheses. Plant Cell Environ. 37, 153–161. doi: 10.1111/pce.12141 Frontiers in Forests and Global Change | www.frontiersin.org 8 January 2020 | Volume 2 | Article 89 Ogaya et al. Stem Mortality Induced by Drought Sheil, D., Burslem, D. F. R. P., and Alder, D. (1995). The interpretation forest drought stress and tree mortality. Nature Clim. Change. 3, 292–297. and misinterpretation of mortality rate measurements. J. Ecol. 83, 331–333. doi: 10.1038/nclimate1693 doi: 10.2307/2261571 Tucker, C. J., Vanpraet, C. L., Sharman, M. J., and Vanittersum, Conflict of Interest: The authors declare that the research was conducted in the G. (1985). Satellite remote-sensing of total Herbaceous biomass absence of any commercial or financial relationships that could be construed as a production in the Senegalese Sahel−1980–1984. Remote potential conflict of interest. Sens. Environ. 17, 233–249. doi: 10.1016/0034-4257(85)9 0097-5 Copyright © 2020 Ogaya, Liu, Barbeta and Peñuelas. This is an open-access article Wang, W., Peng, C., Kneeshaw, D. D., Larocque, G. R., and Luo, Z. distributed under the terms of the Creative Commons Attribution License (CC BY). (2012). Drought-induced tree mortality: ecological consequences, The use, distribution or reproduction in other forums is permitted, provided the causes, and modeling. Environ. Rev. 20, 109–121. doi: 10.1139/a20 original author(s) and the copyright owner(s) are credited and that the original 12-004 publication in this journal is cited, in accordance with accepted academic practice. Williams, A. P., Allen, C. D., Macalady, A. K., Griffin, D., Woodhouse, C. No use, distribution or reproduction is permitted which does not comply with these A., Meko, D. M., et al. (2012). Temperature as a potent driver of regional terms. Frontiers in Forests and Global Change | www.frontiersin.org 9 January 2020 | Volume 2 | Article 89