The Effect of Tree Decline Over Soil Microclimate Largely Controls Soil Respiration Dynamics in a Mediterranean Woodland

1 1The effect of tree decline over soil microclimate largely controls soil respiration 2dynamics in a Mediterranean woodland 3Authors: Alexandra Rodríguez1*, Jorge Durán1,2, Jorge Curiel Yuste3,4, Fernando 4Valladares5,6,7, Ana Rey5 5 6*Corresponding author 71Centre for Functional Ecology, CFE, University of Coimbra, 3000-456 Coimbra, Portugal 82Misión Biolóxica de Galicia, Consejo Superior de Investigaciones Científicas, 36143 9Pontevedra, Spain 10 3BC3 - Basque Centre for Climate Change, Scientific Campus of the University of the Basque 11 Country, 48940 Leioa, Spain 12 4IKERBASQUE - Basque Foundation for Science, Maria Diaz de Haro 3, 6 solairua, 48013 13 Bilbao, Bizkaia, Spain 14 5Department of Biogeography and Global Change, National Museum of Natural Sciences, 15 MNCN, CSIC, 28006 Madrid, Spain 16 6LINCGlobal, Madrid, Spain 17 7Area of Biodiversity and Conservation, ESCET, Rey Juan Carlos University, 28933 Móstoles, 18 Madrid, Spain 19 20 Present postal address of the corresponding author: 21 Centre for Functional Ecology, CFE, Department of Life Sciences, University of Coimbra, 22 Calçada Martim de Freitas, 3000-456 Coimbra, Portugal 23 Full telephone: +351 239240752; E-mail: [email protected] 24 25 Type of article: Research paper. This preprint research paper has not been peer reviewed. Electronic copy available at: https://ssrn.com/abstract=4259956 Preprint not peer reviewed 2 26 Abstract 27 As drought-induced tree defoliation and mortality (i.e. tree decline) in the Mediterranean is 28 expected to worsen with ongoing climate change, it is of paramount importance to understand 29 how, why, and when tree decline affects soil respiration (Rs) and its relationship with soil 30 microclimate and other potential controls. We carried out a novel study exploring the 31 interacting effects of climatic variability and tree decline on soil microclimate and Rs temporal 32 variability in a Mediterranean holm oak woodland. The study further explores the effects of 33 tree decline on the main controls of Rs at the stand scale. We monitored Rs, soil temperature 34 (Tsoil), and soil volumetric water content (SWC) under the canopy of 30 holm oak trees with 35 different defoliation degrees (healthy, affected, and dead) during two years of contrasting 36 precipitation patterns. We estimated different plant structural variables (e.g. DBH, height, and 37 canopy diameter) on those selected trees under whose canopies we also collected soil samples 38 to analyze different soil physicochemical variables. Our results stress the important role of tree 39 health as a modulator of the response of Rs to SWC, with stronger responses of Rs to variations 40 in the amount and distribution of precipitation under healthy than under declining trees. They 41 also suggest that tree decline can significantly increase SWC and decrease Rs but largely 42 depending on the declining stage, the year, and the season. Finally, tree decline also affected 43 the relationship of Rs with soil microclimatic variables, particularly SWC, and the relative 44 importance of the different drivers of Rs, with microclimate variables gaining importance as 45 trees defoliate and die. Altogether, our results point towards a negative impact of drought46 induced tree decline on soil C content and cycling, particularly under forecasted climate change 47 scenarios with dryer and more intense precipitation regimes. 48 49 Keywords: Quercus ilex; forest die-off; climate change; soil functioning; soil CO2 efflux; 50 environmental control. This preprint research paper has not been peer reviewed. Electronic copy available at: https://ssrn.com/abstract=4259956 Preprint not peer reviewed 3 51 1. Introduction 52 Climate models in Mediterranean ecosystems forecast increases in temperature and more 53 intensive and extensive droughts, with more frequent and intense extreme temperature and 54 precipitation events (IPCC, 2021). These changes will affect the structure, composition, and 55 functioning of forests in still unknown ways, with important implications for the carbon (C) 56 balance (Liu et al., 2016; Reichstein et al., 2013; Wang et al., 2012). Forecasted changes in the 57 frequency, intensity, and distribution of rainfall events (IPCC, 2021) will likely increase the 58 importance of rainfall pulses on the C cycle of Mediterranean environments (Rey et al., 2021, 59 2017; Song et al., 2017; Wang et al., 2016), largely influenced by the intra-annual variation in 60 soil water content (Gallardo et al., 2009). Also, the increasingly dry and hot climatic conditions 61 will exacerbate the drought-induced tree defoliation and mortality (hereinafter “forest die-off”) 62 observed over the past two decades (Carnicer et al., 2011; Lloret et al., 2004). 63 Drought-induced forest die-off impacts forest ecosystems by reducing tree 64 productivity, reducing the input of litter and root exudates, and modifying soil microclimatic 65 conditions, and in turn, soil microbial communities (Schlesinger et al., 2016). Accordingly, 66 Rodríguez et al. (2017, 2020) found evidence of a cascade effect of ongoing drought-induced 67 forest die-off in a Mediterranean holm oak (Quercus ilex) woodland ultimately decreasing the 68 content and lability of soil C. Although efforts to understand the effects of forest die-off on 69 important soil processes related to C cycling have increased lately (Avila et al., 2019, 2016; 70 Curiel Yuste et al., 2019; García-Angulo et al., 2020), results are still scarce and contradictory. 71 For instance, some studies have reported decreases in soil respiration (Rs) as a result of reduced 72 root activity (Avila et al., 2016), whereas others have found no changes (Barba et al., 2016) or 73 even increased rates (Barba et al., 2013; Curiel Yuste et al., 2019; Edburg et al., 2012) 74 following changes in microclimate (e.g., increases in soil moisture), increasing litter inputs and 75 secondary successional processes. Thus, the impacts of forest die-off on Rs are complex and This preprint research paper has not been peer reviewed. Electronic copy available at: https://ssrn.com/abstract=4259956 Preprint not peer reviewed 4 76 largely depend on the balance of its effects on roots (autotrophic) and microbial (heterotrophic) 77 components (Avila et al., 2016). Moreover, recent studies suggest that forest die-off can largely 78 modulate the response of microbial respiration to climate change and that forest die-off effects 79 on different soil attributes and processes are seasonal (Avila et al., 2019, 2016; Rodríguez et 80 al., 2019). Still, the interacting effects of tree decline and climatic variability on Rs have been 81 scarcely explored, specifically considering Mediterranean forests and drought as the main 82 driver of tree decline (Barba et al., 2016). 83 Soil respiration is the largest source of CO2 from terrestrial ecosystems and, therefore, 84 a central piece of the global C balance (Schlesinger and Andrews, 2000). However, it is also a 85 very complex process controlled by several physiological, phenological, and environmental 86 processes that vary both in time and space (Rey et al., 2021, 2011; Tang and Baldocchi, 2005). 87 At the global scale, its temporal and spatial variability is mainly controlled by air temperature 88 and precipitation, whereas at micro and stand scales, other factors related to plant community 89 and soil organic matter become more important (Barba et al., 2013; Raich and Schlesinger, 90 1992). Moreover, whereas autotrophic respiration (RA) is mainly controlled by tree physiology 91 and productivity (Högberg et al., 2001; Matteucci et al., 2015), heterotrophic respiration (RH) 92 is largely controlled by soil microclimatic (i.e. water content and temperature) and 93 biogeochemical factors (e.g. organic matter content and quality, microbial community 94 structure) (Curiel Yuste et al., 2007; Tang and Baldocchi, 2005; Zhao et al., 2016). This 95 complexity is at least partially responsible for the large uncertainties associated with 96 predictions of global rates of Rs and Rs responses to climate change (Warner et al., 2019). Thus, 97 there is an urgent need to continue to study this critical ecosystem process under different 98 scenarios and at different temporal and spatial scales. 99 We carried out a novel two-year field study to explore the interacting effects of climatic 100 variability and different stages of tree decline (i.e. healthy, affected, and dead) on soil This preprint research paper has not been peer reviewed. Electronic copy available at: https://ssrn.com/abstract=4259956 Preprint not peer reviewed 5 101 microclimate and Rs in a Quercus ilex Mediterranean woodland. Furthermore, we explored the 102 effects of tree decline alone on the main controls of Rs at the stand scale (i.e. plant variables 103 and soil microclimatic and physicochemical variables). We hypothesized that i) tree decline 104 modifies soil microclimatic conditions and, in turn, Rs rates, but ii) these effects would depend 105 on the season and year. Moreover, assuming changes in the relative contribution of RA and RH 106 associated with tree decline (Curiel Yuste et al., 2019), we hypothesized that iii) the relative 107 importance of plant, soil microclimatic, and physicochemical variables controlling Rs would 108 change considering all tree decline stages (i.e. healthy, affected and dead) together and 109 separately. 110 111 2. Material and Methods 112 2.1 Study site 113 The study was carried out in a holm oak woodland located in the central part of the Iberian 114 Peninsula, southwest of Madrid (40°23′N, 4°11′W; 630-660 m above sea level). The climate is 115 continental Mediterranean with mean annual temperature of 13 °C and mean annual 116 precipitation of 601 mm (Felicísimo et al., 2011). Most rainfall concentrates from autumn to 117 spring, while summers are warm and dry. Soil is a Cambisol, sandy and slightly acid (pH~6.3), 118 with low total C and N content (Table S1). Aboveground vegetation is characterized by a tree 119 density of ~180 trees ha−1, mostly composed of Quercus ilex L. ssp. ballota [Desf.] Samp (holm 120 oak) with scarce Juniperus oxycedrus Sibth. & Sm (cedar). The understory is dominated by 121 Retama sphaerocarpa L., Lavandula stoechas ssp. pedunculata (Mill.) Samp. ex Rozeira, and 122 diverse pasture species (Rodríguez et al., 2017). In 2005, this region suffered a strong event of 123 holm oak defoliation (around 20–30% of the total population) and mortality (15%) due to a 124 severe drought. Since then, the holm oaks of this woodland show different decline stages This preprint research paper has not been peer reviewed. Electronic copy available at: https://ssrn.com/abstract=4259956 Preprint not peer reviewed 6 125 (healthy, affected, and dead) with contrasting soil attributes underneath (Rodríguez et al., 2020, 126 2017) (Table S1). 127 128 2.2 Experimental design 129 In spring 2013, we selected 30 holm oak adult trees of similar size (38 cm and 4 m of diameter 130 at breast height and height on average, respectively, Table S1) but with different levels of 131 canopy decline (10 healthy, 10 affected, and 10 dead), and consequently with significantly 132 different canopy diameter (Table S1; Rodríguez et al. 2017), separated at least 10 m from other 133 trees. Canopy decline status of the different trees was visually determined – healthy <10% 134 defoliated; affected > 50% defoliated; dead 100% defoliated. 135 On the north face of each tree, we established a sampling point below the influence of 136 the tree canopy, 0.3 m away from the trunk (Rodríguez et al., 2017). Then polyvinylchloride 137 soil collars (15 cm in diameter and 7 cm in height) were inserted 3.5 cm into the soil at each 138 sampling point, where they remained for two consecutive annual periods: from June 2013 to 139 May 2014 (year 1) and from June 2014 to May 2015 (year 2). 140 Both annual periods were extremely warm according to the reference period (1971141 2000 and 1981-2010 until and from January 2015, respectively; AEMET), with similar mean 142 annual air temperatures (15.9 oC and 15.8 oC, respectively) but lower precipitation values in 143 year 1 (288.8 mm) than in year 2 (343.3 mm), particularly in late summer and fall (Fig. 1). 144 145 2.3 Soil microclimate and soil respiration measurements 146 We monitored Rs, along with soil temperature (Tsoil) and soil volumetric water content (SWC), 147 in all sampling points from June 2013 to May 2015 at a frequency of approximately twice a 148 month during spring and fall and once a month during summer and winter (30 campaigns in 149 total; Figure 1). To avoid strong diurnal fluctuations, measurements were done during the This preprint research paper has not been peer reviewed. Electronic copy available at: https://ssrn.com/abstract=4259956 Preprint not peer reviewed 7 150 midday period (between 10:00 and 14:00 h). Moreover, special care was taken to fully 151 randomize the sampling sequence across tree decline stages in each measurement campaign. 152 Small plants, excessive litter, insects, and grasses were carefully removed from each collar 153 before each Rs measurement. A portable infrared gas analyzer (IRGA) connected to a soil 154 respiration chamber (EGM-4 and SRC-1; PP Systems, USA) was used to measure Rs rates in 155 situ during 90 s to ensure reliable measurements. At the time of each Rs measurement, Tsoil and 156 SWC were also measured adjacent to each soil collar at 4 cm depth (the most active) inserting 157 vertically a soil digital thermometer (Maxtech, Lokhnath Enterprise, India) and a time-domain 158 reflectometer (Fieldscout TDR 300, Spectrum Technologies, Inc., Plainfield, IL, USA), 159 respectively. We estimated seasonal, annual, and two-year average values of Rs, Tsoil, and SWC 160 considering all measures carried out within each respective period. Whereas our experimental 161 design hardly provides accurate daily, seasonal or annual Rs estimations, it is perfectly valid in 162 terms of comparison among the different decline stages and correlations with the different 163 predictor variables. 164 165 2.4 Plant and soil physicochemical variables 166 At the end of May 2013, we measured tree height, diameter at breast height (DBH), and the 167 canopy diameter of the 30 studied trees by using a clinometer, a DBH tape, and averaging two 168 perpendicular measurements of canopy diameter, respectively. We also estimated a distance169 dependent competition index between trees for each one of the studied trees (Hegyi 1974), 170 considering all the competing holm oaks and cedars within a 5.5 m radius, the DBH of the 171 subject tree and that of its competitors as well as the distance between the subject tree and its 172 competitors (Table S1; Rodríguez et al. 2017). On the same date, we also measured soil 173 compaction in all sampling points by using a soil compaction meter (FieldScout SC 900, USA). This preprint research paper has not been peer reviewed. Electronic copy available at: https://ssrn.com/abstract=4259956 Preprint not peer reviewed 8 174 We carried out two different soil samplings (May 2013 and March 2015) to estimate 175 various physicochemical variables for each decline stage (healthy, affected, and dead). Soil 176 samples were collected close to the soil collars and from the first 10 cm of the soil profile by 177 using a 5 cm (i.d.) metal corer and then kept at 4 °C until analysis. Before analysis, soil samples 178 were sieved (2 mm mesh size), homogenized in field-moist conditions and analyzed for 179 gravimetric moisture by oven-drying a subsample at 60 °C to constant mass. Soil total C and 180 N content, and soil pH, were analyzed in soil samples collected in May 2013 by dry combustion 181 with an elemental analyzer (LECO TruSpec CN, St. Joseph, MI, USA) and using a soil-to182 water ratio of 1:2.5 (m/v), respectively. The possible occurrence of inorganic C was checked 183 by a Dietrich-Fruhling volumetric calcimeter. Given its absence, total C was considered equal 184 to soil organic C. Labile C was determined in the 2015 samples by using the High Gradient 185 Magnetic Separation (HGMS) method as described in Rodríguez et al. (2020). Briefly, 10 186 grams of each sample underwent HGMS by a Frantz Isodynamic Separator (Model L-1, SG 187 Frantz Co, Inc., Trenton, New Jersey, USA) that separated each soil sample in two fractions 188 with different turnover times, a non-magnetic and a magnetic fraction (Chiti et al., 2019). The 189 magnetic fraction has larger contributions from relatively recent C forms than the non-magnetic 190 fraction and thus, is supposed to be more labile. We analyzed total C in the MA fraction (CMA) 191 by dry combustion (see above) and considered it as labile C. Soil bulk density was also 192 estimated in the soil samples collected in 2015 (Table S1). 193 194 2.5 Statistical analyses 195 Linear models were performed to relate seasonal air temperature and precipitation values with 196 our estimated Tsoil and SWC seasonal values, respectively, as indicative of the suitability of our 197 seasonal measurements. This preprint research paper has not been peer reviewed. Electronic copy available at: https://ssrn.com/abstract=4259956 Preprint not peer reviewed 9 198 We used generalized least squares (GLS) models to test the effects of soil microclimatic 199 (interand intra-annual) variability and tree decline status, as well as their interactions, on 200 SWC, Tsoil, and Rs. Pairwise statistical differences among the means of the factor levels were 201 tested using simultaneous tests for general linear hypotheses (multiple comparisons of means: 202 Tukey contrasts). For all the GLS models, the tree was used as a random factor to account for 203 temporal dependencies and, when necessary, no-normality and heteroscedasticity of the 204 residuals were corrected by log-transforming the dependent variable and/or using the argument 205 weights, respectively. 206 We explored the relationship between Rs and Tsoil for the whole sampling period and 207 each decline stage separately using averaged values for each temperature degree. We identified 208 two different periods determined by a temperature threshold from which the relationship of Rs 209 with Tsoil changes (Period I – below the temperature threshold, most of the year except summer; 210 and Period II –from the threshold, summer). For Period I, soil respiration was considered 211 dependent on soil temperature according to an exponential relationship: 212 213 (1) 𝑅 𝑠 = 𝑅 𝑏𝑎𝑠𝑎𝑙 ∗ 𝑒 𝑏 ∗ 𝑇 214 215 where is the soil respiration (µmol C m-2 s-1), is soil temperature (ºC) measured at a depth 𝑅 𝑠 𝑇 216 of 4 cm, and (the basal respiration rate) and b are fitted parameters. The Q10 (the increase 𝑅 𝑏𝑎𝑠𝑎𝑙 217 in the flux rate for a 10 ºC increase in temperature) was calculated as follows: 218 (2) 𝑄 10 = 𝑒 10 ∗ 𝑏 219 220 For Period II, Rs was better explained by soil water content through a logarithmic function: 221 (3) 𝑅 𝑠 = 𝑎 + 𝑏 ∗ 𝑙𝑛 ( 𝑆𝑊𝐶 ) 222 This preprint research paper has not been peer reviewed. Electronic copy available at: https://ssrn.com/abstract=4259956 Preprint not peer reviewed 16 372 lesser extent of soil C quality and/or RA, as drivers of that response. In a previous study on the 373 same site, Rodríguez et al. (2017) found that while potential RH was up to 66% lower under 374 affected than under healthy trees, dead trees showed similar values of RH to those of healthy 375 trees (Rodríguez et al., 2017). Thus, although these previous results do not support the 376 hypothesis of higher RH under dead than under healthy trees, they do suggest a potential 377 recovery of RH after tree death. Although we did not analyze the different components of Rs 378 separately, and the RH values from Rodriguez et al. (2017) do not take into account all the 379 climatic variability of our two-year study, all above suggest that the heterotrophic component 380 could have a key role in the observed response of Rs to tree decline in this Mediterranean 381 woodland. 382 The apparent lack of effect of forest die-off on soil C and nutrient cycling during spring 383 could be largely explained by the herbaceous colonization under declining trees (Rodríguez et 384 al. 2017) and the peak activity of herbaceous roots during this season, which may have helped 385 offset the negative effect of tree decline over the tree and soil microbial activity (Avila et al., 386 2016; Rodríguez et al., 2019; Tang and Baldocchi, 2005). On the other hand, whereas we did 387 not find differences in winter Tsoil and summer SWC among decline stages, we did find them 388 in Rs in both winters and summers, suggesting a high resilience of the RA component to 389 unfavorable climatic conditions in healthy but not in declining trees. In any case, our study 390 shows that the soil effects of tree decline are detectable before tree death. Further, our results 391 suggest that, as mortality increases in this type of woodlands, and provided that conditions for 392 microbial functioning are met, they could still emit a significant amount of CO2 while likely 393 fixing much less C from the atmosphere than healthy woodlands. This situation could lead, at 394 least in the short- (i.e. years) to medium-term (i.e. decades), to a swift in the role of these forests 395 from sinks to sources of C (Baldocchi et al., 2004; Nave et al., 2011; Xiong et al., 2011). The 396 strong negative effects of forest die-off on both soil C content and lability observed in this This preprint research paper has not been peer reviewed. Electronic copy available at: https://ssrn.com/abstract=4259956 Preprint not peer reviewed 17 397 system by Rodríguez et al. (2020) support this hypothesis. Finally, these results reinforce the 398 need to consider both the spatial (i.e. different declining stages) and climatic (seasonal and 399 annual) variability, two components still scarcely explored together, when trying to fully 400 understand the complex effects of tree decline on soil attributes and functioning (Avila et al., 401 2016). 402 403 4.2 Effect of tree decline on the relationship of Rs with soil microclimatic conditions 404 As expected (e.g. Rey et al. 2011; Barba et al. 2016), Rs was better explained by Tsoil up to a 405 determined threshold (Period I), which corresponded with still optimal Tsoil (~17oC) but low 406 SWC (< 7%) conditions. Thus, once SWC reached limiting values, Rs was no longer better 407 explained by Tsoil but by SWC (Period II), highlighting the key role of water availability in the 408 functioning of Mediterranean ecosystems (Reichstein et al., 2002; Rey et al., 2011, 2002). The 409 basal respiration values supported the lower respiration activity under affected than under 410 healthy trees discussed above. Interestingly, the higher Q10 values and variances of Rs explained 411 by Tsoil and SWC under declining than under healthy trees, along with the significant 412 correlations found between the seasonal means of Rs and those of precipitation only under 413 declining trees might suggest that the control of Rs by soil microclimatic variables increases 414 with tree decline. As autotrophic and heterotrophic respirations at the stand level are mainly 415 controlled by plants and soil microclimate, respectively (Chen et al., 2019; Högberg et al., 416 2001; Matteucci et al., 2015), our results might indicate an increase in the relative contribution 417 of the heterotrophic respiration under tree decline. In any case, these results stress the capacity 418 of tree decline to sharply modulate the relationship of Rs with soil microclimatic variables. 419 420 4.3 Main drivers of Rs at the stand scale This preprint research paper has not been peer reviewed. Electronic copy available at: https://ssrn.com/abstract=4259956 Preprint not peer reviewed 18 421 Our results confirmed our third hypothesis about changes in the relative importance of plant, 422 soil microclimatic, and physicochemical variables controlling Rs considering the tree decline 423 stages together and separately. Our results revealed that canopy diameter (a good proxy of tree 424 decline; see methods section) is the most important predictor of Rs, indicating that, at the stand 425 level, healthier trees with larger canopies support higher Rs rates. These results highlight both 426 the important role of tree decline (i.e. health status) as a driver of the ecosystem functioning 427 and the important contribution of the autotrophic component to Rs in our ecosystem. Along 428 with canopy diameter, total N was also a very important predictor of Rs, and the most important 429 predictor under healthy trees, supporting the notion of N as a key nutrient for these ecosystems' 430 functioning (Högberg et al., 2001). Furthermore, the observed negative correlation between 431 canopy diameter and the C:N ratio (Figure S1) supports the previously observed negative effect 432 of forest die-off on soil organic matter quality and N availability (Rodríguez et al., 2020, 2017). 433 Altogether, these results warn about a likely positive feedback between forest die-off and N 434 limitation (Gessler et al., 2016), with important implications for the ecosystem C and N 435 balance. 436 The best models and most important predictors obtained for healthy as compared to 437 affected and dead trees demonstrate that tree decline changes the relative importance of the 438 different Rs, with microclimate variables gaining importance as the defoliation degree 439 increases. Tree defoliation triggers a cascade effect on plant understory and soil microbial 440 communities with important implications for ecosystem C and N budgets, including a decrease 441 in soil organic matter lability (Rodríguez et al. (2017, 2020), which could largely affect RH 442 (Rodríguez et al., 2019; Rui et al., 2016). As discussed above, C:N ratio increased as canopy 443 diameter decreased, supporting the negative effect of tree defoliation and mortality on soil 444 organic matter quality. Accordingly, the best models for both affected and dead trees included 445 labile C as one of the most important Rs predictors. Bulk density and Tsoil were the second and This preprint research paper has not been peer reviewed. Electronic copy available at: https://ssrn.com/abstract=4259956 Preprint not peer reviewed 19 446 third most important predictors for Rs under affected trees, whereas the best model of Rs under 447 dead trees included Tsoil and SWC as the most important predictors. Bulk density is widely 448 recognized to be negatively related to soil organic matter content (Périé and Ouimet, 2008) and 449 roots mass (Daddow and Warrington, 1983), which are among the main controls of RH and RA, 450 respectively. These results would agree with the above-discussed parallel decrease in RA and 451 RH under affected trees and almost complete substitution of Rs for RH under dead trees. Finally, 452 whereas the best models for healthy and affected trees never surpassed 50% of the variance 453 explained, the best model for dead trees reached 93% of the variance explained with just bulk 454 density and both soil microclimatic variables as predictors. This result might suggest a 455 simplification of the control of the soil respiration process under tree decline, which could 456 jeopardize its resilience to the global change drivers, including climate change (Hong et al., 457 2022). 458 459 5. Conclusions 460 Our study adds robust and novel observational evidence of interacting effects between climatic 461 variability and drought-induced tree decline on the spatial-temporal variability of Rs in 462 Mediterranean holm-oak forests. The higher responses of Rs to variations in the amount and 463 distribution of precipitation under healthy than under declining trees stress the important role 464 of tree health as a modulator of the response of Rs to soil moisture conditions. We also show 465 that tree decline strongly influences soil microclimate and Rs in these Mediterranean forests, 466 but the magnitude of this effect depends on other factors such as the declining stage (i.e. 467 affected or dead), the year, and the season. Further, our study demonstrates that tree decline 468 also changes the relative importance of the different drivers of Rs, with microclimate variables 469 gaining importance as the defoliation degree increases. Finally, our study exposed a likely 470 positive feedback between forest die-off and N limitation and a simplification of the soil This preprint research paper has not been peer reviewed. Electronic copy available at: https://ssrn.com/abstract=4259956 Preprint not peer reviewed 20 471 respiration process with tree decline. Altogether, our results point towards negative impacts of 472 drought-induced tree decline on soil C content and cycling, particularly under forecasted 473 climate change scenarios with dryer and more intense precipitation regimes. Thus, more studies 474 investigating the joint effects of drought-induced forest die-off and climatic variability on soil 475 respiration and its main components at different temporal and spatial scales are needed to fully 476 understand the fate of this important ecosystem process. 477 478 Acknowledgments 479 This research was supported by the Spanish National Research Council (CSIC) in the JAE-doc 480 modality co-financed by the European Social Fund (ESF), the ATLANTIS (PID2020481 113244GB-C21) projects funded by the Spanish Government, the Basque Government through 482 the BERC 2022-2025 program, and the Spanish Ministry of Science and Innovation through 483 the BC3 María de Maeztu excellence accreditation (MDM-2017-0714). J.D. and A.R. 484 acknowledge support from the FCT (2020.03670.CEECIND and SFRH/BDP/108913/2015, 485 respectively), as well as from the MCTES, FSE, UE, and the CFE (UIDB/04004/2021) research 486 unit financed by FCT/MCTES through national funds (PIDDAC). The authors are grateful to 487 all the people that at some point helped with fieldwork, particularly to David López. Also, 488 special thanks to Maria José Fernández Alonso and Luis Maria Carrascal for their advice on 489 statistics. This preprint research paper has not been peer reviewed. Electronic copy available at: https://ssrn.com/abstract=4259956 Preprint not peer reviewed 21 490 References 491 Anderegg, W.R.L., Anderegg, L.D.L., Sherman, C., Karp, D.S., 2012. Effects of widespread 492 drought-induced aspen mortality on understory plants. Conserv. Biol. 26, 1082–1090. 493 https://doi.org/10.1111/j.1523-1739.2012.01913.x 494 Avila, J.M., Gallardo, A., Gómez-Aparicio, L., 2019. Pathogen-induced tree mortality 495 interacts with predicted climate change to alter soil respiration and nutrient 496 availability in Mediterranean systems. Biogeochemistry 142, 53–71. 497 https://doi.org/10.1007/s10533-018-0521-3 498 Avila, J.M., Gallardo, A., Ibáñez, B., Gómez-Aparicio, L., 2016. Quercus suber dieback 499 alters soil respiration and nutrient availability in Mediterranean forests. J Ecol 104, 500 1441–1452. https://doi.org/10.1111/1365-2745.12618 501 Baldocchi, D.D., Xu, L., Kiang, N., 2004. How plant functional-type, weather, seasonal 502 drought, and soil physical properties alter water and energy fluxes of an oak–grass 503 savanna and an annual grassland. Agricultural and Forest Meteorology 123, 13–39. 504 https://doi.org/10.1016/j.agrformet.2003.11.006 505 Barba, J., Curiel Yuste, J., Martínez-Vilalta, J., Lloret, F., 2013. Drought-induced tree species 506 replacement is reflected in the spatial variability of soil respiration in a mixed 507 Mediterranean forest. Forest Ecology and Management 306, 79–87. 508 https://doi.org/10.1016/j.foreco.2013.06.025 509 Barba, J., Curiel Yuste, J., Poyatos, R., Janssens, I.A., Lloret, F., 2016. Strong resilience of 510 soil respiration components to drought-induced die-off resulting in forest secondary 511 succession. Oecologia 1–15. https://doi.org/10.1007/s00442-016-3567-8 512 Birch, H.F., 1958. The effect of soil drying on humus decomposition and nitrogen 513 availability. Plant Soil 10, 9–31. https://doi.org/10.1007/BF01343734 514 Bretz, F., Hothorn, T., Westfall, P., 2011. Multiple Comparisons Using R. CRC Press. This preprint research paper has not been peer reviewed. Electronic copy available at: https://ssrn.com/abstract=4259956 Preprint not peer reviewed 22 515 Burnham, K.P., Anderson, D.R., 2010. Model Selection and Multimodel Inference: A 516 Practical Information-Theoretic Approach, 2nd edition. ed. Springer, New York, 517 USA. 518 Carnicer, J., Coll, M., Ninyerola, M., Pons, X., Sánchez, G., Peñuelas, J., 2011. Widespread 519 crown condition decline, food web disruption, and amplified tree mortality with 520 increased climate change-type drought. PNAS 108, 1474–1478. 521 https://doi.org/10.1073/pnas.1010070108 522 Chen, Z., Xu, Y., Castellano, M.J., Fontaine, S., Wang, W., Ding, W., 2019. Soil Respiration 523 Components and their Temperature Sensitivity Under Chemical Fertilizer and 524 Compost Application: The Role of Nitrogen Supply and Compost Substrate Quality. 525 Journal of Geophysical Research: Biogeosciences 124, 556–571. 526 https://doi.org/10.1029/2018JG004771 527 Chiti, T., Certini, G., Marzaioli, F., D’Acqui, L.P., Forte, C., Castaldi, S., Valentini, R., 2019. 528 Composition and turnover time of organic matter in soil fractions with different 529 magnetic susceptibility. Geoderma 349, 88–96. 530 https://doi.org/10.1016/j.geoderma.2019.04.042 531 Curiel Yuste, J., Baldocchi, D.D., Gershenson, A., Goldstein, A., Misson, L., Wong, S., 2007. 532 Microbial soil respiration and its dependency on carbon inputs, soil temperature and 533 moisture. Global Change Biology 13, 2018–2035. https://doi.org/10.1111/j.1365534 2486.2007.01415.x 535 Curiel Yuste, J., Flores-Rentería, D., García-Angulo, D., Hereş, A.-M., Bragă, C., Petritan, 536 A.-M., Petritan, I.C., 2019. Cascading effects associated with climate-change-induced 537 conifer mortality in mountain temperate forests result in hot-spots of soil CO2 538 emissions. Soil Biology and Biochemistry 133, 50–59. 539 https://doi.org/10.1016/j.soilbio.2019.02.017 This preprint research paper has not been peer reviewed. Electronic copy available at: https://ssrn.com/abstract=4259956 Preprint not peer reviewed 23 540 Daddow, R.L., Warrington, G., 1983. Growth-limiting soil bulk densities as influenced by 541 soil texture. Watershed Systems Development Group, USDA Forest Service. 542 Edburg, S., Hicke, J., Brooks, P., Pendall, E., Ewars, B., Norton, U., Gochis, D., Guttman, E., 543 Meddens, A., 2012. Cascading impacts of bark beetle-caused tree mortality on 544 coupled biogeophysical and biogeochemical processes. Frontiers in Ecology and the 545 Environment 10, 416–424. https://doi.org/10.1890/110173 546 Felicísimo, Á.M., Muñoz, J., Villalba, C.J., Mateo, R.G., 2011. Impactos, vulnerabilidad y 547 adaptación al cambio climático de la biodiversidad española. 1. Flora y vegetación. 548 Oficina Española de Cambio Climático, Ministerio de Medio Ambiente y Medio 549 Rural y Marino, Madrid. 550 Gallardo, A., Covelo, F., Morillas, L., Delgado, M., 2009. Ciclos de nutrientes y procesos 551 edáficos en los ecosistemas terrestres: especificidades del caso mediterráneo y sus 552 implicaciones para las relaciones suelo-planta. Revista Ecosistemas 18. 553 https://doi.org/10.7818/re.2014.18-2.00 554 García-Angulo, D., Hereş, A.-M., Fernández-López, M., Flores, O., Sanz, M.J., Rey, A., 555 Valladares, F., Curiel Yuste, J., 2020. Holm oak decline and mortality exacerbates 556 drought effects on soil biogeochemical cycling and soil microbial communities across 557 a climactic gradient. Soil Biology and Biochemistry 107921. 558 https://doi.org/10.1016/j.soilbio.2020.107921 559 Gessler, A., Schaub, M., McDowell, N.G., 2016. The role of nutrients in drought-induced 560 tree mortality and recovery. New Phytol n/a-n/a. https://doi.org/10.1111/nph.14340 561 Högberg, P., Nordgren, A., Buchmann, N., Taylor, A.F.S., Ekblad, A., Högberg, M.N., 562 Nyberg, G., Ottosson-Löfvenius, M., Read, D.J., 2001. Large-scale forest girdling 563 shows that current photosynthesis drives soil respiration. Nature 411, 789–792. 564 https://doi.org/10.1038/35081058 This preprint research paper has not been peer reviewed. Electronic copy available at: https://ssrn.com/abstract=4259956 Preprint not peer reviewed 24 565 Hong, P., Schmid, B., De Laender, F., Eisenhauer, N., Zhang, X., Chen, H., Craven, D., De 566 Boeck, H.J., Hautier, Y., Petchey, O.L., Reich, P.B., Steudel, B., Striebel, M., Thakur, 567 M.P., Wang, S., 2022. Biodiversity promotes ecosystem functioning despite 568 environmental change. Ecology Letters 25, 555–569. 569 https://doi.org/10.1111/ele.13936 570 IPCC, 2021. Climate Change 2021: The Physical Science Basis. Contribution of Working 571 Group I to the Sixth Assessment Report of the Intergovernmental Panel on Climate 572 Change, Masson-Delmotte, V., P. Zhai, A. Pirani, S. L. Connors, C. Péan, S. Berger, 573 N. Caud, Y. Chen, L. Goldfarb, M. I. Gomis, M. Huang, K. Leitzell, E. Lonnoy, J. B. 574 R. Matthews, T. K. Maycock, T. Waterfield, O. Yelekçi, R. Yu and B. Zhou (eds.). 575 ed. 576 Jarvis, P., Rey, A., Petsikos, C., Wingate, L., Rayment, M., Pereira, J., Banza, J., David, J., 577 Miglietta, F., Borghetti, M., Manca, G., Valentini, R., 2007. Drying and wetting of 578 Mediterranean soils stimulates decomposition and carbon dioxide emission: the 579 “Birch effect.” Tree Physiol. 27, 929–940. 580 Liu, Y., Liu, S., Wan, S., Wang, J., Luan, J., Wang, H., 2016. Differential responses of soil 581 respiration to soil warming and experimental throughfall reduction in a transitional 582 oak forest in central China. Agricultural and Forest Meteorology 226–227, 186–198. 583 https://doi.org/10.1016/j.agrformet.2016.06.003 584 Lloret, F., Siscart, D., Dalmases, C., 2004. Canopy recovery after drought dieback in holm585 oak Mediterranean forests of Catalonia (NE Spain). Global Change Biology 10, 586 2092–2099. https://doi.org/10.1111/j.1365-2486.2004.00870.x 587 Lozano-Parra, J., Pulido, M., Lozano-Fondón, C., Schnabel, S., 2018. How do Soil Moisture 588 and Vegetation Covers Influence Soil Temperature in Drylands of Mediterranean 589 Regions? Water 10, 1747. https://doi.org/10.3390/w10121747 This preprint research paper has not been peer reviewed. Electronic copy available at: https://ssrn.com/abstract=4259956 Preprint not peer reviewed 25 590 Matteucci, M., Gruening, C., Goded Ballarin, I., Seufert, G., Cescatti, A., 2015. Components, 591 drivers and temporal dynamics of ecosystem respiration in a Mediterranean pine 592 forest. Soil Biology and Biochemistry 88, 224–235. 593 https://doi.org/10.1016/j.soilbio.2015.05.017 594 Moore, D.J.P., Trahan, N.A., Wilkes, P., Quaife, T., Stephens, B.B., Elder, K., Desai, A.R., 595 Negron, J., Monson, R.K., 2013. Persistent reduced ecosystem respiration after insect 596 disturbance in high elevation forests. Ecology Letters 16, 731–737. 597 https://doi.org/10.1111/ele.12097 598 Nave, L.E., Gough, C.M., Maurer, K.D., Bohrer, G., Hardiman, B.S., Le Moine, J., Munoz, 599 A.B., Nadelhoffer, K.J., Sparks, J.P., Strahm, B.D., Vogel, C.S., Curtis, P.S., 2011. 600 Disturbance and the resilience of coupled carbon and nitrogen cycling in a north 601 temperate forest. J. Geophys. Res. 116, G04016. 602 https://doi.org/10.1029/2011JG001758 603 Périé, C., Ouimet, R., 2008. Organic carbon, organic matter and bulk density relationships in 604 boreal forest soils. Can. J. Soil. Sci. 88, 315–325. https://doi.org/10.4141/CJSS06008 605 Pinheiro, J., Bates, D., DebRoy, S., Sarkar, D., R Core Team, 2020. nlme: Linear and 606 Nonlinear Mixed Effects Models. R package version 3.1-148. 607 R Core Team, 2017. R: A language and environment for statistical computing. R Foundation 608 for Statistical Computing,. Vienna, Austria. 609 Raich, J.W., Schlesinger, W.H., 1992. The global carbon dioxide flux in soil respiration and 610 its relationship to vegetation and climate. Tellus B 44, 81–99. 611 https://doi.org/10.1034/j.1600-0889.1992.t01-1-00001.x 612 Rangel, T.F., Diniz-Filho, J.A.F., Bini, L.M., 2010. SAM: a comprehensive application for 613 Spatial Analysis in Macroecology. Ecography 33, 46–50. 614 https://doi.org/10.1111/j.1600-0587.2009.06299.x This preprint research paper has not been peer reviewed. Electronic copy available at: https://ssrn.com/abstract=4259956 Preprint not peer reviewed 32 725 726 Figure 1 This preprint research paper has not been peer reviewed. Electronic copy available at: https://ssrn.com/abstract=4259956 Preprint not peer reviewed 33 727 728 Figure 2 This preprint research paper has not been peer reviewed. Electronic copy available at: https://ssrn.com/abstract=4259956 Preprint not peer reviewed 34 729 730 Figure 3 This preprint research paper has not been peer reviewed. Electronic copy available at: https://ssrn.com/abstract=4259956 Preprint not peer reviewed 35 731 Supplementary material 732 Table S1. Plant structure and soil physicochemical variables for the different tree decline 733 stages (n = 10 in all cases). Values represent the mean ± 1SE. Different letters next to values 734 within each variable represent significant differences among defoliation degrees (P < 0.05; one735 way ANOVA). Canopy = canopy diameter; DBH = diameter at breast height; CI = competition 736 index; TC, total C; TN, total N; CMA, total carbon in the magnetic fraction (considered as labile 737 C). Plant variables, TC, TN, C:N ratio and pH are data from Rodríguez et al. (2017). Bulk 738 density and CMA are modified data from Rodríguez et al. (2020). Healthy Affected Dead Plant variables Height (m) 4.54 ± 0.20 3.81 ± 0.15 3.82 ± 0.32 Canopy (m) 5.97 ± 0.31a 4.46 ± 0.30b 4.46 ± 0.37b DBH (cm) 45.8 ± 3.86 33.8 ± 5.17 33.8 ± 5.17 CI (x 10-3) 0.8 ± 0.4a 5.5 ± 2.7ab 8.7 ± 3.3b Soil physicochemical variables TC (%) 2.78 ± 0.44 2.27 ± 0.21 2.77 ± 0.31 TN (%) 0.22 ± 0.03 0.17 ± 0.02 0.21 ± 0.02 C:N ratio 12.1 ± 0.38 13.4 ± 0.43 13.1 ± 0.51 CMA (%) 4.86 ± 1.07 2.79 ± 0.40 2.97 ± 0.53 pH (unitless) 6.58 ± 0.11 6.30 ± 0.10 6.56 ± 0.10 Compaction (unitless) 54 ± 15 111 ± 26 114 ± 31 Bulk density (g cm-3) 1.03 ± 0.08 1.09 ± 0.04 1.06 ± 0.06 739 This preprint research paper has not been peer reviewed. Electronic copy available at: https://ssrn.com/abstract=4259956 Preprint not peer reviewed 36 740 Table S2. Seasonal means of precipitation (P), air temperature (Tair), soil water content (SWC), 741 soil temperature (Tsoil) and soil respiration (Rs) for the two studied years. Values represent the 742 mean ± 1SE. Bold P values represent statistically significant effects of year (Y), season (S), as 743 well as significant interactions of both factors. Values with different letters within each year 744 represent significant differences among seasons (P < 0.05). Year Season P (mm)* Tair (°C)* n SWC (%) Tsoil (°C) Rs year 1 Summer 6.45 25.0 30 0.9 ± 0.1c 30.3 ± 0.6a 2.05 ± 0.24c Fall 67.3 16.0 30 3.4 ± 0.3b 16.7 ± 0.2c 2.02 ± 0.11c Winter 147.7 7.10 30 7.7 ± 0.4a 13.5 ± 0.3d 2.41 ± 0.15b Spring 67.3 15.3 30 4.0 ± 0.2b 19.3 ± 0.3b 3.03 ± 0.16a year 2 Summer 21.2 24.3 29 0.9 ± 0.1c 29.1 ± 0.4a 1.87 ± 0.21c Fall 186.8 17.0 29 9.6 ± 0.4a 15.6 ± 0.1c 3.40 ± 0.17a Winter 66.8 6.6 29 11.1 ± 0.7a 7.2 ± 0.2d 1.40 ± 0.10d Spring 68.5 15.3 29 4.7 ± 0.3b 17.1 ± 0.2b 2.86 ± 0.18b Year χ2 77.8 186.2 8.07 P < 0.001 < 0.001 < 0.01 Season χ2 1527.3 7142.6 218.2 GLS models P < 0.001 < 0.001 < 0.001 Y x S χ2 151.7 568.6 251.5 P < 0.001 < 0.001 < 0.001 745 *Getafe (40°17’N 3°43’O) and Cuatro Vientos (40°22’N 3°47’O) stations (AEMET) This preprint research paper has not been peer reviewed. Electronic copy available at: https://ssrn.com/abstract=4259956 Preprint not peer reviewed 37 746 Table S3. Best-fitting regression models and predictors for the two-year average values of soil 747 respiration for each tree decline stage separately (healthy, affected and dead) and all together 748 (All). The best three models (M1-M3), ranked according to their AICc value, and the three 749 most important predictors (Pred.), ranked according to their relative importance (Import) are 750 presented. Coloured cells indicate variables that were included in a particular model (one per 751 row). Canopy, tree canopy diameter; SWC, soil water content; Lab C, labile C; TN, soil total 752 nitrogen; BD, bulk density. Best models Best predictors Canopy SWC Tsoil Lab C TN BD R2 AICc Pred. Import. Healthy M1 0.49 29.4 TN 0.65 M2 0.34 31.9 BD 0.20 M3 0.25 33.1 Lab C 0.16 Affected M1 0.25 6.80 Lab C 0.39 M2 0.22 7.16 BD 0.32 M3 0.09 8.72 Tsoil 0.15 Dead M1 0.93 19.7 Tsoil 0.98 M2 0.77 23.1 Lab C 0.95 M3 0.38 26.8 SWC 0.78 All M1 0.42 60.9 Canopy 0.96 M2 0.45 62.5 TN 0.82 M3 0.44 62.6 Tsoil 0.30 This preprint research paper has not been peer reviewed. Electronic copy available at: https://ssrn.com/abstract=4259956 Preprint not peer reviewed 38 753 754 Figure S1. Heatmap representation of Spearman’s rank correlation matrix among the two-year 755 average values of soil respiration (Rs), plant (Height, Canopy, DBH, CI), and soil microclimatic 756 (SWC and Tsoil) and physicochemical (TC, TN, C:N ratio, Labile C, pH, Compaction, Bulk 757 density) variables for all decline stages together. Only significant values are shown (P < 0.05). 758 SWC, two-years average values of soil volumetric water content; Tsoil, two-years average 759 values of soil temperature; TC, total C; TN, total N. This preprint research paper has not been peer reviewed. Electronic copy available at: https://ssrn.com/abstract=4259956 Preprint not peer reviewed 39 760 761 Figure S2. Relationship between soil temperature and soil respiration for each tree decline 762 stage using averaged values for each temperature degree. Vertical lines show the temperature 763 threshold for each decline stage (i.e., 17C for healthy and dead and 18C for affected). This preprint research paper has not been peer reviewed. Electronic copy available at: https://ssrn.com/abstract=4259956 Preprint not peer reviewed