Montero Et Al 2019 Preprint.Pdf

Home , Mass mortality event

2 DR IGNASI MONTERO-SERRA (Orcid ID : 0000-0003-0284-0591)

3 DR JEAN-BAPTISTE LEDOUX (Orcid ID : 0000-0001-8796-6163)

6 Article type : Research Article

7 Handling Editor: Bret Elderd 8

10 MPAs and warming shape red coral dynamics 11

12 Marine protected areas enhance structural complexity but do

13 not buffer the consequences of ocean warming for an

14 overexploited precious coral

15 16 Ignasi Montero-Serra1*, Joaquim Garrabou2,3, Daniel F. Doak4, Jean-Baptiste 17 Ledoux2,5 & Cristina Linares1

18 1Departament de Biologia Evolutiva, Ecologia i Ciències Ambientals, Institut de Recerca de la Biodiversitat (IRBIO), 19 Universitat de Barcelona, Av. Diagonal 643, 08028 Barcelona, Spain. 20 2Institut de Ciències del Mar, CSIC, Passeig Marítim de la Barceloneta 37-49, 08003 Barcelona, Spain 21 3Aix-Marseille University, Mediterranean Institute of Oceanography (MIO), Marseille, Université de Toulon, CNRS 22 /IRD, France 23 4Environmental Studies Program, University of Colorado, Boulder, CO 80309, USA 24 5CIIMAR/CIMAR, Centro Interdisciplinar de Investigação Marinha e Ambiental, Universidade do Porto, Porto, Portugal. 25 *Correspondence: [email protected] 26 Author Manuscript 27 Authors’ Information:

This is the author manuscript accepted for publication and has undergone full peer review but has not been through the copyediting, typesetting, pagination and proofreading process, which may lead to differences between this version and the Version of Record. Please cite this article as doi: 10.1111/1365-2664.13321

This article is protected by copyright. All rights reserved 28 IMS: E-mail: [email protected]; ORCID: 0000-0003-0284-0591 29 JG: E-mail: [email protected]; ORCID: 0000-0001-9900-7277 30 DFD: E-mail: [email protected]; ORCID: 0000-0001-8980-3392 31 JBL: E-mail: [email protected]; ORCID: 0000-0001-8796-6163 32 CL: E-mail: [email protected]; ORCID: 0000-0003-3855-2743 33 34 35 Abstract 36 1. Global warming and overexploitation both threaten the integrity and resilience of 37 marine ecosystems. Many calls have been made to at least partially offset climate 38 change impacts through local conservation management. However, a mechanistic 39 understanding of the interactions of multiple stressors is generally lacking for habitat- 40 forming species; preventing the development of sound conservation strategies. 41 42 2. We examined the effectiveness of no-take marine protected areas (MPAs) at enhancing 43 structural complexity and resilience to climate change on populations of an 44 overexploited and long-lived octocoral. We used long-term data over eight populations, 45 subjected to varying levels of disturbances, and Integral Projection Models to 46 understand how the interaction between overfishing and mass-mortality events shapes 47 the stochastic dynamics of the Mediterranean red coral Corallium rubrum. 48 49 3. MPAs largely reduced colony partial mortality (i.e. shrinkage), enhancing the structural 50 complexity of coral populations. However, there were no significant differences in 51 individual mortality or population growth rates between protected and exploited 52 populations. In contrast, warming had detrimental consequences for the long-term 53 viability of red coral populations, driving steady declines and potential local extinctions 54 due to sharp effects in survival rates. Stochastic demographic models revealed only a 55 weak compensatory effect of MPAs on the impacts of warming. 56

57 4. Policy implicationsAuthor Manuscript . Our results suggest that marine protected areas (MPAs) are an 58 effective local conservation tool for enhancing the structural complexity of red coral 59 populations. However, MPAs may not be enough to ensure red coral’s persistence 60 under future increases in thermal stress. Accordingly, conservation strategies aiming to

This article is protected by copyright. All rights reserved 61 ensure the persistence and functional role of red coral populations should include 62 management actions at both local (well-enforced MPAs) and global scales (reductions 63 in greenhouse gas emissions). Finally, this study unravels the divergent demographic 64 consequences that can arise from multiple stressors and highlights the key role of 65 demography in better understanding and predicting the consequences of combined 66 impacts for vulnerable ecosystems. 67 68 Keywords: Marine Protected Areas, Climate Change, Integral Projection Models, Multiple 69 Stressors, Heat waves, overfishing Mediterranean Sea, Corallium rubrum, red coral 70 71 72 73 74 75 76 77 78 79 80 81 82 83 84 Introduction 85 The cumulative effects of multiple stressors acting at both local and global scales threaten 86 the integrity and resilience of virtually all ecosystems, and have been particularly of 87 concern for marine ecosystems and their associated services (Worm et al., 2006; Halpern et 88 al., 2008; Ainsworth et al., 2016). The rapid rate of marine biodiversity loss caused by

89 overexploitationAuthor Manuscript and climate change is raising major concerns about the effectiveness of 90 current conservation strategies (Hughes et al., 2010). Global policy agreements to reduce 91 greenhouse gas emissions are urgently needed to slow down the current rate of warming 92 and other drivers of global change such as ocean acidification (Kennedy et al., 2013;

This article is protected by copyright. All rights reserved 93 Anthony et al., 2015). Nonetheless, at local scales, conservation management tools such as 94 fishing restrictions and marine protected areas (MPAs) may enhance ecosystem function 95 and resilience to climate change (Mumby & Harbone, 2010; Bozec et al., 2016; Mellin et 96 al., 2016, Wolff et al., 2018). Given this dichotomy, there are multiple calls for the use of 97 enhanced local conservation efforts to compensate for delayed and ineffective efforts to 98 slow or reverse climate change impacts (Hughes et al., 2010; Roberts et al., 2017). 99 100 At a global scale, climate-induced mass bleaching and mortality events are among the 101 greatest threats faced by tropical and temperate marine benthic ecosystems (Garrabou et al., 102 2009; Wernberg et al., 2016; Hughes et al., 2018). Several comprehensive studies have 103 reported the immediate detrimental impacts of positive thermal anomalies on habitat- 104 forming species such as corals and sponges (Garrabou et al., 2009; Hughes et al., 2017), but 105 much less is known about the long-term consequences of these recurrent events. Individual- 106 based demographic data and modelling tools such as structured population models can 107 improve predictions of the effects of global warming, as well as provide a framework to 108 understand how these effects interact with other stressors. 109 110 In a context of rapid environmental change, a critical uncertainty is how effective local 111 management tools such as no-take marine protected areas (MPAs) are in enhancing the 112 resilience of benthic ecosystems to recurrent warming impacts. Some studies suggest 113 positive effects of MPAs on coral cover and complexity (Linares et al., 2010; Selig & 114 Bruno, 2010), recruitment (Mumby et al., 2007) and resilience (Michelli et al., 2012; Olds 115 et al, 2014; Mellin et al., 2016). But null (Toth et al., 2014; Bruno & Valdivia, 2016; Cox et 116 al., 2016) or even negative relationships between local protection and resistance to climatic 117 disturbances have also been reported (Graham et al., 2008; McClanahan, 2008; Hughes et 118 al., 2017). Developing better methods to assess how local management shapes the 119 stochastic dynamics and structural complexity of habitat-forming species is thus a key 120 research need. Author Manuscript 121 122 The success of local conservation efforts in effectively mitigating for climate change effects 123 will depend on the extent to which different impacts are additive, synergistic or

This article is protected by copyright. All rights reserved 124 compensatory in their combined effects (Figure 1). In marine coastal systems, ecological 125 synergies between fishing and climate change have been commonly hypothesized, but 126 meta-analyses and empirical evaluations have suggested that they are less frequent than 127 expected (Darling et al., 2008; Côté et al., 2016). However, most prior work has been 128 carried out through correlational approaches, preventing a clear dissection of the effects of 129 individual drivers. For example, losses of percent cover in corals, or many other easy to 130 monitor measures of impact, can arise from different processes and lead to counter-intuitive 131 interactions between stressors (Darling et al., 2010). Thus, we still need better mechanistic 132 frameworks for understanding how multiple stressors interact to mitigate or exacerbate 133 their negative consequences on vulnerable ecosystems (Côté et al., 2016). 134 135 While there are different levels of mechanistic analysis, one that can uncover at least some 136 synergies or other interactions is demographic modelling, which tries to tie stressors to 137 impacts on particular vital rates. Yet, obtaining individual-based demographic data can be 138 methodologically challenging in the marine realm, and there are few long-term datasets that 139 allow for an in-depth exploration of the interactive effects of multiple drivers. Here, we 140 apply a demographic analysis to look at the impact of two major threats – ocean warming 141 and fishing– to populations of a long-lived octocoral. We show that, because of their 142 contrasting effects on different vital rates, these threats do not have the strong synergies or 143 compensatory effects that might in theory be expected. Focusing on the Mediterranean red 144 coral, Corallium rubrum, which is a precious long-lived octocoral with an important 145 structural role in Mediterranean hard-bottom communities, we first (1) quantify the effects 146 of fishing (local stressor) and warming-driven mass mortality events (global stressor) to 147 specific vital rates. Then, we (2) predict the long-term consequences of these two different 148 threats. Finally, we (3) evaluate the interactive effects of global change and local protection 149 on the fate of red coral populations. Our results reveal detrimental effects of climate change 150 on red coral dynamics and challenge previous assumptions of MPA capacity to buffer 151 climatic impacts. We also provide insights into how local protection can shape the Author Manuscript 152 structural complexity and dynamics of a long-lived octocoral. Our approach highlights how 153 population models can be used to integrate individual-based demographic data on different 154 impacts to predict the stochastic dynamics of populations threatened by multiple stressors.

This article is protected by copyright. All rights reserved 155 156 Materials and Methods 157 Study System 158 The Mediterranean red coral, Corallium rubrum, is an iconic and long-lived precious 159 octocoral with an important structural role in Mediterranean hard-bottom communities 160 (Ballesteros, 2006; Garrabou et al., 2017). During the last several decades, there has been a 161 growing concern about the conservation of the Mediterranean red coral and other precious 162 corals worldwide due both to unsustainable fishing practices (Tsounis et al., 2010; 163 Bruckner, 2014; Montero-Serra et al., 2015) and to multiple mortality events associated 164 with recent warming in the Mediterranean Sea (Cerrano et al., 2000; Garrabou et al., 2001; 165 2009; Coma et al., 2009). As a result, this species has recently been listed as an endangered 166 species in the IUCN Mediterranean Red List of Threatened Species due to the risks 167 associated with overexploitation and climate change (Otero et al., 2017). 168 169 Study area & coral monitoring 170 Eight red coral populations located in different locations in the NW Mediterranean Sea at 171 relatively shallow depths (15 to 25m), were monitored during periods ranging from 3 to 13 172 years. Five of the studied populations were located within MPAs and three were non- 173 protected and subject to several fishing events during the monitoring (Montero-Serra et al., 174 2015). One of the protected populations, located in the Réserve Natural de Scandola 175 (Corsica, France), has been recurrently affected by warming-induced mass mortality events 176 since 2003 (Garrabou et al., 2009; Coma et al., 2009; for a detailed description of study 177 sites see Table S1 in Supporting Information). 178 179 In each population from 2 to 4 replicate transects were marked using plastic screws 180 attached to the substratum. Colonies were monitored once a year using photographic series, 181 which allow the individual identification of colonies as well as an accurate assessment of 182 life-history processes such as colony growth, partial mortality (e.g. shrinkage and necrosis), Author Manuscript 183 total mortality and recruitment (for more details see Drap et al., 2017; Montero-Serra et al., 184 2018a). 185

This article is protected by copyright. All rights reserved 186 Demographic analyses 187 To estimate the spatial and temporal variability in demography for our study species, we fit 188 a set of annual- and population-level models for size-specific vital rates. These vital rates 189 models included colony size and year as fixed effects and were first fit separately for each 190 population and vital rate: logistic GLMs described probabilities of annual survival and 191 extreme shrinkage (described below) rates, a normally distributed GLM described normal 192 growth data, and a Poisson GLM described fecundity. 193 194 As for many other clonal species, red coral can both grow and shrink over time. Height data 195 from all annual photographs were not available, and growth is so slow in this species that 196 even 7 to 8-year time periods only result in mean growth on the order of 10 to 40mm in 197 maximum height (2.74 ± 0.19 mm·yr-1, Montero-Serra et al., 2018a). Thus, growth data 198 from two populations over variable length periods (7 and 8 years) were used to fit a single 199 model characterizing mean annual growth rate depending on size. Shrinkage or negative 200 growth is a common process in plants and sessile invertebrates (Salguero-Gomez & Casper, 201 2010) that is usually accounted for by assuming normally distributed variation in size 202 changes around a mean rate. However, for many species assuming normally distributed 203 growth variance is problematic, because a few individuals experience sporadic extreme 204 shrinkage events and these results in a highly skewed distribution of growth rates (Shriver 205 et al., 2012). We accounted for these two types of growth following the procedures 206 proposed by Shriver and colleagues (2012), describing normal growth and extreme 207 shrinkage as a two-part process (Montero-Serra et al., 2018a). Size-dependent extreme 208 shrinkage was fit as two parameters (annual probability of shrinking, and then mean and 209 variance if it occurs). Unlike normal growth, extreme shrinkage probabilities were fit to 210 each population and time period. We did not attempt to model the variance around normal 211 growth rates due to the extremely low variance around mean growth. While this is a 212 simplification, our previous work has shown that red coral population growth has relatively 213 low sensitivity to growth rates, which likely minimizes any effects of this modelling Author Manuscript 214 decision (Montero-Serra et al., 2018b).

215

This article is protected by copyright. All rights reserved 216 To directly assess the mean effect of different disturbances on coral survival and extreme 217 shrinkage rates, we also fit a set of Generalized Linear Mixed Models (GLMM) including 218 colony size and disturbance level (protected, unprotected, warming) as fixed factors and 219 population and year as random effects (Table S2 & S3). 220 221 Integral Projection Models 222 Based on the linear models characterizing each vital rate, we derived a set of annual size- 223 structured integral projection models for each population (see Montero-Serra et al. 2018a 224 for details on vital rates estimation and model construction). These sets of models were 225 used to compute the long-term stochastic growth rate at each population, as well as growth 226 rates assuming different combinations of local management and climate effects, as 227 described below. To quantify uncertainty in population growth estimates, we applied a 228 bootstrap resampling to the whole dataset and derived a set of annual- and population-level 229 matrices for each run (n = 1000). We then computed short-term stochastic population

230 growth rate (λs) for each red coral population, based on their observed variation in 231 demography, by choosing each of the annual transition matrices estimated for each 232 population with equal probability over a relevant time horizon of 100 years. This 233 methodology is commonly used in population viability analysis (Genovart et al., 2017). In 234 the simulations, we randomly drew each of the observed annual matrices with equal 235 probability. Simulations were initiated with a population at the stable stage distribution of

236 the mean matrix for the population. λs was estimated as: 237

1 ln ( ( = 100)) ln ( ( = 0)) = , 1000 1000 100 �� − �� 238 � =1

239 where Ntotal,t is the summed population number across all sizes at time t. The resulting

240 distribution of λs values reflect variation due to both parameter uncertainty and stochastic 241 outcomes over a relatively short time horizon. Author Manuscript 242 243 Climate-induced mortality: demographic effects of extreme heat waves

This article is protected by copyright. All rights reserved 244 The effects of heat waves were modelled under current conditions using 13 years of 245 demographic monitoring data in an affected red coral population located in the Scandola 246 Natural Park (Corsica, France). This population has experienced several warming events 247 that led to significant demographic effects. A total of 13 annual-level IPMs were derived 248 from the demographic dataset and were used to forecast the long-term dynamics of this 249 population under current trends. We compared the observed number of coral colonies from 250 2003 to 2016 to 1000 projected simulated trajectories. Simulations were started with the 251 observed initial abundance in 2003 (n = 143 colonies) and colony-size distribution, and 252 were run for 50 years. In addition, to estimate the effects of warming events on size- 253 dependent survival rates (the main vital rate influenced by these events), we fit a logistic 254 GLM to the survival data of 2016, when the highest intensity mass mortality event was 255 recorded. 256 257 Local protection effects on population size and colony-size structure dynamics 258 We tested the effect of local management on the dynamics of coral populations by 259 simulating 1000 trajectories during 50 years using 36 annual- and population-level IPMs 260 derived from the demographic data on 4 protected red coral populations within MPAs (21 261 matrices) and 3 unprotected coral populations (15 matrices) (Table S1). As a starting 262 population vector, we set 100 individuals at the stable stage distribution, which 263 corresponded to the right eigenvector of a mean matrix that included both treatments. In 264 each simulation, we randomly picked one annual transition matrix from the suite of 265 matrices estimated across all years and populations for either protected or unprotected sites. 266 To explore potential effects of transient dynamics and initial conditions to alter the results, 267 we also ran a set of demographic simulations using the observed size distributions at each 268 treatment as well as an average size distribution among treatments; the results showed 269 patterns consistent with those of the main simulation tests (Figure S2 & Figure S2). 270 271 Structural complexity is an important indicator to assess the health status and functional Author Manuscript 272 role of key marine habitat-forming organisms such as corals and gorgonians. Thus, we 273 quantified the proportion of large coral colonies (>100mm) as a proxy for population 274 structural complexity (Linares et al., 2010; Montero-Serra et al., 2018a).

This article is protected by copyright. All rights reserved 275 276 Modelling the combined effect of fishing and warming-induced mass mortality events 277 To explore the interactive effects of local and global stressors (fishing and warming-driven 278 mass mortality events) on the long-term stochastic growth rate and mean time to extinction 279 of red coral populations, we simulated dynamics under 100 scenarios of fishing and 280 warming impacts. To present stable mean results, in these simulations, we only included 281 demographic variation due to local protection and warming events, but not population or 282 annual variation in vital rates. Local management showed no significant effects on the 283 survival rate of coral colonies, but it does strongly impact extreme shrinkage probability 284 (Figure 2). Therefore, we used the predicted mean values from GLMMs for extreme 285 shrinkage probability in protected vs. unprotected coral populations to explore the effects of 286 MPAs. To provide a continuous gradient of local disturbances, we multiplied the effect size 287 estimate of MPA effect by 10 values ranging from 0 to 1 (Table S3 and Figure S5). The 288 lowest level of local disturbance was 0 effect size (corresponding to observed shrinkage 289 within MPAs), while highest level was the shrinkage probability observed in unprotected 290 coral populations. Similarly, we simulated 10 levels of the annual probability of high 291 intensity mass mortality events. These ranged from 0.05 to 0.95 and were set using 292 mortality rates quantified during a high intensity mortality event occurred in 2016 in 293 Palazzu (Figure 4b), and background mortality rates quantified using data from all 294 monitored populations (Table S1). We computed the long-term stochastic population 295 growth rate of the 100 combined scenarios of mass mortality events and local protection 296 using the same methodology described above. We also used stochastic projections to 297 computed mean time to extinction for each combination of stressors by setting an initial

298 population size of N0 = 100 individuals, and a quasi-extinction threshold of 10% that 299 corresponded to 10 individuals. 300 301 Results 302 Demographic consequences of fishing and no-take marine protected areas Author Manuscript 303 All seven red coral populations not subject to mass mortality events had very low 304 extinction risk, with long-term stochastic growth rates near or at those for a stable

305 population (median λs = 1.0088; 95% CI = 0.9941 to 1.0165); and this was consistent

307 = 1.0000 to 1.0165; unprotected sites, median λs = 0.9993, 95% CI = 0.9930 to 1.0167; 308 Figure 2a & Figure S1). Size-dependent mortality rates also did not differ between 309 unprotected sites and protected red coral populations (Logistic-GLMM; P = 0.6810; Figure 310 2b, Table S2). Projected dynamics using IPMs fitted to data from protected and unprotected 311 populations showed very similar patterns regardless of protection status (Figure 2c). 312 313 In contrast, colony shrinkage rates depended strongly on local management status. The best 314 fitted model for shrinkage probability included a significant interaction between coral 315 colony size and protection status, with much higher shrinkage rates for large and 316 unprotected coral colonies (Figure 2d, Logistic GLMM; P < 0.001; Table S3). When 317 considering the proportion of large colonies as proxy of population structural complexity, 318 projected trajectories of coral populations within and outside MPAs showed divergent 319 trajectories: starting at the stable-stage distribution, protected populations maintained their 320 structural complexity while unprotected populations showed strong, rapid decrease in the 321 abundance of large coral colonies (Figure 2e). This caused a change in the relative 322 importance of life-history processes depending on local management schemes. In protected 323 populations, stasis, the probability of surviving but remaining approximately the same size, 324 was the dominant life-history process, with large colonies remaining within the same state 325 and resulting in high structural complexity. On the contrary, recurrent shrinkage events 326 coupled to slow colony growth drive a structural simplification in unprotected populations, 327 which are dominated by small and medium-size coral colonies (Figure 3 & S4. Across 328 colony sizes, stasis (computed by summing all the cell values from the diagonal of the big 329 matrix derived from the IPMs) is the dominant demographic process in both protected and 330 unprotected sites. However, in large colonies (>100mm), summed stasis probabilities 331 decreased from 60% in protected populations to 36% in unprotected ones, while shrinkage 332 (the probability of surviving but declining in size, computed as the sum of all elements 333 below the first row and above the main diagonal of the big matrix) increased from 8% in Author Manuscript 334 protected populations to 38% in unprotected ones (Figure S4). 335 336 Demographic consequences of warming-driven mass mortality events

This article is protected by copyright. All rights reserved 337 Recurrent positive temperature anomalies caused several mass-mortality events in the red 338 coral population at Palazzu (Corsica) during the 13 years of demographic monitoring 339 (Figure 4). These mortality events resulted in a negative long-term stochastic growth

340 (median λs = 0.9729, 95% CI = 0.9573 to 0.9867, Figure S1) and drove a steady decline 341 that may result in the local extinction of the warming-affected population (Figure 4a). The 342 most severe event was observed during the summer of 2016 and caused whole-colony 343 mortality rates of approximately 25% of the total coral colonies (Figure 4b). 344 345 Interactive effects of local (fishing) and global (warming-induced mortalities) stressors 346 Simulated trajectories under multiple scenarios of differing intensities of global warming 347 and fishing revealed that red coral long-term stochastic population growth rate is highly 348 sensitive to changes in frequency of climate-driven mass mortality events, in particular 349 (Figures 5 & 6). Local protection was an important factor enhancing population growth 350 under low mass mortality event probability. However, its effects were biologically 351 negligible under hypothetical scenarios of increased frequency of mortality events, as all 352 growth rates are below one when mass mortality events become common (Figure 5). 353 Overall, we find little evidence for strong synergistic or compensatory interactions among 354 these two stressors. When these extreme events are uncommon, we see some evidence for 355 compensatory effects on growth rates between the two stressors (Figure5a vs. Figure 1f), 356 but in general growth isoclines are largely linear, suggesting additive effects (Figure 5a vs. 357 Figure 1d). Time to extinction shows more consistently convex isoclines (Figure 5b), 358 suggesting somewhat more compensatory interaction between the two stressors, but this 359 was still relatively weak. 360 361 Discussion 362 Predictions of future ecological change are currently challenged by a poor understanding of 363 how multiple stressors interact to shape the dynamics and structure of populations and 364 communities. Here, we show the strong but divergent demographic consequences of Author Manuscript 365 warming-induced mortality events and overfishing on a long-lived octocoral. Recurrent 366 heat waves raise whole-colony mortality rates and compromise the long-term viability of 367 coral populations. By contrast, banning fishing in MPAs did not affect coral mortality

This article is protected by copyright. All rights reserved 368 patterns but did reduce shrinkage rates, allowing colonies to attain large sizes and strongly 369 increasing the structural complexity of populations. However, our results suggest that these 370 local and global scale stressors are mostly additive, with MPAs having only a weak 371 compensatory effect on the population-level effects of warming events. 372

373 Effects of global warming on red coral populations 374 In the Mediterranean, several mass mortality events have impacted multiple organisms, 375 most of them slow-growing species, across vast regions (Cerrano et al., 2000; Garrabou et 376 al., 2009). The immediate destructive effects of these widespread climate-driven mortality 377 events in corals and other habitat-forming marine invertebrates are well known (Garrabou 378 et al., 2009; Hughes et al., 2017). However, predictions of the long-term, population-level 379 consequences of these events are still relatively scarce. In this study, stochastic projections 380 based on individual-level demographic data over a 13-year period in a red coral population 381 subjected to recurrent climatic disturbances revealed clear detrimental consequences for 382 population health (Figure 4). These projections, based on empirical data from 2003 to 2016, 383 are likely to underestimate the impact of future warming in this red coral population. There 384 is still uncertainty around the expected future climatic trends. However, current evidences 385 suggest that the frequency and intensity of heat waves has already increased and is likely to 386 continue following this path during this century (Hughes et al. 2017; 2018). This is 387 especially worrisome for the Mediterranean Sea, where warming is occurring at faster rates 388 than in larger ocean basins (Vargas-Yanez et al., 2008; Macías et al., 2013). Under these 389 trends, local extirpation of shallow red coral populations appears likely in a near future. The 390 predicted declines can be even more adverse for long-lived species such as the red coral, in 391 which rescue effects are very unlikely at ecological time-scales due to highly restricted 392 larval dispersal (Ledoux et al., 2010a, b). For long-lived species with slow population 393 dynamics, catastrophic consequences could go unnoticed by managers when looking at 394 short-term trends (Hughes et al., 2013). Some authors have suggested that adaptive 395

responses can leadAuthor Manuscript to increased thermal resistance of red coral colonies to warming 396 (Ledoux et al., 2015). However, observed increases on the severity of mortality events and 397 their long-lasting consequences provide little empirical support to any adaptive process that 398 could counterbalance these climate change impacts. Furthermore, long generation times for

This article is protected by copyright. All rights reserved 399 habitat-forming species such as the red coral make rapid genetic change unlikely. The 400 largest survey of impacts of the mass bleaching event occurred in the Great Barrier Reef in 401 2016 also refutes any optimistic expectations that corals exposed to previous thermal stress 402 may show increased resistance to future warming impacts (Hughes et al., 2017). 403 404 Local stressors and the role of Marine Protected Areas 405 In the current context of rapid environmental change, understanding the effects of MPAs on 406 population dynamics and extinction risk in marine sessile invertebrates is an essential but 407 unresolved question. Our results show that local protection did not significantly alter red 408 coral densities or overall dynamics. Recurrently fished populations outside of MPAs 409 displayed low extinction risk and their long-term stochastic growth was nearly at 410 equilibrium (Figure 2a). This counterintuitive result could be explained by the clonal nature 411 of the red coral and its capacity to survive and undergo re-growth of new branches after 412 being partially harvested, which is common when fishers leave the colony bases attached to 413 the substratum (Montero-Serra et al., 2015). It is important to note that highly destructive 414 red coral fishing practices (i. e. removal of the whole coral colony, including its base, 415 which prevents recovery) are still observed in some areas of the Mediterranean and they 416 could lead to much worse ecological effects than those reported here (Cattaneo-Vietti et al., 417 2017). A slight increase in total mortality rates due to fishing would result in a dramatic 418 increase on the population extinction risk outside MPAs and result on an enhanced role of 419 local protection. Broad-scale quantification on the extent of these unsustainable practices is 420 essential to improve current management strategies for the conservation of Corallium 421 rubrum and other precious corals. 422 423 In contrast to the lack of effects on abundance, the absence of coral fishing in no-take areas 424 had a strong positive effect on the structural complexity of coral populations by reducing 425 the probability of colony breakage and allowing coral colonies to reach larger sizes (Figure 426 2d). Outside MPAs, structural simplification was observed with absence of large colonies, Author Manuscript 427 driven by both the natural slow colony growth of the species and recurrent colony breakage 428 events due to fishing. Thus, local management can shape the relative importance of the 429 demographic processes underlying red coral dynamics, enhancing the structural complexity

This article is protected by copyright. All rights reserved 430 and functionality of red coral populations (Figure 3 & S2). These results are consistent with 431 previous observations that found that populations within MPAs have larger biomass than 432 unprotected sites (Linares et al., 2010; 2012; Bavestrello et al., 2015; Garrabou et al., 433 2017). Here, we provide a mechanistic understanding on how differential shrinkage but 434 similar whole colony mortality rates shape the consistent broad scale patterns observed in 435 colony-size structures and population persistence of the red coral across the Mediterranean 436 Sea (Linares et al., 2010; Garrabou et al., 2017, Figure 2 & 3). 437 438 Interactive effect of local management and global warming on the viability of red coral 439 populations 440 It remains unclear how effective local management tools such as Marine Protected Areas 441 (MPAs) are in building resilience of benthic ecosystems to climate change. Local 442 protection enhanced the size-structure of red coral populations and, since climate-driven 443 whole colony mortality is size-dependent in the red coral (with smaller colonies being more 444 susceptible, Figure 4b), a compensatory effect of MPAs against climatic perturbations 445 could be expected. However, our results show that MPAs provided only a very weak 446 compensatory effect. This slows down the rate of population declines along a thermal stress 447 gradient but does so only under low to moderate frequencies of mass mortality events 448 (MME) (Figure 5 & 6). When recurrent MME occurred at high frequencies, MPAs had a 449 little buffering effect on the long-term persistence of coral populations. To date, only few 450 studies provide data over relevant time-periods to assess the potential role of MPAs at 451 buffering climatic impacts. Micheli and colleagues (2012) showed that, in a temperate 452 abalone species, MPAs enhanced population resilience to climatic disturbances by 453 increasing the presence of larger individuals that drove a major reproductive output as well 454 as larvae export potential. Similarly, in tropical coral reefs, Mellin and colleagues (2016) 455 reported a strong effect of MPAs at enhancing resilience of benthic communities to external 456 perturbations, although the mechanisms underlying these patterns were not demonstrated. 457 Author Manuscript 458 In contrast, Graham and colleagues (2008) found that reefs under local protection had been 459 more impacted by El Niño bleaching event in 1998. More recently, an unprecedented mass 460 bleaching event occurred at the Great Barrier Reef, with more than 90% of the reefs

This article is protected by copyright. All rights reserved 461 affected irrespective of their level of local protection (Hughes et al., 2017). Our results add 462 support to the latter findings in showing that local protection could be effective at lower 463 levels of thermal stress but have weak to non-existent compensatory effects under potential 464 future scenarios of increased ocean warming. 465 466 Conclusions 467 Long-term studies provide a unique window to understand and predict the responses of 468 marine populations and communities to multiple stressors. This study demonstrates that 469 MPAs are a good tool to enhance the structural dynamics of economically and ecologically 470 important species such as the red coral. However, warming represents a new threat 471 originated at a global scale that can drive either subtle or more abrupt declines that will 472 result in the “extinction debt” of these key habitat-forming species (Hughes et al., 2013). 473 While the red coral and other long-lived habitat-forming species have persisted millennia of 474 human pressures, novel climatic threats may lead to unanticipated and massive 475 consequences for the integrity of temperate reefs. Unfortunately, local protection may not 476 be enough to ensure the persistence of shallow populations under future scenarios of higher 477 frequency and intensity of mass mortality events. A rapid and drastic reduction of 478 greenhouse gas emissions seems the only effective way to prevent the widespread 479 extinction of shallow red coral populations and other long-lived marine invertebrates. In the 480 context of a variety of multi-scale stressors, any conservation action aimed to preserve the 481 functional role and ensure the persistence of benthic communities will need to consider 482 conservation actions at both local and global scales. 483 484 Authors’ contributions 485 IMS, CL, DFD, and JG designed the study. CL, JBL, IMS, and JG collected the 486 demographic data in the field. IMS and DFD performed analyses. IMS led the writing of 487 the manuscript, all authors contributed critically to drafts and gave final approval for 488 publication. Author Manuscript 489 490 Acknowledgements

This article is protected by copyright. All rights reserved 491 This study was partially funded by the Spanish Ministry of Economy and Innovation 492 Biorock project (CTM2009-08045), Smart project (CGL2012-32194), TOTAL Foundation 493 Perfect Project, Prince Albert II of Monaco Foundation MIMOSA project and the European 494 Union’s Horizon 2020 research and innovation programme under grant agreement No 495 689518 (MERCES). This output reflects only the author’s view and the European Union 496 cannot be held responsible for any use that may be made of the information contained 497 therein. IMS was supported by a FPI grant (BES-2013-066150), CL by a Ramon y Cajal 498 (RyC-2011-08134), and JBL by a Postdoctoral grant (SFRH/BPD/74400/2010) from the 499 Portuguese Foundation for Science and Technology. Support for DFD came from National 500 Science Foundation awards 1340024 and 1242355. IMS, CL, JBL and JG are part of the 501 Marine Conservation research group (2017 SGR 1521) from the Generalitat de Catalunya. 502 The authors declare no conflict of interests 503 504 Data Accessibility 505 Data available via the Dryad Digital Repository https:// doi:10.5061/dryad.785sc4d 506 (Montero-Serra et al. 2018c).

507

508 References 509 Ainsworth, T. D., Heron, S. F., Ortiz, J. C., Mumby, P. J., Grech, A., Ogawa, D., ... & 510 Leggat, W. (2016). Climate change disables coral bleaching protection on the Great Barrier 511 Reef. Science, 352(6283), 338-342.

512 Anthony, K., Marshall, P. A., Abdulla, A., Beeden, R., Bergh, C., Black, R., ... & Green, A. 513 (2015). Operationalizing resilience for adaptive coral reef management under global 514 environmental change. Global Change Biology, 21(1), 48-61. 515 516 Ballesteros E (2006) Mediterranean coralligenous assemblages: a synthesis of present

517 knowledge. OceanogryAuthor Manuscript and Marine Biology: An Annual Review, 123–195 518 519 Bavestrello, G., Bo, M., Bertolino, M., Betti, F., & Cattaneo‐Vietti, R. (2015). Long‐ term 520 comparison of structure and dynamics of the red coral metapopulation of the Portofino

This article is protected by copyright. All rights reserved 521 Promontory (Ligurian Sea): a case‐study for a M arine Protected A rea in the M editerranean 522 Sea. Marine Ecology, 36(4), 1354-1363. 523 524 Bozec, Y. M., O'Farrell, S., Bruggemann, J. H., Luckhurst, B. E., & Mumby, P. J. (2016). 525 Tradeoffs between fisheries harvest and the resilience of coral reefs. Proceedings of the 526 National Academy of Sciences USA, 113(16), 4536-4541. 527 528 Bruckner, A. W. (2014). Advances in management of precious corals in the family 529 Corallidae: are new measures adequate?. Current Opinion in Environmental Sustainability, 530 7, 1-8. 531

532 Bruno, J. F., & Valdivia, A. (2016). Coral reef degradation is not correlated with local 533 human population density. Scientific Reports, 6, 29778. 534

535 Cattaneo-Vietti, R., Bavestrello, G., Bo, M., Canese, S., Vigo, A., & Andaloro, F. (2017). 536 Illegal ingegno fishery and conservation of deep red coral banks in the Sicily Channel 537 (Mediterranean Sea). Aquatic Conservation: Marine and Freshwater Ecosystems, 27(3), 538 604-616. 539 540 Cerrano, C., Bavestrello, G., Bianchi, C. N. (2000). A catastrophic mass‐mortality episode 541 of gorgonians and other organisms in the Ligurian Sea (North-western Mediterranean), 542 summer 1999. Ecology Letters, 3(4), 284-293. 543 544 Coma, R., Ribes, M., Serrano, E. et al. (2009). Global warming-enhanced stratification and 545 mass mortality events in the Mediterranean. Proceedings of the National Academy of 546 Sciences, 106(15), 6176-6181. 547 Author Manuscript 548 Côté, I. M., Darling, E. S., & Brown, C. J. (2016). Interactions among ecosystem stressors 549 and their importance in conservation. Proceedings of the Royal Society B, 283 (1824): 550 20152592

This article is protected by copyright. All rights reserved 551 552 Cox, C., Valdivia, A., McField, M., Castillo, K., & Bruno, J. F. (2017). Establishment of 553 marine protected areas alone does not restore coral reef communities in Belize. Marine 554 Ecology Progress Series, 563, 65-79. 555

556 Darling, E. S., McClanahan, T. R., & Côté, I. M. (2010). Combined effects of two stressors 557 on Kenyan coral reefs are additive or antagonistic, not synergistic. Conservation Letters, 558 3(2), 122-130. 559

560 Darling, E. S., & Côté, I. M. (2008). Quantifying the evidence for ecological synergies. 561 Ecology Letters, 11(12), 1278-1286. 562

563 Drap, P., Royera, J. P., Nawaf, M. M., Saccone, M., Merad, D., López-Sanz, A., ... & 564 Garrabou, J. (2017). Underwater photogrammetry, coded target and plenoptic technology: a 565 set of tools for monitoring red coral in the Mediterranean sea in the framework of the 566 ”PERFECT” project. ISPRS-International Archives of the Photogrammetry, Remote 567 Sensing and Spatial Information Sciences, 275-282. 568 569 Garrabou, J., Perez, T., Sartoretto, S. et al. (2001). Mass mortality event in red coral 570 Corallium rubrum populations in the Provence region (France, NW Mediterranean). Marine 571 Ecology Progress Series, 217, 263-272. 572 573 Garrabou, J., Coma, R., Bensoussan, N. et al. (2009). Mass mortality in Northwestern 574 Mediterranean rocky benthic communities: effects of the 2003 heat wave. Global Change 575 Biology, 15(5), 1090-1103. 576

577 Garrabou, J., Sala,Author Manuscript E., Linares, C., Ledoux, J. B., Montero-Serra, I., Dominici, J. M., ... & 578 Harmelin, J. G. (2017). Re-shifting the ecological baseline for the overexploited 579 Mediterranean red coral. Scientific Reports, 7, 42404. 580

This article is protected by copyright. All rights reserved 581 Genovart, M., Doak, D. F., Igual, J. M., Sponza, S., Kralj, J., & Oro, D. (2017). Varying 582 demographic impacts of different fisheries on three Mediterranean seabird species. Global 583 Change Biology, 23(8), 3012-3029. 584 585 Graham N.A.J., McClanahan T.R., MacNeil M.A. et al. (2008) Climate warming, marine 586 protected areas and the ocean-scale integrity of coral reef ecosystems. PLoS ONE, 3:e3039 587 588 Halpern, B. S., Walbridge, S., Selkoe, K. A. et al. (2008). A global map of human impact 589 on marine ecosystems. Science, 319(5865), 948-952. 590 591 Hughes, T. P., Graham, N. A., Jackson, J. B. et al. (2010). Rising to the challenge of 592 sustaining coral reef resilience. Trends in Ecology & Evolution, 25(11), 633-642. 593 594 Hughes, T. P., Linares, C., Dakos, V. et al. (2013). Living dangerously on borrowed time 595 during slow, unrecognized regime shifts. Trends in Ecology & Evolution, 28(3), 149-155. 596 597 Hughes, T. P., Kerry, J. T., Álvarez-Noriega, M. et al. (2017). Global warming and 598 recurrent mass bleaching of corals. Nature, 543(7645), 373-377. 599

600 Hughes, T. P., Anderson, K. D., Connolly, S. R., Heron, S. F., Kerry, J. T., Lough, J. M., ... 601 & Claar, D. C. (2018). Spatial and temporal patterns of mass bleaching of corals in the 602 Anthropocene. Science, 359(6371), 80-83. 603 604 Kennedy, E. V., Perry, C. T., Halloran, P. R., Iglesias-Prieto, R., Schönberg, C. H., 605 Wisshak, M., ... & Mumby, P. J. (2013). Avoiding coral reef functional collapse requires 606 local and global action. Current Biology, 23(10), 912-918.

607 Author Manuscript 608 Ledoux, J. B., Garrabou, J., Bianchimani, O., Drap, P., Féral, J. P., & Aurelle, D. (2010a). 609 Fine‐scale genetic structure and inf erences on popul ation biology in the hreatenedt 610 Mediterranean red coral, Corallium rubrum. Molecular Ecology, 19(19), 4204-4216.

612 Ledoux, J. B., Mokhtar-Jamaï, K., Roby, C., Féral, J. P., Garrabou, J., & Aurelle, D. 613 (2010b). Genetic survey of shallow populations of the Mediterranean red coral [Corallium 614 rubrum (Linnaeus, 1758)]: new insights into evolutionary processes shaping nuclear 615 diversity and implications for conservation. Molecular Ecology, 19(4), 675-690. 616

617 Ledoux, J. B., Aurelle, D., Bensoussan, N., Marschal, C., Féral, J. P., & Garrabou, J. 618 (2015). Potential for adaptive evolution at species range margins: contrasting interactions 619 between red coral populations and their environment in a changing ocean. Ecology and 620 Evolution, 5(6), 1178-1192. 621 622 Linares, C., Bianchimani, O., Torrents, O. et al. (2010). Marine Protected Areas and the 623 conservation of long-lived marine invertebrates: the Mediterranean red coral. Marine 624 Ecology Progress Series, 402, 69-79. 625 626 Linares, C., Garrabou, J., Hereu, B. et al. (2012). Assessing the effectiveness of marine 627 reserves on unsustainably harvested long-lived sessile invertebrates. Conservation Biology, 628 26(1), 88-96. 629 630 Macias, D., Garcia-Gorriz, E., & Stips, A. (2013). Understanding the causes of recent 631 warming of Mediterranean waters. How much could be attributed to climate change?. 632 PLOS ONE, 8(11), e81591. 633 634 McClanahan, T. R. (2008). Response of the coral reef benthos and herbivory to fishery 635 closure management and the 1998 ENSO disturbance. Oecologia, 155(1), 169-177. 636 637 Mellin, C., Aaron MacNeil, M., Cheal, A. J. et al. (2016). Marine protected areas increase

638 resilience amongAuthor Manuscript coral reef communities. Ecology Letters, 19(6), 629-637. 639

This article is protected by copyright. All rights reserved 640 Micheli, F., Saenz-Arroyo, A., Greenley, A., Vazquez, L., Montes, J. A. E., Rossetto, M., 641 & De Leo, G. A. (2012). Evidence that marine reserves enhance resilience to climatic 642 impacts. PloS ONE, 7(7), e40832. 643 644 Montero-Serra, I., Linares, C., García, M., Pancaldi, F., Frleta-Valić, M. et al. (2015). 645 Harvesting effects, recovery mechanisms, and management strategies for a long-lived and 646 structural precious coral. PloS ONE, 10(2), e0117250. 647 648 Montero-Serra, I., Garrabou, J., Doak, D. F., Figuerola, L., Hereu, B., Ledoux, J. B., & 649 Linares, C. (2018a). Accounting for life-history strategies and timescales in marine 650 restoration. Conservation Letters. 11(1), e12341. 651

652 Montero-Serra, I., Linares, C., Doak, D. F., Ledoux, J. B., & Garrabou, J. (2018b). Strong 653 linkages between depth, longevity and demographic stability across marine sessile species. 654 Proceedings of the Royal Society B, 285(1873), 20172688

655

656 Montero-Serra, I., Garrabou, J., Doak, D. F., Ledoux, J. B., & Linares, C. (2018c). Data 657 from: Marine protected areas enhance structural complexity but do not buffer the 658 consequences of ocean warming for an overexploited precious coral. Dryad Digital 659 Repository. https:// doi:10.5061/dryad.785sc4d

660

661 Mumby, P. J., Harborne, A. R., Williams, J., Kappel, C. V., Brumbaugh, D. R., Micheli, F., 662 & Blackwell, P. G. (2007). Trophic cascade facilitates coral recruitment in a marine 663 reserve. Proceedings of the National Academy of Sciences, 104(20), 8362-8367. 664

665 Mumby PJ, HarborneAuthor Manuscript AR (2010) Marine reserves enhance the recovery of corals on 666 Caribbean reefs. PLoS ONE, 5, e8657. 667

This article is protected by copyright. All rights reserved 668 Olds, A. D., Pitt, K. A., Maxwell, P. S., Babcock, R. C., Rissik, D., & Connolly, R. M. 669 (2014). Marine reserves help coastal ecosystems cope with extreme weather. Global 670 Change Biology, 20(10), 3050-3058. 671

672 Otero, M.M., Numa, C., Bo, M., Orejas, C., Garrabou, J., et al., (2017). Overview of the 673 conservation status of Mediterranean anthozoans. IUCN, Malaga, Spain. 73 pp. 674 675 Roberts, C. M., O’Leary, B. C., McCauley, D. J., Cury, P. M., Duarte, C. M., Lubchenco, 676 J., ... & Worm, B. (2017). Marine reserves can mitigate and promote adaptation to climate 677 change. Proceedings of the National Academy of Sciences, 114(24), 6167-6175. 678

679 Salguero‐Gómez, R., & Casper, B. B. (2010) . K eeping plant shrink age in het demographi c 680 loop. Journal of Ecology, 98(2), 312-323. 681 682 Selig E.R. & Bruno J.F. (2010) A global analysis of the effectiveness of marine protected 683 areas in preventing coral loss. PLoS ONE 5(2): e9278.Toth, L. T., Van Woesik, R., 684 Murdoch, T. J. T. et al. (2014). Do no-take reserves benefit Florida’s corals? 14 years of 685 change and stasis in the Florida Keys National Marine Sanctuary. Coral Reefs, 33(3), 565- 686 577. 687

688 Shriver, R. K., Cutler, K., & Doak, D. F. (2012). Comparative demography of an epiphytic 689 lichen: support for general life history patterns and solutions to common problems in 690 demographic parameter estimation. Oecologia, 170(1), 137-146. 691 692 Toth, L. T., Van Woesik, R., Murdoch, T. J. T., Smith, S. R., Ogden, J. C., Precht, W. F., & 693 Aronson, R. B. (2014). Do no-take reserves benefit Florida’s corals? 14 years of change 694 and stasis in the Florida Keys National Marine Sanctuary. Coral Reefs, 33(3), 565-577. Author Manuscript 695

This article is protected by copyright. All rights reserved 696 Tsounis, G., Rossi, S., Grigg, R., Santangelo, G., Bramanti, L., & Gili, J. M. (2010). The 697 exploitation and conservation of precious corals. Oceanography and Marine Biology: An 698 Annual Review. 48, 161-212 699

700 Vargas-Yanez, M., García, M.J., Salat, J., García-Martínez, M.C., Pascual, J.,and Moya, F. 701 (2008).Warming trends and decadal variability in the Western Mediterranean shelf. Glob. 702 Planet. Chang. 63,177–184.doi:10.1016/j.gloplacha.2007.09.001 703

704 Wernberg, T., Bennett, S., Babcock, R. C., de Bettignies, T., Cure, K., Depczynski, M., ... 705 & Harvey, E. S. (2016). Climate-driven regime shift of a temperate marine ecosystem. 706 Science, 353(6295), 169-172. 707

708 Wolff, N.H., Mmumby, P.J., Devlin, M. & Anthony, K.R.N. (2018) Vulnerability of the 709 Great Barrier Reef to climate chnage and local pressures. Global Change Biology, 1-14 710 711 Worm, B., Barbier, E. B., Beaumont, N. et al. (2006). Impacts of biodiversity loss on ocean 712 ecosystem services. Science, 314(5800), 787-790. 713

714

715

716

717

718

719 Author Manuscript 720

721 FIGURES

722

This article is protected by copyright. All rights reserved 723 Figure 1. Conceptual models for two interacting stressors. Potential interactions can be 724 additive (left), in which the total impact is equivalent to the sum of the two individual 725 stressors, synergistic (center), when the combined effect leads to a higher impact than the 726 expected addition of the two individual sressors, and compensatory (right), when the 727 combined effects of the two stressors lead to a lower impact than the expected sum of the 728 two individual stressors.

729

730 Figure 2. Local protection effects on coral population dynamics and structural 731 complexity. Unprotected sites affected by recurrent fishing events (red) and protected sites 732 within MPAs where fishing is not allowed (blue). (a) Long-term stochastic population 733 growth rate at each site. (b, d) MPA effects in vital rates: (b) total mortality (% of annual 734 dead colonies depending on maximum colony size) and (d) partial mortality (annual 735 shrinkage probability depending on colony size ). (c, e) Simulated stochastic dynamics in 736 abundance (c) and structural complexity (e), represented as % of large colonies (>100mm).

737 738 Figure 3. Effects of local protection and overfishing on demographic processes and 739 structural complexity of red coral populations. The conceptual diagram shows how 740 shrinkage rates and stasis of large colonies are shaped by local protection (left) and fishing 741 (right). Protection from fishing results in more complex red coral populations with presence 742 of large coral colonies, whereas recurrent fishing events outside MPAs cause a structural 743 simplification of populations which are dominated by small to medium size coral colonies 744 (stasis and shrinkage probabilities for large colonies are shown in Figure S4).

745 746 Figure 4. Demographics effects of warming-driven mass mortality events. A. Long- 747 term stochastic dynamics on a warming-affected red coral population. Dots show the 748 observed abundance during the 13yrs of monitoring. Grey lines represent 1000 stochastic

749 projections usingAuthor Manuscript IPMs and yellow area represents the 95% CI around the predicted 750 trajectories (quantiles 2.5% and 97%). B. Logistic GLMMs describe undisturbed mortality 751 rates across all populations and years (excluding mass mortality events [MME]) (green) and

754

755 Figure 5. Interactive effects of local and global stressors on the stochastic population 756 dynamics and extinction risk of Mediterranean red coral populations. A. Long-term 757 stochastic growth rate (λs) and B shows the mean time to extinction. Vertical axis 758 represents levels of local disturbance intensity (see methods and Figure S5) and horizontal 759 axis represents levels of probability of mass mortality events.

760

761 Figure 6. Demographic simulations show an additive interaction between local 762 disturbances (overfishing) and global warming (probability of mass mortality event) on the 763 long-term stochastic growth rate of Mediterranean red coral populations. Author Manuscript

esnopser lacigoloib esnopser low stressor B low stressor B low stressor B

high stressor B high stressor B high stressor B biological response biological biological response biological

low stressor Ahigh low stressor Ahigh low stressor A high

(d) (e) (f)

high high high (+) λ < 1 λ < 1 λ < 1 B B B or or or λ = 1 λ = 1 λs stress stress stress

Author Manuscript λ = 1 λ > 1 < 1 λ > 1 λ low low low (-) high low high This article is protected by copyright. All rights low stressor A high reservedlow stressor A stressor A jpe_13321_f2.pdf

Author Manuscript

0.00

−0.05

−0.10 Local disturbance levels

Low (protected)

High (overfished)

−0.15

0.05Author Manuscript 0.15 0.25 0.35 0.45 0.55 0.65 0.75 0.85 0.95

This article is protected by copyright. All rights Probability of Mass Mortalityreserved Event