Upload
others
View
3
Download
0
Embed Size (px)
Citation preview
Title:1
Meetingfisheries,ecosystemfunction,andbiodiversitygoalsinahuman2 dominatedworld3
4
Onesentencesummary:Simultaneouslymeetingfisheries,ecosystem5 function,andbiodiversitygoalsforcoralreefsispossiblethroughstrategically6 placedmarinereservesandfisheriesrestrictions.7
8
9
Authors:JoshuaE.Cinner1,*,JessicaZamborain-Mason1,GeorginaG.Gurney1,10
NicholasA.J.Graham1,2,M.AaronMacNeil3,AndrewS.Hoey1,CamiloMora4,11
SébastienVilléger5,EvaMaire1,2,5,TimR.McClanahan6,JosephM.Maina6,7,12
JohnN.Kittinger8,9,ChristinaC.Hicks1,2,StephanieD’agata5,6,7,CindyHuchery113
,MicheleL.Barnes1,DavidA.Feary11,IvorD.Williams12,MichelKulbicki13,14
LaurentVigliola10,LaurentWantiez14,GrahamJ.Edgar15,RickD.Stuart-15
Smith15,StuartA.Sandin16,AlisonL.Green17,MariaBeger18,19,AlanM.16
Friedlander20,21,ShaunK.Wilson22,23,EranBrokovich24,AndrewJ.Brooks25,17
JuanJ.Cruz-Motta26,DavidJ.Booth27,PascaleChabanet28,MarkTupper29,18
SebastianC.A.Ferse30,31,U.RashidSumaila32,MarahJ.Hardt33,34,David19
Mouillot1,620
21
Affiliations:22
2
1AustralianResearchCouncilCentreofExcellenceforCoralReefStudies,23 JamesCookUniversity,Townsville,QLD4811Australia24 2LancasterEnvironmentCentre,LancasterUniversity,Lancaster,LA14YQ,UK25 3DepartmentofBiology,DalhousieUniversity,Halifax,NSB3H3J5Canada26 4DepartmentofGeography,UniversityofHawai‘iatManoa,Honolulu,Hawai‘i27 96822USA28 5MARBEC,Univ.Montpellier,CNRS,Ifremer,IRD,Montpellier,France.29 6WildlifeConservationSociety,GlobalMarineProgram,Bronx,NY10460USA30 7DepartmentofEnvironmentalSciences,MacquarieUniversity,NorthRyde,31 NSW2109Australia32 8ConservationInternational,CenterforOceans&BettyandGordonMoore33 CenterforScience,Honolulu,HI96825USA34 9ArizonaStateUniversity,CenterforBiodiversityOutcomes,JulieAnn35 WrigleyGlobalInstituteofSustainability,TempeAZ85281USA36 10InstitutdeRecherchepourleDéveloppement,UMRIRD-UR-CNRS37 ENTROPIE,Laboratoired’ExcellenceLABEXCORAIL,BPA5,98848Nouméa38 Cedex,NewCaledonia39 11MRAGLtd,LondonW1J5PN,UK40 12EcosystemScienceDivision,NOAAPacificIslandsFisheriesScienceCenter,41 Honolulu,HI96818USA42 13UMREntropie,LabexCorail,–IRD,UniversitédePerpignan,66000,43 Perpignan,France44 14UniversityofNewCaledonia,BPR498851Noumeacedex,NewCaledonia45 15InstituteforMarineandAntarcticStudies,UniversityofTasmania,Hobart,46 Tasmania,7001Australia47 16ScrippsInstitutionofOceanography,UniversityofCalifornia,SanDiego,La48 Jolla,CA92093USA49 17TheNatureConservancy,48MontagueRoad,SouthBrisbane,QLD.4101.50 Australia51 18SchoolofBiology,FacultyofBiologicalSciences,UniversityofLeeds,Leeds,52 WestYorkshireLS29JT,UnitedKingdom53 19CentreforBiodiversityandConservationScience,UniversityofQueensland,54 Brisbane,Queensland4072,Australia55 20NationalGeographicSociety,PristineSeasProgram,114517thStreetN.W.56 Washington,D.C.20036-4688,USA57 21FisheriesEcologyResearchLab,DepartmentofBiology,Universityof58 Hawaii,Honolulu,HI96822,USA59 22OceansInstitute,UniversityofWesternAustralia,Crawley,WA6009,60 Australia61
3
23MarineScienceProgram,DepartmentofBiodiversity,Conservationand62 Attractions,Kensington,PerthWA6151Australia63 24NaturalResourcesAdministration,MinistryofEnergy,BankIsrael764 Jerusalem,Israel65 25MarineScienceInstitute,UniversityofCalifornia,SantaBarbara,CA93106-66 6150,USA67 26DepartamentodeCienciasMarinas.,RecintoUniversitariodeMayaguez,68 UniversidaddePuertoRico,00680,PuertoRico69 27SchoolofLifeSciences,UniversityofTechnologySydney2007Australia70 28UMRENTROPIE,IRD-UniversitédeLaRéunion-CNRS,LABEXCORAIL,CS71 41095,97495SainteClotilde,LaRéunion(FR)72 29ACCORD,CentreforMaritimeandOceanStudies,TheUniversityofTrinidad73 andTobago,2ndAvenueNorth,WesternMainRoad,Chaguaramas,Trinidad74 andTobago75 30LeibnizCentreforTropicalMarineResearch(ZMT),Fahrenheitstrasse6,D-76 28359Bremen,Germany77 31DepartmentofMarineEcology,FacultyofBiologyandChemistry(FB2),78 UniversityofBremen,Bibliothekstrasse1,D-28359Bremen,Germany79 32FisheriesEconomicsResearchUnit,InstitutefortheOceansand80 Fisheries/LiuInstituteforGlobalIssues,UniversityofBritishColumbia,220281 MainMall,Vancouver,B.C.,V6T1Z4,Canada82 33FutureofFish,Bethesda,MD,20814,USA83 34OceanInkLLC,Lafayette,CO,USA84 85
*Correspondenceto:[email protected] 87
Abstract:88
The worldwide decline of coral reefs necessitates targeting management89
solutionsthatcansustainnotonlyreefsbutalsothelivelihoodsofthemillions90
ofpeoplewhodependonthem.Yetlittleisknownaboutthecontextinwhich91
different reef management tools can help to achieve multiple social and92
ecologicalgoals.Duetonon-linearitiesinthelikelihoodofachievingcombined93
4
fisheries,ecologicalfunction,andbiodiversitygoalsalongagradientofhuman94
pressure, relatively small changes in the context where management is95
implemented could have dramatic implications on whether these goals are96
likely to bemet or not. Critically,marine reserves and fisheries restrictions97
could provide substantial conservation benefits to the majority of reefs for98
fisheriesandecologicalfunction,butnotbiodiversitygoals.99
100
MainText:101
At the forefront of ongoing efforts to sustain coral reef ecosystems in the102
current period of intense social and environmental change is an increasing103
need to simultaneously manage for multiple goals, including fisheries,104
ecosystemfunctioning,andbiodiversity(1–3).Yet,criticalgapsremaininour105
capacity to effectively implement this typeof ecosystem-basedmanagement106
approach,wheremultiplegoalsaresimultaneouslypursued(4).Inparticular,107
little is known about: (i) the context under which key goals can be108
simultaneouslymet,and(ii)thedegreetowhichlocalmanagementeffortscan109
helptomeetthem.110
111
5
Here,we compiled data from~1800 tropical reef sites across 41 countries,112
states, and territories to examine the conditions under which reefs113
simultaneously support three ecological metrics reflecting key fisheries,114
ecologicalfunction,andbiodiversitygoals(Fig.1,TablesS1-2,Methods).These115
are, respectively: (1) potential stocks available for multi-species coral reef116
fisheries, calculated as the biomass of fishes >20 cm total length (Fig. 1,117
Methods, Table S2); (2) scraping potential, reflecting a unique ecological118
functionperformedbyparrotfishthatiscriticalfortheremovalofalgalbiomass119
and the provision of bare substrate for coral settlement (5, 6) (Table S2;120
Methods);and(3)thediversityofspeciestraits(i.e.homerange,bodysize,diet,121
diurnalactivity,schoolingbehavior,positioninthewatercolumn),whichcan122
underpin aspects of biodiversity such as community assembly processes,123
ecosystemproductivity,andstability(7).Wemeasuredtraitdiversityusinga124
generalization of the Shannon entropy index accounting for both the125
dissimilarityoftraitvaluespresentinareeffishcommunityandthespreadof126
biomassacrossthesetraitvalues(8)(Methods,TableS2).Ouranalysisshows127
thatthethreemetricsarenotstronglyrelatedtoeachother(r<0.54;FigS1).128
129
6
130
Figure 1| Meeting multiple goals on coral reefs. The distribution of (A)131 biomassof reef fish>20cm (innatural log, n=1798), (B)parrotfish scraping132 potential(innaturallog,n=1662),and(C)traitdiversity(n=1662),corrected133 forsampling(Methods).Differencesinthenumberofsitesarebecauseonedata134 provider collected data at the family level, which could not be used in135 calculatingparrotfish scrapingpotential or trait diversity. Parrotfisheswere136 notdetectedat31%ofourreef sites (Fig.S1). (D)Sites thatsimultaneously137 have fishbiomass, parrotfish scrapingpotential, and trait diversity at >75%138 (purple), 50-75% (dark pink), 25-50% (light pink), and <25% (black) of139 referenceconditions(Methods).Pointsarejitteredtoallowforvisualizationof140 overlappingreefsites.141 142
7
Simultaneouslymeetingmultiplegoals143
Toelucidatethecapacityofreefstosimultaneouslysupportmultiplegoals(i.e.144
fisheries,ecological function,andbiodiversity)wefirstdevelopedaseriesof145
aspirational targets (i.e. 25, 50, and 75% of reference conditions) for each146
metric to serve as benchmarks. Reference conditions (also called reference147
points)areakeyconceptinfisheriesandconservation(9,10),butarenascent148
incoralreefscience(11,12).Askeyreferenceconditions,weusedthetop10%149
value for eachmetric (corrected for sampling), but also included additional150
referenceconditions(i.e. thetop5%and20%)inthesupplementalmaterial151
(Methods).We then set aspirational targetsof25,50, and75%of reference152
conditions. For example, in fisheries, a key reference condition is unfished153
biomass, while an aspirational management target is often maximum154
sustainable yield, which is expected to be achieved at 25-50% of unfished155
biomass in multispecies fisheries (11, 13, 14). Thus, our 25% and 50% of156
referenceconditionsforpotentialfisheriesstocksapproximatethelowerand157
upper bounds of expectedmultispeciesmaximum sustainable yield; 75%of158
reference conditions reflects a more ambitious conservation target. When159
lookingattheseaspirationaltargetsacrossmultiplegoals,wefoundthatonly160
5%ofreefsitessimultaneouslyhadfishbiomass,parrotfishscraping,andtrait161
diversity at 75% of reference conditions (Fig. 1D). These sites, though162
8
reasonablyrare,weregeographicallyspread through the Indian,Pacific,and163
Atlanticoceanbasins(Fig1D).Wefoundthat12.5%ofsitessimultaneouslymet164
the50%target,and29.3%ofsitesmetthe25%target(Fig.1D)165
166
Conditionsunderwhichmultiplegoalscanbemet167
Toexaminethecontextunderwhichkeygoalscanbemet,wefirstdevelopeda168
series of Bayesian hierarchical models (15) that quantify how the three169
ecological metrics are related to key socioeconomic drivers of resource170
exploitation, while controlling for environmental conditions and sampling171
techniques (16–18) Fig. S2;Table S3;Methods).We thenused theposterior172
distributions from these models to calculate how the probability of173
simultaneously meeting multiple goals changes along a gradient of human174
pressure,whileholdingothercovariatesconstant(Fig.2,S3,S4,Methods).Our175
measureofhumanpressureinthesurroundingseascapewasadaptedfromthe176
economic geography concept of gravity (19, 20) and displayed the most177
consistent negative relationships to our response variables (Fig. S2). The178
distribution of human pressure and other key socioeconomic and179
environmental covariates among our surveyed reefs closelymatches that of180
reefs globally (Fig. S5). The probability of openly fished reef sites181
simultaneouslyhavingallthreemetricsdeclinedwithourmeasureofhuman182
9
pressureandtheambitiousnessoftheconservationtarget(Fig.2A).Inother183
words,onopenlyfishedreefsitisextremelyunlikelythatallthreegoalswillbe184
simultaneously met where human pressure is intense, but this likelihood185
increaseswhere human pressure is low, particularly for the 25% and 50%186
targets.187
188
10
189
11
Figure2|Theestimatedprobabilityofopenlyfishedreefsiteshaving25,190 50, and 75% of reference conditions (light,medium, and dark purple,191 respectively) for(A)acombinationof fishbiomass(>20cm),parrotfish192 scraping potential, trait diversity, and (B-D) eachmetric, respectively,193 along a gradient of human pressure (gravity). Separate estimates are194 providedforreefsitesinmarinereserveswherefishingisprohibited(E-H)and195 with restricted fishing (I-L). To highlight how the potential benefits of196 management change along a gradient of human pressure (gravity), we197 extracted the difference in the probability of achieving each target between198 marine reserves and openly fished sites (M-P), restricted and openly fished199 areas(Q-T),andmarinereservesandrestrictedareas(U-X). Weplottedthe200 partialeffectoftherelationshipbetweengravityandeachtargetbysettingall201 othercontinuouscovariatesto0(becausetheywereallstandardized)andall202 categorical covariates to theirmost common category (i.e. 4-10m for depth,203 slopeforhabitat,standardbelttransectforcensusmethod).Gravity(xaxis)is204 standardized,withanaverageof0.205 206
Importantly, there was considerable variability in how the probability of207
meetingindividualgoalschangedalongagradientofhumanpressure(Fig.2B-208
D). For example, it is extremely unlikely that openly fished reefs meet any209
fisheriestarget,exceptwherehumanpressureislow.Incontrast,the25and210
50%targetsmaybemetacrossabroaderspectrumofhumanpressureforthe211
parrotfishscrapingpotentialandtraitdiversitygoals(Fig.2C,D).212
213
Benefitsfrommanagement214
A critical gap remains in understanding the context inwhich different local215
management tools canhelp to simultaneouslyachievekeygoals (21–23).To216
address this, we first examine the probability of reef sites in both marine217
12
reserves(wherefishingisprohibited)andrestrictedfishingareas(wherethere218
arelimitationsonfishinggearsusedandwhocanaccessthefishinggrounds)219
inachievingkeytargetsfortheindividualandcombinedecologicalmetrics(Fig220
2E-L). We then calculated the ‘conservation gains’ from employing these221
differentformsofmanagementalongagradientofhumanpressure(22)(Fig.222
2M-X). By conservation gain, we refer to the difference in probability of223
achieving a specific target (e.g. 25% of reference condition biomass) when224
marine reserves or fishery restrictions are implemented relative to openly225
fished areas. This concept gets at the idea that contexts with maximal226
conservationgainshighlightthebestopportunitiesformanagementtohavethe227
biggest impact; conversely, implementing management in contexts with228
minimal conservation gains (either because goals are already beingmet or229
becausetheyareunlikelytobemetregardlessofmanagement)providesfew230
returnsforlimitedconservationresources(24).231
232
Critically,wefindthatbothmarinereservesandrestrictedfishingareashave233
thepotentialtoprovideconservationgains,butthecontextunderwhichthese234
gainscanbemaximizedishighlyvariabledependingonboththegoalandtarget235
(Fig.2M-X).Forsimultaneouslymeetingallthreegoals,maximalconservation236
gainsarefrommarinereservesinthelowesthumanpressurelocationsforthe237
13
most ambitious target (75%of reference conditions), but as targetsbecome238
less ambitious, conservation gains peak where human pressure is more239
intermediate(Fig.2M).Inotherwords,whentheobjectiveissimultaneously240
meetingfisheries,function,andbiodiversitygoals,moreambitioustargetscan241
bestbemetbyreservesunderlowhumanpressure,butlessambitioustargets242
are best met by placing reserves in locations where human pressure is243
intermediate. This is because the difference between marine reserves and244
openlyfishedreefsishighestatintermediatehumanpressureforthe25and245
50% targets. For all three targets, there are minimal conservation gains in246
locations where human pressure is most intense, which means that in this247
context,managementisunlikelytohelpmeetthesegoals.Foreachindependent248
goal, the context under which conservation gains can be maximized varies249
considerably (Fig2).Of note is that trait diversity is the least responsive to250
management,withconservationgainsneverreachingabove0.4.251
252
Wethensimulatedhowthenumberofouropenly fishedsitesachievingkey253
conservation targets would change if a marine reserve (Fig. 3) or fisheries254
restrictions (Fig S6)were implemented, given the existing conditions at our255
reefsites.Ouranalysisrevealsbothkeyopportunitiesandconstraints inthe256
capacityforlocalmanagementtosimultaneouslymeetmultiplegoals.Onone257
14
hand,formorethan50%ofourfishedsites,theimplementationofamarine258
reserveispredictedtohelpachievemultiplegoals(Fig.3A).Ontheotherhand,259
less than 1% of the sites starting below 25% of reference conditions are260
predictedtoachievethe75%ofreferenceconditionstarget,highlightinghow261
the broader seascape contextmay stunt reservepotential in degraded reefs262
(22).Indeed,morethanhalfofthe87.4%ofopenlyfishedreefsstartingbelow263
25%ofreferenceconditionsarepredictedtoremaininthethatsamecategory264
(Fig3A).Additionally,ouranalysisshowsthatevenwherefishablebiomassis265
very low, scraping potential and trait diversity are often >25%of reference266
conditions(Fig.3B-D);afindingsupportedbypreviousresearchshowingthat267
herbivoresandadiversityoftraitscanstillpersistondegradedreefs(25).268
269
270
15
Fig.3.Conservationtargetoutcomesfromsimulatingtheimplementation271 ofmarinereservesinopenlyfishedsites.Alluvialplotsshowthechangein272 the number of sites expected to achieve key conservation targets if marine273 reserveswereimplementedinouropenlyfishedsitesfor(A)simultaneously274 meetingfishbiomass,parrotfishscrapingpotential,andtraitdiversity,and(B-275 D)eachgoal,respectively.276 277
Thegoalofmarinereservesistoprohibitfishing,whichcanbeindirectconflict278
withachievingcertainfisheriesgoals(26).Wefoundthatfisheriesrestrictions279
provideasimilarpattern,buttypicallylowermagnitude,ofconservationgains280
than marine reserves, particularly for achieving the combined goal and281
fisheriesgoal(Fig2Q-X,FigS6).Ofnoteisthatforparrotfishgrazingpotential,282
fishing restrictionsprovide the sameconservationgainsasmarine reserves,283
providingmultiplewaystoachievethatspecificgoal(Fig.2W).284
285
Together,ourfindingsprovideguidanceonwhatcanberealisticallyachieved286
with various forms of local management regarding key fisheries, ecological287
function,andbiodiversitygoalsoncoralreefs.Wehighlightkeyprosandcons288
of placing management in different areas by demonstrating how potential289
conservationgainsvarynotonlybygoal,butalsoarestronglydependenton290
both the ambitiousness of the target and the context (Fig. 2, S3, S4). In291
particular, the potential for local management to help in meeting goals is292
stronglyrelatedtotheamountofhumanpressureinthesurroundingseascape293
16
(Fig.2,S2).Akeyfindingisthatconservationgainstendtochangenon-linearly294
withhumanpressure,whichmeansthatrelativelysmallchangesinthecontext295
where management is implemented could have dramatic implications on296
whetherkeygoalsare likely tobemetornot (Fig.2M-X).Thisnotonlyhas297
importantimplicationsfortheplacementofnewmarinereserves,butisalso298
relevant to how future socioeconomic changes, such as infrastructure299
development and population growth may impact the efficacy of reef300
conservation. However, the impacts of these changes could potentially be301
buffered bymakingmanagementmore effective, for example, by leveraging302
insightsaboutusingsocialnormsandcognitivebiasestoimprovecompliance303
(22,27)andlearninglessonsaboutkeypracticesandprocessesfromlocations304
thathavedefiedexpectationsof global reefdegradation (28,29).Our global305
analysis makes clear the limitations of local management, especially in306
promoting certain aspects of biodiversity like trait diversity. While307
internationalactiononclimatechangewillbecrucialtoensuringafuturefor308
coral-dominated reefs (1, 2), good governance that promotes effective309
management will also be critical to sustaining reefs and the millions of310
livelihoodsthatdependonthem. 311
17
References312
1. T. P. Hughes, M. L. Barnes, D. R. Bellwood, J. E. Cinner, G. S. Cumming, J. B. C. 313 Jackson, J. Kleypas, I. A. van de Leemput, J. M. Lough, T. H. Morrison, S. R. Palumbi, E. 314 H. van Nes, M. Scheffer, Coral reefs in the Anthropocene. Nature. 546, 82–90 (2017). 315
2. A. V Norström, M. Nyström, J. Jouffray, C. Folke, N. A. J. Graham, F. Moberg, P. 316 Olsson, G. J. Williams, Guiding coral reef futures in the Anthropocene. Front. Ecol. 317 Environ. 14, 490–498 (2016). 318
3. S. Zhou, A. D. M. Smith, A. E. Punt, A. J. Richardson, M. Gibbs, E. A. Fulton, S. Pascoe, 319 C. Bulman, P. Bayliss, K. Sainsbury, Ecosystem-based fisheries management requires a 320 change to the selective fishing philosophy. Proc. Natl. Acad. Sci. 107, 9485–9489 (2010). 321
4. E. K. Pikitch, C. Santora, E. A. Babcock, A. Bakun, R. Bonfil, D. O. Conover, P. Dayton, 322 P. Doukakis, D. Fluharty, B. Heneman, E. D. Houde, J. Link, P. A. Livingston, M. 323 Mangel, M. K. McAllister, J. Pope, K. J. Sainsbury, Ecosystem-based fishery 324 management. Science (80-. ). (2004), , doi:10.1126/science.1098222. 325
5. D. R. Bellwood, T. P. Hughes, C. Folke, M. Nyström, Confronting the coral reef crisis. 326 Nature. 429, 827–833 (2004). 327
6. Y.-M. Bozec, S. O’Farrell, J. H. Bruggemann, B. E. Luckhurst, P. J. Mumby, Tradeoffs 328 between fisheries harvest and the resilience of coral reefs. Proc. Natl. Acad. Sci. 113, 329 4536–4541 (2016). 330
7. J. E. Duffy, J. S. Lefcheck, R. D. Stuart-Smith, S. A. Navarrete, G. J. Edgar, Biodiversity 331 enhances reef fish biomass and resistance to climate change. Proc. Natl. Acad. Sci. 113, 332 6230–6235 (2016). 333
8. A. Chao, C. H. Chiu, S. Villéger, I. F. Sun, S. Thorn, Y. C. Lin, J. M. Chiang, W. B. 334 Sherwin, An attribute-diversity approach to functional diversity, functional beta diversity, 335 and related (dis)similarity measures. Ecol. Monogr. (2019), doi:10.1002/ecm.1343. 336
9. M. A. MacNeil, The politics, science and policy of reference points for resource 337 management. Environ. Conserv. 40, 297–301 (2013). 338
10. G. M. MACE, R. LANDE, Assessing Extinction Threats: Toward a Reevaluation of 339 IUCN Threatened Species Categories. Conserv. Biol. (1991), , doi:10.1111/j.1523-340 1739.1991.tb00119.x. 341
11. T. R. McClanahan, N. A. J. Graham, M. A. MacNeil, N. A. Muthiga, J. E. Cinner, J. H. 342 Bruggemann, S. K. Wilson, Critical thresholds and tangible targets for ecosystem-based 343 management of coral reef fisheries. Proc. Natl. Acad. Sci. 108, 17230–17233 (2011). 344
12. T. R. McClanahan, R. E. Schroeder, A. M. Friedlander, L. Vigliola, L. Wantiez, J. E. 345 Caselle, N. A. J. Graham, S. K. Wilson, G. J. Edgar, R. D. Stuart-Smith, R. M. Oddenyo, 346 J. E. Cinner, Global baselines and benchmarks for fish biomass: Comparing remote reefs 347 and fisheries closures. Mar. Ecol. Prog. Ser. 612, 167–192 (2019). 348
13. M. B. Schaefer, Some aspects of the dynamics of populations important to the 349 management of the commercial marine fisheries. Inter-American Trop. Tuna Comm. Bull. 350 1, 23–56 (1954). 351
14. B. Worm, R. Hilborn, J. K. Baum, T. A. Branch, J. S. Collie, C. Costello, M. J. Fogarty, 352 E. A. Fulton, J. A. Hutchings, S. Jennings, O. P. Jensen, H. K. Lotze, P. M. Mace, T. R. 353 McClanahan, C. Minto, S. R. Palumbi, A. M. Parma, D. Ricard, A. A. Rosenberg, R. 354 Watson, D. Zeller, Rebuilding Global Fisheries. Science (80-. ). 325, 578–585 (2009). 355
15. A. Gelman, J. Hill, Data analysis using regression and multilevel/hierarchical models 356 (Cambridge university press, 2007). 357
18
16. H. J. Geist, E. F. Lambin, Proximate Causes and Underlying Driving Forces of Tropical 358 DeforestationTropical forests are disappearing as the result of many pressures, both local 359 and regional, acting in various combinations in different geographical locations. 360 Bioscience. 52, 143–150 (2002). 361
17. R. York, E. A. Rosa, T. Dietz, Footprints on the earth: The environmental consequences 362 of modernity. Am. Sociol. Rev. 68, 279–300 (2003). 363
18. J. E. Cinner, C. Huchery, M. A. MacNeil, N. A. J. Graham, T. R. McClanahan, J. Maina, 364 E. Maire, J. N. Kittinger, C. C. Hicks, C. Mora, E. H. Allison, S. D’Agata, A. S. Hoey, D. 365 A. Feary, L. Crowder, I. D. Williams, M. Kulbicki, L. Vigliola, L. Wantiez, G. J. Edgar, 366 R. D. Stuart-Smith, S. A. Sandin, A. L. Green, M. J. Hardt, M. Beger, A. M. Friedlander, 367 S. J. Campbell, K. E. Holmes, S. K. Wilson, E. Brokovich, A. J. Brooks, J. J. Cruz-Motta, 368 D. J. Booth, P. Chabanet, C. Gough, M. Tupper, S. C. A. Ferse, U. R. Sumaila, D. 369 Mouillot, Bright spots among the world’s coral reefs. Nature. 535, 416–419 (2016). 370
19. J. E. Anderson, The Gravity Model. Annu. Rev. Econom. (2011), doi:10.1146/annurev-371 economics-111809-125114. 372
20. F. Lukermann, P. W. Porter, Gravity and Potential Models in Economic Geography. Ann. 373 Assoc. Am. Geogr. 50, 493–504 (1960). 374
21. R. S. Steneck, P. J. Mumby, C. MacDonald, D. B. Rasher, G. Stoyle, Attenuating effects 375 of ecosystem management on coral reefs. Sci. Adv. 4, eaao5493 (2018). 376
22. J. E. Cinner, E. Maire, C. Huchery, M. A. MacNeil, N. A. J. Graham, C. Mora, T. R. 377 McClanahan, M. L. Barnes, J. N. Kittinger, C. C. Hicks, S. D’Agata, A. S. Hoey, G. G. 378 Gurney, D. A. Feary, I. D. Williams, M. Kulbicki, L. Vigliola, L. Wantiez, G. J. Edgar, R. 379 D. Stuart-Smith, S. A. Sandin, A. L. Green, M. J. Hardt, M. Beger, A. M. Friedlander, S. 380 K. Wilson, E. Brokovich, A. J. Brooks, J. J. Cruz-Motta, D. J. Booth, P. Chabanet, C. 381 Gough, M. Tupper, S. C. A. Ferse, U. R. Sumaila, S. Pardede, D. Mouillot, Gravity of 382 human impacts mediates coral reef conservation gains. Proc. Natl. Acad. Sci. 115, E6116–383 E6125 (2018). 384
23. D. A. Gill, M. B. Mascia, G. N. Ahmadia, L. Glew, S. E. Lester, M. Barnes, I. Craigie, E. 385 S. Darling, C. M. Free, J. Geldmann, S. Holst, O. P. Jensen, A. T. White, X. Basurto, L. 386 Coad, R. D. Gates, G. Guannel, P. J. Mumby, H. Thomas, S. Whitmee, S. Woodley, H. E. 387 Fox, Capacity shortfalls hinder the performance of marine protected areas globally. 388 Nature (2017), doi:10.1038/nature21708. 389
24. P. J. Ferraro, S. K. Pattanayak, Money for nothing? A call for empirical evaluation of 390 biodiversity conservation investments. PLoS Biol. (2006), , 391 doi:10.1371/journal.pbio.0040105. 392
25. J. P. W. Robinson, S. K. Wilson, J. Robinson, C. Gerry, J. Lucas, C. Assan, R. Govinden, 393 S. Jennings, N. A. J. Graham, Productive instability of coral reef fisheries after climate-394 driven regime shifts. Nat. Ecol. Evol. (2019), doi:10.1038/s41559-018-0715-z. 395
26. R. Hilborn, K. Stokes, J. J. Maguire, T. Smith, L. W. Botsford, M. Mangel, J. Orensanz, 396 A. Parma, J. Rice, J. Bell, K. L. Cochrane, S. Garcia, S. J. Hall, G. P. Kirkwood, K. 397 Sainsbury, G. Stefansson, C. Walters, When can marine reserves improve fisheries 398 management? Ocean Coast. Manag. (2004), doi:10.1016/j.ocecoaman.2004.04.001. 399
27. B. J. Bergseth, M. Roscher, Discerning the culture of compliance through recreational 400 fisher’s perceptions of poaching. Mar. Policy (2018), doi:10.1016/j.marpol.2017.12.022. 401
28. G. Post, J. Geldmann, Exceptional responders in conservation. Conserv. Biol. (2018), 402 doi:10.1111/cobi.13006. 403
19
29. J. E. Cinner, C. Huchery, M. A. MacNeil, N. A. J. Graham, T. R. McClanahan, J. Maina, 404 E. Maire, J. N. Kittinger, C. C. Hicks, C. Mora, E. H. Allison, S. D’Agata, A. S. Hoey, D. 405 A. Feary, L. Crowder, I. D. Williams, M. Kulbicki, L. Vigliola, L. Wantiez, G. J. Edgar, 406 R. D. Stuart-Smith, S. A. Sandin, A. L. Green, M. J. Hardt, M. Beger, A. M. Friedlander, 407 S. J. Campbell, K. E. Holmes, S. K. Wilson, E. Brokovich, A. J. Brooks, J. J. Cruz-Motta, 408 D. J. Booth, P. Chabanet, C. Gough, M. Tupper, S. C. A. Ferse, U. R. Sumaila, D. 409 Mouillot, Bright spots among the world’s coral reefs. Nature. 535, 416–419 (2016). 410
30. T. M. Daw, J. Maina, J. E. Cinner, J. Robinson, A. Wamukota, C. Gerry, C. Abunge, P. 411 Thoya, K. E. Abernethy, M. Cedras, The spatial behaviour of artisanal fishers: 412 Implications for fisheries management and development (Fishers in Space). Final Report 413 (2011). 414
31. R. Froese, D. Pauly, FishBase (2015), (available at http://www.fishbase.org). 415 32. A. S. Hoey, D. R. Bellwood, Limited Functional Redundancy in a High Diversity System: 416
Single Species Dominates Key Ecological Process on Coral Reefs. Ecosystems. 12, 1316–417 1328 (2009). 418
33. R. M. Bonaldo, A. S. Hoey, D. R. Bellwood, Oceanography and Marine Biology (CRC 419 Press, 2014; https://www.taylorfrancis.com/books/9781482220667), vol. 52. 420
34. R. M. Bonaldo, D. R. Bellwood, Size-dependent variation in the functional role of the 421 parrotfish Scarus rivulatus on the Great Barrier Reef, Australia. Mar. Ecol. Prog. Ser. 360, 422 237–244 (2008). 423
35. L. Ong, K. N. Holland, Bioerosion of coral reefs by two Hawaiian parrotfishes: species, 424 size differences and fishery implications. Mar. Biol. 157, 1313–1323 (2010). 425
36. J. H. Bruggemann, A. M. Van Kessel, J. M. Van Rooij, A. M. Breeman, Bioerosion and 426 sediment ingestion by the caribbean parrotfish Scarus vetula and Sparisoma viride: 427 Implications of fish size, feeding mode and habitat use. Mar. Ecol. Prog. Ser. 134, 59–71 428 (1996). 429
37. D. Mouillot, S. Villéger, V. Parravicini, M. Kulbicki, J. E. Arias-González, M. Bender, P. 430 Chabanet, S. R. Floeter, A. M. Friedlander, L. Vigliola, D. R. Bellwood, Functional over-431 redundancy and high functional vulnerability in global fish faunas on tropical reefs. Proc. 432 Natl. Acad. Sci. 111, 13757–13762 (2014). 433
38. C. Mora, ECOLOGY: Enhanced: Coral Reefs and the Global Network of Marine 434 Protected Areas. Science (80-. ). 312, 1750–1751 (2006). 435
39. U. N. F. + A. P.-F. A. O. & C. I. de A. T.-C. (NASA S. D. and A. C. (SEDAC) Center for 436 International Earth Science Information Network - CIESIN - Columbia University, 437 “Gridded Population of the World, Version 4 (GPWv4): Population Count, Revision 11” 438 (Palisades, NY, 2015), , doi:10.7927/H4JW8BX5. 439
40. M. A. MacNeil, S. R. Connolly, 12 Multi-scale patterns and processes in reef fish 440 abundance. Ecol. fishes coral reefs, 116–126 (2015). 441
41. J. E. Anderson, A theoretical foundation for the gravity equation. Am. Econ. Rev. 69, 106–442 116 (1979). 443
42. E. G. Ravenstein, The Laws of Migration. J. Stat. Soc. London. 48, 167 (1885). 444 43. E. A. Bright, P. R. Coleman, A. N. Rose, M. L. Urban, LandScan. Oak Ridge Natl. Lab. 445
(2012). 446 44. E. Maire, J. Cinner, L. Velez, C. Huchery, C. Mora, S. Dagata, L. Vigliola, L. Wantiez, 447
M. Kulbicki, D. Mouillot, How accessible are coral reefs to people? A global assessment 448 based on travel time. Ecol. Lett. (2016), doi:10.1111/ele.12577. 449
20
45. D. Pauly, D. Zeller, Sea Around Us Concepts, Design and Data (2015), (available at 450 http://www.seaaroundus.org). 451
46. W. C. UNEP-WCMC WRI, TNC, Global distribution of warm-water coral reefs, 452 compiled from multiple sources including the Millennium Coral Reef Mapping Project. 453 Version 3.0. Incl. Contrib. from IMaRS-USF IRD (2005), IMaRS-USF Spalding al. 454 (2001). Cambridge UN Environ. World Conserv. Monit. Centre. (2010), (available at 455 http://data.unep-wcmc.org/datasets/1). 456
47. J. M. Gove, G. J. Williams, M. A. McManus, S. F. Heron, S. A. Sandin, O. J. Vetter, D. 457 G. Foley, Quantifying Climatological Ranges and Anomalies for Pacific Coral Reef 458 Ecosystems. PLoS One. 8, e61974 (2013). 459
48. J. Maina, T. R. McClanahan, V. Venus, M. Ateweberhan, J. Madin, Global Gradients of 460 Coral Exposure to Environmental Stresses and Implications for Local Management. PLoS 461 One. 6, e23064 (2011). 462
49. W. Bank, The World Bank (2015), (available at http://www.govindicators.org). 463 50. P. C. Bürkner, brms: An R package for Bayesian multilevel models using Stan. J. Stat. 464
Softw. (2017), doi:10.18637/jss.v080.i01. 465 51. A. Vehtari, A. Gelman, J. Gabry, Practical Bayesian model evaluation using leave-one-out 466
cross-validation and WAIC. Stat. Comput. 27, 1433 (2017). 467 52. N. C. Krueck, G. N. Ahmadia, A. Green, G. P. Jones, H. P. Possingham, C. Riginos, E. A. 468
Treml, P. J. Mumby, Incorporating larval dispersal into MPA design for both conservation 469 and fisheries. Ecol. Appl. 27, 925–941 (2017). 470
53. S. M. W. Reddy, A. Wentz, O. Aburto-Oropeza, M. Maxey, S. Nagavarapu, H. M. Leslie, 471 Evidence of market‐driven size‐selective fishing and the mediating effects of biological 472 and institutional factors. Ecol. Appl. 23, 726–741 (2013). 473
54. N. K. Dulvy, R. P. Freckleton, N. V. C. Polunin, Coral reef cascades and the indirect 474 effects of predator removal by exploitation. Ecol. Lett. 7, 410–416 (2004). 475
55. D. R. Bellwood, A. S. Hoey, T. P. Hughes, Human activity selectively impacts the 476 ecosystem roles of parrotfishes on coral reefs. Proc. R. Soc. B Biol. Sci. 279, 1621–1629 477 (2012). 478
56. G. J. Edgar, R. D. Stuart-Smith, T. J. Willis, S. J. Kininmonth, S. C. Baker, S. Banks, N. 479 S. Barrett, M. A. Becerro, A. T. F. Bernard, J. Berkhout, C. D. Buxton, S. J. Campbell, A. 480 T. Cooper, M. Davey, S. C. Edgar, G. Försterra, D. E. Galván, A. J. Irigoyen, D. J. 481 Kushner, R. Moura, P. E. Parnell, N. T. Shears, G. A. Soler, E. M. A. Strain, R. J. 482 Thomson, Global conservation outcomes depend on marine protected areas with five key 483 features. Nature. 506, 216–220 (2014). 484
57. T. P. Hughes, M. J. Rodrigues, D. R. Bellwood, D. Ceccarelli, O. Hoegh-Guldberg, L. 485 McCook, N. Moltschaniwskyj, M. S. Pratchett, R. S. Steneck, B. Willis, Phase Shifts, 486 Herbivory, and the Resilience of Coral Reefs to Climate Change. Curr. Biol. 17, 360–365 487 (2007). 488
58. R. D. Stuart-Smith, A. E. Bates, J. S. Lefcheck, J. E. Duffy, S. C. Baker, R. J. Thomson, J. 489 F. Stuart-Smith, N. A. Hill, S. J. Kininmonth, L. Airoldi, M. A. Becerro, S. J. Campbell, 490 T. P. Dawson, S. A. Navarrete, G. A. Soler, E. M. A. Strain, T. J. Willis, G. J. Edgar, 491 Integrating abundance and functional traits reveals new global hotspots of fish diversity. 492 Nature. 501, 539–542 (2013). 493
59. G. D. Tilman, The Influence of Functional Diversity and Composition on Ecosystem 494 Processes. Science (80-. ). 277, 1300–1302 (1997). 495
21
60. N. Gross, Y. Le Bagousse-Pinguet, P. Liancourt, M. Berdugo, N. J. Gotelli, F. T. Maestre, 496 Functional trait diversity maximizes ecosystem multifunctionality. Nat. Ecol. Evol. 1, 132 497 (2017). 498
61. S. Dı́az, M. Cabido, Vive la différence: plant functional diversity matters to ecosystem 499 processes. Trends Ecol. Evol. 16, 646–655 (2001). 500
62. S. D’Agata, D. Mouillot, L. Wantiez, A. M. Friedlander, M. Kulbicki, L. Vigliola, Marine 501 reserves lag behind wilderness in the conservation of key functional roles. Nat. Commun. 502 7, 12000 (2016). 503
63. D. Mouillot, N. A. J. Graham, S. Villéger, N. W. H. Mason, D. R. Bellwood, A functional 504 approach reveals community responses to disturbances. Trends Ecol. Evol. 28, 167–177 505 (2013). 506
507 508 509 Acknowledgments510
General:TheARCCentreofExcellenceforCoralReefStudies,J.E.C.’sPewFellowshipin511
MarineConservationandARCFutureFellowship,andUniversityofMontpellierfunded512
workinggroupmeetings.ThankstoM.Hardt,S.Pardede,andBlueVenturesfordata513
contributions.ThisisaSocial-EcologicalResearchFrontiers(SERF)workinggroup514
contribution.515
516
Funding:JECissupportedbytheAustralianResearchCouncil(CE140100020,517
FT160100047,DP110101540,DP0877905),thePewCharitableTrust,thePaulM.Angell518
FamilyFoundation,andtheWorldFishFISHCRPproject.NAJGissupportedthrougha519
RoyalSocietyUniversityResearchFellowship(UF140691).520
521
Authorcontributions:J.E.C.conceivedofthestudywithsupportfromD.M,C.M,E.M.,522
N.A.J.G,T.R.M,J.K,C.H,M.L.B.,M.A.M,andC.C.H;JZM,G.G.,J.E.C.,D.M.,andE.Mdeveloped523
andimplementedtheanalyses;J.E.C.ledthemanuscript.Allotherauthorscontributedto524
datacollectionandmadesubstantivecontributionstothetext.525
22
526
527
Competinginterests:none528
529
Dataandmaterialsavailability:datawillbemadeavailableviaalinktotheJamesCook530
UniversityTropicalDataHubandthecodeusedforthispaperwillavailablefromGitHub531
532
Supplementalmaterial533
Methods534
Scalesofdata535
Ourdatawereorganizedat four spatial scales: survey (n=4399), reef site (n=1797), reef536
cluster(n=734),andnation/state(n=41).537
i) surveyswereoursmallestscaleofdata–seedetailsaboutsurveymethodsbelow.538
ii) reefsiteswereaggregationsofreplicatesurveyswithina fewhundredmeters.539
Therewereanaverageof2.4replicatesurveysperreefsite.540
iii) reefclusters-Weclusteredreefsitestogetherthatwerewithin4kmofeachother,541
and used the centroid of these reef clusters to estimate certain social and542
environmentalcovariates(TableS3).Tomakereefclusters,wefirstestimatedthe543
lineardistancebetweenallreefsites,thenusedahierarchicalanalysiswiththe544
complete-linkageclusteringtechniquebasedonthemaximumdistancebetween545
reef sites.We set the cut-off at 4 km to selectmutually exclusive reef clusters546
where reef sites cannot be more distant than 4 km. The choice of 4km was547
informedbya3-yearstudyof thespatialmovementpatternsofartisanalcoral548
23
reef fishers, corresponding to the highest density of fishing activities on reefs549
basedonGPS-derivedeffortdensitymapsofartisanalcoralreeffishingactivities550
(30).ThisclusteringanalysiswascarriedoutusingtheRfunctions ‘hclust’and551
‘cutree’,resultinginanaverageof2.7reefsites/reefcluster.552
iv) Nation/state (nation, state, or territory). A larger scale in our analysis was553
‘nation/state’, which are jurisdictions that generally correspond to individual554
nations(butcouldalsoincludestates,territories,overseasregions,orextremely555
remote areas within a state such as the Hawaii or the British Indian Ocean556
Territory; Table S1),withinwhich reef clusters and reef siteswere nested for557
analysis.558
559
Reeffishsurveymethods560
Estimateswere based on instantaneous visual counts from 4399 surveys collected from561
1798tropicalreefsites(i.e.,within23.5latitudedegrees).Allsurveysusedstandardbelt-562
transects,distancesampling,orpoint-counts,andwereconductedbetween2004and2013.563
Foreachsite,habitattype(i.e.,slope,crest,flat,lagoon/backreef),depthrange(i.e.,0-4m,4-564
10mand>10m)and totalsamplingareawererecorded.Wheredata frommultipleyears565
wereavailable forasinglereefsite,we includedonlydata fromtheyearclosestto2010.566
Withineachsurveyarea,reef-associatedfisheswereidentifiedtospecieslevel,abundance567
counted, and total length (TL) estimated, with the exception of one data provider who568
measuredbiomassatthefamilylevel.Aspartofourstandardizationprocess,we:569
24
i) Retainedfamiliesthatwereconsistentlystudiedandwereaboveaminimumsize570
cut-off.Thus,weretainedcountsof>10cmnon-crypticreeffishesfromfamilies571
thatareresidentonthereef(TableS4).572
ii) Directly accounted for depth, survey method, survey area, and habitat as573
covariatesinthemodel.574
575
Keyecologicalmetrics576
Wethenusedthesesurveystocalculatethreekeyreeffishecologicalmetrics:577
i)Biomassofreeffishabove20cm.Wecalculatedtotalbiomassoffishabove20cm578
(TL)oneachreefsite(n=1798)usingstandardpublishedspecies-levellength-weight579
relationship parameters or those available on FishBase (31).When length-weight580
relationshipparameterswerenotavailableforaspecies,weusedtheparametersfor581
acloselyrelatedspeciesorgenus.IncludedfamiliesarespecifiedinTableS4.582
ii). Parrotfish Scraping Potential. Scraping rates (area grazed per minute) for583
parrotfishesateachreefsite(n=1662)werecalculatedastheproductofparrotfish584
fishdensity,feedingrate,andbitedimension(area)(32).Size-specificfeedingrates585
werederivedfrombest-fitregressionsofbiterate(bitesmin-1)andfishlength([TL],586
cm) for each species or closely related congener. Bite rates for Indo-Pacific587
parrotfishes were quantified at three locations (Great Barrier Reef, Australia;588
Indonesia;andtheRedSea)duringwhichTLwasestimatedandthenumberofbites589
on different benthic substrata (primarily epilithic algal matrix and live corals)590
recordedandconvertedtobitesmin-1.Individualfishwerefollowedforaminimum591
of 3-minutes and 19-126 individuals (mean = 41 individuals) were observed per592
25
species. These values were supplemented with published length-feeding rate593
relationships,includingforAtlanticparrotfishes(reviewedin(33)).Size-specificbite594
dimensions(mm2)weretakenfromtheliterature(32,34–36).595
iii).TraitDiversity.Traitdiversitywascomputedforeachreefsite,consideredasa596
localfishcommunity(n=1662).First,weusedthetraitdatabaseontropicalreeffishes597
fromMouillotetal.(37)todescribespeciestraits.Thesixtraitsconsideredwere:(1)598
size(observedlengthofeachindividualfish)codedusing5orderedcategories:10-599
15 cm, 15.1-30 cm, 30.1-50 cm, 50.1-80 cm, >80 cm; (2) mobility coded using 3600
orderedcategories:sedentary,mobilewithinareef,andmobilebetweenreefs;(3)601
period of activity coded using 3 ordered categories: diurnal, both diurnal and602
nocturnal,andnocturnal; (4)schoolingcodedusing5orderedcategories:solitary,603
paired, or living in small (3-20 individuals),medium (20-50 individuals), or large604
groups(>50groups);(5)verticalpositioninthewatercolumncodedusing3ordered605
categories: benthic, bentho-pelagic, and pelagic; (6) diet coded using 7 trophic606
categories: herbivorous-detritivorous, macro-algal herbivorous, invertivorous607
targeting sessile invertebrates, invertivorous targeting mobile invertebrates,608
planktivorous,piscivorous,andomnivorous,(i.e.fishesthatfeedonbothvegetaland609
animalmaterial).Sincealltraitswerecategorical,specieswithidenticaltraitswere610
grouped into entities.We then computed theGowerdistancebetweenall pairs of611
entities.Finally,foreachfishcommunitywecomputedtrait-diversityusingtheChao’s612
FDq=1index(Chaoetal2019):613
614
26
𝐹𝐷#$% = exp +−-𝑝/ × 𝑙𝑜𝑔 +1 −-1−min(𝑑/:,𝑚𝐷)
𝑚𝐷 × 𝑝:/>:
?@
/$%
?615
616
wherepiandpjaretherespectiverelativebiomassesofthetwoentitiesiandjinthe617
community,dijistheGowerdistancebetweenentitiesiandj,mDistheaverageofall618
Gowerdistancesbetweentheentitiespresentintheglobalpoolofspecies.Thisindex619
is expressed as an equivalent numbers of species (Chao et al 2019). Hence, it is620
minimalandequals1whenallbiomassissupportedbythesameentity(i.e.whenone621
species isultra-dominantorwhenall specieshave thesame traitvalues)and it is622
maximalandequalsthenumberofspecieswhenallspeciespairshavedissimilarities623
higherthantheaveragedissimilarityintheglobalspeciespoolandequalbiomasses.624
625
Weusedspecies-leveldatatocalculateparrotfishscrapingpotentialandtraitdiversity.Thus,626
data from the one providerwho only recorded family level datawere not used in those627
responsevariables.628
629
Socialandenvironmentalpotentialdrivers630
1.Management:Foreachreefsite,wedeterminedifitwas:i)unfished-whetheritfellwithin631
thebordersofahighcomplianceno-takemarinereserve;ii)restricted-whethertherewere632
activerestrictionsongears(e.g.bansontheuseofnets,spearguns,ortraps)orfishingeffort633
(which could have included areas under customary tenure, where ‘outsiders’ were634
effectivelyexcluded,aswellasinsidemarineparksthatwerenotnecessarilynotake);oriii)635
openly fished - regularly fished without effective restrictions. To determine these636
27
classifications,weusedtheexpertopinionofthedataproviders,andtriangulatedthiswith637
a global database of marine reserve boundaries (38). As a sensitivity analysis, we also638
conductedanalyseswithasubsetofreservesthatwere>2km2andthathavebeenprotected639
formorethan4years(seeanalysissectionbelow).640
641
2.LocalPopulationGrowth:Wecreateda100kmbufferaroundeachreefclusterandused642
this to calculate human population within the buffer in 2000 and 2010 based on the643
Socioeconomic Data and Application Centre (SEDAC) gridded population of the world644
database(39).Populationgrowthwastheproportionaldifferencebetweenthepopulation645
in2000and2010.Wechosea100kmbufferasareasonablerangeatwhichmanykeyhuman646
impactsfrompopulation(e.g.,land-useandnutrients)mightaffectreefs(40).647
648
3.Gravity:Weadaptedtheeconomicgeographyconceptofgravity(19,22,41,42)toexamine649
theamountofhumanpressurewithinthesurrounding500kmofareef.Thegravitymodel650
hasbeenusedbyeconomistsandgeographerssincethe1880stomeasureawiderangeof651
economic interactions such as trade and migration flows (19). To calculate gravity, we652
gathereddataonbothpopulationestimatesandasurrogatefordistance:traveltime.653
654
Populationestimations655
Wegatheredpopulationestimatesforevery1-by-1kmpopulatedcellwithina500km656
radiusofeachreefsiteusingtheLandScanTM2011database(43).Wechosea500km657
radiusfromthenearestsettlementasthemaximumdistanceanynon-marketfishing658
activitiesforfreshreeffisharelikelytooccur.659
28
660
Traveltimecalculation661
Foreachpopulatedcellwithin500km,wethenusedacost-distancealgorithmthat662
computestheleast‘cost’(inminutes)oftravellingtothereefsite.Costwasbasedon663
arastergridoflandcover,roadnetworks,andshorelinesdataandestimatedtravel664
timeoverdifferentsurfaces(44).665
666
Gravitycomputation667
Wefirstcalculatedavalueforthe“gravitationalpull”exertedbyeachpopulatedcell668
within500kmofa reefsite,bydividing thepopulationof thatcellby thesquared669
travel timetothereefsite.Wethensummedthegravityvalues forallcellswithin670
500kmofeachreefsitetomeasurethetotal“gravitationalpull”ofhumanpressure671
thatagivenreefisexperiencing.Thisapplicationofthegravityconceptinfersthat672
potentialinteractionsincreasewithpopulationsize,butdecaynon-linearlywiththe673
effectivedistance.Althoughdifferentexponentscanbeused,weusedthetraditional674
application of dividing by squared distance (in our case travel time)(19). This675
application emphasizes a non-linear decay in the propensity for interactions as676
distancefrompeopletothereefincreases.Ourrationaleforcalculatinggravityusing677
squaredtraveltimeinthedenominator(asopposedtojusttraveltime)isbasedon678
theideathatourreefsiteislikelyonlyoneofmultiplereefsthatcouldpotentiallybe679
harvested,andthatthenumberofpotentialalternativereefsthatcouldbeharvested680
shouldincreasewiththeareacoveredbyaradiusfromanypopulatedcell(i.e.,based681
on area not linear distance). Since the decision to fish on a given reef is likely682
29
dependentonhowthatreefcompareswithallotheralternatives,itmakessensethat683
fishingpressureatanyreefsitewillalsodeclinebydistancesquared(i.e.comparing684
with all other reefs within a similar distance) rather than linear distance (i.e.685
comparingonlywithotherreefsalongthesamepath).Totestwhetherthisrationale686
tousesquaredtraveltimeissupportedbyourdata,wedevelopedgravitymetrics687
usingarangeofexponents(^1,^2,^3)andusedleave-one-outcross-validationfor688
modelselectiontodeterminethebestfit.Forfishbiomassandtraitdiversity,squared689
traveltimeperformedbest,butforparrotfishgrazing,traveltime(i.e.exponent1)690
performed slightly better (though in the parrotfish grazing models, all three691
exponents were within the standard error). Given that two out of three of our692
responsevariablesfavoredthetraveltimesquared,thissupportsourdecisiontouse693
that for our analysis. However, due to the potential ambiguity in the parrotfish694
grazing potential, we ran a sensitivity test, calculating how the probabilities of695
achievinggoals changealongagradientofhumanpressureusingagravitymetric696
calculatedusingthefirstexponent(i.e.traveltimeinthedenominator).Therewere697
nodiscernibledifferencesbetweenourresults,suggestingthatourdecisiontouse698
travel time squared as opposed to travel time in the denominator did not699
meaningfullyimpactourresults.700
701
4. Human Development Index (HDI): HDI is a summarymeasure of human development702
encompassing: life expectancy, education, and per capita income. We obtained the HDI703
measurefromtheUnitedNationsDevelopmentProgramfor2010.IncaseswhereHDIvalues704
werenotavailablespecific to theState(e.g.Hawaii),weusedthenational (e.g.USA)HDI705
30
value,andinothercases(e.g.MarshallIslands)wehadtocalculateHDIfromlifeexpectancy,706
education,andpercapitaincomestatistics.707
708
5.Populationsize.Foreachnation/state,wedeterminedthesizeofthehuman709
Populationin2010.DatawerederivedmainlyfromthenationalcensusreportsCIAfactbook710
(https://www.cia.gov/library/publications/the-world-711
factbook/rankorder/2119rank.html), and Wikipedia712
(https://en.wikipedia.org/wiki/Main_Page).713
714
6.NationalReefFishLandings:Reconstructedreef fishcatchestimates (inmetric tonnes)715
were obtained from the Sea Around Us Project (SAUP) catch database716
(http://www.seaaroundus.org)(45). We used estimates corresponding to 2010 and only717
included reef associated species.Wecalculated the catchperunit area (catch/km2/y)by718
dividinganation/state’scatchbytheitsestimatedreefarea(46).719
720
7.Oceanicproductivity:Weexaminedoceanicnetproductivity foreachreef followingthe721
proceduredescribedby(47).Wedelimiteda100kmbufferaroundeachofourreefclusters,722
we removed shallowwaters pixels (those that intersected orwere containedwithin the723
depth contour of 30m from the General Bathymetric Chart of the Oceans 2014724
(http://www.gebco.net/), a global gridded bathymetry dataset) and then calculated the725
averageofmonthlychlorophyll-aconcentration(proxy forphytoplanktonbiomass)using726
dataprovidedata4km-resolutionbyAquaMODIS(ModerateResolutionImagingSpectro-727
radiometer)foryears2005to2010.728
31
729
8.Climatestress.Weincludedanindexofclimatestressforcorals,developedby(48),which730
incorporated11differentenvironmentalconditions,includingthemeanandvariabilityof731
sea-surfacetemperature,tidalrange,ultravioletradiation,adoldrumindex,andchlorophyll.732
733
Analyses734
We first looked for collinearity among our covariates using bivariate correlations and735
variance inflation factor estimates. This led to the exclusion of several covariates (not736
described above): i) Gross Domestic Product (purchasing power parity); ii) Rule of Law737
(WorldBankgovernanceindex);iii)ControlofCorruption(WorldBankgovernanceindex738
(49));iv)Sedimentation;v)Tourism(touristarrivalsfromtheWorldTourismOrganization’s739
CompendiumofTourismStatisticsrelative to landarea);vi)Atoll(i.e., abinarymetricof740
whether the reef site was on an atoll or not); vii) Frequency of storms since 1980741
(http://weather.unisys.com/hurricane); viii) Environmental performance index (EPI)742
(https://epi.envirocenter.yale.edu/). ; and ix) theGINI index (measureof anation/state’s743
inequality).AlthoughtheGINIindexwasnotstronglycorrelatedwithothercovariates,there744
werenumerousmissingvalues,sothatpotentialcovariatewasremoved.Allothercovariates745
hadcorrelationcoefficientslowerthan0.6andVarianceInflationFactorscoreslessthan2746
(indicatingmulticollinearitywasnotaconcern).Caremustbetakenincausalattributionof747
covariatesthatweresignificantinourmodels,butdemonstratedcollinearitywithcandidate748
covariatesthatwereremovedduringtheaforementionedprocess.Critically,ourmetricof749
totalgravitywascolinearwithatoll (i.e.,mostremoteor lowgravityreefsareatolls)but750
whenwerestrictedtheanalysestoonlynon-atollstheresultsdidnotchange.Additionally,751
32
correlationsbetweenmeanbodysizeofthefishassemblage(length,cm)andourresponse752
variables:biomass(r=0.73),parrotfishscrapingpotential(r=0.2),andtraitdiversity(r=0.4)753
suggestthatmeanbodysizeisonlypredictiveofbiomass.754
755
Multilevelmodels756
Toquantifythemulti-scalesocial,environmental,andeconomicfactorsaffectingthethree757
ecologicalmetrics,wemodelledeachresponsevariableseparatelyusingmultilevelmodels758
thatexplicitlyrecognizedthethreescalesofspatialorganization:reefsite,reefclusterand759
nation/state.ModelswererunusingaBayesianapproachusingtheHamiltonianMonteCarlo760
algorithmimplementedinStanthroughthebrmspackage(50)for10000iterations,anda761
9000burnin.Thisleft4000samplesintheposteriordistributionofeachparameter(four762
chains). We did not have a priori information about parameter distributions; thus, the763
posterior estimates were informed by the data alone (i.e. weakly informative priors).764
Convergence was monitored by running four chains from different starting points,765
examiningposterior chains anddistribution for stability, and checking that thepotential766
scale reduction factor (also termed R_hat) was close to 1. We employed a gaussian767
distribution to analyze biomass of reef fish above 20 cm (log +1 transformed) and trait768
diversity,andusedahurdle-lognormaltoanalyzeparrotfishscrapingpotentialbecausethe769
dataforthismetriccontainedalargenumberofzeros(31%).Thehurdlemodelisatwo-770
partmodelcomposedof(i)abinomialdistributionandalogitlinkfunctiontopredictthe771
probability of observing the herbivory function (i.e., whether the response outcome is772
positiveorzero)and(ii)alognormaldistributionforthenon-zerodata.773
774
33
Foreachmodel,wesetreefclusterandnation/stateasrandomeffectstoaccountforthe775
hierarchicalnatureofthedata(i.e.reefsitesnestedinreefclusters,reefclustersnestedin776
nations/states).Foreachmetric,wetestedtwoalternatemodels:anullmodel,consisting777
onlyofthehierarchicalunitsofobservation(thatis,intercepts-only)andafullmodelthat778
includedallofourcovariates (potentialdrivers)of interest.Weused thenullmodelasa779
baselineagainstwhichwecouldensurethroughleave-one-outcross-validationinformation780
criteria(LOOIC)(51)thatourfullmodelperformedbetterthanamodelwithnocovariate781
information. To account for any methodological effects, sampling area, census method,782
sampledhabitatanddepthwerealsoincludedinallthemodelsascovariates.Tocontrolfor783
samplingeffects,wemarginalizedresponsevariablesbysubtractingtheestimatedsampling784
standardizedmeanmodeleffectstotheobservedresponsevariables.Foralltheanalyses,785
continuous covariates were standardized (mean centered and divided by 2 standard786
deviations).Toexaminemodelfitandhomoscedasticity,weconductedposteriorpredictive787
checks, checked residuals against fitted values and ensured residuals followed expected788
distributions around zero (e.g., for the gaussian distribution models we checked that789
residualswerenormallydistributedaroundzero).Wealsocheckedtheresidualsagainstall790
covariatesincludedinthemodels,andthecovariatesdescribedabovethatwerenotincluded791
in themodels (primarily due to collinearity). The residuals of each of the threemodels792
showed no patternswith these covariates, suggesting theywould not explain additional793
informationinourmodels.Additionally,toaccountforthepotentialeffectthatreservesize794
andagecouldhaveonourresponsevariableswerantwodifferentanalyses:(i)wherewe795
includedallthehighcompliancereservesinourdatairrespectiveofsizeandage(N=106reef796
sites);and(ii)whereweonlyretainedreservesitesthatwereaboveaminimumthreshold797
34
ofat least2km2andolder than4years, (N=61 reef sites).These inclusion criteriawere798
informedbytheliteratureonreserveeffectiveness,whichsuggeststhatadiameterof1-2km799
(1-3km2)isrequiredtoachievepartialprotection(52), butwerealsoconstrainedbyour800
sample;amoreconservativecutoffofsay10km2and10yearswouldhaveleftonly16reef801
sites.Inthemainmanuscript,wereport(i),buthighlightthedifferencesbetween(i)and(ii)802
inFig.S7.AllanalyseswereundertakenusingR(3.02)statisticpackage.803
804
Referenceconditionsandtargets805
Wedefined reference conditions for each ecologicalmetric using the 0.9 quantile of the806
marginalized response variables accounting for sampling, habitat sampled, and sampling807
location(i.e.,responsevariablesminustherandomeffectsandthemodelestimatedeffect808
sizesofdepthcategory,reefhabitatandsamplingmethod).Thus,referenceconditionsare809
foraveragesamplingareaand“Slopes”,“4-10m”and“Standardbelttransects”.Asexpected,810
the90%referencepointvaluesforthefisheriestarget(biomassabove20cm)wasslightly811
belowtheexpectedtotalbiomassinremotelocations(12).Consequently,wethensettargets812
of25,50,and75%ofthesereferencepointconditions,thelowertwoofwhichcorrespond813
to typical standing biomass levels of multispecies maximum sustainable yields814
(hypothesizedtobebetween25-50%ofunfishedbiomassestimates(11,14)).Meanwhile815
75% of reference conditions is considered a more stringent conservation target. For816
consistency,weused the same reference conditions and targets (i.e.25,50, and75%of817
reference conditions) for parrotfish scraping potential and trait diversity, although818
established ecological significance of these figures remains untested, and establishing819
benchmarksfortheseisanimportantareaoffutureresearch,asisdevelopingregion-specific820
35
referenceconditions.Toavoidbeingoverlyprescriptive,wealsoranouranalysesforarange821
ofreferenceconditions,basedonthe0.8and0.95quantilesoftheresponsevariables,and822
incorporatedtheresultsinthesupplementalinformation(Fig.S3-S4).823
824
Toestimatetheprobabilityofpassingdifferentthresholdsunderagradientofgravity(e.g.,825
Fig.2),foreachresponsevariable,wesimulatednewdatafromthemodelposteriorswhere826
onlygravitywasmodified(i.e.,maintainingalltheothercovariatesataverageconditions,for827
slopes,4-10mofdepthandstandardbelttransectsandnotincludingtherandomeffects)828
andestimatedtheprobabilityoftheposteriorsamplesbeingaboveorbelowthetargets.To829
determine the probability of all three response variables passing the targets (i.e., co-830
occurrenceofmetrics),weusedthesubsetof1662reefsitesthathadallthreeecological831
metricsandmultipliedtheprobabilities(i.e.,assumingindependence).832
833
Potentialgainsfrommanagementforourreefsites834
Toestimatethenumberoffishedsitesthatwouldpassdifferenttargetsifmanagement(i.e.,835
highcompliancemarinereservesorrestrictions)wereimplemented,wesimulatednewdata836
fortheposteriordistributionsmaintainingsamplingconsistent(i.e.,samplingmethodand837
samplingarea)butallowingindividualsitestohavetheirownsocio-ecologicalcontext(e.g.,838
habitat,depth,HDI,randomeffects).Then,wechangedtheirprotection(fromopenlyfished839
tohighcompliancemarinereservesorrestricted)andsimulatedanewsetofdatabasedon840
thatcondition.Thisallowedustoestimatethenumberofoursitesthatcouldpotentiallypass841
differentthresholdsifmanagementwasimplementedgiventheeffectofmanagementinour842
model and a site’s own environmental and socio-economic context. We report the high843
36
compliance marine results in the main manuscript and the restricted fishing in the844
supplementalinformation.845
846
37
Supplementalfigures847
848 849
FigureS1|Correlationsbetweenthethreekeyecologicalmetricssupportedby fish850
communitiesoncoralreefs851
852
853
854
log(Biomass >20cm+1)
12
34
5
0.54
0 2 4 6 8 10
0.45
1 2 3 4 5
●● ●●
●
●
●●
●
●
●
●
● ●
●
● ●
●
●
●●
●
●
●●
●
●
●●
●
●
●
●
●●
●●
●
●●●●
●●
●
●
●
●
●
●●
●
●●
●
●●
●
●●
●
●
●
● ●
● ●
●●
●
● ●
●●
●●
●
●
●
●
●
●
●
●
●
●
●
●●
●
●
●
●●
●
●
●
●
●●
●
●
● ●
●
●●
●
●
●●
●●
●
●
●●
●
●●
●
●●
● ●
●
●●
●
●
●
●
●
●
●
●
●
●● ●
●●
●
●
●
●
●
●
●●
●
●●
●
●
●●
●
●
●●
●
●● ●
●
●
●●
●
●●
●
● ●●
●
●●
●●●
●●
●
●
●
●
●
●
●
● ●
●
●
● ●
●
●●
●●
● ●●
●
●
●
●
●
●●● ●●
●
●●
●●
● ●●●
●
●●● ●● ●
●
●●●
●●
●
●
●
●
●●
●
●
●
●
●
●
●
●
●
●
●
●
●
●
●
●
●
●
●●
●
●
●●●
●
●
●
●
●
●●
●
●
●
●●
●
●
● ●
●
●
●
●
●
●
●●
●
●
●●
●
●
●●
●
●
●
●●●
●
●
●
●●●
●
●
●
●
●
●
●
●
●
●
●
●
●
● ●●●
●● ●
●
● ●
●
●
●
●
●●
●●●
●
●
●●
●
●
●
●
●
●
●
●
●
●
●
●
●
●●
●
●
●
●
●
●
●
●
●
●●● ●● ●
●
●●●
●
●● ●
●
●
●● ●
●● ●
●
●●
●
●●
●●
●
●●●
●●
●
●●
●
●
●
●●
●
●
●
●
●●
●
●
●●
●●
●
●
●
●
●
●
●●
●
●
●
●
●
●
●
●
●
●
●
●●
●
●●
●
●
●
●●
●●
●
●
●
●
●
●
●●
●● ●
●
●
●
●
●
●
●
●
●
●●
●
●
●●●
●●
●
●
●
●
●●
●●
●●
●●
●
●
●●
●
●
●
●
●●
●
●
● ●●
●
●
●●
●
●● ●●●
●
●● ●
●
●
●
●
●
●
●
●
●
●
●
●
●
●
●
●
●
●
●
●
●
●
●●●
●
●
●
●●
●
●
●
●●
●
●
●
●
●●
●
●
●
●●
●
●
●
●
●●
●
● ●
●●
●
●●
●
● ●
●
●
●●
● ●●●●
●
●●
●
●
●
●●●
●
●● ●
●●
●
●
●
●●
●
●●
●
●
●●
●
●
●●●
●●
●
●
●
●●●
●
● ● ●●
●
●
●
●
●
●
●
●●
●
● ●●
●●
●
●
●
●
●
●
●
●
●
●
●
●
●
●
● ●
●
●
●
●
●●
●
●
●
●
●
●●
●
●
●
●
●
●
●
●
●●
●●
●
●●
●
●●
●
●
●
●
●
●
●●
●
●●
●
●● ●
●●
●
●●
●●
●●
●
●
●
●
●●●
●
●
●●●●
●
●●
●
●●●●●
●
●
●
●●
●
●
●
●
●●●
●
●
●●
●
●
●
●
●
●
●
●
●
●●
●
●
●
●●
●
●
●
●●
●
●
● ●
●
●
●
●
●
●
●●
●●
●●
●
●
●
●
●
●
●
● ●●
● ●
●
●● ●●
●
●
●
●
●
●
●
●● ●●
●
●
●
●
●
●
●
●
●
●
●
●
●●
●
●
●
●●
●
● ●
●●
●●
●
●
●
●
●
●
●
●
●
●
●
●
●●
●
●
●
●
●
●
●
●
●
●
●
●
●●●
●
●
●●●
●
●
●
●
●●●
●
●
●
●●
●
●●
●
●
● ●
●
●●
● ●
●●
●
●
●●●
● ●
●
●
●●
● ●
●
●
●
●●●
●
●
●
●
●
●●
●
●
●●
●
●
●
●
●
●
●●
●
●●
●●
●
●●
●
●
●
● ●
●●
●●
●
●
●●
●
●
●●
●
●
●
●
●●●
●
●
●
●
●
●
●
●●
●
●●●●●●
●●
●
●
●●
●
●
● ●●
●
●●
●
●
●
●
●
●
●
●●
●●
● ●●
●●●
● ●
●
●
●●
●
●
●
●
●●
●
●
●
●
● ●
●
●
●●
●
●
●
●
●
●
●
●
●
●
●
● ●●
●
●●
●
●●
●●
●
●
●●● ●
●
●
●
●●
●
●●●
●
●●●
●
●
●
●●●
● ●
●
●
●
●
●
●
●
●
●
●
●
●
●
●
●
●
●
●
●
●●
●
●
●
●
●
●
●
●●●
●●
●
●
●
●
●
●
●
●
●
●
●●●
●
●
●
●
●
●
●
●
●
●●
●
●
●
●
●
●
●
●
●
●●
●
●
●●
●
●●
●●
●
●
●
●
●
●
●
●
●
●
●
●
●
●
●
●●
●
●
●●
●
●
●
●
●
●●
●●
●
●
●●
●
●●
●
●
●
●
●
●
●
●
●
●
●
●
●●
●
●
●●
●
●
●●
●
●
●
●
●
●
●
●
●
●
●
●
●
●●
●●
●
●●
●
●
●
●●●
●●
●●● ●
●●
●
●
●
●●
●
●●
●
●
●
●
●
●
●
●●
●
●
●
●
●●
●
●
●
●
●
●
●
●
●
●
●
●
●
●
●
●
●
●
● ●
●
●
●
●
●●
●
●
●
●
●
●
●
●
●●
●
● ●
●●
●
●
●
●●
●
●●
●
●●
●
●
●
●
●
●●
●
●
●●
●
●
●
●
●
● ●●
●
●
●
●
●
● ●●●
●
●
●
●
●
●
●
●
●
●●
●
●● ●
●
●
●●
●
●
●●
●
●
●
●
●
●●
●
●
●●●
●●
●
●
●
●●
●
●
●●
●
●
●
● ●
●
●●
●●
● ●
●
●
●●
●
●
●
●
●
●●
●
●
●
●●
●
●
●
●
●
●
●
●
●
●
●●
●●
●●
●
●
●
●
● ●●
●●
●
●●
●●
●
●
●
●
●●
●
●
●
●● ●
●
●●
●
●
●●
●
●
●
●
●
●●
●
●●
●●●
●
●
●
●●●
●
●
●
●
●
●
●
●●
●
●●● ●
●●
●
●
●●
● ●
●
●●
●
● ● ●
●
●●
●
●●
●●
●
●
●
●
●
●
●●
●
●●
●
●
●
●
●●
●
●
●
●
●
●
●
●
●
●
● ●
●
●
●
●
●
●
●●
●
●
●
●●
●
●
●
●
●●
●
●
●
●
●
●
●●
●●
●
●●
●
●
●
●
●●
●
●●
●
●
●
●
●
●●
●●
●
●
●
●●
●
●●
●●
● ●●
●● ●
●
●●
●
●
●
Trait diversity
0.32
02
46
810
●●● ●
●
●
● ●
●
●
●
●
●●
●
●●
●
●
● ●
●
●
●●
●
●
● ●
●
●
●
●
●●
●●
●
●● ● ●
●●
●
●
●
●
●
●●
●
●●
●
●●
●
●●
●
●
●
●●
● ●
●●
●
●●
●●
●●
●
●
●
●
●
●
●
●
●
●
●
●●
●
●
●
●●
●
●
●
●
● ●
●
●
●●
●
●●
●
●
●●
●●
●
●
●●
●
●●
●
●●
●●
●
● ●
●
●
●
●
●
●
●
●
●
●● ●
●●
●
●
●
●
●
●
● ●
●
●●
●
●
●●
●
●
●●
●
●●●
●
●
●●
●
●●
●
● ●●
●
●●
● ●●
●●
●
●
●
●
●
●
●
●●
●
●
●●
●
●●
●●
●●●
●
●
●
●
●
●● ●●●●
●●
●●
● ●● ●
●
●●
● ●● ●
●
● ● ●
● ●
●
●
●
●
●●
●
●
●
●
●
●
●
●
●
●
●
●
●
●
●
●
●
●
●●
●
●
●●●
●
●
●
●
●
● ●
●
●
●
●●
●
●
●●
●
●
●
●
●
●
●●
●
●
● ●
●
●
●●
●
●
●
●●●
●
●
●
●●●
●
●
●
●
●
●
●
●
●
●
●
●
●
●●●●
●●●
●
●●
●
●
●
●
●●
●●●
●
●
● ●
●
●
●
●
●
●
●
●
●
●
●
●
●
●●
●
●
●
●
●
●
●
●
●
●●●●●●
●
● ●●
●
●●●
●
●
● ●●
●● ●
●
●●
●
●●
●●
●
●● ●
●●
●
●●
●
●
●
●●
●
●
●
●
●●
●
●
●●
●●
●
●
●
●
●
●
●●
●
●
●
●
●
●
●
●
●
●
●
●●
●
●●
●
●
●
●●
●●
●
●
●
●
●
●
●●
●● ●
●
●
●
●
●
●
●
●
●
●●
●
●
●●●
●●
●
●
●
●
●●
●●
●●
●●
●
●
●●
●
●
●
●
●●
●
●
● ●●
●
●
●●
●
● ●●●●
●
●●●
●
●
●
●
●
●
●
●
●
●
●
●
●
●
●
●
●
●
●
●
●
●
●●●
●
●
●
●●
●
●
●
●●
●
●
●
●
●●
●
●
●
●●
●
●
●
●
●●
●
●●
●●
●
●●
●
●●
●
●
●●
● ●●● ●
●
● ●
●
●
●
●●●
●
● ●●
●●
●
●
●
●●
●
● ●
●
●
●●
●
●
● ●●
●●
●
●
●
● ●●
●
● ●●●
●
●
●
●
●
●
●
●●
●
● ●●●●
●
●
●
●
●
●
●
●
●
●
●
●
●
●
● ●
●
●
●
●
●●
●
●
●
●
●
●●
●
●
●
●
●
●
●
●
●●
●●
●
●●
●
●●
●
●
●
●
●
●
● ●
●
●●
●
●●● ●
●
●
●●
●●
●●
●
●
●
●
● ●●
●
●
● ●● ●
●
●●
●
● ●●
●●
●
●
●
●●
●
●
●
●
●●
●
●
●
●●
●
●
●
●
●
●
●
●
●
● ●
●
●
●
●●
●
●
●
●●
●
●
●●
●
●
●
●
●
●
●●
●●
●●
●
●
●
●
●
●
●
● ● ●
● ●
●
●● ●●
●
●
●
●
●
●
●
● ●●●
●
●
●
●
●
●
●
●
●
●
●
●
●●
●
●
●
●●
●
●●
●●
●●
●
●
●
●
●
●
●
●
●
●
●
●
●●
●
●
●
●
●
●
●
●
●
●
●
●
● ●●
●
●
●●●
●
●
●
●
●●
●
●
●
●
●●
●
●●
●
●
● ●
●
●●
●●
●●
●
●
●●●
● ●
●
●
●●
●●
●
●
●
● ●●
●
●
●
●
●
●●
●
●
●●
●
●
●
●
●
●
●●
●
●●
●●
●
●●
●
●
●
●●
●●
●●
●
●
●●
●
●
●●
●
●
●
●
●●●
●
●
●
●
●
●
●
●●
●
●●●●●
●
●●
●
●
●●
●
●
●●●
●
●●
●
●
●
●
●
●
●
●●
●●
● ●●
● ●●
●●
●
●
●●
●
●
●
●
●●
●
●
●
●
●●
●
●
●●
●
●
●
●
●
●
●
●
●
●
●
● ● ●
●
●●
●
●●
●●
●
●
●●● ●●
●
●
●●
●
●●
●
●
●●●
●
●
●
●●●
●●
●
●
●
●
●
●
●
●
●
●
●
●
●
●
●
●
●
●
●
●●
●
●
●
●
●
●
●
●●●
●●●
●
●
●
●
●
●
●
●
●
●●●
●
●
●
●
●
●
●
●
●
●●
●
●
●
●
●
●
●
●
●
●●
●
●
●●
●
●●
●●
●
●
●
●
●
●
●
●
●
●
●
●
●
●
●
●●
●
●
●●●
●
●
●
●
●●
●●
●
●
●●
●
●●
●
●
●
●
●
●
●
●
●
●
●
●
●●
●
●
●●
●
●
●●
●
●
●
●
●
●
●
●
●
●
●
●
●
●●
●●
●
●●
●
●
●
●● ●
●●
●●●●
●●
●
●
●
●●
●
●●
●
●
●
●
●
●
●
●●
●
●
●
●
●●
●
●
●
●
●
●
●
●
●
●
●
●
●
●
●
●
●
●
●●
●
●
●
●
●●
●
●
●
●
●
●
●
●
●●●
● ●
●●
●
●
●
●●
●
●●
●
●●
●
●
●
●
●
●●
●
●
●●
●
●
●
●
●
●●●
●
●
●
●
●
●●●●
●
●
●
●
●
●
●
●
●
●●
●
●●●
●
●
●●●
●
●●
●
●
●
●
●
●●
●
●
●● ●
● ●
●
●
●
●●
●
●
●●
●
●
●
●●
●
●●
●●
●●
●
●
●●
●
●
●
●
●
●●
●
●
●
●●
●
●
●
●
●
●
●
●
●
●
●●
●●
●●
●
●
●
●
●●●
●●
●
●●
●●
●
●
●
●
●●
●
●
●
● ●●
●
●●
●
●
●●
●
●
●
●
●
●●
●
●●
●●●
●
●
●
●●●
●
●
●
●
●
●
●
●●
●
●●● ●
●●
●
●
● ●
●●
●
● ●
●
●●●
●
●●
●
●●
●●
●
●
●
●
●
●
●●
●
●●
●
●
●
●
●●
●
●
●
●
●
●
●
●
●
●
●●
●
●
●
●
●
●
●●
●
●
●
●●
●
●
●
●
●●
●
●
●
●
●
●
● ●
●●
●
●●
●
●
●
●
● ●
●
●●
●
●
●
●
●
●●
●●
●
●
●
● ●
●
●●
●●
●●●
●●●
●
●●
●
●
●
●
●
●
●
●
●
● ●
●
●
●
●
●
●
●
●
●
●
●
●
●
●
●
●
●
●
●
●
●
●
●
●●
●
●
●
●
●
●
●
●
●
●
●
●
●
●●
●
●
●●
●
●
●
●●
●
●
●
●
●●
●
●
●●●
●●
●
●
●
●
●
●
●●
● ●
●●
●
● ●
●
●●
●
●
●●
●
●●
●
●
●
●
●
●●
●
●
●●
●●
●
●●
●
●
●
●
●
●
●
●●
●
●
●●●
●
●●
●
●
●
●● ●●
● ●
●
●
●
●●
●
●●
●
●●
● ●
●
●
●
●
●●●
●
●
●
●
●
●
●
● ●
●
●
●●
●●
●
●
●
●●
●●
●
●
●
●
●
●
●
●●
●
● ●
●
●
●●
●
●
●
●
●●
●
●
●
●
●
●
●
●
●
●●
●
●
●
●
●
●
●
●
●
●
●●
●
●●●
●
●
●
●
●
●
●
●●
●
● ●
●
●
●
●
●
●
●
●
●
●
●
●
●
●
●
●
●
●
●
●
●
●
●
●
●
●
●●
●
●
●
●
●
●
●
●
● ●
●
●
●
●
●
●●
●
●
●
●●●
●
●
●
●
●
●
●
●
●
●
●
●
●
●
●
●
●
●●
●
● ●
●
●
●
●●
●
●
●
●
●
●●
●
●
●
●
●
●
●
●
●
●
●
●
●
●
●
●
●
●●
●
●
●
●
●
●
●
●●●
●
●
●
●
●
●
●
●
●●
●
●
●
●
● ●
●
●
●
●●
●
●
●
●
●
●
●
●
● ●
●
●
●
●●
●●
●
●
●
●
●
●
●
●
●
●
●●●
●
●
●
●
●
●●●●
●
●●●●
●
●
●
●●
● ●
●
●
●
●●
●●
●
●
●
●
●●
●
●
●
●
●
●
●
●●●
●
●●
●
●
●
●
●
●
●
●
●
●
●●
●
●●
●
●
●
●
●
●
● ●
●●●
●
●●
●
●
●
●
●
●
●●
●
●
●
●
●
●
●
●●
●
●●
●
●●
●
●
●
●●
●●●
●
●
●
●
●●
●
●
●
●
●
●
●
●
●
●●
●
●
●
●
●●
●
●
●
●
●
●
●
●
●
●
●
●
●
●
●
●
●
●
●
●
●
●
●
●
●
●
●● ●
●●
●
●
●
●
●
●
●●●
●
●
●
●
●
●
●●
●
●
●
●
●
●
●
●
●
●●
● ●
●
●
●●
●
●● ●
●
●
●●
●
●
●
●●
●
●
●
●●
●
●●
●
●
●
●
●
●
●
●
●
●
●●
●
●● ●
●
●
●●
●
●
●
●
●
●
●●
●
●●
●
●●● ●●
●
●●
●
●
●
●●
●
●
●
●
●●
●
●
●
●
●
●
●
●
●
●
●●
●
●●●
●●
●
●
●
●●
●●
●
●●●●
●
●●
●
●
●
●
●
●
●
●
●● ●●
●
●
●
●
●
●
●
●
●
●●
●
●●●
●
●●
●
●
●
●
●
●
● ●●●
●
●
●
●
●
●● ●●
●
●
●
●
●
●
●●
●●
● ●
●
●●
●
●
●
●
●●
●
●
●●
●
●
●
●
●
●
●
●●
●
●
●
●
●●
●
●
●●
●
●
●● ●
●●
●
●
●
●●●
●
●●
●
● ●
●
●
●
●●
●
●
●
●
●●
●
●
●
● ●
●
●●
●●
●
●
●
●
●
●
●
●
●●
●
● ●
●
●
● ●
●
●
●●
●
●
●
●
●
●
●
●
●
●
●
●●
●
●
●●
●
●
●
●●
●
●
● ●●
●
●
●●
●
●
●
●
●
●
●
●●
●
●●
●
●
●●
●●
●
●
●
●
●
●
●
●
●●● ●
●
●
●
●
●
●●
●
●
●
●
●
●
●
● ●
●
●
●●
●
●
●
●
●●
●
●● ●
●●
●●●
●
●●
●
●
●
●
●●
●
●●
●
●
●
●
●●
●●
●
●●
●
●●
●●
●
●
●
●
●●
●●
●
●●
●
●
●● ●
● ●
●
●
●
●●●
●
●
●●
●
●
●
●
●
●
●
●
●
●
●●
●●
●
●
●
●
●
●
●●
●
●
●
●
●
●
●
●
●
●
●
●
●
●
●
●
●
●
●
●
●
●
●●
●
●
●
●
●
●
●
●●
●
●
●
●
●
●
●
●
●
●●
●
●
●
●
●●
●
● ●
●
●
●
●●
●●●
●
●●
●
●
●●
●
●●
●
●
●
●
●
●
●
●●
●
●●●●
●
●
●
●
●
●
●
●
●●●●
●
●
●●
●
●●
●
●●
●
●
●●●●
●
●
●●●
●
●
●●●
●●●●●●●●●●●
●
●●●
●●
●
●
●●
●●●
●
●●●
●
●
●
●●
●●●●●●●
●●●●
●●
●●
●
●
●●●
●
●●
●
●●
●
●●
●●●●●
●●
●●●
●
●●●
●
●
●
●
●●●●●
●
●
●
●
●
●
●●●●●●
●
●
●●
●
●
●
●
●
●
●
●
●
●
●
●
●
●
●
●●
●
●
●
●
●
●●
●
●
●
●
●●
●
●
●
●
●
●
●
●
●●
●
●
●
●
●
●
●
●
●
●
●
●
●
●
●●
●
●
●
●
●
●
●
●
●
●
●
●
●
●
●
●
●●
●
●
●
●
●
●
●
●
●
●
●●● ●
●
●
● ●●
●
●●
●
●
●
●
●
●
●
●
●
●●
●
●
●
●
●
●
●
●
●
●●
●
●
●
●
●● ●
●
●
●
●
●
●
●
●
●
●●
●
●
●
●
●
●
●●
● ●
●
●
●
●●
●●
●
●
●
●
●
●
●
●
●
●
●
●
●●
●
●
●●
●
●
●
●
●
●
●●
●
●
●
●
●
●
●
●
●
●●
●
●
●
●
●
●
●
●
●
●
●
●
●
●
●
●
●
●
●
●●
●
●
●●
●
●
●
●
●
●
●
●●
●
●●
●
●
●●
●
●●
●●
●
●●
●
●
●
●
●
●●
●●
●● ●
●
●
●●
●
●
●●
●
●
●
●
●
●
●
●
●●
●
●
●
●
●
●
●
●
●●
●
●
●
●
●
●
●
●●
●●
●
●
●
●
●● ●
●
●
●
●●
●●●
●●
●
●
●●●
●
●
●
●
●●
●
●
●
●
●
●
●
●
●
●
●
●
●
●
●
●
●
●
●●
●
● ●
●
● ●
●
●
●
●
●
●
●
●
●●
●
●
●
●●
●
●●
● ●
●
●
●
●
●●
●
●
●●●
●
●●●●
●
●
●● ●●
●
●
●●
●
● ●●
●
●
●●
●
●
●
●
●
●
●
0 2 4 6 8
02
46
8
log(Parrotfish scraping+1)
38
855 Figure S2| Effect size of eight socioeconomic drivers, management, sampling, and856 environmental conditions on three fishmetrics. (A) biomass of reef fish >20cm. (B)857 parrotfish scraping potential. (C) trait diversity. Total gravity was the most consistent858 socioeconomiccovariate,demonstratingstrongnegativerelationshipswithfishbiomassand859 trait diversity, and a weaker negative relationship with parrotfish scraping potential860 (posterior slope had 65.4% of the samples negative). Continuous covariates were861 standardized (mean centered and divided by 2 standard deviations), while response862 variables were not. Thus, effect sizes are standardized within columns only. Parameter863 estimatesareBayesianposteriormeanvaluesand95%uncertaintyintervals(UI).Redor864 greendotsindicatenegativeorpositiverelationships,respectively,wherethe95%UIdoes865 notoverlap0.AHurdlemodelwasusedforparrotfishscraping(b).866 867
868
869
870 871 872 873
39
874 875 FigureS3 |Theestimatedprobabilityofopenly fishedreefsiteshaving25,50,and876 75% of reference conditions (light, medium, and dark purple, respectively)877 benchmarked from remote reefs >10 hours from the nearest settlement for (A) a878 combinationof fishbiomass (>20cm),parrotfish scrapingpotential, traitdiversity,879
40
and (B-D) eachmetric, respectively, along a gradient of humanpressure (gravity).880 Separateestimatesareprovidedforreefsitesinmarinereserves(E-H)andwithrestricted881 fishing(I-L).Tohighlighthowthepotentialbenefitsofmanagementchangealongagradient882 ofhumanpressure(gravity),weextractedthedifferenceintheprobabilityofachievingeach883 targetbetweenmarinereservesandopenlyfishedsites(M-P),restrictedandopenlyfished884 areas(Q-T),andmarinereservesandrestrictedareas(U-X).Weplottedthepartialeffectof885 the relationship between gravity and each benchmark by setting all other continuous886 covariatesto0(becausetheywereallstandardized)andallcategoricalcovariatestotheir887 mostcommoncategory(i.e.4-10mfordepth,slope forhabitat,standardbelt transect for888 censusmethod).889
41
890 FigureS4 |Theestimatedprobabilityofopenly fishedreefsiteshaving25,50,and891
75% of reference conditions (light, medium, and dark purple, respectively)892
benchmarked from remote reefs >10 hours from the nearest settlement for (A) a893
combinationof fishbiomass (>20cm),parrotfish scrapingpotential, traitdiversity,894
and (B-D) eachmetric, respectively, along a gradient of humanpressure (gravity).895
42
Separateestimatesareprovidedforreefsitesinmarinereserves(E-H)andwithrestricted896
fishing(I-L).Tohighlighthowthepotentialbenefitsofmanagementchangealongagradient897
ofhumanpressure(gravity),weextractedthedifferenceintheprobabilityofachievingeach898
targetbetweenmarinereservesandopenlyfishedsites(M-P),restrictedandopenlyfished899
areas(Q-T),andmarinereservesandrestrictedareas(U-X).Weplottedthepartialeffectof900
the relationship between gravity and each benchmark by setting all other continuous901
covariatesto0(becausetheywereallstandardized)andallcategoricalcovariatestotheir902
mostcommoncategory(i.e.4-10mfordepth,slope forhabitat,standardbelt transect for903
censusmethod).904
905
906
907 Fig.S5.Thescaleddistributionofcovariatesforoursampleofreefs(blue)andforall908
tropical reefs globally (grey). Our sampled reefs display a reasonably similar909
distributionandrangeformostcovariatesNotethattheglobalgravityvalueswere910
onlyavailableroundedtothenearestinteger,thereforetodirectlycomparewithour911
site level values, we used a log+1 transformation, rather than log+minimum912
transformationasusedintherestofthemanuscript. 913
0.00
0.25
0.50
0.75
1.00
0.0 2.5 5.0 7.5 10.0 12.5Total Gravity (log+1)
scal
ed
0.00
0.25
0.50
0.75
1.00
−2 0 2Ocean productivity (log)
0.00
0.25
0.50
0.75
1.00
0.25 0.50 0.75 1.00Climate stress
0.00
0.25
0.50
0.75
1.00
−50 0 50 100Local population growth
0.00
0.25
0.50
0.75
1.00
0 2 4 6 8Reef fish landings (log+1)
scal
ed
0.00
0.25
0.50
0.75
1.00
0.4 0.6 0.8HDI
0.00
0.25
0.50
0.75
1.00
0 5 10 15 20Population size (log+1)
43
914
915 916
Fig.S6.Conservationtargetoutcomesfromsimulatingtheimplementationoffishing917
restrictionsinopenlyfishedsites.Alluvialplotsshowthechangeinthenumberofsites918
expectedtoachievekeyconservationtargetsiffisheriesrestrictionswereimplementedin919
our openly fished sites for (A) simultaneouslymeeting fish biomass, parrotfish scraping920
potential,andtraitdiversity,and(B-D)eachgoal,respectively.921
922
923
44
924
925 Fig.S7.Differenceinprobabilityofachievingspecifictargetsbetweentherestricted926
subsetofmarinereserves(>2km2and4yearsold,n=61)andallmarinereservesin927
our sample (n=106) for (A) simultaneously meeting fish biomass, parrotfish scraping928
potential, and trait diversity, and (B-D) each goal, respectively. Alluvial plots show the929
change in thenumberofsitesexpected toachievekeyconservation targets if themarine930
reserves>2km2and4yearsold(basedonourrestrictedsubset)wereimplementedinour931
openly fished sites for (E) simultaneously meeting fish biomass, parrotfish scraping932
potential,andtraitdiversity,and(F-H)eachgoal,respectively.933
934 935
45
SupplementalTables936
937
TableS1|Listof‘Nation/states’coveredinstudy.Inmostcases,938
nation/statereferstoanindividualcountry,butcanalsoincludestates(e.g.939
Hawaii),territories(e.g.BritishIndianOceanTerritory),orotherjurisdictions.940
941
Nation/StatesAmericanSamoaAustraliaBelizeBrazilBritishIndianOceanTerritoryCaymanIslandsColombiaCommonwealthoftheNorthernMarianaIslandsComoroIslandsCubaEgyptFederatedStatesofMicronesiaFijiFrenchPolynesiaGuamHawaiiIndonesiaJamaicaKenyaKiribatiMadagascarMaldivesMarshallIslandsMauritiusMayotteMexicoMozambiqueNetherlandsAntillesNewCaledoniaOman
46
PalauPanamaPapuaNewGuineaPhilippinesPRIAReunionSeychellesSolomonIslandsTanzaniaTongaVenezuela
942
943
Table S2| Justification of ecological metrics944 Biomass of fish above 20 cm
Large fish are both key to sustain ecosystem functioning and common fishery targets. We selected a 20 cm cut-off point because it includes large fish and “plate-sized” fish, targeting not only the most valuable fish but also the fish destined to food consumption (53). Additionally, large fish exert top-down control on ecosystems, regulating the structure and functions of reef ecosystems (54). Biomass captures both the size and number of fish above 20 cm in the system, which dictates the magnitude of the function (55). Biomass of fish above 20 cm is expected to decline rapidly as human impacts intensify (11), and there is empirical evidence that management can allow the recovery of large species (56).
Parrotfish scraping
Herbivory mediates the competition between corals and algae. Bioerosion removes dead reef structures, providing suitable substrate for coral recruitment. Parrotfish are among the most important groups of herbivorous fish on coral reefs performing processes of algae removal and contributing to bioerosion, hence maintenance of good condition for reef growth. Herbivory is expected to decline as human impacts intensify (55) and respond positively to management (57).
Trait diversity The diversity of ecological traits supported by species can represent the range of potential ecological roles present in a given community (58, 59). A broader range of traits are assumed to provide a greater contribution to key ecosystem processes (e.g. biomass production, nutrient cycling) and cultural services (e.g. aesthetic value) than a smaller range of traits (59–61). We estimated trait diversity (TD) using the Chao’s FDq=1 index which is a generalization of the taxonomic Shannon’s entropy index (Chao et al 2019). This index is high when both the dissimilarity of species’ traits (e.g. diet, size) and the spread of biomass across these traits are high. We posit that TD should generally decrease as human impacts increase, because activities such as fisheries selectively target species with specific traits, which can reduce the trait space occupied and the balance of biomass among traits, and thus TD (62, 63).
945
946
47
TableS3|Summaryofsocialandenvironmentalcovariates.Further947
detailscanbefoundinMethods.Thesmallestscaleistheindividualreefsite.948
Reefclustersconsistofclustersofreefsiteswithin4kmofeachother.949
Nation/statesgenerallycorrespondtocountry,butcanalsoincludeor950
territoriesorstates,particularlywhengeographicallyisolated(e.g.Hawaii).951
952
Covariate Description Scale Keydatasources
Localpopulation
growth
Differenceinlocalhuman
population(i.e.100km
bufferaroundourreef
clusters)between2000-
2010
Reefcluster SocioeconomicDataand
ApplicationCentre
(SEDAC)gridded
populationofthework
database(39)
‘Gravity’ofhuman
pressure
Foreachpopulatedcell
withina500kmradiusof
areefsite,wedividedthe
populationofthatcellby
thesquaredtraveltime
betweenthereefsiteand
thecelltogetagravity
value(i.e.howmuch
“gravitationalpull"that
populationwasexerting
onthereefsite).Thiswas
thensummedforallcells
togetthetotalgravityof
humanpressure.
Reefsite Humanpopulationsize,
landcover,road
networks,coastlines
48
Managementstatus Whetherthereefsiteis
openlyfished,restricted
(e.g.effectivegearbans
oreffortrestrictions),or
unfished
Reefsite Expertopinion,global
mapofmarineprotected
areas.
HumanDevelopment
index
Asummarymeasureof
humandevelopment
encompassing:alongand
healthylife,being
knowledgeableandhave
adecentstandardof
living.Weusedlinearand
quadraticfunctionsfor
HDI.
Nation/state UnitedNations
DevelopmentProgramme
PopulationSize Totalpopulationsizeof
thejurisdiction
Nation/state WorldBank,census
estimates,Wikipedia
Fishlandings Landingsofreeffish
(tons)perKm2ofreef
Nation/state SeaAroundUsProject
(PaulyandZeller(45))
Climatestress Acompositemetric
comprisedof11different
environmentalvariables
thatarerelatedtocoral
mortalityfrombleaching
Reefcluster Mainaetal.(48)
Productivity Themonthlyaverage
(2005-2010)oceanic
productivity
Reefcluster Goveetal.2013(47),
AquaMODIS
49
Habitat Whetherthereefsiteisa
slope,crest,flat,orback
reef/lagoon
Reefsite Primarydata
Depth Depthoftheecological
survey(<4m,4.1-10m,
>10m)
Reefsite Primarydata
Samplingtechnique Whetherthedata
collectorusedpoint
count,linetransects,or
distancesampling
Reefsite Primarydata
AreaSampled Thesizeofthearea
sampledbythedata
provider(inm2)
Reefsite Primarydata
953
954
955
956
957
958
50
TableS4|Listoffishfamiliesincludedinthisstudyforboththetrait959
diversityandthebiomassabove20cmresponsevariables.960
Fishfamily Commonfamilyname
Acanthuridae SurgeonfishesBalistidae TriggerfishesCarangidae JacksDiodontidae PorcupinefishesEphippidae BatfishesHaemulidae SweetlipsKyphosidae DrummersLabridae WrassesLethrinidae EmperorsLutjanidae Snappers
Monacanthidae FilefishesMullidae Goatfishes
Nemipteridae CoralBreamsPinguipedidae SandperchesPomacanthidae Angelfishes
Scaridae WrassesandParrotfish
Serranidae GroupersSiganidae RabbitfishesSparidae Porgies
Synodontidae LizardfishesTetraodontidae Pufferfishes
Zanclidae MoorishIdol
961
962