Download to read offline
![2
natureresearch|reportingsummaryApril2020
Field-specific reporting
Please select the one below that is the best fit for your research. If you are not sure, read the appropriate sections before making your selection.
Life sciences Behavioural & social sciences Ecological, evolutionary & environmental sciences
For a reference copy of the document with all sections, see nature.com/documents/nr-reporting-summary-flat.pdf
Ecological, evolutionary & environmental sciences study design
All studies must disclose on these points even when the disclosure is negative.
Study description This study compared the richness, density, and biomass of fish from 23 protected river reaches and paired adjacent unprotected
reaches. To evaluate the effect of protection on fish, we treated the protected and unprotected reaches at each site (n = 23) as
experimental units within which each of two observers made fish observations at each of two plots per reach (n=184 total: 23 sites x
2 reaches per site x 2 plots per reach x 2 observers per plot x 1 observation per observer). In mixed effects models, protection status
was coded as a binary fixed effect (N = no community protection, Y = community protection), and site (n = 23) was used as a random
effect to account for site-level variation among paired reaches. Fish were enumerated during visual surveys conducted using mask
and snorkel along longitudinal transects, as well as supplemental lateral transects at 4 of 23 sites where the river size was wide (>15
m width). Lengths of fish were estimated by eye for up to 10 individuals of each species; where >10 individuals were observed, only
10 lengths were estimated from a representative, haphazardly selected subset of individuals. Estimated lengths were used to
calculate estimated biomass of each fish using species-specific regressions from captured conspecifics. Reserve characteristics were
determined using GIS analysis, field measurements, and interviews with community leaders.
Research sample We sampled (observed) only fish (Class: Actinopterygii) along 50-m-long transects, enumerating all fish observed within A 2-m-wide
transect. In all study reaches, 100 square meters was sufficient to include observations from all major habitat types. Samples
consisted of underwater fish counts made along transects by [two] observers wearing masks and snorkels. Because we made
underwater observations of fish, it was not possible to determine the age range of fishes nor their sexes. While it is possible to
estimate the age of fish based on their lengths, these statistical relationships have not been determined for nearly all of the species
we observed, making it impossible to accurately determine their ages from estimated body length data. The sample population of
fish observed along each transect is interpreted to represent the local fish community within the reach, whether it was an protected
or unprotected reach.
Sampling strategy We did not perform sample size calculations. We selected two locations (plots) within each protected and unprotected reach within
each site. Plots were typically separated by 50+ meters along the river length, and individual plots were selected to capture a range
of habitat types. Observers moved in an upstream direction. These replicate observations in every reach are adequate to represent
the local fish community at the scale of our study. Similar methods using 50 m transects and underwater counts have been used to
estimate fish community composition for marine protected areas that are far larger than the systems we studied (e.g. Edgar et al.,
2014 in references). Our paired observations from protected and unprotected reaches at each site were then used to calculate a
suite of metrics of fish responses to reserve protection based on response ratios. Response ratios were then compared across all 23
sites.
Data collection Data collection in the field was performed by Aaron Koning and Martin Perales. Fish count data were collected using the methods
described above under Sampling strategy. Additional data collected included habitat survey: substrate classification, stream width
and depth measurements, and stream discharge. Substrate analysis was done by categorizing the substrate particle size (using a
modified Wentworth particle size classification) at 10 points along 6 lateral stream transects at each survey plot (n = 60). Stream
widths was measured 6 times (at fish count transect meters 0, 10, 20, 30, 40, 50), and depth was measured at ten points evenly
spaced across each of these width transects (n = 60 total). Stream discharge was measured once at each site using a USGS pygmy
flow meter following standard methods. Reserve feature information was collected from community members. Prior to surveying any
location, we explained the purpose of our study and requested approval from community leaders. At this point we collected
information from community members regarding the age of the local reserve, the penalty associated with illegal fishing activity, etc.
Reserve size, connectivity, and proximity to the nearest village and nearest road were all done using GIS.
Timing and spatial scale Sampling was conducted between December 2017 and March 2018. Sampling was conducted during the dry season, which extends
from November to April or early May. During this time, water clarity increases and depth decreases sufficiently to allow for visual
snorkel surveys. The Mae Ngao River Basin, in which each of our 23 survey sites is located, is approximately 1,000 square kilometers
in area. Our 23 sites were distributed throughout this valley (see Figure 1), and sampling at each site (including both protected and
unprotected reaches) extending along approximately 1 km of channel length.
Data exclusions No data were excluded from this analysis.
Reproducibility There was no experimental treatment used in this study. The protection status of each reach was considered to be a treatment for
statistical analysis. This designation of each reach as protected or unprotected was made by local communities, not by researchers.
The results reported herein are qualitatively similar to our observations from other years and locations. All aspects of this study could
be replicated, given community approvals to do so. Given limitations of timing and funding, we made the decision to sample a wider
number of sites for our study only once rather than sampling fewer sites multiple times. This decision enabled us to increase the
inferential power regarding which features or characteristics of reserves were most important for predicting outcomes of community
protection for fish diversity and abundance.
Randomization The 23 sites used in this study represent a gradient in the predictor variables of interest. Thus, while sites weren't randomly selected,
they did span differences in key parameters including reserve length, river size, local population size, years of protection, etc. Due to](https://image.slidesharecdn.com/koningetal2020-201202151822/85/Koning-etal-2020-15-320.jpg)
Uploaded byVictor Ha-Kã Azevedo
Koning etal 2020
1) A network of 23 small, community-based reserves in Thailand's Salween River basin protected tropical river fish diversity and enhanced fish communities inside reserve boundaries. 2) Reserves increased fish species richness by 27%, density by 124%, and biomass by over 2,000% compared to adjacent fished areas. Larger bodied and herbivorous fish benefited most. 3) Reserve characteristics like size, enforcement, and connectivity predicted the magnitude of ecological benefits, following principles from marine reserve design. Larger reserves with strong enforcement near villages had greater positive impacts.
More Related Content
![[Ostrom, 2009] a general framework for analyzing sustainability of social-e...](https://cdn.slidesharecdn.com/ss_thumbnails/ostrom2009-ageneralframeworkforanalyzingsustainabilityofsocial-ecologicalsystems-211108205623-thumbnail.jpg?width=640&height=640&fit=bounds)
Koning etal 2020
- 1. Nature | www.nature.com | 1 Article Anetworkofgrassrootsreservesprotects tropicalriverfishdiversity Aaron A. Koning1,2,3 ✉,K. Martin Perales1 , Etienne Fluet-Chouinard1,4 & Peter B. McIntyre1,5 Intensivefisherieshavereducedfishbiodiversityandabundanceinaquatic ecosystemsworldwide1–3 .‘No-take’marinereserveshavebecomeacornerstoneof marineecosystem-basedfisheriesmanagement4–6 ,andtheirbenefitsforadjacent fisheriesaremaximizedwhenreservedesignfosterssynergiesamongnearby reserves7,8 .Theapplicabilityofthismarinereservenetworkparadigmtoriverine biodiversityandinlandfisheriesremainslargelyuntested.Hereweshowthatreserves createdby23separatecommunitiesinThailand’sSalweenbasinhavemarkedly increasedfishrichness,density,andbiomassrelativetoadjacentareas.Moreover,key correlatesofthesuccessofprotectedareasinmarineecosystems—particularly reservesizeandenforcement—predictdifferencesinecologicalbenefitsamong riverinereserves.Occupyingacentralpositioninthenetworkconfersadditional gains,underscoringtheimportanceofconnectivitywithindendriticriversystems. Theemergenceofnetwork-basedbenefitsisremarkablegiventhatthesereservesare young(lessthan25yearsold)andarosewithoutformalcoordination.Freshwater ecosystemsareunder-representedamongtheworld’sprotectedareas9 ,andour findingssuggestthatnetworksofsmall,community-basedreservesoffera generalizablemodelforprotectingbiodiversityandaugmentingfisheriesasthe world’sriversfaceunprecedentedpressures10,11 . Overharvesting of fisheries threatens thousands of species and the food and nutrition security of hundreds of millions of people around the world12–14 . Over the past several decades, no-take marine reserves havebecomecentralmanagementtoolsforconservingmarineecosys- temsandsustaininglocalfisheries5,7 .Thewidespreadsuccessofmarine reservesinenhancingtheabundance,size,andbiomassoffishes4 has guided the distillation of design principles that maximize both con- servation efficacy within reserves8,15 and the export of harvestable biomassacrosstheirboundaries7,16,17 .Thesuccessofindividualmarine reservesinprovidingecologicalbenefitsvarieswidely18 ,buttypically increases with vigorous enforcement, complete bans on harvest, time since establishment, large size, and high degree of isolation8,15 . As catchable fish disperse across reserve borders, they can augment nearbyfisheriesandlocallivelihoods,therebyprovidinganincentive to respect reserve boundaries16 . Furthermore, gains in diversity and biomasswithinareservecanbeamplifiedbyexchangeamongnearby reserves7,16 , which has motivated the expansion of reserve networks worldwide. To date, the reserve network paradigm remains untested forhaltinghighratesofbiodiversityloss11 andbolsteringhumanfood supplies from the world’s freshwater ecosystems. Despiteholdingroughlyhalfofallfishspeciesandprovidingacces- sible sources of animal protein and critical micronutrients to many poor and undernourished populations12–14 , freshwater ecosystems are not well represented by existing protected areas9 . Indeed, formal protectionofriversandlakeshasbeenlargelylimitedtotheirincidental inclusion within terrestrial reserves19,20 . Although terrestrial reserves can provide watershed-level benefits, they often inadequately repre- sentfreshwaterbiodiversity21 andrarelyaddressoverfishing3 .Bycon- trast,recentmarinereserveparadigmshaveexplicitlyaimedtobalance protectionofbiodiversityinsidereserveboundarieswithsustainable harvests beyond them7 . Although the effectiveness of reserves as a large-scale management strategy continues to be debated18,22 , there is widespread evidence that small, community-based reserves can improvethesustainabilityofcoastalfisheries23 ,particularlyfornations withlimitedfisheriesmanagementcapacity24 .Suchanapproachoffers greatpotentialtoaddresshumanandecosystemneedsinlow-income countries where biodiverse rivers contribute disproportionately to inland fisheries12,13 , frequently suffer from overharvesting3 , and defy efforts to apply traditional fisheries management strategies12 . Southeast Asia is the only region of the world where riverine reservesarecommonplace;hundredsofcommunitieshavedesignated no-fishingzonesthattogetherformdefactoreservenetworksinrivers throughout the region25,26 . Here, we use one such network of small, community-based reserves located within the Salween River basin in northern Thailand (Fig. 1) to test how fish communities respond to protection. We studied 23 reserves that, despite a common cultural context(see SupplementaryInformation),representgradientsinage, size, enforcement, isolation, and network connectivity. Each of these factorscontributestothesuccessofmarinereserves7,8,17 ,sowetested their influence on the responses of fish species richness, density, and https://doi.org/10.1038/s41586-020-2944-y Received: 14 January 2020 Accepted: 5 October 2020 Published online: xx xx xxxx Check for updates 1 Center for Limnology, University of Wisconsin-Madison, Madison, WI, USA. 2 Cornell Atkinson Center for Sustainability, Cornell University, Ithaca, NY, USA. 3 Global Water Center, Department of Biology, University of Nevada, Reno, Reno, NV, USA. 4 Department of Earth System Science, Stanford University, Stanford, CA, USA. 5 Department of Natural Resources, Cornell University, Ithaca, NY, USA. ✉e-mail: akoning@unr.edu
- 2. 2 | Nature | www.nature.com Article biomass to protectionfrom fishing in rivers. These reserves are sur- rounded by reaches in which intensive harvests discriminate little among fish species, sizes, or trophic positions27 , providing a strong test of the overall effectiveness of community-based reserves as well as the design principles that can maximize their ecological benefits. Despite their small size, grassroots reserves enhanced the spe- cies richness, density, and biomass of protected fish communities enormously (Fig. 2). Relative to adjacent fished areas with compara- ble water depth and substrate composition (see Methods), reserves held an average of 27% more fish species (95% confidence interval (CI), 9–47%), 124% higher fish density (118–130%), and 2,247% higher biomass (1,460–3,433%). Whereas richness and density responses at our study sites were similar to those reported for marine reserves, mean biomass differences were six times higher than those typically observed in marine protected areas4 . These large biomass responses reflectdefaunationofthemajorityoftheriverbyfishers,whichcreates suchstarkdisparitiesthatunmarkedreserveboundariescanbereadily discernedbyeyebecausethevisibilityoflargefishesfromabovewater shifts sharply (Supplementary Fig. 1). The strength of responses to reserves varied among fish species accordingtobodysizeandtrophicguild.Relativelylargefishes(maxi- mumlength 200 mmormore;ExtendedDataFig. 1)disproportionately benefited from protection; their species richness increased by 59% (31–93%),densityby245%(234–257%),andbiomassby4,326%(2,490– 7,463%). By contrast, reserves had no effect on the richness, density, orbiomassofsmaller-bodiedfishspecies(maximumlengthlessthan 200 mm;P > 0.05).Amongtrophicguilds,herbivores(trophicposition (TP) ≤ 2.5) had the highest gains in richness (96% (22–221%)), density (117%(105–128%)),andbiomass(3,536%(1,688–7,293%))insidereserves. Omnivores (2.5 < TP ≤ 3.5) also benefited substantially; their richness increased 18% (1–39%), density 132% (125–140%), and biomass 2,327% (1,527–3,522%) relative to adjacent fished areas. Although predators (TP > 3.5)increasedindensityby179%(58–421%)andbiomassby136% (26–340%), we found no change in their richness (P > 0.05). Theresponsesofriverfishtoprotectionaccordwellwithseveralprin- ciplesofmarinereservedesign,butthedegreeofbenefitwasalsomedi- atedbyfunctionaltraits.Forinstance,therichnessresponse(Rr)across allspecieswasrarelycorrelatedwithreservecharacteristics,butsize, enforcement, and connectivity were all associated with higher Rr for particularfunctionalgroups(Fig. 3).Increasingreserveareaincreased Rr forallfishes,largefishes,andomnivores(Fig. 4),suggestingthatspe- cies–arearelationshipsapplyeventothesesmallhabitatpatches.For predatoryandherbivorousfishes,reservesinhigh-dischargereaches wereparticularlyimportantforrichnessgains(Fig. 4),probablyowing to the greater availability of deep-water habitats favoured by large species in these trophic guilds. Density responses (Dr) to reserves were strongly influenced by enforcement metrics. Specifically, Dr for all fishes and large fishes decreasedwithincreasedisolation,suggestingthatvillageproximity encouragescommunityvigilanceagainstillegalharvest(Figs. 3,4).The boundariesofseveralreservescorrespondedwiththemostupstream and downstream homes in the village, allowing near-constant sur- veillance while minimizing travel distances to adjacent fishing areas. Although close proximity to a village could increase fishing intensity outside reserves, village size—another proxy for potential fishing intensity—had little effect on observed outcomes (Fig. 3). The Dr for small fish was higher in reserves that had explicitly stated penalties forillegalharvest,regardlessofseverity,thaninthosethatlackedsuch community-imposed penalties (Fig. 4). Thus, it appears that merely havinganagreed-uponenforcementpolicycanbolsterreserveprotec- tion for some functional groups of river fishes28 . Reserve size was among the best predictors of reserve biomass response (Br) across all categories except for small fishes (Fig. 3), and village proximity remained important for Br of large fishes and herbivores(Fig. 4).Biomassgainsareparticularlyimportantforfisher- ies, as larger fish may be expected to provide both greater yields on fishing effort and disproportionately high reproductive capacity to Outlet to Salween River 2,000 m 1,000 m 250 m Reserve length Permanent streams Thailand China Myanmar Salween River basin 5 km Ngao River basin N Fig.1|RiverinereservenetworkwithintheMaeNgaobasin.Networkof52 community-basedreservesdocumentedin2018withinthe1,000-km2 Mae NgaoRiver,atributaryoftheSalweenRiverinnorthernThailand(inset). Reservesaredepictedbycirclesscaledaccordingtoreservelength.Purple denotesthe23reserveswherefishcommunitieswereprofiledforthisstudy, whichrepresentastratifiedsubsetofallknownreserves.Riverbasinand streamdelineationaredetailedintheMethods. –2 0 2 4 6 8 10 Reserve−non-reserve Rr Dr –0.5 0 0.5 1.0 1.5 2.0 2.5 log10 Reserve Non-reserve Br +1,000% +10,000% +124% +2,247% No change Fig.2|Fishresponsestoprotectioninreserves.Responsesoffish communitiestoprotectionfromfishing,intermsofspeciesrichness(Rr), density(Dr),andbiomass(Br).Pointsaremeanpairwisecomparisons(Rr isthe arithmeticdifferencebetweenreserveandnon-reservespeciesrichness;Dr andBr arelog10 responseratioscomparingreserveandnon-reserveareas).Grey rectangles,responsemeans;redcircles,responsemedians.
- 3. Nature | www.nature.com | 3 sustainreservepopulationsandseedunprotectedareasbyexporting juveniles16,17 . Ourresultsdemonstratethatsmallreserveshavegreatbenefitsfor intensively harvested fishesin this tropical river, even though their collectiveareaencompassesonly2%ofthechannelinourstudycatch- ment. The areas of individual reserves ranged from just 0.2 to 2.2 Ha (2,003–21,629m2 ),butbothfishrichnessandbiomassoutcomesscaled with reserve size. The increase in Br with area may result from greater resourceavailabilityprovidedbylargerreserves7 (Fig. 4),andthestrong area dependenceofRr forlargefishessuggestsaroleforthescalingof homerangerequirementswithbodysize29 .Notably,thelinearscaling ofgainsinrichnessandbiomasswithreservearea(Fig. 4)suggestthat modestexpansionsoftheboundariesofexistingsmallreservesmight yield considerable ecological benefits. Thesuccessofthisreservenetworkhasemergedthroughthevoluntary actionsofnumerouscommunities.Inreturnforcreatingandenforcing areserve,thelocalpopulationstandstobenefitdirectlyfromenhanced fishpopulations,therebycompletingavirtuouscyclethatencourages continuedcommunityaction30 .Successfulcommunity-basedinlandfish- erymanagementeffortshavebeendocumentedintheAmazon31–33 ,Bang- ladesh34 ,andelsewhereinSoutheastAsia25 ,butoursystematicanalysis acrossreplicatereservesintheSalweenbasinelucidateshowdecisions regarding size, location, and enforcement affect the ecological out- comesofprotection.Inparticular,twoaspectsofcommunityenforce- mentmaximizereservebenefits:closeproximitybetweenreserveand village, which increased density and biomass outcomes (Fig. 4), and explicitpenalties,whethermonetary(about$15–$300peroffence)or non-monetary(animalsacrificeorlibations).Thesefindingsunderscore thefactthatempoweringcommunitiestomanagelocalresourcescan achieveconservationandecosystemserviceoutcomesmoreeffectively than top-down, centralized management32,35,36 . Indeed, the grassroots reserves of the Salween basin, individually and collectively, exemplify thecriticalhumandimensionsofachievingecological,economic,and socialsuccessinnaturalresourcemanagement23 . Notably,reserveagewasnearlyalwaysnegativelyassociatedwiththe magnitudeofrichness,density,andbiomassbenefits(Fig. 3).However, ourdatagivenoindicationthatthepositiveeffectsofreserveseventu- ally wane. Negative responses were found in only six reserves (Fig. 2), each of which was relatively newly established (mean age, 5.3 years; median, 5.0 years),andsimplelinearregressionsshowedthatagehad no independent effect on any reserve response. In the best averaged models, age was strongly important only in conjunction with other variables that had higher importance scores, such as size, connectiv- ity,orenforcementvariables(Fig. 3,ExtendedDataTables 1–3).Thus, weinferthatthegradualdeclineinreservebenefitswithtimereflects improvements in fish communities in unprotected areas adjacent to reservesofsufficientsize,connectivity,andenforcement.Themecha- nismthatunderliesthereduceddisparitiesbetweenadjacentprotected and fished areas is presumably a rise in density-dependent spillover with reserve age (Extended Data Fig. 2). This is a critical insight with regardtobothconservationandlocalfoodsecurityintropicalrivers. Growingevidencefrommarinereservessuggeststhatspillovereffects aregreatestamongnetworksofmanysmallreserves,akintoourstudy system, rather than fewer large ones7,37 . Furthermore, estimates of spillover distance from marine reserve boundaries range from 102 to 103 m (ref. 37 ), which broadly matches the distances over which inten- sive fishing pressure occurs between reserves among our study sites. There is also circumstantial evidence that spillover of fish from our study reserves enhances local fisheries: fishers regularly harvest fish ofmanyspeciesthatarelargerthanwehaveobservedoutsidereserves during years of field work. The ongoing proliferation of new reserves suggests that subsistence fishing communities expect this conserva- tion approach to improve their own food security. Network-level benefits have emerged across our study reserves despite a lack of formal planning or coordination among communi- ties during creation of this de facto reserve network (see Supplemen- tary Information). Connectivity, measured as how central a reserve is withinthenetwork,wasimportantforRr ofallfishesandlargespecies, suggesting the importance of dispersal among protected areas for maintainingmetapopulations(Fig. 3,ExtendedDataFig. 3).Indendritic rivernetworks,dispersalpathwaysareeffectivelyone-dimensional,but centrallylocatedreservescanlinkpopulationsinmultipletributaries, thereby stabilizing metapopulations through time38 . Nonetheless, certain rare predators were found only in downstream reserves that had high discharge (Fig. 4). For large-bodied (>1 m) and far-ranging species, persistence may depend on immigration from larger rivers downstream,andindividualreservesarelikelytobetoosmalltosustain a viable population. Despitetheapparenteffectivenessofsmallreservesforprotecting the richness, density, and biomass of river fishes, questions about (Aikake weight) 1.0 0.5 0 1.0 0.5 Positive correlation Negative correlation Allfishes Large Sm all Predators O m nivores H erbivores Years protected Village distance Road distance Houses Penalty Closest reserve distance Distance to mouth Betweenness centrality Discharge Reserve area S S C C C E I I I,E A Rr Dr Br Allfishes Large Sm all Predators O m nivores H erbivores Allfishes Large Sm all Predators O m nivores H erbivores Importance Fig.3|Reservefeaturesvaryinbenefitsforfish.Richness(Rr),density(Dr), andbiomass(Br)gainsforallfishesandparticularfunctionalgroupsinbest averagedmodels.Cellcolouridentifiespositive(blue)ornegative(red)model coefficientsanddeepercoloursaturationsignifiesgreaterimportanceas Akaikeweightsummedacrossmodels.Letterson theleftindicatethemajor categoriesofreservefeatures:S,size;C,connectivity;E,enforcement;I, isolation;A,age.
- 4. 4 | Nature | www.nature.com Article theirlong-termimpactremain.Ifprotectedsub-populationsbecome isolated by intensivefishing between reserves, metapopulation per- sistence is unlikely39 . In highly seasonal rivers such as the Salween— tropical Asia’s longest remaining free-flowing river10 —connectivity between reserves is most likely during the wet season, when rising water levels simultaneously cue spawning migrations by some spe- cies and decrease the efficacy of all fishing methods. Relaxing these constraints on movement may enable a critical influx of individuals and alleles, alleviating demographic bottlenecks and inbreeding39 . It remainsunclear,however,whetherseasonalmovementamongreserves canovercomethelikelihoodofextinctiondebtswheneachindividual protected area is so small40 . Additionally, for long-distance migrants such as catadromous eels (Anguilla bengalensis), reserves can offer onlypartialprotectioninthefaceofproposedhydropowerandwater diversion projects throughout the Salween basin. Thedemonstrablebenefitsofreservesize,enforcement,andconnec- tivityinthisgrowingnetworkofgrassrootsriverinereservesshowthat manyprinciplesthatweredevelopedformarinereservesareapplicable torivers.However,ourfindingsalsorevealtheneedforcareinadapting marinereservedesignstosuitthephysicalstructureofrivernetworks, thedistinctivelifehistoryrequirementsoftheirfauna,andthecultures oflocalcommunitiesandfishers.Forexample,largereservesareclearly important, but maximizing area by exclusively creating reserves in widedownstreamriverreacheswouldleaveuniqueheadwaterspecies unprotected.Rather,riverinereservenetworksshouldbedesignedto encompass both high- and low-order segments38 . Additionally, while explicit fines did improve reserve outcomes, communal surveillance resultingfromclosevillageproximitymaybeevenmoreimportantfor deterringpoaching,becauseenforcementissochallenginginremote areas within large watersheds. More generally, cultural context must be accounted for in the design of effective enforcement strategies. AlthoughcommunitiesinnorthwesternThailandhavecreatedriverine reserves of their own accord, such grassroots networks may require encouragement and even subsidies in other locations. From an eco- logical perspective, reserves located centrally within networks may benefit many resident species and functional groups, but upstream– downstream connectivity along the entire river continuum will be required to conserve the migratory species that are critical in many river fisheries10,41 . While surely not a panacea for all that ails the world’s rivers, the reserves of the Salween basin demonstrate the benefits of a community-based model for protecting the biodiversity that under- pins ecosystem resilience and local subsistence fisheries. Like many rivers worldwide, the Salween is threatened by a combination of land usechange,intensifyingagriculturalpractices,non-nativespecies,flow modification, and intensive and indiscriminate fishing pressure11,42 . Networks of riverine reserves directly address only one stressor, but thebiologicaldiversityandecologicalprocessesthattheyprotectare likelytoaugmentlocalecosystemresilienceandmoderatetheimpacts of other stressors43 . Given the variety and magnitude of positive outcomes observed in each of these small reserves, as well as the emergent benefits of posi- tion within the ad hoc network, we expect that even greater fishery and conservation gains could be achieved through a more deliberate approach.Manyoftheprinciplesthathavebeendistilledfromsuccess- fulmarinereservesappeartoapplytorivers,andfurtherlessonsfrom marine and terrestrial networks of protected areas should be sought in order to strengthen ecosystem-based protection for freshwater systems.Accountingforbothlocalandnetwork-basedfeatureswillbe criticalforplanninganoptimalportfolioofriverinereserves.Perhaps evenmorecentraltoconservationsuccessarethehumanmotivations for establishing, enforcing, and harvesting from reserves. Our find- ings from the Salween prove that riverine reserves, when embedded in a conducive cultural context, can yield impressive benefits for the imperilledfreshwaterecosystemsuponwhichhundredsofmillionsof people depend worldwide. Onlinecontent Anymethods,additionalreferences,NatureResearchreportingsum- maries, source data, extended data, supplementary information, acknowledgements, peer review information; details of author con- tributions and competing interests; and statements of data and code availabilityareavailableathttps://doi.org/10.1038/s41586-020-2944-y. 1. Worm, B. et al. Rebuilding global fisheries. Science 325, 578–585 (2009). 2. Costello, C. et al. Global fishery prospects under contrasting management regimes. Proc. Natl Acad. Sci. USA 113, 5125–5129 (2016). 3. Allan, J. D. et al. Overfishing of inland waters. Bioscience 55, 1041–1051 (2005). 4. Lester, S. E. et al. Biological effects within no-take marine reserves: a global synthesis. Mar. Ecol. Prog. Ser. 384, 33–46 (2009). 5. Halpern, B. S., Lester, S. E. & McLeod, K. L. Placing marine protected areas onto the ecosystem-based management seascape. Proc. Natl Acad. Sci. USA 107, 18312–18317 (2010). 6. Carr, M. H. et al. Marine protected areas exemplify the evolution of science and policy. Oceanography (Wash. D.C.) 32, 94–103 (2019). 7. Gaines, S. D., White, C., Carr, M. H. & Palumbi, S. R. Designing marine reserve networks for both conservation and fisheries management. Proc. Natl Acad. Sci. USA 107, 18286–18293 (2010). Partialresiduals 0 1.25 2.50 –4 0 6 Reserve area (ha) 0 9 18 27 –10 –5 0 5 Years protected 0 1 2 –2 0 2 Distance to village (km) 10 40 70 –1.5 0 1.5 Distance to mouth (km) 0 1.25 2.50 –2.5 0 2.5 Reserve area (ha) 0 1 2 –2 0 2 Distance to village (km) All fishes Large fishes Small fishes Predators Omnivores Herbivores P < 0.05 P < 0.1 Rr Dr Br Fig.4|Scalingofbenefitswithkeyreservefeatures.Partialresidualplots illustratingrelationshipsbetweenimportantpredictorsofrichness(Rr), density (Dr),andbiomass(Br)responsestono-takereservesinbestaveraged models.Symbologyindicatesalternativegroupingsoffishspeciesbybodysize andtrophicgroup.FullmodelresultsarereportedinExtendedDataTables 1–3.
- 5. Nature | www.nature.com | 5 8. Edgar, G.J. et al. Global conservation outcomes depend on marine protected areas with five key features. Nature 506, 216–220 (2014). 9. Abell, R., Lehner, B., Thieme, M. & Linke, S. Looking beyond the fenceline: assessing protection gaps for the world’s rivers. Conserv. Lett. 10, 384–394 (2017). 10. Grill, G. et al. Mapping the world’s free-flowing rivers. Nature 569, 215–221 (2019). 11. Vörösmarty, C. J. et al. Global threats to human water security and river biodiversity. Nature 467, 555–561 (2010). 12. McIntyre, P. B., Reidy Liermann, C. A. & Revenga, C. Linking freshwater fishery management to global food security and biodiversity conservation. Proc. Natl Acad. Sci. USA 113, 12880–12885 (2016). 13. Fluet-Chouinard, E., Funge-Smith, S. & McIntyre, P. B. Global hidden harvest of freshwater fish revealed by household surveys. Proc. Natl Acad. Sci. USA 115, 7623–7628 (2018). 14. Golden, C. D. et al. Nutrition: Fall in fish catch threatens human health. Nature 534, 317–320 (2016). 15. Botsford, L. W., Micheli, F. & Hastings, A. Principles for the design of marine reserves. Ecol. Appl. 13, 25–31 (2003). 16. Gell, F. R. & Roberts, C. M. Benefits beyond boundaries: the fishery effects of marine reserves. Trends Ecol. Evol. 18, 448–455 (2003). 17. Hastings, A. & Botsford, L. W. Comparing designs of marine reserves for fisheries and for biodiversity. Ecol. Appl. 13, 65–70 (2003). 18. Pendleton, L. H. et al. Debating the effectiveness of marine protected areas. ICES J. Mar. Sci. 75, 1156–1159 (2018). 19. Chessman, B. C. Do protected areas benefit freshwater species? A broad-scale assessment for fish in Australia’s Murray–Darling Basin. J. Appl. Ecol. 50, 969–976 (2013). 20. Lawrence, D. J. et al. National parks as protected areas for U.S. freshwater fish diversity. Conserv. Lett. 4, 364–371 (2011). 21. Abell, R. et al. Concordance of freshwater and terrestrial biodiversity. Conserv. Lett. 4, 127–136 (2011). 22. Hilborn, R. Policy: Marine biodiversity needs more than protection. Nature 535, 224–226 (2016). 23. Halpern, B. S. et al. Achieving the triple bottom line in the face of inherent trade-offs among social equity, economic return, and conservation. Proc. Natl Acad. Sci. USA 110, 6229–6234 (2013). 24. Gill, D. A. et al. Capacity shortfalls hinder the performance of marine protected areas globally. Nature 543, 665–669 (2017). 25. Baird, I. G., Flaherty, M. S. & Baird, I. G. Mekong River fish conservation zones in southern Laos: assessing effectiveness using local ecological knowledge. Environ. Manage. 36, 439–454 (2005). 26. Loury, E. K. et al. Salty stories, fresh spaces: lessons for aquatic protected areas from marine and freshwater experiences. Aquat. Conserv. 28, 485–500 (2018). 27. McCann, K. S. et al. Food webs and the sustainability of indiscriminate fisheries. Can. J. Fish. Aquat. Sci. 73, 656–665 (2016). 28. Gutiérrez, N. L., Hilborn, R. & Defeo, O. Leadership, social capital and incentives promote successful fisheries. Nature 470, 386–389 (2011). 29. Kramer, D. L. & Chapman, M. R. Implications of fish home range size and relocation for marine reserve function. Environ. Biol. Fishes 55, 65–79 (1999). 30. Morrison, S. A. A framework for conservation in a human-dominated world. Conserv. Biol. 29, 960–964 (2015). 31. Campos-Silva, J. V. & Peres, C. A. Community-based management induces rapid recovery of a high-value tropical freshwater fishery. Sci. Rep. 6, 34745 (2016). 32. Castello, L., Viana, J. P., Watkins, G., Pinedo-Vasquez, M. & Luzadis, V. A. Lessons from integrating fishers of arapaima in small-scale fisheries management at the Mamirauá Reserve, Amazon. Environ. Manage. 43, 197–209 (2009). 33. Pinho, P. F., Orlove, B. & Lubell, M. Overcoming barriers to collective action in community-based fisheries management in the Amazon. Hum. Organ. 71, 99–109 (2012). 34. Thompson, P. M., Sultana, P. & Islam, N. Lessons from community based management of floodplain fisheries in Bangladesh. J. Environ. Manage. 69, 307–321 (2003). 35. Alexander, S. M., Epstein, G., Bodin, Ö., Armitage, D. & Campbell, D. Participation in planning and social networks increase social monitoring in community-based conservation. Conserv. Lett. 11, e12562 (2018). 36. Ostrom, E. A general framework for analyzing sustainability of social-ecological systems. Science 325, 419–422 (2009). 37. Halpern, B. S., Lester, S. E. & Kellner, J. B. Spillover from marine reserves and the replenishment of fished stocks. Environ. Conserv. 36, 268–276 (2009). 38. Campbell Grant, E. H., Lowe, W. H. & Fagan, W. F. Living in the branches: population dynamics and ecological processes in dendritic networks. Ecol. Lett. 10, 165–175 (2007). 39. Palumbi, S. R. Population genetics, demographic connectivity, and the design of marine reserves. Ecol. Appl. 13, 146–158 (2003). 40. Tilman, D., May, R. M., Lehman, C. L. & Nowak, M. A. Habitat destruction and the extinction debt. Nature 371, 65–66 (1994). 41. McIntyre, P. B. et al. in Conservation of Freshwater Fishes (eds Closs, G. P. et al.) 324–360 (Cambridge Univ. Press, 2015). 42. Dudgeon, D. et al. Freshwater biodiversity: importance, threats, status and conservation challenges. Biol. Rev. Camb. Philos. Soc. 81, 163–182 (2006). 43. Roberts, C. M. et al. Marine reserves can mitigate and promote adaptation to climate change. Proc. Natl Acad. Sci. USA 114, 6167–6175 (2017). Publisher’s note Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations. © The Author(s), under exclusive licence to Springer Nature Limited 2020
- 6. Article Methods Studysite We surveyed fishassemblages within the Mae Ngao tributary of the Salween River, the largest free-flowing river in Southeast Asia (Fig. 1). The Mae Ngao River basin of northwestern Thailand encompasses approximately1,000km2 ofmixeddeciduoussecondaryforest,swid- denandlowlandagriculturallands,andover80villages(>8,000people) of predominately ethnic Karen people. A land use survey conducted by the Land Development Department of Thailand in 2009 estimated basin-wideforestcovertobe70%,withagriculturecomprisingapproxi- mately27%oftheland,thoughtherehasbeensubstantialforestclear- ing for intensified forms of agriculture within the basin over the past decade44 .Duringthistimecommunitieshavemovedfromdiversified, low-inputswiddenagriculturalpracticestorowcrops,predominately soy,requiringsubstantialchemicalherbicideandpesticideinputs.The MaeNgaoRiverhasastronglyseasonalhydrograph,withasinglerainy seasonextendingfromMaytoOctoberandadryseasonfromNovember toApril.Duringtherainyseason,thedepthoftheNgaoRiverincreases by over 4 m relative to dry season baseflows. During dry season, the riverbecomesclear,allowingcensusoffishusingvisualmethods(with maskandsnorkel).Whilefishingeffortandgearsusedvaryseasonally, thecollectivefishingeffortofcommunitiesdispersedthroughoutthe valley encompasses nearly all unprotected waters, access to which is facilitatedbymotorbikesandextensivestreamsidetrails. Fishsurveys Toestimatethepotentialforreservestoincreasefishrichness,density, andbiomass,wesurveyed23pairedreserveandnon-reservelocations during a single dry season between December 2017 and March 2018. Individualreserveswereselectedtorepresentarangeinage,size,isola- tion, size of nearest village, stream order (first to fifth), and network position.Non-reservesurveyswereconducteddownstreamofreserves at all locations except one, to which there was only upstream access. Where stream segments were sufficiently narrow and shallow (19 of 23 sites), two researchers wearing dive masks and snorkels carried outfishcensusesbyswimmingorcrawlingalong50-mtransectsfrom downstream to upstream and enumerating all fish within a 2-m band centred on each observer. Inthefourlargemainstemsites(meanwidth>20m),oneresearcher surveyed 50-m-long reaches by systematically counting fish while movingfrombanktobankinanupstreamdirection.Eachsurveylasted 20 min, which approximated the average survey duration for shallow reaches.Additionally,toaccountforbenthicorcrypticspeciesinshal- lowwater(<80cm)atthesefoursites,asecondresearcherconducted four lateral belt transects at each survey reach. Belt transects were demarcatedwithachainplacedonthesubstrateperpendiculartoflow. To allow disturbance effects to dissipate, we waited for 15 min after chain placement to begin surveys, then enumerated all fish within a 2-mbandcentredonthechainforupto20mofstreamwidth.Foreach reserve–non-reservesite,atotaloffoursurveyswereconductedattwo reserve and two non-reserve reaches. To estimate site-specific biomass, each researcher estimated fish totallengthsduringeachtransect.Whenfewerthantenindividualsofa specieswereobserved,lengthswereestimatedforeachindividual.For species with more than ten observations in a survey, researchers esti- matedtenlengthsrepresentingthesizedistributionobservedforthat survey.Estimatedlengthswerecross-validatedbetweenresearchersin thefieldusingsubmergedmeasuringtapesasreference.Weestimated site-specificbiomassforreserveandnon-reservefishbyevaluatingthe meanobservedlengthforeachsitewithalength–weightrelationship developedfrompreviousworkintheNgaoRiver(A.A.K.,unpublished data) and supplemented with literature values45 . For the four large mainstemsites,wecombineddensityestimatesusingcount-weighted averages of both survey techniques for each survey location. Ethics oversight for the handling of animals was provided and the methods approved by the University of Wisconsin Research Animals ResourcesandComplianceandtheInstitutionalAnimalCareandUse Committee under protocol number L00447-0-02-12. Habitatvariables Ateachsurveysitewemeasuredkeyaspectsofreachhabitatsthatcould affect fish communities, then tested for habitat differences within (reserve vs. non-reserve) and among areas. At each transect location, wemeasureddepthandsubstratecompositionatsixlateraltransects correspondingto0,10,20,30,40,and50mmarks.Depthandsubstrate type were recorded at ten evenly spaced locations across the stream width.SubstratetypesfollowedtheWentworthclassification:silt(<62.5 μm); sand (62.5 μm–2 mm); pea gravel (2–8 mm); gravel (8–32 mm); pebble (32–64 mm); cobble (64–256 mm); boulder (>256 mm); bed- rock(>4,000cm).Formainstemsiteswherebothsnorkelandbenthic counts were employed, we conducted five lateral transects from the chain counts and snorkel surveys for a total of ten transects and 100 benthicpointsamples.Tocalculatemedianparticlesizeateachsite,we used the median size of each particle class for each observation, then calculatedthemedianacrossall60point-samples.Wealsomeasured discharge at each reserve–non-reserve location to account for the effectofsegmentsizeonreserveoutcomesusingstandardmethods46 . From these measurements we calculated mean depth, maximum depth, mean width, median substrate particle size, and three metrics of substrate diversity: Simpson’s diversity index of substrate types, andtheloadingscoresforthefirsttwoaxesofaprincipalcomponent analysisofsubstratetypesbysite.Simpson’sdiversityindexandprin- cipal component analyses were conducted in the ‘vegan’47 package in R48 . We tested for differences in each habitat variable using mixed effects models with reserve protection as a fixed categorical variable and each reserve–non-reserve location as a random term. Across all variables, reserve habitats did not differ from non-reserves for our study transects (P > 0.05). On the basis of these results, we ruled out any potential contribution of habitat differences to our analysis of reserve–non-reserve effects. Reservefeatures Toquantifythosereservefeaturesweconsideredpotentiallyimportant for predicting reserve success, we either made direct field measure- ments,extracteddatafromdigitizedmaps,orinterviewedcommunity membersateachsurveysite.Riversizequantifiedasdischarge(m3 s−1 ) was measured in the field using standard methods. Reserve size was quantified using field measurements of river width, multiplied by reserve length determined as river length between upstream and downstream reserve boundaries. To evaluate the spatial metrics of the reserve network, we digitized the Mae Ngao River network from a Google Earth49 satellite imagery base map in ArcGIS 10.350 and mapped all potential stream courses regardlessofthepresenceofvisiblesurfacewater,asnoexistinghydro- logicalmapswereatsufficientresolutionforourstudyregion.Weused fieldobservationstoconstrainthedigitizedstreamnetworkbasedon in situobservationofthepresenceofwaterin20tributariesduringthe height of dry season (early May 2016). Using our digital river network andahydrologicallyconditioneddigitalelevationmodelof90-mreso- lution51 ,weextractedtheupstreamcatchmentarea,thenestimatedan upstream area threshold that best separated wet from dry locations (pROC package52 ). We estimated an upstream extent of 1.02 km2 to bestdelineateperennialflows(receiveroperatingcharacteristicarea under the curve (ROC-AUC) = 0.89, n = 20) and trimmed our digitized stream network accordingly to a total length of 827 km of perennial rivers. We also delineated all roads and villages within the Ngao River Valley from the same satellite imagery, which allowed us to calculate Euclidean distances between each reserve and the nearest road and the nearest village as metrics of reserve isolation.
- 7. Considering these reservesas a network, we calculated three addi- tional parameters from our stream network that could influence eco- logicalresponses:riverdistancetothenearestreserve,riverdistance tomainstemconfluence,andbetweennesscentrality.Distancesamong reserves and to the river confluence were calculated from reserve boundariesandmeasuredalongtheperennialrivernetwork.Between- nesscentrality(BC)isanindexusedinnetworkanalysisthatdescribes therelativeimportanceofeachnode(thatis,reserve)tooverallconnec- tivitywithinthenetwork.Specifically,thestandardizedbetweenness centrality for a node i is calculated as: g i g N N BC = 2 × ∑ ( )/ ( − 1)( − 2)i j k jk jk< wherei≠j≠k,gjk isthenumberofequallyshortestpathsbetweennodes j and k, gjk (i) is the number of these paths that include node i, and the denominatorrepresentstwicethetotalnumberofnodepairswithout node i53 . We determined the reserve age (numbers of years since estab- lishment), enforcement (explicit penalty for illegal harvest: yes or no), and village size (number of households in sponsoring village) from more than 35 interviews with village leaders and community members. Dataanalysis To test for differences in species richness and abundance between reservesandunprotectedareas,weusedmixed-effectsPoissonregres- siontomodelspeciescounts,accountingforareasurveyedusinganoff- set.Wetestedbiomassdifferencesbetweenreservesandnon-reserves bymodelledbiomassperunitareausingamixed-effectslinearmodel, again with reserve as the lone predictor and site as a random effect. Mixed-effect models were analysed using the ‘lme4’ package54 in R48 . Totesttheeffectsofreservefeaturesonfishspeciesrichness,density, andbiomass,wecalculatedareserveresponseindexforeachoutcome variable. For species richness (S), we calculated a reserve response index (Rr) as Rreserve − Rnon-reserve, where R is the mean number of species observed during two surveys55 . We calculated the reserve response indices for fish density (Dr) and biomass (Br) as: log10(reserve mean/ non-reserve mean). In addition to testing the significance of overall reserve responses across sites, we used functional trait categorization of species to test for differential responses by size and trophic position. Both the mag- nitudeanddirection(±)oforganismalresponsestoreserveprotection have been shown to vary with life history and ecological traits56 . We dividedfishintolarger-bodiedandsmaller-bodiedcategoriesusinga threshold of 20 cm maximum length from observations across all 23 reserves. This cutoff was derived empirically from the size frequency distributionacrossallspeciesobserved(SupplementaryFig. 1).Trophic positions were estimated using our own nitrogen stable isotope data fromthestudyareawhenavailable,supplementedwithliteraturedata for 10 of 38 species45 . Weregressedthereserveresponseindicesforrichness(Rr),density (Dr),andbiomass(Br) againstallpredictorvariables.Westandardized all model coefficients, then evaluated all factorial models (n = 1,024) andquantitativelycomparedthesubsetofmodelswhosecumulative sum was 0.95 of the total Aikake weight, which corresponds to a 95% credibleintervalforbestmodels57 .Forthesubsetofbest-fittingmod- els,wesummedAikakeweights(wi)foreachpredictorandusedthem to estimate their relative importance57 . All analyses and model fitting were conducted using the ‘MuMIn’ package in R58 . Full model results are reported in Extended Data Tables 1–3. Reportingsummary Further information on research design is available in the Nature Research Reporting Summary linked to this paper. Dataavailability The datasets used and/or analysed during the current study are avail- able in the Environmental Data Initiative repository (https://portal. edirepository.org/nis/mapbrowse?packageid=edi.513.1). Codeavailability The R code used for the analyses presented here is available from GitHub (https://github.com/aakoning/riv_res_2020). 44. Koning, A. A., Moore, J., Suttidate, N., Hannigan, R. & McIntyre, P. B. Aquatic ecosystem impacts of land sharing versus sparing: nutrient loading to Southeast Asian rivers. Ecosystems (N. Y.) 20, 393–405 (2017). 45. Froese, R. & Pauly, D. (eds) FishBase http://www.fishbase.org (2019). 46. Lamberti, G. A. & Hauer, F. R. Methods in Stream Ecology (Academic, 2017). 47. Jari Oksanen, F. et al. vegan: Community Ecology Package. R package version 2.5-6 https://CRAN.R-project.org/package=vegan (2019). 48. R Core Team. R: A Language and Environment for Statistical Computing https://www. R-project.org/ (R Foundation for Statistical Computing, 2019). 49. Google Earth (November, 2015). Mae Ngao, Thailand. https://www.google.co.uk/earth/ (2020). 50. Environmental Systems Research Institute (ESRI). ArcGIS Release 10.3 (2015). 51. Yamazaki, D. et al. A high-accuracy map of global terrain elevations. Geophys. Res. Lett. 44, 5844–5853 (2017). 52. Robin, X. et al. pROC: an open-source package for R and S+ to analyze and compare ROC curves. BMC Bioinformatics 12, 77 (2011). 53. Jordán, F., Liu, W. C. & Andrew, J. D. Topological keystone species: measures of positional importance in food webs. Oikos 112, 535–546 (2006). 54. Bates, D., Maechler, M. & Ben Bolker, S. W. Fitting linear mixed-effects models using lme4. J. Stat. Softw. 67, 1–48 (2015). 55. Claudet, J. et al. Marine reserves: size and age do matter. Ecol. Lett. 11, 481–489 (2008). 56. Claudet, J. et al. Marine reserves: fish life history and ecological traits matter. Ecol. Appl. 20, 830–839 (2010). 57. Johnson, J. B. & Omland, K. S. Model selection in ecology and evolution. Trends Ecol. Evol. 19, 101–108 (2004). 58. Barton, K. MuMIn: Multi-Model Inference. R package version 1.43.6 https://CRAN.R-project. org/package=MuMIn (2019). Acknowledgements We thank the communities of the Mae Ngao River basin for their participation in and support of this research, and the International Sustainable Development Studies Institute (Chiang Mai) for logistical support. Earlier drafts of the manuscript were improved by the McIntyre laboratory groups at the University of Wisconsin-Madison and Cornell University, and by comments from I. Baird, B. Peckarsky, K. Winemiller, E. Stanley, T. Ives, R. Abell and M. Thieme. Funding was provided by National Science Foundation grants DGE-0718123, DGE-1144752, and DEB-15011836 to A.A.K. and DGE-1144752 to K.M.P., a Harvey Fellowship and a Cornell Atkinson Center for Sustainability Postdoctoral Fellowship to A.A.K., and a David and Lucille Packard Fellowship to P.B.M. Additional funding provided by USAID’s ‘Wonders of the Mekong’ Cooperative Agreement No: AID-OAA-A-16-00057. Author contributions A.A.K. conceptualized the research, acquired funding, performed fieldwork, conducted analyses, and wrote the manuscript. K.M.P. performed fieldwork, assisted in methodological development, and edited the manuscript. E.F.-C. conducted analyses, contributed to creating figures, and edited the manuscript. P.B.M. contributed to research conceptualization and methodological development, and edited the manuscript. Competing interests The authors declare no competing interests. Additional information Supplementary information is available for this paper at https://doi.org/10.1038/s41586-020- 2944-y. Correspondence and requests for materials should be addressed to A.A.K. Peer review information Nature thanks Edward Allison and the other, anonymous, reviewer(s) for their contribution to the peer review of this work. Reprints and permissions information is available at http://www.nature.com/reprints.
- 8.
- 9.
- 10.
- 11. Extended Data Table1 | Results of model averaging for richness reserve response (Rr) Aikake weight (wi), coefficient (β), and p-value (p) for two-sided z-test of predictors from best averaged models. All factorial models were fit (n = 1,024). The results reported are for the best aver- aged model obtained from the subset of models (# Models) having highest Akaike weights (wi), corresponding to a 95% credible interval for best fitting models.
- 12. Article Extended Data Table2 | Results of model averaging for density reserve response (Dr) Aikake weight (wi), coefficient (β), and p-value (p) for two-sided z-test of predictors from best averaged models. All factorial models were fit (n = 1,024). The results reported are for the best aver- aged model obtained from the subset of models (# Models) having highest Akaike weights (wi), corresponding to a 95% credible interval for best fitting models.
- 13. Extended Data Table3 | Results of model averaging for biomass reserve response (Br) Aikake weight (wi), coefficient (β), and p-value (p) for two-sided z-test of predictors from best averaged models. All factorial models were fit (n = 1,024). The results reported are for the best aver- aged model obtained from the subset of models (# Models) having highest Akaike weights (wi), corresponding to a 95% credible interval for best fitting models.
- 14. 1 natureresearch|reportingsummaryApril2020 Corresponding author(s): AaronA. Koning Last updated by author(s): Sep 8, 2020 Reporting Summary Nature Research wishes to improve the reproducibility of the work that we publish. This form provides structure for consistency and transparency in reporting. For further information on Nature Research policies, see our Editorial Policies and the Editorial Policy Checklist. Statistics For all statistical analyses, confirm that the following items are present in the figure legend, table legend, main text, or Methods section. n/a Confirmed The exact sample size (n) for each experimental group/condition, given as a discrete number and unit of measurement A statement on whether measurements were taken from distinct samples or whether the same sample was measured repeatedly The statistical test(s) used AND whether they are one- or two-sided Only common tests should be described solely by name; describe more complex techniques in the Methods section. A description of all covariates tested A description of any assumptions or corrections, such as tests of normality and adjustment for multiple comparisons A full description of the statistical parameters including central tendency (e.g. means) or other basic estimates (e.g. regression coefficient) AND variation (e.g. standard deviation) or associated estimates of uncertainty (e.g. confidence intervals) For null hypothesis testing, the test statistic (e.g. F, t, r) with confidence intervals, effect sizes, degrees of freedom and P value noted Give P values as exact values whenever suitable. For Bayesian analysis, information on the choice of priors and Markov chain Monte Carlo settings For hierarchical and complex designs, identification of the appropriate level for tests and full reporting of outcomes Estimates of effect sizes (e.g. Cohen's d, Pearson's r), indicating how they were calculated Our web collection on statistics for biologists contains articles on many of the points above. Software and code Policy information about availability of computer code Data collection Stream pathways were digitized using Google Earth satellite imagery version 7.1 (Google Earth, 2015). The digital elevation map used to generate upstream catchment data for trimming stream lengths is available upon request at http://hydro.iis.u-tokyo.ac.jp/~yamadai/ MERIT_Hydro/ (Yamazaki et al, 2017). Data analysis All statistical analyses and model fittings were conducted in R version 3.6.1 (R Core Team 2019) using the packages 'lme4' version 1.1-21 (Bates et al. 2015) or 'MuMIn' version 1.43.6 (Barton, 2019). Habitat diversity was analyzed using the 'vegan' package version 2.5-6 in R (Oksanen et al, 2019). Spatial analyses was conducted using ArcGIS version 10.3 (ESRI 2015). Custom R code for analyses are available on GitHub (https://github.com/aakoning/riv_res_2020). For manuscripts utilizing custom algorithms or software that are central to the research but not yet described in published literature, software must be made available to editors and reviewers. We strongly encourage code deposition in a community repository (e.g. GitHub). See the Nature Research guidelines for submitting code & software for further information. Data Policy information about availability of data All manuscripts must include a data availability statement. This statement should provide the following information, where applicable: - Accession codes, unique identifiers, or web links for publicly available datasets - A list of figures that have associated raw data - A description of any restrictions on data availability The data sets used and/or analyzed during the current study are available in the Environmental Data Initiative repository, (doi:10.6073/pasta/ e81a8c6eceff001e3e8b7062ad2ea483).
- 15. 2 natureresearch|reportingsummaryApril2020 Field-specific reporting Please selectthe one below that is the best fit for your research. If you are not sure, read the appropriate sections before making your selection. Life sciences Behavioural & social sciences Ecological, evolutionary & environmental sciences For a reference copy of the document with all sections, see nature.com/documents/nr-reporting-summary-flat.pdf Ecological, evolutionary & environmental sciences study design All studies must disclose on these points even when the disclosure is negative. Study description This study compared the richness, density, and biomass of fish from 23 protected river reaches and paired adjacent unprotected reaches. To evaluate the effect of protection on fish, we treated the protected and unprotected reaches at each site (n = 23) as experimental units within which each of two observers made fish observations at each of two plots per reach (n=184 total: 23 sites x 2 reaches per site x 2 plots per reach x 2 observers per plot x 1 observation per observer). In mixed effects models, protection status was coded as a binary fixed effect (N = no community protection, Y = community protection), and site (n = 23) was used as a random effect to account for site-level variation among paired reaches. Fish were enumerated during visual surveys conducted using mask and snorkel along longitudinal transects, as well as supplemental lateral transects at 4 of 23 sites where the river size was wide (>15 m width). Lengths of fish were estimated by eye for up to 10 individuals of each species; where >10 individuals were observed, only 10 lengths were estimated from a representative, haphazardly selected subset of individuals. Estimated lengths were used to calculate estimated biomass of each fish using species-specific regressions from captured conspecifics. Reserve characteristics were determined using GIS analysis, field measurements, and interviews with community leaders. Research sample We sampled (observed) only fish (Class: Actinopterygii) along 50-m-long transects, enumerating all fish observed within A 2-m-wide transect. In all study reaches, 100 square meters was sufficient to include observations from all major habitat types. Samples consisted of underwater fish counts made along transects by [two] observers wearing masks and snorkels. Because we made underwater observations of fish, it was not possible to determine the age range of fishes nor their sexes. While it is possible to estimate the age of fish based on their lengths, these statistical relationships have not been determined for nearly all of the species we observed, making it impossible to accurately determine their ages from estimated body length data. The sample population of fish observed along each transect is interpreted to represent the local fish community within the reach, whether it was an protected or unprotected reach. Sampling strategy We did not perform sample size calculations. We selected two locations (plots) within each protected and unprotected reach within each site. Plots were typically separated by 50+ meters along the river length, and individual plots were selected to capture a range of habitat types. Observers moved in an upstream direction. These replicate observations in every reach are adequate to represent the local fish community at the scale of our study. Similar methods using 50 m transects and underwater counts have been used to estimate fish community composition for marine protected areas that are far larger than the systems we studied (e.g. Edgar et al., 2014 in references). Our paired observations from protected and unprotected reaches at each site were then used to calculate a suite of metrics of fish responses to reserve protection based on response ratios. Response ratios were then compared across all 23 sites. Data collection Data collection in the field was performed by Aaron Koning and Martin Perales. Fish count data were collected using the methods described above under Sampling strategy. Additional data collected included habitat survey: substrate classification, stream width and depth measurements, and stream discharge. Substrate analysis was done by categorizing the substrate particle size (using a modified Wentworth particle size classification) at 10 points along 6 lateral stream transects at each survey plot (n = 60). Stream widths was measured 6 times (at fish count transect meters 0, 10, 20, 30, 40, 50), and depth was measured at ten points evenly spaced across each of these width transects (n = 60 total). Stream discharge was measured once at each site using a USGS pygmy flow meter following standard methods. Reserve feature information was collected from community members. Prior to surveying any location, we explained the purpose of our study and requested approval from community leaders. At this point we collected information from community members regarding the age of the local reserve, the penalty associated with illegal fishing activity, etc. Reserve size, connectivity, and proximity to the nearest village and nearest road were all done using GIS. Timing and spatial scale Sampling was conducted between December 2017 and March 2018. Sampling was conducted during the dry season, which extends from November to April or early May. During this time, water clarity increases and depth decreases sufficiently to allow for visual snorkel surveys. The Mae Ngao River Basin, in which each of our 23 survey sites is located, is approximately 1,000 square kilometers in area. Our 23 sites were distributed throughout this valley (see Figure 1), and sampling at each site (including both protected and unprotected reaches) extending along approximately 1 km of channel length. Data exclusions No data were excluded from this analysis. Reproducibility There was no experimental treatment used in this study. The protection status of each reach was considered to be a treatment for statistical analysis. This designation of each reach as protected or unprotected was made by local communities, not by researchers. The results reported herein are qualitatively similar to our observations from other years and locations. All aspects of this study could be replicated, given community approvals to do so. Given limitations of timing and funding, we made the decision to sample a wider number of sites for our study only once rather than sampling fewer sites multiple times. This decision enabled us to increase the inferential power regarding which features or characteristics of reserves were most important for predicting outcomes of community protection for fish diversity and abundance. Randomization The 23 sites used in this study represent a gradient in the predictor variables of interest. Thus, while sites weren't randomly selected, they did span differences in key parameters including reserve length, river size, local population size, years of protection, etc. Due to
- 16. 3 natureresearch|reportingsummaryApril2020 the pairing ofprotected and unprotected reaches at each site, we were able to rigorously control for larger-scale spatial variation in drawing inferences about the outcomes of community protection. Blinding Blinding was not relevant to our study; under field conditions, observers could not be prevented from perceiving whether a reach was protected or not. However, the same pair of investigators collected data in parallel from every reach, minimizing risk of bias during observations. Did the study involve field work? Yes No Field work, collection and transport Field conditions This study was conducted in the dry season, which extends from November through May. During this time water clarity allows for visual census of fish. Daily temperatures are generally cooler (highs:26º-30ºC, lows: 9º-12ºC) in December, January, and February, and become increasingly warm in March, April, and May (highs: 31º-34ºC, lows: 14º-20ºC). Surveys were conducted during daylight hours to ensure maximal fish visibility. Flow and habitat conditions at each site are reported in the supplemental information. Location This study was conducted in the Mae Ngao River Basin, in northwestern Thailand. The mouth of the river is located at N17.865980 E97.963357, an elevation of 145 m. Access & import/export There were no samples imported or exported for this study. We worked with community members to acquire permission to access community protected and unprotected areas of the Mae Ngao River. Disturbance Our study caused little to no disturbance because we used entirely non-destructive sampling (i.e., snorkel counts, habitat measurements, etc.). We minimized our disturbance by accessing locations using only existing paths and trails to access each reach. In the stream, we made efforts to avoid disturbing substrates, avoided causing bankside erosion, and ensured all flagging and marking materials were removed upon completion of our surveys. The individual surveys were of relatively short duration 20-45 minutes, allowing fishes to return to their habitats quickly. Reporting for specific materials, systems and methods We require information from authors about some types of materials, experimental systems and methods used in many studies. Here, indicate whether each material, system or method listed is relevant to your study. If you are not sure if a list item applies to your research, read the appropriate section before selecting a response. Materials & experimental systems n/a Involved in the study Antibodies Eukaryotic cell lines Palaeontology and archaeology Animals and other organisms Human research participants Clinical data Dual use research of concern Methods n/a Involved in the study ChIP-seq Flow cytometry MRI-based neuroimaging Animals and other organisms Policy information about studies involving animals; ARRIVE guidelines recommended for reporting animal research Laboratory animals No laboratory animals were used in this study. Wild animals Only fish (Class: Actinopterygii) were observed in the field. The species observed include: Parambassis vollmeri, Danio albolineatus, Pethia stoliczkana, Folifer brevifilis, Hara filamentosa, Nemacheiline loach, Chagunius bayeli, Xenentodon cancila, Barilius ornatus, Glyptothorax sp., Mystacoleucus argenteus, Systomus rubriprinnis, Garra salweenica, Poropuntius sp., Scaphiodonichthys burmanicus, Hampala salweenensis, Hypsibarbus salweenensis, Crossocheilus burmanicus, Mastacembelus armatus, Neolissochilus stracheyi, Tor sinensis, Raimas guttatus, Bangana devdevi, Sperata acicularis, Hemibagrus micropthalmus. Age-length relationships are unknown for nearly all of these species, and thus we were unable to determine their ages. However, from previous work we had already established length-weight regression relationship for nearly all species. For those species that we lacked robust length-weight relationships, we used publicly available published relationships (Froese and Pualy 2019). Additionally, determining the sex of these fish requires handling them (either to express gametes or dissection), which was not possible in the study due to local regulations that forbid catching fish in reserve areas. Field-collected samples This study did not involve laboratory study of field-collected samples. Ethics oversight Ethics oversight for the handling of animals was provided and the methods approved by the University of Wisconsin Research Animals Resources and Compliance and the Institutional Animal Care and Use Committee under protocol number L00447-0-02-12. Note that full information on the approval of the study protocol must also be provided in the manuscript.