EcologicalModelling222 (2011) 1471–1478
ContentslistsavailableatScienceDirect
Ecological

Modelling
journalhomepage:doc.guandang.net

l
ApplicationofaRandomForestalgorithmtopredictspatialdistributionofthepotentialyieldofRuditapesphilippinarumintheVenicelagoon,Italy
SimoneVincenzia, ,MatteoZucchettab,PieroFranzoib,MichelePellizzatoc,FabioPranovib,GiulioA.DeLeod,PatriziaTorricellib
a
DipartimentodiScienzeAmbientali,UniversitàdegliStudidiParma,VialeG.P.Usberti33/A,I-43125Parma,Italy
DipartimentodiScienzeAmbientali,InformaticaeStatistica,UniversitàCa’FoscariVenezia,Castello2737/B,30122Venezia,Italyc
AGRI.TE.COscAmbienteProgettoTerritorio,ViaCarloMezzacapo15,30175Marghera,Italyd
DipartimentodiScienzeAmbientali,UniversitàdegliStudidiParma,VialeG.P.Usberti11/A,I-43125Parma,Italy
b
articleinfoabstract
WepresentamodellingframeworkthatcombinesmachinelearningtechniquesandGeographicInfor-mationSystemstosupportthemanagementofanimportantaquaculturespecies,Manilaclam(Ruditapesphilippinarum).WeusetheVenicelagoon(Italy),the rstsiteinEuropefortheproductionofR.philip-pinarum,toillustratethepotentialofthismodellingapproach.ToinvestigatetherelationshipbetweentheyieldofR.philippinarumandasetofenvironmentalfactors,weusedaRandomForest(RF)algo-rithm.TheRFmodelwastunedwithalargedataset(n=1698)andvalidatedbyanindependentdataset(n=841).Overall,themodelprovidedgoodpredictionsofsite-speci cyieldsandtheanalysisofmarginaleffectofpredictorsshowedsubstantialagreementamongthemodelledresponsesandavailableecolog-icalknowledgeforR.philippinarum.Themostin uentenvironmentalfactorsforyieldestimationwerepercentageofsandinthesediment,salinity,andwaterdepth.Ourresultsagreewith ndingsfromotherNorthAdriaticlagoons.Theapplicationofthe ttedRFmodeltocontinuousmapsofalltheenvironmen-talvariablesallowedestimatesofthepotentialyieldforthewholebasin.Suchaspatialrepresentationenabledsite-speci cestimatesofyieldindifferentfarmingareaswithinthelagoon.Wepresentapos-siblemanagementapplicationofourmodelbyestimatingthepotentialyieldunderthecurrentfarmingdistributionandcomparingittoaproposedre-organizationofthefarmingareas.Ouranalysissuggestsareductionoftotalyieldislikelytoresultfromtheproposedre-organization.
© 2011 Elsevier B.V. All rights reserved.
Articlehistory:
Received25June2010
Receivedinrevisedform20January2011Accepted6February2011
Available online 4 March 2011Keywords:
RuditapesphilippinarumVenicelagoonRandomForestYield
Habitatsuitability
1.Introduction
TheManilaclamRuditapesphilippinarum(AdamandReeve,1850),whichisofIndo-Paci corigin,wasintroducedintheVenicelagoon(Fig.1)inthe80sasaculturespecies(CesariandPellizzato,1985)andradicallychangedtheexploitationoflivingresourcesinthelagoon.Withinafewyears,R.philippinarumbecamethemostimportantexploitedspeciesinthelagoon,withaproductionreach-ingapeakofover40,000ty 1attheendofthe90s,estimatedfromvarioussourcesandusingexpertknowledgebyPellizzatoandDaRos(2005).Noof cial sherylandingsdataforthewholelagoonisavailableandyieldpotentialislargelyunknown,despitetherelevantsocial,economicandenvironmentalconsequenceoftheexploitationactivities.
Sincetheintroduction,theexploitationofR.philippinarumhasbeencarriedoutinaregimeoffreeaccess.In1999,theProvince
Correspondingauthor.Tel.:+390521905696;fax:+390521906611.E-mailaddress:simone.vincenzi@nemo.unipr.it(S.Vincenzi).0304-3800/$–seefrontmatter© 2011 Elsevier B.V. All rights reserved.doi:10.1016/j.ecolmodel.2011.02.007
ofVenicebeganagradualshifttoaconcessionregime,i.e.,toasystemwhereharvestingareasaredividedbytheregulatoryagencyamonganumberofconcessions,eachmanagedbylocalclam shermenunderastrictsetofrulesonaccesslimitationandexploitationeffort.Technically,concessionsaredividedinfarm-ing(i.e.,whereclamsareseeded)and shing(i.e.,whereclamsarenaturallyrecruited)areas.Inthefollowing,wewillusethewordconcessionwithoutfurtherdifferentiatingbetweenfarmingand shingareas.In2007,about42km2oftheVenicelagoonweregiveninconcessionto shermenforharvestingofR.philippinarum(Fig.2aandTable1).
However,thetransitionfromuncontrolled shingtoa“culture-based shery”basedoncorrectandsustainablerearingprocedures,whilebeingsuccessfulinreducingproductionofR.philippinarum,revealedtobemorecomplexthanexpectedandcannotbecon-sideredsuccessfullycompleted(PellizzatoandDaRos,2005).TheProvinceofVeniceiswillingtoreducethenumberof sher-menoperatinginthelagoon(fromabout900to600)andtoremodelandreducetheareasgiveninconcessiontoclam sh-ermen(G.R.A.L.,2006,2009;ProvinceofVenice,2009)toreduce
connecting with the specific applicationof ...The multivariate normal distribution in n-dimensions... but the Naive Bayes algorithm is still ...
Application of Different... 11页 免费 EM-algorithm for Trainin... 暂无...was the ability to predict the spatial distribution of soil moisture content...
Application of homogenous continuous Ant Colony Optimization algorithm to_电子...random number randu uniform random number in [0, 1] s spatial position T...
so difficult to obtain in practical application....However, the spatial distribution of these classes...to predict the outputs of these test samples ...
A note on spatial partition models, with application to Bayesian mapping ...derive the joint posterior distribution of all the involved random quantities....
Spatial mixtures of Poisson distributions and application to disease mapping ...Thus the joint distribution of all variables corresponding to the Potts model...
A Receiver Algorithm of TDD-LTE Physical Random Access_专业资料 暂无评价
Predict how the fire will grow or urban ...Provides the same distribution options supported by...? ? ? ? What is the application? What do ...
SPATIAL DISTRIBUTION AND MERGING HISTORY OF DARK ...algorithm (PINOCCHIO, PINpointing Orbit-Crossing ... to predict accurately the formation and evolution...
? The bootstrap algorithm for estimating standard ...functional the distribution of thetahat that is ...Application: random forest Complete data: K ...
热门文档