different lengthsandvisualtextures.Intheseinvestigations, the analyzed inseveralflyingor walking insects,usingchannelsof one ofthefourpairseyes,i.e.ALEs. component ofthedisplacementseemstobeprocessedthrough only other eyeshadnoeffect. Therefore, itwouldappearthattheangular was preventedwhentheALEsweremasked,althoughmasking the dispersion ofthedirectionalbearings(Ortega-Escobar, 2011) that therewasasignificant white grating)wasrotatedby90deg, (ALEs). Whenthevisualstructureofsubstratum(black-and- and thatthisinputisonlyobtainedthroughitsanteriorlateral eyes needs somevisualinputfortheintegrationofangularcomponent Escobar, 2002b;Ortega-Escobar, 2006)haveshownthatthisspider Muñoz-Cuevas, 1999).Underindoorconditions,we(Ortega- the functioningofanteriormedianeyes(Ortega-Escobarand measured usingapolarized-lightcompass,whichisassociatedwith natural conditions,theangularcomponentofthisdisplacementis implies themeasurementofangularandlineardisplacement.Under Lycosidae) homesbymeansofpath integration.Thismechanism the anteriorlateraleyes. difference intheresolutioncapacityofposteriorlateraleyesand lateral opticflowovertheventralcanbeexplainedby lateral fieldofviewbeingthemostimportant.Thepreferencea by themotionofimageasspiderwalks,in burrow atveryvariabledistances.Thisvisualinformationiscreated eyeswerecovered,thespiderstendedtosearchfor the ’s visual informationcontributestodistanceestimationbecausewhen grating waschangedfrom the ventralfieldofview. Finally, theperiodoflateralorventral *Author ([email protected]) for correspondence © 2014.PublishedbyTheCompanyofBiologistsLtd|JournalExperimentalBiology(2014)217,395-401doi:10.1242/jeb.091868 Received 31May2013; Accepted25September2013 1 The wolfspider KEY WORDS:Orientation,,Opticflow black-and-white grating( eyes.Secondly,spider’s wechangedtheopticflowsuppliedbya manipulated theperceptionofvisualstimulibycoveringall experimental channelsplacedinanindoorsetup.Firstly, we after beingdisplacedfromitsburrowwasinvestigatedusing investigation. TheabilityofL. component ofthedisplacementwasnotknownpriortothis median eyes.However, how polarized-light compassassociatedwiththefunctioningofanterior angular componentofthedisplacementismeasuredusinga The wolfspider J. Ortega-Escobar (Araneae: Lycosidae) Visual odometryinthewolf spider RESEARCH ARTICLE INTRODUCTION ABSTRACT Faculty of Psychology,Faculty Madrid,28049-Madrid, Spain. of University Autónoma of The useoflineardisplacementinhoming(odometry)hasbeen tarantula 1, * andM.A.Ruiz λ 2cm) perceivedeitherinthelateralor =2 λ 2cmtoλ =2 L. tarantula homes usingpathintegration.The 1cm.Ourresultsindicatethat =1 (Linnaeus 1758)(Araneae, to gaugethedistancewalked 1 estimates thelinear spider colleagues (Seyfarthetal.,1982)investigatedhowthewandering channels toinvestigateaspectsoftheirorientation.Seyfarthand al., 2012)]. 2007; Narendraetal.,2008;SchwarzandCheng,2011; Schwarzet 2002) andMelophorusbagoti 2004; Stecketal.,2009;Wittlinger etal.,2007;Wohlgemuth et al., and Wehner, 1995;Ronacheretal.,2000;SommerandWehner, Eckles etal.,2012)anddesertants[ 2010), stinglessbees(Meliponaseminigra and Zhang,2004),bumblebees( Esch etal.,2001;Si2003;Srinivasan1997; 1996) orflying(Bairdetal.,2005;DackeandSrinivasan,2007; studied arehoneybees( regard tothevisualcontrolofflightspeed.Amongspecies using thismethodnotonlyinrelationtoodometrybutalsowith parts modified.Severalhymenopteranspecieshavebeenstudied which therelevantstimulihadbeenchanged,orsomeinsectbody insects weretrainedinonechannelandtestedadifferent onein carried outinorder toascertainwhethertheinformation wasgauged the eyesofspider was changed. would happeniftheself-induced opticflowperceivedbythespiders effect ofblindingthespidersontheirestimatedistance, orwhat the searchdensitydistributions. However, theydidnotstudythe changing avisuallandmarkplacedneartheburrowhadnoeffect on distance fromtheburrowbyidiotheticinformationand that of release).Theauthorsdeducedthat cmfromthepoint cm(theburrowwas30 mean distanceof27.6±9.3 displacement, andwithoutavisuallandmark,thespiderswalked a landmark neartheburrow, onthedistancewalked.Afteractive passive displacement,andthepresenceorabsenceofavisual longitudinal channelstoinvestigatetheinfluenceofactiveversus 2009) intheirstudyofthewolfspider walked. Reyes-Alcubillaandcolleagues(Reyes-Alcubillaet al., suggesting thatnovisualinputwasnecessarytogaugethedistance (allofwhichwereblinded)remainedquiteaccurate, Regarding odometry, thedistanceestimatedbyexperimental trajectories butmostofthemdidnotreturntotheirstartingpoint. Srinivasan, 1981).Theoperatedspidershadsimilarmultiple-step loop ashasbeendescribedindesertants(Turner, 1907;Wehner and the prey, andwhentheywereneartotheirgoalmadeaTurner’s trajectory thattookthemneartothepointwheretheyhadabandoned retrace theirstepsbutwalkedbackfollowingamultiple-step removed andonegroupofintactspiders.Thespidersdidnot (cuticular proprioceptors)onalllegfemorahadbeensurgically divided intotwogroups:onegroupinwhichthelyriformorgans or throughasemicircularcorridor. All thespiderswereblindedand some prey, whenthespidersweremovedawayeitherrectilinearly Lycosa tarantula There haveonlybeenafewstudiescarriedoutonspidersusing In thisstudy, weanalyzedtheeffect oftemporarilyblindingall Cupiennius salei L. tarantula estimated thedistancetoagoal,inthiscase Apis mellifera (Narendra, 2007;Narendraetal., Bombus terrestris) (Bairdetal., on thedistancewalked.Thiswas Cataglyphis fortis L. tarantula ) eitherwalking(Schöne, ) (Hrnciretal.,2003; L. tarantula,used could gaugethe (Ronacher 395

The Journal of Experimental Biology 396 information availableastothekindofbehavior This experimentwasusedasacontrolbecausethereno respectively. (interquartile range)ofthedistancewalkedandlatencyperiod, conditions, thecentraltendency(median)anddispersion 1and2represent,forallexperimentsexperimental Figs perceived bydifferent eyes. whether thismodificationwasimportantinbothfieldsofviewas of thismodificationontheodometryspiderandtoascertain flow perceivedeitherlaterallyorventrally, toinvestigatetheeffect solely byidiotheticmeans.We alsomodifiedtheself-inducedoptic RESEARCH ARTICLE gauge thedistancewalked? to Experiment 1:doesthespiderneedvisualinformation RESULTS n alt h te Fg 3,‘alleyes’ condition).Displacements in one walltothe other (Fig. the channelandthispatternwas intermixedwithdisplacementsfrom dissecting microscope. the eyecoveringswereremoved andtheirintegritycheckedusinga submitted to10testtrials.After thecompletionofexperiment, described previously(Ortega-Escobar, 2006)andthespiderswere uncovered, alltheireyeswerecoveredfollowingprocedures burrow, thedistancewalkedwasscoredaspositive. turnandwalkedtowardsthevirtual positionofthe made a180deg was scoredasnegative.However, ifthespider inthetestchannel the samedirectiontoendoftestchannel,distancewalked ‘test channel’ wereplainwhite.Ifthespidercontinuedwalkingin to thetestchannel.Thewallsofboth‘spiderchannel’ andthe away fromtheirburrow, placedinaclearglasscupandtransferred cm with alltheireyesuncovered;theyweremovedadistanceof30 but withtheireyesmasked.Theanimalsweresubmittedto10 trials Subsequently, thisexperimentwasrepeatedusingthesamespiders were uncoveredandconsequentlyhadcompletevisualinformation. walked. Initially, odometrywasstudiedusingspiderswhoseeyes exhibit intheabsenceofvisualinformationtoestimatedistance Turning point (cm) In general,thespiderswalkedin contactwithoneofthewalls After 10controltrialsinthetestchannelwiththeireyes –10 10 20 30 40 50 0 l ysNoeyes All eyes Position: Grating: Frequency: No grating x.1Ep x.3Ep Exp.5 Exp.4 Exp.3 Exp.2 Exp. 1 ***** ogtdnlCosLniuia rs rs rs rs Cross Cross Cross Cross Cross Longitudinal Cross Longitudinal λ=2 al usrtmWlsSubstratum Walls Substratum Walls λ=2 λ=2 L. tarantula λ=2 λ=2 would λ=1 eoetrigo erhn f2.±1cm( before turningorsearchingof26.5±11 during displacement. condition). Thespidersdidnotmakeexploratorylegmovements 3,‘noeyes’ that includedcrossingsfromonewalltotheother(Fig. homeward pathwasamultiple-steptrajectoryincontactwithwall that ofthesameanimalswhentheireyeswereuncovered; locomotive behaviorofthespiderswithcoveredeyeswassimilarto performing eitherthereturnlooporsearchpatterns.The step trajectory, i.e.theywalked,stoppingseveraltimesuntil were notobserved.Spidersmadetheirinboundpathinamultiple- which thespiderswalkedthroughcentralpartofchannel ( cm when theireyeswerecoveredthemeandistancewas18.4±18.8 condition’ asafixedeffect, eyes werecovered(mixedeffects modelanalysiswith‘eye cm)inaplacenearertothefictiveburrowthanwhenalltheir at 30 with alltheireyesuncoveredsearchedforthefictiveburrow(placed respectively). compared withthe‘noeyes’ condition(0.06%and25%, the spiderchannel,althoughthiswasaverysmallproportion when walking inthesamedirectiontheywerewhentaken from Spiders withuncoveredeyeswerealsoobservedtocontinue was insufficient forthemtobeablereturnthevirtualburrow. virtual burrow, whichsuggeststhattheproprioceptiveinformation Sometimes theydidnotturnandthereforewalkedawayfromthe when thespidershadalltheireyescovered(‘noeyes’ condition). variation inthedistancetosearchingmovementsorturningpoint observed inthe control condition. behavior whileintheotherthree thisbehaviorwassimilartothat five animalshadaverylarge distributionofburrow-searching the placewherevirtualburrow was.Inthe‘noeyes’ condition, of twoanimals,thesearchfor the burrowwasmainlyconfinedto cm.With theexception walked ameandistanceslightly lessthan30 cm,andfiveanimalsthatsearchedorturnedafterhaving almost 30 for theburroworturnedafterhavingwalkedameandistance of t 79 λ=2 The spiderswithalltheireyesuncoveredwalkedameandistance Regarding theturningpointsforinboundsearches,spiders In thecontrolcondition,therewerethreeanimalsthatsearched =–5.55, λ=1 The JournalofExperimentalBiology(2014)doi:10.1242/jeb.091868 P 001.A a ese nFg 1,therewasmore <0.001). AscanbeseeninFig. ( circles representoutliers.*Significantdifference whiskers correspondtotheextremevaluesand substratum. Boxesshowmediansandinterquartiles, the different channelswithgratingsinthewallsor control (‘alleyes’)and‘noeyes’ conditions,andin spiders. Dataareshownfor Turning pointsofinboundsearchesby Fig.1. condition. on therightcorrespondingtoexperimental corresponding tothecontrolconditionandone summarized bytwoboxes,theoneonleft separate different experiments.Eachexperimentis position ofthevirtualburrow;verticaldashedlines P <0.05). The dashed line at 30cm indicatesthe <0.05). Thedashedlineat30cm F 1,101 =18.019, P <0.001). t 79 Lycosa tarantula =–2.87, P <0.05) but in the

The Journal of Experimental Biology spider tosearchfortheburrow. covering thespider’s eyesdidnotalterthegeneralmotivationof RESEARCH ARTICLE the wallsof spiderchannelwerealwayslined withagratingof Materials andmethods(see‘General procedure’).Inthisexperiment the PME,cannotbeusedreliably toobtainopticflow. possibilities becauseweconsider thataneyelooksfrontally, like In thisexperimentwedidnot discriminatebetweenthesetwo would perceivethestripesmoving immediatelylateraltothespider. lateral(Land,1985),andbythePLEs,which frontal to60–80deg a certaindistance,giventhatPMEvisualfieldsextendfrom 0deg could beperceivedbothbythePMEsforstripesplacedfrontally at the lateralfieldofview. Inourexperimentusinglateralgrating,this distance fromit.ThePMEslookforwardandthePLEstowards substratum andwhenthespiderwalkstheymaintainaconstant of thestimuliusedinthisstudy. TheALEsareorientedtowardsthe zenith anddonothavearoleindetectingthevisualcharacteristics 1992). Theanteriormedianeyes(AMEs)aredirectedtowards the L. tarantula eyes (PLEs)(Land,1985;Kovooretal.,1992).Thevisualfields of mainly bytheposteriormedianeyes(PMEs)andlateral to itsself-displacement. experiments wemodifiedtheopticflowperceivedbyspiderdue information togaugethedistancewalked.Inthisandfollowing implies thatthespidersareusingsomeotherkindofsensory marked effect onthedistanceestimatedbyspiders.Thisresult As experiment1demonstrated,coveringalloftheeyeshada effects modelanalysis, period towalkinthetwoconditionswasnotsignificant(mixed 2).Thedifference betweenthelatency eyes werecovered(Fig. whenthespider’s spider’s eyeswereuncoveredand538.9±272.3s estimate thedistancewalked? Experiment 2:does Latency (s) The spidersweredisplacedinthe samechannelsasdescribedin When stimuliarepresentedinthelateralfield,theyperceived The mean latency periods to walk were 556.6±283.6s whenthe The meanlatencyperiodstowalkwere556.6±283.6s 1000 1250 250 500 750 0 l ysNoeyes All eyes are different foreacheye(Land, 1985;Kovooretal., Position: Grating: Frequency: No grating x.1Ep x.3Ep Exp.5 Exp.4 Exp.3 Exp.2 Exp. 1 NS L. tarantula L. F 1,139 ogtdnlCosLniuia rs rs rs rs Cross Cross Cross Cross Cross Longitudinal Cross Longitudinal =0.170, λ=2 al usrtmWlsSubstratum Walls Substratum Walls NS use lateralopticflow to λ=2 P =0.681). Thissuggeststhat λ=2 NS λ=2 ϫ ϫ λ=2 NS λ=1 ϫ F effect overthedistancewalked(mixedeffects model analysis, P each conditionfor eachoneoftheeightfemale spiders. the channel(cross-stripedgrating). Ten trialswerecarried outin the stripeorientationwasperpendicular tothelongitudinalaxisof substrates ofbothchannelswere longitudinallystriped.Forthetest, parallel tothelongitudinalaxis ofthechannel.Fortraining, always consistedofablack-and-white grating( channel wereplainwhite.The substratumofthespiderchannel Kovoor etal.,1992). visual fieldsaredirectedtowardsthesubstratum(Land,1985; Stimuli intheventralvisualfieldareperceivedbyALEs,whose needed to( and searchedfortheburrowfurtherthanpointwherethey cm.Thespiderswalked striped testchannel(training)was34±6.6 3). displacements fromonewalltotheother(Fig. walls ofthechannelandthispatternwasintermixedwith channel. Ten trialswerecarriedoutineachconditionforspider. same gratingbutorientatedperpendiculartothelongaxisof grating. Forthetest,wallsoftestchannelwerelinedwith spider andtestchannelswerelinedwiththesamelongitudinal-stripe that suppliesverylittleopticflow. Fortraining,thewallsofboth orientated inthelongitudinaldirectionofchannel,acondition cmand black-and-white stripes(widthofthestripe=1 ts)tema itnewle a 4981cm( (test) themeandistancewalkedwas24.9±8.1 estimate thedistancewalked? orientation (mixedeffects model analysis, was nosignificantdifference inthelatencytowalkduestripe fortrainingandtestconditions,respectively. There 387.9±259.5s Experiment 3:does 1,124 <0.001). Thefixedeffect ‘stripeorientation’ hadasignificant λ=2 In thisexperiment,thewallsofspiderchannelandtest The meandistancethespiderswalkedinlongitudinally As inexperiment1,thespiderswalkedcontactwithoneof Mean latencyperiodstowalkwere380.9±217.2and =72.567, * λ=1 The JournalofExperimentalBiology(2014)doi:10.1242/jeb.091868 t 79 =4.98, P <0.001). condition. on therightcorrespondingtoexperimental corresponding tothecontrolconditionandone summarized bytwoboxes,theoneonleft the different experiments.Eachexperimentis difference ( interquartile rangesfromtheboxedge.*Significant crosses representoutliersatmorethan1.5 the extremevalues,circlesrepresentoutliersand medians andinterquartiles,whiskerscorrespondto under differentconditions. Latency periodtowalkbyallthegroups 2. Fig. P L. tarantula L. <0.001). Inthecross-stripedtestchannel P <0.05). Vertical dashedlinesseparate use ventral opticflow to F 1,108 The boxesshow =0.416, λ 2cm)orientated =2 P t 79 =0.520). λ =–5.35, 2cm) =2 397

The Journal of Experimental Biology channel (mixedeffects modelanalysis, orientation’ had asignificanteffect overthedistancewalkedin analysis, had asignificanteffect onthedistancewalked(mixedeffects model λ for eachspider. previous experiments,10trialswerecarriedoutineachcondition channel werelinedwithcross-stripesofperiod cross-stripes ofperiodλ 398 ( the wallsofspiderchannelwerealwayslinedwithcross-stripes Materials andmethods(see‘Generalprocedure’).Inthisexperiment cm)inthetestchannel. width=0.5 λ width ofthechannel,wealteredperiodcross-stripes from for thevisualsystem.Inthisexperiment,insteadofchanging the smaller (widerchannel)orlarger (narrowerchannel)imagemotion cross-striped channels,thestripewidthwaschangedtoinducea In someexperimentscarriedoutusinginsects(e.g.honeybees)in RESEARCH ARTICLE cross-stripe grating thespiderslookedfor burrowatashorter cm( 27.0±9.1 9491cm( 29.4±9.1 significant (mixedeffects modelanalysis, mean difference betweenthetwolatenciestowalkwasnot fortrainingandtestconditions,respectively. The 219.0±159.5s of the cross-stripes liningthewalls? thecross-stripes of 3182cm( 33.1±8.2 other. with thewallsandalternatingmovementsfromonewallto turn,thespidersalsowalkedkeepingincontact making a180deg experiment (paperoveraplasticsheetinbothchannels).After the wayspiderswalkedoverartificialsubstratumusedinthis Experiment 4:can the spiders changed from 30.8±7.1cm ( the spiderschangedfrom30.8±7.1cm λ 2cm to =2 2c srp it= cm)inthespiderchannelto cm(stripewidth=1 =2 All eyes 2cm).Fortraining,thewallsoftestchannelwerelinedwith =2 When thestripeperiodof lateral gratingwaschangedfrom The spidersweredisplacedinthesamechannelsasdescribed in Mean latencyperiodstowalkwere256.2±136.5and The meandistancewalkedbythespidersintrainingtrialswas Compared withexperiments1and2,therewasnodifference in F Experiment 1 1,123 λ Point ofrelease Virtual burrow 1cm inthetestchannel,meandistancewalkedby =1 t =10.576, 79 t t 79 79 =–2.94, =3.37, =–0.54, L. tarantula L. P No eyes P=0.004). Thefixedeffect ‘stripeperiod’ <0.001). Thisshowsthatwitha half-period P 2cm.Forthetest,wallsoftest =2 <0.001) andinthetesttrialsitwas P =0.590). Thefixedeffect ‘stripe distinguish the grating period distinguish thegrating F 1,127 F 1,109 t 79 =11.258, Longitudinal =1.06, stripes =2.236, λ λ 1cm.Asin =1 1cm(stripe =1 P P Experiment 2 =0.292) to <0.001). P Point ofrelease Virtual burrow =0.138). stripes Cross walk (mixedeffects modelanalysis, fixed effect ‘stripeperiod’ hadnosignificanteffect onthelatencyto respectively. The conditions was325.9±145.5and329.0±186.7s, further decreasedthedistancespiderswalked. in imagemotionontheeyeduetohalf-periodofgrating observed inexperiment2(seeabove),thisanincrease distance. Bearinginmindthelarge reductioninthedistancewalked temperatures. predators, wouldbeabletostalk itspreyandcouldavoidhigh and entertheburrow, aplacewhereitwouldfeelsecure from form offood.Inourstudythe reward forL.tarantula insects (honeybees,desertants),wheretherewasarewardin the These studiesaredifferent to thosethathavebeencarriedoutusing ( each conditionforspider. ( ( cm training, thestripewidthinsubstratesofbothchannelswas1 the substratumalwaysconsistedofacross-stripedgrating.For substratum fromλ In thisexperimentwechangedtheperiodofcross-stripesin turned duringthe outwardjourney(Ortega-Escobar, 2006). The to estimatethedistancewalked astheyneedtoestimatetheangles P liningthesubstratum? thecross-stripes of 2186cm( 32.1±8.6 latency towalk(mixedeffects modelanalysis, The fixedeffect ‘stripeperiod’ hadasmall, significanteffect onthe respectively. and testconditionswas307.5±230258.9±196.6s, DISCUSSION Experiment 5:can F walked wassmallbutsignificant(mixedeffects modelanalysis, λ λ t 1,120 =0.048). 79 2c) o h et h tiewdhi h etcanlws05cm cm).Forthetest,stripewidthintestchannelwas0.5 =2 1cm).Asinpreviousexperiments, 10trialswerecarriedoutin =1 The meanlatencyperiodtowalkforspidersundercontrolandtest Our studyshowsthatL.tarantula cm For training,themeandistancewalkedwas33.6±8.3 The wallsofthespiderchannelandtestwereplainwhite; The meanlatencyperiodtowalkforthespidersundertraining =3.84, =4.797, P The JournalofExperimentalBiology(2014)doi:10.1242/jeb.091868 crosses fromonewalltotheother. In bothexperimentstherearetrajectoriesinwhichtheanimal which thespiderswalkedindirectionofoutboundpath. stops. Inthe‘noeyes’ conditiontherearethreetrajectoriesin the pointofrelease.Theblackdotsrepresentwherespider rectangle (whichrepresentsatopviewofthetestchannel)at condition areshown.Trajectories beginatthetoppartof experiment 2.Foreachexperiment,sixtrajectoriesfor The trajectoriesofexperiments3–5aresimilartothose 3.Examplesofrealtrajectoriesexperiments1and2. Fig. <0.001) whilefortestthemeandistancewalkedwas P t =0.030). 79 =2.20, 2cmtoλ =2 L. tarantula L. P =0.030); thedifference in the distance 1cm. =1 distinguish the grating period distinguish thegrating females needvisualinformation F 1,83 =0.303, P =0.584). F was tofind 1,99 =4.022,

The Journal of Experimental Biology RESEARCH ARTICLE for PMEs and 0.78deg forPLEs. Clearly, theALEsareless for PMEsand 0.78deg forALEs,0.63deg Cuevas, 1997):infemalestheangles are1.70deg light intensity, duringtheday. However,(40lx) inindoorconditionswithahigher 1992) underindoorconditionswherethereisreducedlightintensity show nocturnalbehavior(Ortega-Escobar, 2002a;Ortegaetal., this species(Seyfarthetal.,1982). this idiotheticorientationin success ratewaslessthan50%.Seyfarthandcolleaguesassociated colleagues destroyedthelyriformorgans onallthefemora, obtained byitsproprioceptors.Consequently, whenSeyfarthand masked, thisindicatesthatalltheinformationmusthavebeen where itwasmovedaftercapturingitsprey. Asallitseyeswere a similarway. contrast tothatshownby constantly movingupanddown. illumination notperceivedbythespider, theforelegswere (Schmid, 1997)forCupienniussalei movements wasobserved.ThisiscontrarytothefindingsofSchmid However, inthisconditionnodifference inthepatternofleg all theireyesweremasked,thedistancewalkedwasveryvariable. looking fortheburrowincorrectdirection.Furthermore,when turnand they hadwalkedpreviously, instead ofmakingthe180deg of displacements(25%).Thespiderswalkedinthesamedirection behavior ofthespidersbecamemorevariableinahighpercentage burrow position.Whenalltheireyesweremasked,thesearching all theireyesbegantosearchforitatapointlittlebeforethefictive spiders thatsearchedfortheburrowwithvisualinformationthrough calculated for (Land, 1985;LandandNilsson, 2002).Thisanglehasbeen resolution istheinter-receptor anglesubtendedbytworeceptors retinas withfinemosaicsofphotoreceptors. Onemeasureofthe different eyes?Inordertoresolvefinedetails,theeyesmust have marked. cm)beforetheactualsite,buteffect wasless at adistance(3.7 in theventralfieldofview, thespidersalsosearchedforburrow the spidertolocateburrow. Whentheopticflowwasgenerated this group.Thismeansthatlateralopticalflowisveryimportant for distance theyhadtowalk.Thiseffect wasseeninalltheanimals cmbeforethesiteofburrow, approximatelyone-thirdofthe 9.6 to thewalkingdirection,spiderssearchedforfictiveburrow there wasnotasignificantdifference inthelatencytowalkperiod. masked lookedfortheburrowandwerenotlessmotivatedbecause followed byaTurner’s loop.Eventhosespiderswithalltheireyes of themtriedtofindtheburrowafteramultiple-stepdisplacement polarized lightcompass,thespiderswerewellmotivatedbecauseall absence ofanycompassinput,forexampleasunor fields ofview. abundance ofmovingvisualstimulibothinthelateralandventral not forageinafeaturelesslandscapebutonewhichthereisan desert antC.fortis,andsimilarlytothehoneybee, equates welltonormaldistancesforoutdoorconditions.Unlikethe unpublished). Therefore,thedistanceusedintheseexperiments cm(J.O.-E., returning totheburrowfromdistancesof~30–40 of theburrow, andduringhottemperaturestheycanbeobserved autumn seasons,thefemalescanalsobeobservedinupperpart unpublished). Inoutdoorconditions,duringthespring,summeror cmfromit(J.O.-E., part oftheburroworatadistance3–4 The behaviorof Could thiseffect berelated to theresolutioncapabilitiesof When lateralopticflowwasinducedbyagratingperpendicular Although ourstudywascarriedoutinindoorconditionsthe L. tarantula Cupiennius salei L. tarantula .tarantula L. C. salei females andmales(Kovoor Muñoz- females havebeenobservedintheupper C. salei was abletoreturnthepointfrom with allitseyesmaskedwasin (Seyfarth etal.,1982)studiedin Lycosa tarantula with thenocturnalbehaviorof where, underinfrared L. tarantula females also does functioning inthis case. the learnedruleofwalkingtowards theendofchannelisnot they didnot,withtheexception ofexperiment5.Thissuggeststhat overshot thelocationoffictive burrow, butinthetestconditions that inthetrainingconditionsof experiments2,3and5thespiders learned rulebefunctioninginour studywithL.tarantula?Itistrue along thechannelbecausetheirgoal isattheendofit.Couldasimilar fact alearnedrule,andfollowingthistheantsshould run enclosed byarectangularbox(Schwarzetal.,2012).Theycalled this nest, wasduetothetrainingconditions,whichcompriseda nest systematic drift,sometimesaccompaniedbyovershootingthefictive search behavioroftheseants.Schwarzandcolleaguesfoundthat the and colleagues(Narendraetal.,2008)isnotastereotypicalpart of the towards thefictivenestobservedinlongitudinalchannelsbyNarendra colleagues (Schwarzetal.,2012)haveshownthatthesystematic drift localization inthecentralnervoussystemcanonlybespeculated. functioning ofastrideintegratorwhosesensoryinput and underestimated it(Wittlinger etal.,2007). Theauthorssuggestthe overestimating thehomingdistancewhilestumpsants and stumpsantsmisjudgedthedistancetowalk,withstilts where theyhadbeentrained.Ontheirreturnjourneyboththestilts in leglengthwerecarriedoutaftertheantshadarrivedatafeeder shortening them(theso-called‘stumpsants’).Thesemodifications manipulated byelongatingthelegs(theso-called‘stiltsants’)or 2007; Wolf, 2011). Inthese studies,legandstridelengthswere information tomeasurethedistancetravelled(Wittlinger etal., was uncovered. statistically different fromthatofthegroupinwhichventralhalf was coveredbuttheyfoundthatthedistancewalkednot experimental groupinwhichtheventralhalfofcompoundeye fortis they analyzedtheeffect ofamputatingtwothewalkinglegs obtained byWittlinger andWolf inthecourseofastudywhich affect thedistancewalkedinatestchannel.Similarresultswere of theeyessomeantsandfoundthatthisconditiondidnot on thedistancetravelled.Inaddition,theycoveredventralhalf mm.Therewasnosignificantinfluence ofthespatialfrequency 20 the antsusingstationarypatternsinwhichstripewidthwas5or mmblack-and-white-stripes( grating of10 (Ronacher andWehner, 1995).Theytrainedantsusingastationary similar tothisstudy, placing visualpatternsonthesubstratum et al.,2000).RonacherandWehner carriedoutanexperiment sufficient nornecessaryforcorrectdistanceestimation’ (Ronacher results, theopticflowinlateralvisualfieldofantis‘neither Ronacher etal.,2000;Wittlinger andWolf, 2013).Contrarytoour carried outusingdesertants . Similarstudieswithself-inducedopticflowhavebeen shall relateourresultstothoseobtainedforotherwalking in themillimeterrangeorofrandomdotpatterns. we proposeinfutureexperimentstoinvestigatetheeffect ofgratings have aninfluenceongaugingthedistancetoburrow. Therefore, distance totheburrow. These experimentsshowthatopticflowdoes not knowwhetherthiswouldbeanimportantfactoringaugingthe study carriedoutonthisspiderusingkindofstimulusandwedid finer thanthoseusedinthisinvestigation.However, thisisthefirst that thePLEretinaof the ventralopticflowislesseffective thanthelateralopticflow. discriminative thanthePMEsandPLEs,thiscouldexplainwhy Using theAustraliandesertantMelophorusbagoti,Schwarzand Other studieshaveclearlyshownthat In theabsenceofsimilarresearchconductedusingspiders,we From previousdata(KovoorandMuñoz-Cuevas,1997)wecansee (Wittlinger andWolf, 2013).Inthisstudytheyhadone The JournalofExperimentalBiology(2014)doi:10.1242/jeb.091868 L. tarantula C. fortis is capableofresolvinggratings (Ronacher andWehner, 1995; C. fortis λ 2 mm)andthentested =20 uses proprioceptive 399 C.

The Journal of Experimental Biology 400 cm cup andtransferred tothesamepointintestchannel (channel3),30 the burrowintheir channel(channels1or2),placed inatransparentglass test channelwasthoroughlycleaned withabrushaftereachtest. were nosilkthreads,whichcould have actedasacue,thesubstratein release, therewasapointknownas thevirtualburrow. To ensurethatthere cmfromthepoint of the burrowshouldhavebeen,at a distanceof30 burrow inthemiddletestchannel(channel 3).However, intheplacewhere around theburrowonlyduringtheirnocturnaldisplacements.Therewas no cmindiameterwasbuilt. deep and2.5 cm from oneendofthelateralchannels,anartificialburrowmeasuring 17 cm cmabovetheterrarium.At12 fluorescentbulbspositioned160 36W channels forthedurationofexperiment.Lightwasprovidedby four, beforethebeginningofstudyandthey livedinthese terrarium 3days cmhighwalls.Thespiderswereplacedinthe cmwide,with10 9.5 cmlongand 4).Thechannelswere52 of thesamelengthandwidth(Fig. The spiderswereplacedinaterrariumthatwasdividedintothreechannels Adult virgin femalesof RESEARCH ARTICLE proprioceptive methods. displacements arealsocontrolledbyvisualinformationoronly under infraredlighttodeterminewhethertheirnocturnal of thespidersbychangingtheirphotoperiodandobservingthem through theALEs. and toalesserdegreebytheventralregionofvisualfield the lateralregionsofvisualfield,probablythroughPLEs, tarantula modified bytheperiodofgrating.Asinhoneybee, honeybee odometerreliesprimarilyonvisualinformationandisnot that linedthewallsoftunnel.Thiswouldsuggest distribution wasnotalteredbychangingtheperiodofgrating tunnel (Srinivasanetal.,1997).Theyalsofoundthatthesearch grating) theylookedforthetrainedrewardthroughoutentire were trainedandtestedinatunnelwithlittleopticflow(longitudinal Zhang, 2004).Srinivasanandcolleaguesfoundthatwhenhoneybees et al.,2001;Si2003;Srinivasan1997;and visual stimuli(Bairdetal.,2005;DackeandSrinivasan,2007;Esch carried outonflyingbeestrainedtoflyinchannelswithdifferent estimation ofdistance. cases, therewasaninfluenceoftheventralopticflowon increase oftheopticflow, the oppositesituationwasfound.Inboth walking speedandover-estimated thedistance.Whentherewasan When therewasadecreaseintheopticflow, thebeesincreasedtheir grating couldbemovedeitherinthebee’s directionoragainstit. is stationaryandthegratingmoved.Inbothcasesflooror stripes (λ the channelismovedoverastationarygratingofblack-and-white were twowaysofinducingopticflow:(1)thetransparentfloor walked alongachannelthatledthemtothehiveentrance.There (Schöne, 1996),anotherwell-studiedinsect.Inthisstudy, thebees General procedure General MATERIALS ANDMETHODS 40°32′ animals hadbeencapturedfromawildpopulationinMadrid(centralSpain; vomitoria) andgivenwatertwiceaweek. move aroundanddigburrows.Theywerefedblowflies( cmwithsufficient substratum(earth)to containers measuring17×13×8 conducted aftermaturation.Thespidersweremaintainedinindividual laboratory; theageofallanimalswasverysimilarandtrialswere nalteeprmns h pdr eegnl uhd3 cmawayfrom In alltheexperiments,spiders were gentlypushed30 The spidersleftsilkthreadsallalongthespiderchannels,1and2, and It isnecessarytocarryoutsimilarstudiesinthesubjectivenight However, themajorityofworkonodometryinbeeshasbeen Schöne carriedoutexperimentsusingwalkinghoneybees N, 3°42′ 1c ratraiey05o cm);(2)thetransparentfloor cmoralternatively 0.5or2 =1 the odometerisalsomainlyvisualandsensedoverallby W) andhadbeenthroughthefinaltwotothreemoltsin L. tarantula were usedforallexperiments.These Calliphora L. For eachtrial,the distancewalkedbythespiderbefore searchingaswellits experiment, 80training trialsand80experimental trials werecarriedout. submitted to10trainingtrials and 10experimentaltrials.Ineach motivation tosearchfortheburrow. measured. Thislatencyperiodis consideredtobethemeasureof channel substratumandthemoment itbegantowalk(latency)wasalso loop). Thetimelapsebetweenthemoment thespiderwasplacedontest changeindirection(Turner’s searching, orifitmadeacomplete(180deg) if itmadesearchmovementsusinglegpairsIandII,typicalofburrow was consideredthataspiderhadwalkedthecorrectdistancetoitsburrow overshot itandproceededtowardstheendnearesttovirtualburrow). It turnandwalkedtowardsthevirtualburroworeven did makea180deg cm(ifthey to theendofchannelfurthestfromvirtualburrow)and42 turninthedirectionofvirtualburrowandproceeded not makea180deg and Muñoz-Cuevas,1997).Spiderscouldwalkbetween (Kovoor PLEs (inter-receptor anglesubtendedbytworeceptorsis0.78deg) any case,eachstripecouldbeperceivedfromtheneareroneby of the In cmstripeequivalentto29deg). coarser intheeyenearesttowall(a1 and cmstripeequivalentto8deg) in theeyefurthestfromwall(a1 the visualinformationgatheredthrougheachPLEwasdifferent, beingfinest of thesubstratumtomeasuredistancewalked.Asaresultthigmotaxis, one wallofthechannel(thigmotaxis),apaperrulerwasplacedincenter camera positionedabovetheterrarium.Asspidersmovedincontactwith displacement ofthespiderswasfilmedusingaPanasonicSDR-H80video was takenviatheglasscupandputbackneartoitsburrow. The it see clearlythroughtheglasscup.Ifspiderdidnotmoveafter20min, from thevirtualburrow. Duringtransfertothetestchannel,spiderscould longitudinally stripedbutitwascross-stripedfortesting. the gratingwasplacedonsubstratum.Fortraining, burrow. (B)Exampleofatopviewchannel3foranexperimentinwhich cmfromthe point ofreleasethetestchannel(channel3)positioned30 channel’. Spiderswalkedalongtheirchannelandweretransferredtothe ‘spider channels’ each containingonespider. Channel3wasthe‘test terrarium dividedintothreechannels(1–3).Channels1and2werethe 4.Set-upusedtostudyodometrybyL.tarantula. Fig. In eachexperiment,eightfreshfemale spiderswereused.Eachspiderwas B A 2 3 1 The JournalofExperimentalBiology(2014)doi:10.1242/jeb.091868 Virtual burrow Virtual burrow Virtual burrow 30.0 cm 52.0 cm Training Test Point ofrelease Point ofrelease Point ofrelease (A) Top viewofthe − 10cm (if theydid 10cm

9.5 cm

The Journal of Experimental Biology RESEARCH ARTICLE commercial, ornot-for-profitsectors. This researchreceivednospecificgrantfromanyfundingagencyinthepublic, reviewed themanuscript.Bothauthorsreadandapprovedfinal manuscript, participatedintheinterpretationofdataandstudyresults,critically statistical dataanalysis,participatedinthedraftinganddevelopmentof animals. M.A.R.contributedtoreviewthedesignofexperiments,carriedout and inreviewing.J.O.-E.wasresponsibleforthecapturemaintenanceof interpretation offindingsandthedevelopmentmanuscript,bothindrafting experiments, collectedthedata,carriedoutinterimdataanalysis,andguided J.O.-E. developedtheconceptionofstudy, designedandcarriedoutthe The authorsdeclarenocompetingfinancialinterests. improved themanuscriptsubstantially. in thismanuscript.Ourthanksalsogotothetwoanonymousreviewerswho We wouldliketothankEmilio Ortega-Escobarforhispreciseworkwiththefigures this effect hasonlytwolevels(training–test). conditions wereinterpretedandno considered arandomeffect. Onlydifferences betweenthe twoexperimental considered arandomeffect, andthedifferent animalsmeasuredwerealso considered asafixedeffect, repeatedmeasurementswithinconditionwere condition factor. Inthemodel,experimentalconditionfactorwas measurements foreachanimalshouldbenestedwithintheexperimental each ,givingatotalof20measurements.Hence,repeated measures. Within eachexperimentalcondition10trialsweremeasuredfor experimental conditions(trainingversustest)aretreatedasrepeated any visualconditioninthetestchannel.Inproposedmodeltwo repeated measuresANOVA) wascarriedouttostudytheeffect ofchanging cmforallanimals). the pointofrelease(establishedasafixeddistance30 performed tocomparetheinbounddistanceagainstburrowfrom percentiles, depictingthedispersionofdata.One-sample range (25%and75%percentiles)thewhiskersare10th90th whisker plots,wherethecenterismedian,spreadinterquartile period arereportedasmeans±s.d.andtheyalsoillustratedusingbox-and- levels ofsignificanceweresetto All datawereanalyzedusingthestatisticalsoftwareIBMSPSS19.0. a previouspaper(Ortega-Escobar, 2006). latency periodweremeasured.Theeye-coveringprocedureisdescribedin ak,M n rnvsn M.V. Srinivasan, and M. M. Dacke, Dacke, and T. Kornfeldt, E., Baird, A. Cowling, and S. Zhang, M.V., Srinivasan, E., Baird, References Funding Author contributions Competing interests Acknowledgements Data andstatistics analysis cls .A,Rui,D .adNe,J.C. Nieh, and D. W. Roubik, M.A., Eckles, oor .adMñzCea,A. Muñoz-Cuevas, and J. Kovoor, F. G. Barth, and R. Zucchi, S., Jarau, M., Hrncir, sh .E,Zag . rnvsn .V n at,J. Tautz, and M.V. Srinivasan, S., Zhang, H.E., Esch, guided groundspeedcontrolinbumblebees. speed inhoneybees. the thirddimension. 133-145. of lynxspiders(Oxyopidae)anditsrelationtohabitatbehaviour. odometry toestimatebothheightanddistance. 768. seminigra communicate distancesmeasuredbyopticflow. In allexperimentsamixedeffects longitudinallinearmodel(similartoa ) usesopticflowtoestimateflightdistances. J. Exp.Biol. J. Exp.Biol. (2007). Honeybeenavigation:distanceestimationin 210 208 (1997). Comparativestructureofthevisualsystem α <0.05. Thedistanceswalkedandthelatency , 845-853. , 3895-3905. othoc post (2010). Minimumviewingangleforvisually J. Exp.Biol. (2012). A stinglessbeecan use visual J. Exp.Biol. Nature (2003). A stinglessbee( tests werenecessarybecause J. Comp.Physiol.A (2005). Visual controlofflight 411, 581-583. 213 (2001). Honeybeedances 215 , 1625-1632. , 3155-3160. Zool. Anz. t -tests were 189 Melipona , 761- 235 , itigr .adWl,H. Wolf, and M. Wittlinger, W S.W. Zhang, and M.V. Srinivasan, itigr . enr .adWl,H. Wolf, and R. Wehner, M., Wittlinger, M.V. Srinivasan, and R. Wehner, H. Wolf, and M. Wittlinger, K., Steck, unr C.H. Turner, aeda . hn,K n enr R. Wehner, and K. Cheng, A., Narendra, rnvsn .V,Zag .W n iwl,N.J. Bidwell, and S.W. Zhang, M.V., Srinivasan, omr .adWhe,R. Wehner, and S. S. Sommer, Zhang, and M.V. Srinivasan, A., Si, efrh .A,Hrerdr . be,H n at,F. G. Barth, and H. Ebbes, R., Hergenröder, E.-A., Seyfarth, rea . uz .adFrádzMnrvt,C. Fernández-Montraveta, and M. Ruiz, J., Ortega, R. Wehner, and D. Sulikowski, K., Cheng, A., Narendra, A. Narendra, reaEcbr J. Ortega-Escobar, reaEcbr J. Ortega-Escobar, ad .adNlsn D.-E. Nilsson, and M. Land, J. Ortega-Escobar, and A. Muñoz-Cuevas, J., Kovoor, cwr,S,Shlhis .adCeg K. Cheng, and P. Schultheiss, S., Schwarz, A. Schmid, reaEcbr J. Ortega-Escobar, ad M. Land, of H. Wolf, cwr,S n hn,K. Cheng, and S. Schwarz, H. Schöne, Ronacher J. Ortega-Escobar, reaEcbr .adMñzCea,A. Muñoz-Cuevas, and J. Ortega-Escobar, oahr .adWhe,R. Wehner, and B. Ronacher, J. Ortega-Escobar, and M.A. Ruíz, C., Reyes-Alcubilla, hgmt,S,Rnce,B n enr R. Wehner, and B. Ronacher, S., ohlgemuth, integrator thataccountsforstridelengthandwalkingspeed. Paris remains unaffected bydisturbanceofwalkingbehaviourandvisualinput. Cataglyphis Comp. Neurol. Exp. Biol. 210 207. odometry inhoneybees. Annu. Rev. Neurosci. experiments withdesertants, the visuallydriven‘odometer’. ants, J. Exp.Biol. memories oftheoutbounddistanceinAustraliandesertant bagoti distance. orientation ofawanderingspider:compensationdetoursandestimatesgoal artificial channels. locomotor activityinalycosidspider. ants inlandmark-richhabitats. odometry inguidingthesearchbehaviorofdesertants, Biol. does notinfluencedistanceestimationinthedesertant Rhythm Res. tarentula premiers ganglionsoptiques. fasciiventris tarentula Lycosa third dimensionindesertants. walking honeybees. 861-870. Naturforsch. the substratum.J.Exp.Biol. Lycosidae) areusedduringorientationtodetectchangesinthevisualstructureof tarentula Lycosidae) needsvisualinputforpathintegration. memory retrievalintheAustraliandesertant optic flowtomeasuredistancestravelled. visual landmarks. spider electroretinographic analysis. tarentula , 1798-1803. 203 aalpi fortis Cataglyphis 107 . I.Proportionalpath-integrationtakestheanthalf-wayhome. (2011). Odometryandinsectnavigation. yoatarantula Lycosa . alzi . oleuh .adWhe,R. Wehner, and S. Wohlgemuth, K., Gallizzi, B., , (1985). Themorphologyandopticsofspidereyes.In , 1113-1121. (Araneae, Lycosidae) duringpathintegration. , 130-136. Behav. Ecol.Sociobiol. (Araneae, Lycosidae) detectpolarizedlight:behavioralexperimentsand 212 (Araneae, Lycosidae) andthepathwaysofocularentrainment. (1997). A visuallyinducedswitchinmodeoflocomotionaspider. (ed. F. G.Barth),pp.53-78.Berlin:Springer-Verlag. (1996). Optokineticspeedcontrolandestimationoftraveldistancein (1907). Thehomingofants:anexperimentalstudyantbehavior. (Formicidae, Hymenoptera). 210 The JournalofExperimentalBiology(2014)doi:10.1242/jeb.091868 52C (2007). HomingstrategiesoftheAustraliandesertant 33, 561-576. , 2893-2901. 17, 367-434. , 570-577. , 124-128. Naturwissenschaften (2002b). Evidencethatthewolfspider (2006). Roleoftheanteriorlateraleyeswolfspider Behav. Processes (2002a). Circadianrhythmsoflocomotoractivityin (2011). Anteriorlateraleyesof J. Comp.Physiol.A 27, 679-696. , remainsunaffected bydisturbanceofwalkingbehaviour. J. Exp.Biol. (Araneae, Lycosidae): theroleofactivelocomotionand (2013). Homingdistanceindesertants, (2002). 214 (2011). Visual discrimination,sequentiallearningand (2004). Theant’s estimation ofthedistancetraveled: Ann. Sci.Nat.Zool. (1995). Desertants J. Arachnol. (Araneae, Lycosidae). I.Organisationdesnerfset aalpi ots J.Comp.Physiol.A fortis. Cataglyphis J. Exp.Biol. J. Comp.Physiol.A J. Comp.Physiol.A , 2375-2380. 11 (1981). Searchingbehaviourofdesertants,genus Animal Eyes , 139-148. J. Interdiscipl.CycleRes. (2009). Estimationofhomingdistanceindesert 91, 298-303. 200 (2004). Visual motorcomputations ininsects. 96, 485-494. (2003). Honeybeenavigation:propertiesof 179 , 2513-2522. J. Comp.Physiol. (2007). Thedesertantodometer:astride (2007). Acquiring,retainingandintegrating J. Comp.Physiol.A 27, 663-671. 206 (1999). Anteriormedianeyesof , 587-592. , 1265-1273. J. Exp.Biol. eohrsbagoti Melophorus . Oxford:OxfordUniversityPress. 13, 25-36. (2002). Distanceestimationinthe (2012). Visual cuelearning and aalpi fortis Cataglyphis J. Arachnol. 194 188 , 929-938. , 273-281. J. Arachnol. (1992). Lesystèmevisualde yoatarantula Lycosa (2008). Searchstrategiesof yoatarentula Lycosa (1997). Visually mediated (2009). Hominginthewolf (1992). Dailypatternsof aalpi ots J.Exp. fortis. Cataglyphis 142 214 (2000). Lateralopticflow eohrsbagoti Melophorus J. Exp.Biol. 177 30, 481-486. 23, 295-301. , 315-338. , 1629-1641. eohrsbagoti. Melophorus , 21-27. . aalpi fortis Cataglyphis (1982). Idiothetic 34, 51-61. Anim. Cogn. use self-induced Neurobiology of 190 J. Exp.Biol. , 1-6. Melophorus J. Physiol. (Araneae, (Araneae: 210 Lycosa Lycosa Lycosa , 198- 401 Biol. , in 14, Z. J. J. ,

The Journal of Experimental Biology