Risso's dolphins plan foraging dives

RESEARCH ARTICLE Risso’s dolphins plan o aging di es Pa icia A anz 1,2, *, Kelly J. Benoi -Bi d 3 , B andon L. Sou hall 4,5 , John Calambokidis 6 , A i S. F iedlaende 4,7 and Pe e L. Tyack 1 ABSTRACT Humans emembe he pas and use ha in o ma ion o plan u u e ac ions. Lab expe imen s ha es memo y o he loca ion o ood show ha animals ha e a simila capabili y o ac in an icipa ion o u u e needs, bu less wo k has been done on animals o aging in he wild. We hypo hesized ha planning abili ies a e c i ical and common in b ea h- hold di e s who adjus each di e o o age on p ey a ying in quali y, loca ion and p edic abili y wi hin cons ain s o limi ed oxygen a ailabili y. We equipped Risso’s dolphins wi h sound-and-mo ion eco ding ags o e eal whe e hey ocus hei a en ion h ough hei ex e nally obse able echoloca ion and how hey ine une sea ch s a egies in esponse o expec ed and obse ed p ey dis ibu ion. The in o ma ion om he dolphins was in eg a ed wi h synop ic p ey da a ob ained om echosounde s on an unde wa e ehicle. A he s a o he di es, whales adjus ed hei echoloca ion inspec ion anges in ways ha sugges planning o o age a a pa icula dep h. Once en e ing a p oduc i e p ey laye , dolphins educed hei sea ch ange compa able o he scale o pa ches wi hin he laye , sugges ing ha hey we e using echoloca ion o selec p ey wi hin he pa ch. On ascen , hei sea ch ange inc eased, indica ing ha hey decided o s op o aging wi hin ha laye and s a ed sea ching o p ey in shallowe laye s. In o ma ion abou p ey, lea ned h oughou he di e, was used o plan o aging in he nex di e. Ou esul s demons a e ha planning o u u e di es is modula ed by spa ial memo y de i ed om mul i-modal p ey sampling (echoic, isual and cap u e) du ing ea lie di es. KEY WORDS: P eda o –p ey dynamics, Pe cep ual ange, G ampus g iseus, Animal decision making, Episodic-like memo y, Fo aging beha iou INTRODUCTION Animals use pas expe iences o make decisions abou u u e e en s (Os a h and Ma in-O das, 2014; P ei e and Fos e , 2013). The abili y o emembe in o ma ion abou esou ce dis ibu ion and use his o plan o aging allows hem o educe hei sea ch ime and inc ease o aging e iciency, ep esen ing a po en ial a ge o na u al selec ion (Benhamou, 1994; Saye s and Menzel, 2012). This abili y o an indi idual o ecall empo ally da ed e en s and empo al–spa ial ela ionships among hese e en s is called ‘episodic memo y’in humans (Tul ing and Donaldson, 1972). The abili y o go back in ime (episodic memo y) and o p ojec in o he u u e is e med ‘men al ime a el’(Suddendo and Co ballis, 1997; Cheke and Clay on, 2010) and was ega ded, un il ecen ly, as an exclusi ely human abili y. This emains a con en ious subjec in compa a i e cogni ion (She lewo h, 2007; Vonk and Shackel o d, 2012; Robe s e al., 2012). While challenging o demons a e in non-linguis ic animals, h ee beha iou al c i e ia, namely con en , s uc u e and lexibili y, ha e been p oposed o es ing episodic- like memo y and u u e planning in animals (Clay on e al., 2003). Resea ch in he las decade has sough o explo e he capaci y o in o med o esigh in non-humans, pa icula ly whe he an animal’s beha iou can be d i en by he an icipa ion o a u u e need ha is di e en om i s cu en mo i a ional s a e (Suddendo and Co ballis, 1997). A se ies o inno a i e s udies ha e shown ha cap i e bi ds lea n o p o ide o upcoming needs, ca ching ood whe e hey ha e lea ned ha i will no be a ailable when hey a e hung y in he u u e (Raby e al., 2007). Ra s modi y consump ion o one ood ype in an icipa ion o access o ano he (C ys al, 2012). Apes ained in he lab o selec an objec du ing a sho ime window can o ego selec ing an objec ha would sa is y an immedia e mo i a ion and ins ead selec a ool ha hey plan o use o ob aining some hing be e hou s la e (Mulcahy and Call, 2006; Os a h and Os a h, 2008). Ai -b ea hing ma ine p eda o s ha o age a dep h mus al e na e be ween wo spa ially seg ega ed esou ces: ood a dep h and oxygen a he su ace. Thus, we hypo hesized ha hey ha e he abili y o plan hei ac i i ies acco ding o pas expe iences and hei needs o bo h esou ces. E idence o his is ound in penguins and pinnipeds, which inc ease hei oxygen s o e when hey expec o o age deepe (Wilson, 2003; Sa o e al., 2011; Gallon e al., 2007). Simila ly, male spe m whales in high-la i ude habi a s adjus hei echoloca ing sampling beha iou a he s a o di es ela i e o he expec ed ange o p ey (Fais e al., 2015). While hese examples sugges planning in di e en species, hey lack da a abou bo h p io explo a o y sensing beha iou and he en i onmen al con ex o animal decisions –speci ically, he ine-scale dis ibu ion o epheme al p ey –da a ha a e impo an o expe imen al demons a ions o u u e planning. Echoloca ing p eda o s acqui e senso y in o ma ion by scanning he en i onmen using di ec ional sona pulses and e alua ing he ime delay om pulse o echo o es ima e ange (Su lykke e al., 2009; Seibe e al., 2013). We call he ange associa ed wi h he ound- ip a el ime o he in e -click in e al he ‘inspec ion ange’. To a oid pulse–echo o e lap, which may esul in ambiguous ange es ima ions, echoloca ing species gene ally adjus hei click a e o allow echoes o a i e be o e he nex pulse is emi ed (Simmons, 1973; Kadane and Penne , 1983). This means ha h ough ac i e acous ic sensing, hey e eal whe e hey a e ocusing hei a en ion o na iga e and ind p ey. Mo eo e , hey Recei ed 21 June 2017; Accep ed 18 Decembe 2017 1 Sea Mammal Resea ch Uni , School o Biology, Uni e si y o S And ews, Eas Sands, S And ews KY16 8LB, UK. 2 Depa men o Animal Biology, Uni e si y o La Laguna, A da. As o isico Fco Sanchez s/n, La Laguna 36200, Tene i e, Spain. 3 Mon e ey Bay Aqua ium Resea ch Ins i u e, 7700 Sandhold Road, Moss Landing, CA 95039, USA. 4 Sou hall En i onmen al Associa es, 9099 Soquel D i e, Sui e 8, Ap os, CA 95003, USA. 5 Long Ma ine Labo a o y, Ins i u e o Ma ine Sciences, Uni e si y o Cali o nia San a C uz, 1156 High S , San a C uz, CA 95064, USA. 6 Cascadia Resea ch Collec i e, 218 1/2 4 h A e W, Olympia, WA 98501, USA. 7 Depa men o Fishe ies and Wildli e, Ma ine Mammal Ins i u e, 2030 Ma ine Science D i e, Newpo , OR 97365, USA. *Au ho o co espondence ([email p o ec ed]) P.A., 0000-0001-8998-5149 1 © 2018. Published by The Company o Biologis s L d | Jou nal o Expe imen al Biology (2018) 221, jeb165209. doi:10.1242/jeb.165209 Jou nal o Expe imen al Biology p oduce as click se ies e med ‘buzzes’when hey a emp o cap u e p ey (Mille e al., 2004; Johnson e al., 2004; A anz e al., 2016), p o iding an indica o o o aging choices. This makes biosona an e ec i e senso y sys em o explo ing he manne in which animals ac i ely seek in o ma ion h ough hei senses and how hey p ocess, e ain and decide o ac on i , when combined wi h he esul ing o aging beha iou and da a on p ey dis ibu ion. The ma ine en i onmen is cha ac e ized by a high le el o pa chiness and ood esou ces in shallow wa e s can be epheme al and occu a e y di e en densi ies (S eele, 1976). In con as , small midwa e animals ha se e as impo an p ey o many species o en o m ‘deep sca e ing laye s’ ha ex end ho izon ally o ens o kilome es o mo e wi hin a ela i ely cons an dep h ange. While p edic able, hese laye s may ep esen a less- accessible esou ce o an animal ha mus hold i s b ea h o o age a dep h. The p esence o unp edic able shallow esou ces wi h mo e p edic able deep p ey in oduces he p oblem o decision making o ai -b ea hing ma ine p eda o s. One solu ion o hese p eda o s would be o sample p ey oppo unis ically (She lewo h, 1998), hen plan hei o aging di es acco dingly (Dunlap and S ephens, 2012) wi hin he cons ain s imposed by limi ed oxygen a ailabili y. Echoloca ing di e s may be able o assess cu en esou ces using hei long- ange biosona (Au, 1993; A anz e al., 2016), which could be a c i ical capabili y needed o decide ea ly in di es whe e o o age (Au e al., 2000). The maximum sona ange o delphinids is hough o be <100 m o indi idual p ey i ems (Madsen e al., 2007), bu i dense p ey laye s p o ide s onge sona a ge s, his could allow hem o be de ec ed a g ea e anges (Au and Lamme s, 2016). While sea ch ime migh be educed by echoloca ion on descen , o aging e iciency may be u he inc eased by emembe ing in o ma ion abou p ey loca ion, ype and alue om he las di e and using i o plan he nex o age be o e s a ing he nex di e (Benhamou, 1994; Saye s and Menzel, 2012; Ba aquand e al., 2009). I so, echoloca ing ma ine p eda o s ep esen a p omising model o s udying whe he animals can ecall pas e en s o s a egize o he u u e. We in es iga ed whe he he echoloca ing ma ine p eda o Risso’s dolphin, G ampus g iseus (Cu ie 1812), uses p io in o ma ion abou p ey dis ibu ion o plan o aging. Mo e speci ically, we explo ed whe he hese dolphins use in o ma ion abou p ey ea u es om p e ious di es o plan he nex o aging di e and how hese plans a e a ec ed by p ey encoun e a es du ing he o aging di e. As analysis o he ela ionship be ween en i onmen al inpu and beha iou al ou pu is equi ed o unde s and he p ocesses ha media e be ween hem (She lewo h, 2001), he e we used ine-scale p eda o beha iou al da a eco ded wi h suc ion-cup-a ached sound- and-mo ion eco ding ags (Johnson and Tyack, 2003) coupled wi h independen and concu en p ey measu emen s ob ained om shipboa d and unde wa e ehicle-based echosounde sys ems. We p esen an in eg a i e s udy linking pe cep ion, memo y and senso imo o con ol in o ma ion om a ee- anging p eda o o in si u p ey ields, p o iding he con ex in which animal decisions a e execu ed unde na u al condi ions. We used hese da a o in es iga e he unde lying cogni i e mechanisms p eda o s use o ind p ey and o age e ec i ely in dynamic en i onmen s. MATERIALS AND METHODS Da a collec ion P eda o s G ampus g iseus, o San Clemen e Island, CA, USA, we e equipped wi h high- esolu ion, sound-and-mo emen eco ding ags (DTags; Johnson and Tyack, 2003) be ween 2011 and 2016. The agging p ocedu e is desc ibed in de ail in A anz e al. (2016). Focal ollows o agged animals we e conduc ed om he ag boa using VHF adio acking equipmen wi hin a minimum ange o 25 m. The acks o agged dolphins we e geo- e e enced when possible om isual obse a ions using GPS (Fig. 1). Acous ic da a we e sampled in s e eo wi h a 16-bi esolu ion a 240 kHz. P essu e senso , i-axial accele ome e and magne ome e da a we e sampled a 200 Hz pe channel and decima ed o 25 Hz o analysis. Pi ch and dep h o he agged dolphins we e de i ed om o ien a ion and p essu e senso s on he ags (Johnson and Tyack, 2003). Tag da a p ocessing and analysis we e ca ied ou using MATLAB (h p://www.ma hwo ks.es/), DTag oolbox and cus om unc ions. O he 33 agged dolphins, 18 we e exposed o playbacks o acous ic s imuli as pa o he Sou he n Cali o nia Beha io al Response S udy (SOCAL-BRS; Sou hall e al., 2012), bu only beha iou al da a eco ded be o e he onse o he playbacks we e analysed o exposed dolphins. Di es we e de ined as e ical excu sions >20 m dep h. Di e bou s we e de ined as a g oup o di es ending wi hin 10 min o he s a o he nex one. These dep h –118.326 –118.281 –118.237 –118.192 –118.147 –118.326 –118.281 –118.237 –118.192 –118.147 33.304 33.335 33.367 33.304 33.335 33.367 Fig. 1. Map o hyd oacous ic su ey and dolphin agging expe imen s o Ca alina Island. Lines ep esen acous ic ansec s, whi e and g ey o he ship and yellow o he obo ; do ed lines ep esen he acks o he agged dolphins, in e ed om su ace obse a ions. 2 RESEARCH ARTICLE Jou nal o Expe imen al Biology (2018) 221, jeb165209. doi:10.1242/jeb.165209 Jou nal o Expe imen al Biology and ime c i e ia we e selec ed based on he change o slope o he log-su i o ship plo o maximum di e dep hs and in e -di e in e als, espec i ely (Sla e and Les e , 1982). Di es ha s a ed wi hin 15 min o a achmen o he ag we e excluded o emo e da a po en ially a ec ed by he agging p ocedu e. Incomple e di es a he s a o end o he eco d we e also excluded. Fi e ag eco dings did no ha e su icien da a o mee ou c i e ia and we e no u he analysed. P ey Synch onous da a on he dis ibu ion o p ey we e ob ained o wo agged G. g iseus (gg13_266b and gg13_267) om ship- and au onomous unde wa e ehicle (AUV)-based hyd oacous ic su eys (spli -beam Sim ad EK60s a 38 and 120 kHz). An echosounde in eg a ed in o a REMUS 600 AUV (Moline e al., 2015) sampled a dep hs whe e di ing p eda o s we e eeding, allowing indi idual p ey animals ( a ge s) o be esol ed. Ins umen speci ica ions and calib a ion me hods we e acco ding o hose in Moline e al. (2015). Vessel-based echosounde ansec s we e conduc ed du ing he day ime a an a e age speed o 2 m s −1 , co e ing he sides o a squa e o ∼12 km 2 o e lapping in space and ime wi h agged dolphin acks (Fig. 1). The AUV sampled a a speed o ∼1.5 m s −1 a he dep hs o iden i ied sca e ing ea u es. Hyd oacous ic da a we e p ocessed using Eco iew (h p://we labs. com/so wa e/eco iew-0). The maximum acous ic in ensi y a ei he 38 o 120 kHz was in eg a ed in o 50 cm dep h by 1 min ime bins be o e he de ec ion o laye s. Fo each bin, a unning 25 m median cen ed on he bin was calcula ed e ically o de ine he backg ound sca e ing. The emo al o he backg ound om he maximum acous ic in ensi y o each bin e ealed he dep h dis ibu ion o sca e ing laye s. The edges o laye s we e de ined as he i s and las loca ions in a con iguous ea u e ha exceeded he backg ound. The uppe and lowe bounda ies o each dep h laye o e he sampling a ea we e de ined as he hi d qua ile o each edge’s dep h ac oss he samples. Ta ge s eng h o indi idual p ey wi hin sca e ing ea u es measu ed wi h he AUV was es ima ed om echoes o single a ge s (a single e u n ecei ed pe acous ic e e be a ion olume o each pulse; Sawada e al., 1993), a 38 and 120 kHz, which acili a ed coa se axonomic classi ica ion (by equency esponse) and size es ima ion. In addi ion, hese da a we e used o examine in e -indi idual and in e -g oup spacing o p ey in laye s. Tag da a analysis Da a om he wo dolphins o which p ey da a we e a ailable we e used o de ine di e classes by looking a he maximum di e dep h and he dis ibu ion o he sca e ing ea u es. Mos di es had a maximum dep h ha ell wi hin one o he laye s. Di es wi h a maximum dep h ha did no all in o one o he laye s (15%) we e classi ied as pe aining o he nea es shallowe laye . In shallow di es, he descen and ascen phases we e no de ined because he shallow bounda y o he p ey laye (25 m) was jus 5 m below he 20 m de ini ion o he s a o he di e. Fo buzz a e analysis in ol ing he la ge da a se (N=174 di es om 28 dolphins), including he 26 dolphins whe e he p ey laye s we e no measu ed, he bo om phase was de ined as deepe han 70% o he maximum dep h (A anz e al., 2016). Echoloca ion clicks and buzzes we e isola ed on he eco dings om he agged dolphins ollowing he me hods desc ibed in A anz e al. (2016). Fo aging di es we e de ined as di es ha con ained one o mo e buzzes eco ded om he agged dolphin, because buzzes indica ed p ey sea ch and cap u e (A anz e al., 2016). The in e -click in e al o he dolphins a a ce ain pi ch angle was used o es ima e hei inspec ion slan ange. The e ical componen o his inspec ion ange equalled he slan ange imes he sin (pi ch). The inspec ion dep h was calcula ed by adding he dep h o he dolphin plus he e ical componen o he inspec ion ange. When aken om he i s clicks in he di e, he in e -click in e al can be used as an indica o o he expec ed ange o p ey o an echoloca ing p eda o based on in o ma ion ga he ed on he p e ious di e, be o e new in o ma ion is gained ia echoloca ion (Fais e al., 2015). The ini ial inspec ion ange was compu ed om he maximum in e -click in e al o he i s h ee clicks emi ed in he di e by he agged dolphin. We used Spea man’s ank ρ o check o mono onic co ela ions be ween dolphin and p ey a iables, as he assump ion o linea i y ailed a e compa ing hem wi h a i ed e sus esidual plo . S a is ical signi icance was judged a a c i ical P- alue o 5%. To compa e ela i e mo emen s o he animals a he bo om phase o di es associa ed wi h o aging in di e en p ey laye s, we econs uc ed he dead- eckoned ack o he dolphins (sensu Johnson and Tyack, 2003). Swim speed was app oxima ed om he e ical eloci y ( a e o change o dep h) and pi ch angle o he dolphins using a Kalman il e . The ack was gene a ed by combining swim speed du ing he bo om phase o he di e wi h he pi ch and heading o he dolphin a a 1 Hz sampling a e. The a e age pi ch angle a he bo om phase o di es was small (7±5 deg, mean±s.d.). We acknowledge ha he eliabili y o his speed es ima e dec eases a such low pi ch angles and ha we a e igno ing wa e cu en s ha may a ec he speed o he animal o e he g ound. Howe e , as we we e compa ing ela i e mo emen s ac oss sca e ing ea u es, he e ec o cu en s may be negligible and a ia ions in he absolu e alues may ha e li le impac on he o e all conclusions. Pe mi de ails Expe imen s we e pe o med unde he US Na ional Ma ine Fishe ies Se ice (NMFS; pe mi no. 14534-2), Channel Islands Na ional Ma ine Sanc ua y (pe mi no. 2010-003) (B.L.S., p incipal in es iga o o bo h) and IACUC pe mi s issued o he p ojec in es iga o s. RESULTS A o al o 9 h o synch onous p eda o and p ey da a, comp ising sound and mo emen DTag eco dings om 37 di es pe o med by wo dolphins and hyd oacous ic mapping o p ey wi hin hei o aging a ea, we e analysed. Addi ional DTag da a om 26 G. g iseus agged in he same gene al a ea, amoun ing o 83.2 h o da a and 174 di es, we e used o in es iga e p eda o sea ch beha iou in ela ion o o aging a es. A di e was conside ed as s a ing he i s ime he dolphins exceeded 20 m dep h a e a su acing and ending a he nex su acing. Di es wi h one o mo e buzzes, i.e. p ey cap u e a emp s (A anz e al., 2016), om he agged dolphins, hence o h e e ed o as o aging di es, las ed o 1–10 min and con ained on a e age 5 ( ange: 1–11) buzzes. Di e and p ey da a in eg a ion Hyd oacous ic su eys concu en wi h ag da a e ealed a s ong seg ega ion o biomass e ically in sca e ed pa ches 50–150 m ac oss be ween 100 and 200 m dep h, and in h ee sound-sca e ing laye s, each o which was ho izon ally con inuous (Fig. 2). The h ee sca e ing laye s we e iden i ied as ollows: ‘shallow’ (30–90 m minimum and maximum dep hs o he laye bounda ies a e aged o he wo da ase s, ounded o he nea es 10); ‘midwa e ’ (200–300 m, mig a ing e ically in 24 h cycles) and ‘deep’ 3 RESEARCH ARTICLE Jou nal o Expe imen al Biology (2018) 221, jeb165209. doi:10.1242/jeb.165209 Jou nal o Expe imen al Biology (350–450 m, no diu nal mig a ion). Al hough he e ical dis ibu ion o hese laye s emained ela i ely cons an o e he du a ion o he wo ag deploymen s, when moni o ed o e longe ime scales, he p esence o he shallow laye a ied in an unp edic able ashion when compa ed wi h he mo e eliable midwa e and deep sca e ing laye s. In he ho izon al domain, midwa e and deep laye s had a ubiqui ous dis ibu ion o e a scale o ens o kilome es wi h a complex he e ogeneous inne s uc u e composed o small-scale, disc e e agg ega ions o animals o simila sizes and axonomic g oups app oxima ely 100 indi iduals ac oss, and adjacen o agg ega ions o animals o a di e en size and/o g oup (Benoi -Bi d e al., 2017). The sca e ed pa ches ound be ween 100 and 200 m dep h ( axonomy and size o animal wi hin unknown) ep esen ed only 5% o he wa e olume wi hin hese dep hs and had a less p edic able occu ence han he pa ches ha o med disc e e laye s a g ea e dep hs. Di e ypes Dolphins o aged in bou s o 7–11 di es co e ing a wide dep h ange, wi h only 1–3 min spen a he su ace be ween di es wi hin he same bou . The maximum dep h o di es a ied by 260 m ac oss bou s and 120 m wi hin bou s (median o 3 di e bou s pe ag). The maximum dep h o mos di es ell wi hin he bounda y o one o he ou obse ed p ey ea u es (co esponding o ei he laye s o sca e ed pa ches; Fig. 3). Di es we e classi ied in o ou di e ypes based on he deepes sca e ing ea u e isi ed. O e all, dolphins pe o med 57% o di es o no deepe han he shallow laye , 30% o he sca e ed pa ches, 17% o he midwa e laye and 12% o he deep laye (Fig. 3). Each di e was di ided in o h ee phases based on he bounda y o p ey ea u es: (i) descen – he pe iod be ween when he dolphin le he su ace and en e ed he deepes p ey laye isi ed du ing he di e; (ii) ascen – he pe iod om when he dolphin le he deepes p ey laye isi ed o when i eached he su ace; and (iii) bo om – he pe iod be ween he end o he descen and s a o he ascen , excep o shallow di es whe e he e was no descen o ascen and he whole di e was conside ed a bo om phase. Mos buzzes (112/181) occu ed du ing he bo om phase o di es; o e all, 31% occu ed in he shallow laye , 21% in he sca e ed pa ches, and 33% and 15% in he midwa e and deep laye s. These p opo ions di e ed om he pe cen age o di es o each o hese p ey ea u es. Di es ha eached he midwa e and deep laye s egula ly ea u ed buzzes on he ascen , accoun ing o 30% o he o al numbe o buzzes, mos o hem (66%) wi hin sca e ed pa ches, wi h 26% and 7% in shallow and midwa e laye s, espec i ely. The emaining 7% o buzzes (13/181) we e pe o med on di e descen s, p ima ily in he shallow laye (N=6) and sca e ed pa ches (N=6). Ad anced selec ion o o aging laye To explo e whe he dolphins planned o aging wi hin a speci ic dep h ange be o e gaining new in o ma ion by sensing he en i onmen in he cu en di e, we i s in es iga ed whe he planning would cause animals o wai o s a clicking un il hey swam close o he planned laye a he han nea he su ace. 500 11.5 12 12.5 13 13.5 12 13 15 Time local (h) Dep h (m) 14 16 18 17 450 400 350 300 250 200 150 100 50 A B 0 500 450 400 350 300 250 200 150 100 50 0 Shallow Sca e ed pa ches Midwa e Deep Shallow Sca e ed pa ches Midwa e Deep Fig. 3. Fo aging ac i i y o he dolphins and synch onous ime–dep h dis ibu ion o p ey agg ega ions. The colou ed laye s ep esen , om da ke o ligh e blue, he mean dep h, hi d qua ile and 95% con idence in e al o each edge’s dep h laye in eg a ed ac oss he sampling a ea. The di e p o ile (black) is shown wi h buzzes ( ed ci cles) ha indica e a emp s o cap u e p ey o dolphin one (A) and wo (B). 500 –35 –45 –55 –65 –75 012 Dis ance (km) Dep h 3 400 300 200 100 0 s (dB e. 1 m−1) Fig. 2. Longi udinal sec ion o he acous ic backsca e olume o Ca alina Island, CA, USA, showing ela i e dis ibu ion and composi ion o p ey agg ega ions. The colou map o acous ic backsca e uses g ey o ep esen low alues, blue o in e media e, and g een and ed o high alues. F om shallow o deep wa e s, based on a ge s eng h measu emen s, he e was a ai ly dilu e shallow laye o small nek on, a zone o sca e ed pa ches o unknown composi ion, and midwa e and deep sound-sca e ing laye s composed o o ganisms o la ge sizes. 4 RESEARCH ARTICLE Jou nal o Expe imen al Biology (2018) 221, jeb165209. doi:10.1242/jeb.165209 Jou nal o Expe imen al Biology Pooling da a om wo dolphins, he mean dep h o he i s click in di es was 23±13 m, indica ing ha he dolphins always s a ed echoloca ing ea ly in di es, ega dless o how deep hey s a ed o aging. Mo eo e , he dep h a which he i s click occu ed was only weakly co ela ed o he mean dep h o he sca e ing ea u e in which he i s o aging a emp o he di e was eco ded (Spea man’sρ=0.27, P=0.11, N=33 di es). This sugges ed ha e en when dolphins planned o o age a dep h, hey s ill s a ed sampling in shallowe laye s. Second, we in es iga ed whe he he inspec ion ange o he i s clicks on descen co ela ed wi h he dep h o he i s o aging a emp . The ini ial inspec ion ange occu ed o e a wide b ead h (50–460 m) and showed a signi ican ly posi i e co ela ion wi h he dep h o he u hes edge o he i s p ey laye whe e buzzes we e eco ded (Spea man’sρ=0.64, P=0.0001, N=33 di es), consis en wi h ou planning hypo hesis. Sampling adjus ed o p ey ea u es Based on he dis ibu ion o po en ial p ey, we p edic ed h ee ypes o sea ch pa e ns each cha ac e ized by how he inspec ion ange a ied o e he cou se o descen s. I a dolphin planned o o age in a shallow laye o shallow sca e ed pa ches and ound he expec ed p ey he e, hen we p edic ed a sho inspec ion ange ha would no change as a unc ion o dep h. I dolphins planned o o age in a shallow p ey ea u e bu encoun e ed ewe p ey o lowe quali y p ey han expec ed, we p edic ed hey may swi ch hei a en ion o a new, deepe p ey laye , esul ing om emembe ing he dep h o o he p ey ea u es. A sudden inc ease in he inspec ion ange was expec ed i such a decision was made, compa able o he ange o he new expec ed p ey laye . Al e na i ely, i a dolphin was planning a he beginning o a di e o o age on a medium o deep laye , we p edic ed ha he inspec ion ange would s a a long anges and educe g adually as he dolphin descended o ack he sca e ing ea u e on which i planned o o age. To es his hypo hesis, we i s ook he descen phase o indi idual di es as he uni o analysis and, o each di e ype, assessed he co ela ion be ween he maximum dep h sea ched by he dolphin and he dep h o he deepes p ey laye isi ed. In shallow di es (N=12), dolphins exhibi ed a ela i ely sho and cons an inspec ion ange o e he di e (mean inspec ion ange be o e he i s buzz, 97±52 m). In 11 o he 12 shallow di es, he inspec ion ange in he pe iod om he emission o he i s echoloca ion click and he i s buzz was no associa ed wi h he dis ance o he u hes edge o he shallow sca e ing laye (mean Spea man’sρ=0.1, P>0.05, N=12 di es). Simila ly, in 7 o 8 di es a ge ing sca e ed pa ches, he dolphins employed a ela i ely sho inspec ion ange (112±60 m) and he e was no co ela ion wi h he dis ance o he u he edge o he dep h ange o sca e ed pa ches (mean Spea man’sρ=0.03, P>0.05, N=8 di es; Fig. 4A). A e iew o midwa e and deep di es e ealed wo dis inc i e pa e ns: one pa e n (Fig. 4B) sugges ed he in en ion o di e o deepe dep hs om he ou se , while he o he (Fig. 4C) was consis en wi h he dolphin ini ially planning a shallow di e bu la e swi ching i s a en ion o a deepe laye . Midwa e and deep di es wi h Spea man’sρ<0.5 (N=6) had an ini ially sho inspec ion ange (140±90 m) and o en ea u ed buzzes du ing he descen (1±1 buzz eco ded in 5 o 8 di es). In he descen phase o hese di es, dolphins main ained a ela i ely cons an and sho inspec ion ange, oughly equi alen o ha obse ed in di es a ge ing he shallow laye and sca e ed pa ches, un il he i s p ey cap u e a emp occu ed, usually be ween 100 and 200 m dep h (85% o he buzzes). Sho ly a e he buzz, he dolphins inc eased hei inspec ion ange. This dis ance was compa able o he ange o he laye ul ima ely a ge ed du ing he di e and he inspec ion ange employed wi hin he same dep h ange in di es in which laye acking was appa en (i.e. di es wi h s ong co ela ions o ini ial inspec ion ange and dis ance o he u he edge o he chosen p ey laye ) (Fig. 4C). In hese cases, he dolphins ended up o aging a he bo om phase o he di e in a deepe laye . This swi ch om sho o long sea ches du ing he descen and subsequen deepe o aging was obse ed in 40% o he midwa e and deep di es. In con as , in midwa e and deep di es wi h Spea man’sρ>0.5 (N=8), he ini ial inspec ion ange was long 0 ABC 100 Sca e ed pa ches Midwa e Deep 200 Dep h (m)Inspec ion ange (m) 300 400 500 400 300 200 100 00123 Di e ime (min) 0246802468 0 100 200 300 400 500 400 300 200 100 0 0 100 200 300 400 500 400 300 200 100 0 Fig. 4. Examples o he sampling s a egies o he dolphins in ela ion o p ey laye s in h ee di e ypes. Uppe panel: di e p o ile wi h clicks (whi e) and buzzes ( ed), and mean dep h o he bounda ies o he p ey laye a ge ed, (A) sca e ed pa ches, (B) midwa e and (C) deep laye s. Lowe panel: dolphin inspec ion ange (black do s) aken om he in e -click in e al mul iplied by one/hal he speed o sound as a unc ion o di e ime. The g eya ea ep esen s he bo om phase o he di e and he ed lines indica e buzzes. The co ela ion be ween inspec ion ange and ange o he deepe bounda y o he i s p ey laye wi h o aging buzzes was assessed isually o exempli y di es wi h none (A), weak (C) and s ong (B) co ela ions. No e he buzz du ing he descen o he deep di e (C) and associa ed changes in he inspec ion ange a e wa ds. 5 RESEARCH ARTICLE Jou nal o Expe imen al Biology (2018) 221, jeb165209. doi:10.1242/jeb.165209 Jou nal o Expe imen al Biology (400±100 m) and g adually dec eased as hey app oached he o age laye in a pa e n consis en wi h ini ially planning o o age in he deep laye s. Dolphins did no p oduce buzzes du ing he descen o hese di es. Fo midwa e and deep di es in which dolphins swi ched hei a en ion o a new p ey laye o e he cou se o he descen (i.e. di es wi h buzzes du ing descen , N=5), we measu ed whe he dolphins could ha e de ec ed he new expec ed laye based upon clicking while a shallow dep hs o in sca e ed pa ches. The a e age di e ence be ween he ange o he shallowes edge o he new o aging laye and he maximum inspec ion dep h be o e he i s buzz eco ded o e he descen was −348±108 m. This sugges ed ha he dolphins we e sampling he new o aging laye be o e he las buzz in shallowe laye s a he han simply elying on in o ma ion om p e ious deep di es. Sampling adjus ed o p ey pa ch Once en e ing he o aging laye , dolphins adop ed a mean inspec ion ange ha was simila o sca e ed pa ches (127± 15 m), midwa e (127±22 m) and deep di es (127±7 m) and ended o dec ease in shallow di es (105±7 m). On a e age, his ange was oughly simila o he ho izon al dis ance co e ed by he dolphins while o aging in he laye (99±47 m). Bo h we e compa able o he mean size, in he ho izon al dimension, o mono-speci ic agg ega ions (Scalab in e al., 1996) con aining he la ges a ailable p ey in each laye (113±34 m), sugges ing ha dolphins use echoloca ion o selec p ey wi hin hese agg ega ions. As epo ed in Benoi -Bi d e al. (2017), he a ge s eng h o indi idual p ey in he shallow laye was signi ican ly lowe han ha in deepe agg ega ions and shallow p ey we e gene ally in smalle agg ega ions, which co ela es wi h he use o sho e inspec ion anges by he dolphins. Ascen sampling based on p ey expec a ions Op imal o aging heo y p edic s ha b ea h-hold di e s would maximize he ime spen o aging a dep h and minimize he ime spen in ansi (Mo i, 1998). This led us o p esume ha dolphins will o age o as long as possible in hei chosen o aging laye and cease sea ching o p ey on he di e ascen once hey ha e le he laye . Al e na i ely, dolphins may echoloca e all he way up o o ien a ion, o because hey expec o gain om cap u ing shallowe p ey du ing he ascen , o o gain new in o ma ion abou p ey o planning he nex di e. To es hese hypo heses, we examined he dep h o he las click in di es and he sampling s a egy used by he dolphins on di e ascen s in ela ion o p ey ea u es and buzz occu ences. Pooling da a om he wo dolphins o which p ey da a we e a ailable, he mean dep h o he las click in a di e was 37±36 m, indica ing hey echoloca ed almos all he way up o he su ace, i espec i e o he ype o di e. Fu he mo e, he buzz a es, i.e. numbe o buzzes pe minu e, du ing di e ascen s we e up o six imes highe han hose du ing descen s (signed- ank P=0.005, N=14 pai ed compa isons o buzz a e du ing descen e sus ascen phases). Dolphins emi ed buzzes du ing he ascen in 40% o di es. In hose di es, clicking con inued a e he las buzz eco ded on he ascen . All midwa e and deep di es wi h no buzz on ascen and one wi h buzz on ascen had buzzes on descen (4 o 14 di es). A e age ascen du a ion was 58±31 and 76 ±42 s o di es wi hou and wi h buzzes on he ascen , espec i ely, whe eas bo om ime was educed om 158±61 o 110±50 s du ing he same di es. To es whe he , on ascen , dolphins we e seeking mo e e icien o aging in uppe laye s, we explo ed whe he he buzz a e du ing he bo om phase was highe han ha du ing he ascen . Buzz a es a he bo om phase we e wo imes highe han hose du ing ascen (signed- ank P=3.2e−9, N=130 pai ed compa isons o buzz a e du ing bo om e sus ascen phases), sugges ing ha dolphins we e o aging mo e e icien ly a he bo om o he di e compa ed wi h o aging on ascen . Du ing ascen s o mos di es, an inspec ion ange consis en wi h acking he nex p ey ea u e was obse ed, sugges ing he dolphins ocused hei sea ch on expec ed shallowe sca e ing ea u es. The e was only one deep di e (ou o 4) du ing which he dolphin did no p oduce buzzes on he ascen and his was he only di e o which he ini ial ascen inspec ion ange was adjus ed o a ange consis en wi h acking he su ace. The i e buzzes eco ded a he bo om phase o his di e we e compa able o he a e age numbe o buzzes eco ded a he bo om phase o deep di es in which buzzes on ascen we e also eco ded (3.5±1.5 buzzes). Fo midwa e di es, we es ed whe he he dolphins we e mo e likely o swi ch o a longe sea ch ange a he s a o ascen han while hey we e sea ching o p ey wi hin he laye . Fo his es , we compa ed he mean maximum inspec ion ange o he las h ee clicks eco ded a he bo om phase (177±86 m) wi h he i s h ee clicks o he ascen phase (366±160 m). The same a iables measu ed in deep di es we e 198±44 and 370±244 m, espec i ely. Bo h in midwa e and in deep di es, he e was a s ong co ela ion be ween he e ical inspec ion ange measu ed om he i s h ee clicks o he dolphins emi ed on he ascen and he dep h di e ence be ween he dolphin and he shallowe edge o he nex shallowe laye whe e buzzes we e eco ded (Spea man’sρ=0.66, P=0.01, N=14 di es). The absolu e di e ence be ween hese wo a iables was small (16±14 and 17±11 m in midwa e and deep di es, espec i ely), suppo ing he hypo hesis ha dolphins we e sampling p ey a a pa icula dep h, on hei way up o he su ace. To explo e whe he dolphins sensed p ey loca ed abo e hem du ing he bo om phase o he di e o emembe ed hei loca ion om be o e ha phase, we es ed whe he he e ical inspec ion ange o long- ange clicks (i.e. es ima ed inspec ion ange >200 m) eco ded o e he bo om phase o he di e ma ched he dep h di e ence be ween he dolphin, when emi ing he click, and he shallowe edge o he nex shallowe laye whe e buzzes we e eco ded on ascen . The a e age absolu e di e ence be ween he wo a iables was 5±78 m in midwa e di es and 40±85 m in deep di es, sugges ing ha dolphins could sense shallowe p ey ea u es loca ed abo e hem while a he bo om o he di es. P ey-dependen planning To es whe he dolphins plan he laye in which o o age based on pe cei ed p ey p o i abili y in he p e ious di e, we looked a he ela ionship be ween he buzz a e a he bo om phase o he p e ious di e and he ini ial inspec ion ange in he nex di e, pooling da a o 28 dolphins. I a p e ious di e sugges ed he bes pa ches we e shallow, and he animal used his in o ma ion o plan he nex di e, hen we expec ed ha i would s a sampling a a sho ange. In con as , i pas expe iences indica ed ha shallow pa ches we e o low p o i , we expec ed he animal would s a sea ching a a longe ange. The esul s suppo ed his hypo hesis; a e shallow di es when buzz a es we e mode a e o high (>3 buzzes min −1 ), dolphins s a ed sampling a a sho ange (169±158 m) du ing he nex di e (Fig. 5). A e shallow di es wi h low buzz a es (<3 buzzes min −1 ), dolphins, on a e age, chose o sample a longe anges du ing he nex di e (249±204 m). A e e se pa e n was obse ed a e deep di es: when buzz a es we e mode a e o high (>3 buzzes min −1 ), dolphins s a ed sampling a a long ange (414±157 m) du ing he nex di e (Fig. 5); when buzz a es we e 6 RESEARCH ARTICLE Jou nal o Expe imen al Biology (2018) 221, jeb165209. doi:10.1242/jeb.165209 Jou nal o Expe imen al Biology low (<3 buzzes min −1 ), dolphins sampled a sho e anges (181 ±110 m) in he nex di e, consis en wi h dolphins using in o ma ion on p ey om a p e ious di e o plan he o aging dep h in he nex one. DISCUSSION Each ime an ai -b ea hing aqua ic animal di es o o age, i mus make decisions o adjus i s physiology o di e en leng hs and dep hs o di es. In en i onmen s wi h epheme al pa ches o p ey a close shallow anges and p edic able esou ces a mo e cos ly deepe anges, he animal may bene i om using in o ma ion om p e ious di es o plan he nex o aging di e. Ye , we a ely can ob ain e idence o how an indi idual’s decisions a e a ec ed by p io knowledge and modula ed by compa isons o eal- ime in o ma ion and balancing a ailable al e na i es, and his has na owed p og ess in s udies o animal planning and decision making unde na u al se ings. Echoloca ing di e s a e a good axon o use o add ess hese ques ions because hey p o ide us wi h in o ma ion on hei sea ch anges (i.e. wha in o ma ion is a ailable in o de o make a decision) and when and whe e p ey cap u e a emp s occu (i.e. ac ual choices). In eg a ion o his in o ma ion wi h con empo aneous ine-scale da a om he p ey ields, sampled a he same dep hs and imes whe e he p eda o s occu , p o ides a unique pe spec i e on he choices a ailable o a p eda o in a h ee- dimensional ma ine sys em. He e, we shed ligh on how p eda o s adjus hei sea ch based upon sensing he cu en dis ibu ion o p ey o using in o ma ion ga he ed in p e ious di es o om ea lie phases o he same di e. These p ocesses we e iewed om sub- mesoscales (i.e. how hey plan and choose o o age on di e en p ey agg ega ions) o mic oscales (i.e. how hey adjus hei acous ic gaze o sea ch wi hin a pa ch). The dolphins explo ed hei en i onmen based on pas expe iences o in o m upcoming o aging decisions o di e- ime alloca ion and p ey choice. They used a long inspec ion ange o di es when hey we e planning o o age in he s able deep pa ches, and swi ched o a sho inspec ion ange once hey loca ed a pa ch in which o o age. Once his decision was made, he inspec ion ange emained oughly cons an and compa able o he scale o p ey pa ches con aining he la ges p ey, sugges ing ha inspec ion ange was adjus ed o selec p ey wi hin he expec ed size o p ey pa ches. When dolphins pe o med well in one pa ch, as judged by high buzz a es, hey planned o a ge he same laye in he nex di e. Obse ed di e ences in maximum di e dep h o di es wi hin bou s we e p obably d i en by a ia ions in p ey selec ion by he dolphins on a di e-by-di e basis. The e was an appa en misma ch be ween he p opo ion o di es and buzzes pe o med wi hin each p ey laye , as judged by he g adual dec ease in he pe cen age o di es pe o med pe p ey ea u e agains he mo e Gaussian dis ibu ion o buzz coun s. This may be because he la e is ela ed o he ela i e cap u e a e in each laye , whe eas he o me may be d i en by ene ge ic cos s o di ing o di e en dep hs. La ge deep-di ing odon oce es, such as spe m o beaked whales, emi be ween 3 and 5 imes mo e buzzes han G. g iseus a he bo om phase o di es (A anz e al., 2011; Wa wood e al., 2006; his s udy). This appa en , highe o aging e iciency o la ge species p obably e lec s di e ences in o aging equi emen s as well as hei g ea e di ing capaci y and lowe cos o anspo (Williams e al., 1999). G ampus g iseus a e small- o medium-sized delphinids o which ex ending he ime o aging a dep h o compensa e o long ansi imes (Thompson and Fedak, 2001) may be limi ed by he a ailable oxygen s o es. Animals conside ing po en ial p ey al e na i es wi h a iable bene i s as well as a iable cos s o locomo ion o a el o hese pa ches may need o ack he en i onmen ela i ely closely o 0 0 250 500 246 02 Buzz a e (p e ious di e) (buzzes min–1) Ini ial inspec ion ange (cu en di e) (m) 46 024602 46 Sca e ed pa ches Midwa e Deep Shallow Fig. 5. Ini ial inspec ion ange adap ed o he bes o aging dep h encoun e ed on he p e ious di e, om 174 di es by 28 dolphins. (A) Illus a ion o an indi idual’s o aging scena io showing buzzes ( ed s a s) emi ed in he p e ious di e in a gi en p ey ea u e and he inspec ion ange a he s a o cu en di e. (B) Ini ial inspec ion ange o he cu en di e (y-axis) as a unc ion o buzz a e in he las di e (x-axis). No e he endency o he dolphin o a ge he dep h laye whe e i had he highes buzz a e on he p e ious di e. Di e ypes a e classi ied using he dep h dis ibu ion o p ey ea u es desc ibed in Fig. 1. 7 RESEARCH ARTICLE Jou nal o Expe imen al Biology (2018) 221, jeb165209. doi:10.1242/jeb.165209 Jou nal o Expe imen al Biology upda e exis ing in o ma ion in o de o espond quickly o sho - e m changes in cos and bene i (Dunlap and S ephens, 2012; Mo i, 1998). He e, in con as o o he deep-di ing odon oce es (A anz e al., 2011; Wa wood e al., 2006), he dolphins s a ed echoloca ing ea ly and h oughou he di es, p obably o gain in o ma ion on he dep h dis ibu ion and a ailabili y o p ey and o espond swi ly o apid changes in habi a s uc u e a di e en dep hs. We ound e idence ha dolphins upda ed hei o aging plans based on in o ma ion abou p ey ga he ed du ing he di e, sugges ing ha animals we e e isi ing plans egula ly, compa ing al e na i es based on memo y wi h hose based on cu en assessmen o esou ce dis ibu ion ia senso y in o ma ion acqui ed om sampling h oughou he di e. Be o e ascen , dolphins planned when o o age on he way up and when no o, as e idenced by hei inspec ion ange a he s a o he ascen . Sampling du ing descen s o o aging dep hs may guide ini ial decisions abou which p ey ea u es o sample on ascen , as demons a ed by he jump in he inspec ion ange when dolphins decided o lea e he bo om o aging laye . S ikingly, when dolphins a emp ed o ca ch p ey (demons a ed by he p oduc ion o buzzes) on descen , hey did no p oduce buzzes on ascen , pe haps because he inc eased ime on descen in e e ed wi h oppo uni ies o o age on ascen , gi en he ime cons ain s imposed by b ea h- hold di ing. Dolphins a imes slowed down hei ascen , aking 20% longe in o de o sea ch o and cap u e p ey in shallowe laye s. These esul s sugges ha while dolphins may sample p ey on descen , hey make a decision abou whe he o sho en o aging a he bo om o o age on ascen based upon in o ma ion abou p ey and oxygen a ailabili y. The endency o dolphins o o age on ascen e sus descen may esul om he s abili y o p ey pa ches and he abili y o an indi idual o aging in he bo om laye o weigh he bene i s los by b eaking o om o aging a he bo om o o age on ascen . Fo aging on ascen con adic s mos o aging heo ies, which emphasize minimizing he ime in di e ansi s o and om o aging dep hs o ex end bo om ime and he e o e he p obabili y o inding p ey (Mo i, 1998; Thompson and Fedak, 2001). Mos di ing mammals s udied o da e, including spe m, beaked, blue and humpback whales, do no show o aging ac i i y on he di e ascen (A anz e al., 2011; Wa wood e al., 2006; Hazen e al., 2015; Doniol-Valc oze e al., 2011). Mo eo e , mos deep-di ing echoloca ing species s op clicking ea ly in he ascen . Howe e , he e is conside able selec ion p essu e o sea ch s a egies ha inc ease he encoun e a e wi h ood and he e o e hei o aging e iciency (Cowie, 1977; F iedlaende e al., 2016; Ydenbe g and Hu d, 1998). Spe m whales o aging in high-la i ude habi a s ha e been obse ed sea ching o and cap u ing p ey on di e ascen s (Fais e al., 2015), as he Risso’s dolphins did in his s udy. This sugges s ha such a s a egy migh ep esen an adap i e beha iou o ai -b ea hing p eda o s in mul i-laye ed habi a s con aining bo h shallow- and deep-wa e p ey o a iable p edic abili y. In his pape , we ha e shown ha a he s a o a o aging di e, dolphins selec ed an echoloca ion ange ha a ge ed he bes o aging dep h encoun e ed on he las di e. This can be in e p e ed as dolphins ecalling in o ma ion om he p e ious di e o plan he nex o aging di e (i.e. episodic-like memo y). Al e na i ely, one can a gue ha ha ing spa ial memo y abili ies is su icien o allow such beha iou (i.e. seman ic knowledge). Howe e , u he con incing e idence o planning comes om si ua ions in which an animal beha es in a way ha igno es i s cu en mo i a ional s a e in o de o mee a need o an expec ed u u e mo i a ional s a e (Naqshbandi and Robe s, 2006). He e, we ound ha dolphins emi clicks h oughou hei ascen o he su ace a e eeding a he bo om laye , when he dolphins ha e lowe mo i a ion o eed and highe mo i a ion o b ea he. The unc ion o his beha iou appea s o be o sample p ey a di e en laye s in an icipa ion o hei u u e di e in o de o upda e hei knowledge on p ey dis ibu ion, suppo ing he lexible deploymen o in o ma ion in no el si ua ions (Clay on e al., 2003). In his si ua ion, he dolphins appea ed o igno e hei mo e u gen mo i a ion o b ea he in o de o imp o e hei o aging on he nex di e, by po en ially modi ying hei sea ch s a egy; howe e , a e b ea hing, hei p ima y mo i a ion again became o o age. The da a p esen ed he e on he spa ial dis ibu ion o p ey and du a ions o e which dis ibu ion was s able, coupled wi h ou da a on how b ea h-hold di e s make use o senso y in o ma ion o make o aging decisions, p o ide unique insigh in o how hese animals plan dynamically, wi h a balance o s a egy and lexibili y, o cope wi h a ying p edic abili y in he dis ibu ion o hei ood. Acknowledgemen s The au ho s hank he SOCAL-BRS pa icipan s, pa icula ly ag ope a o s Alison S impe and Ann Allen, o hei assis ance in he ield. We also g a e ully acknowledge ChloeYzoa d o dolphin illus a ions and one anonymous e iewe o cons uc i e c i icism on he manusc ip . Compe ing in e es s The au ho s decla e no compe ing o inancial in e es s. Au ho con ibu ions Concep ualiza ion: P.A., K.J.B., B.L.S., P.L.T.; Me hodology: K.J.B., B.L.S.; So wa e: K.J.B.; Valida ion: P.L.T.; Fo mal analysis: P.A., K.J.B.; In es iga ion: P.A., K.J.B., J.C., A.S.F.; Resou ces: J.C., A.S.F.; Da a cu a ion: P.A., K.J.B.; W i ing - o iginal d a : P.A.; W i ing - e iew & edi ing: P.A., K.J.B., B.L.S., J.C., A.S.F., P.L.T.; Supe ision: P.L.T.; P ojec adminis a ion: B.L.S., J.C., P.L.T.; Funding acquisi ion: K.J.B., B.L.S., J.C., P.L.T. Funding Funding o he SOCAL-BRS p ojec was p o ided by he Chie o Na al Ope a ions En i onmen al Readiness Di ision, he U.S. Na y’s Li ing Ma ine Resou ces P og am, and he O ice o Na al Resea ch Ma ine Mammal P og am. The S a egic En i onmen al Resea ch and De elopmen P og am ia a U.S. A my Co ps o Enginee s Con ac (K.J.B. and B.L.S.) p o ided unding o da a collec ion and p ey analysis. This s udy was suppo ed by he MASTS pooling ini ia i e (Ma ine Alliance o Science and Technology o Sco land). MASTS is unded by he Sco ish Funding Council (g an e e ence HR09011) and con ibu ing ins i u ions. Da a a ailabili y Da a a e a ailable om he D yad Digi al Reposi o y (A anz e al., 2017): h p:// doi:10.5061/d yad.48 q4 Re e ences A anz, P., Aguila de So o, N., Madsen, P. T., B i o, A., Bo des, F. and Johnson, M. P. (2011). Following a o aging ish- inde : diel habi a use o Blain ille’s beaked whales e ealed by echoloca ion. PLoS ONE 6, e28353. A anz, P., DeRui e , S. L., S impe , A. K., Ne es, S., F iedlaende , A. S., Goldbogen, J. A., Visse , F., Calambokidis, J., Sou hall, B. L. and Tyack, P. L. (2016). Disc imina ion o as click-se ies p oduced by agged Risso’s dolphins (G ampus g iseus) o echoloca ion o communica ion. J. Exp. Biol. 219, 2898-2907. A anz, P., Benoi -Bi d, K. J., Sou hall, B. L., Calambokidis, J., F iedlaende , A. S. and Tyack, P. L. (2017). Da a om: Risso’s dolphins plan o aging di es. D yad Digi al Reposi o y. h ps://doi.o g/10.5061/d yad.48 q4 Au, W. L. (1993). The Sona o Dolphins. New Yo k, US: Sp inge . Au, W. L. and Lamme s, M. O. (2016). Biosona de ec ion ange o mesopelagic pa ches by spinne dolphins in Hawaii. J. Acous . Soc. Am. 140, 3130-3130. Au, W. L., Poppe , A. N. and Fay, R. R. (2000). Hea ing by Whales and Dolphins. New Yo k, US: Sp inge -Ve lag. Ba aquand, F., Inchaus i, P. and B e agnolle, V. (2009). Cogni i e abili ies o a cen al place o age in e ac wi h p ey spa ial agg ega ion in hei e ec on in ake a e. Anim. Beha . 78, 505-514. Benhamou, S. (1994). Spa ial memo y and sea ching e iciency. Anim. Beha . 47, 1423-1433. 8 RESEARCH ARTICLE Jou nal o Expe imen al Biology (2018) 221, jeb165209. doi:10.1242/jeb.165209 Jou nal o Expe imen al Biology Benoi -Bi d, K. J., Moline, M. A. and Sou hall, B. L. (2017). P ey in oceanic sound sca e ing laye s o ganize o ge a li le help om hei iends. Limnol. Oceanog . 62, 2788-2798. Cheke, L. G. and Clay on, N. S. (2010). Men al ime a el in animals. Wiley In e discip. Re . Cogn. Sci. 1, 915-930. Clay on, N. S., Bussey, T. J. and Dickinson, A. (2003). Can animals ecall he pas and plan o he u u e? Na . Re . Neu osci. 4, 685-691. Cowie, R. J. (1977). Op imal o aging in g ea i s (Pa us majo ). Na u e 268, 137-139. C ys al, J. D. (2012). P ospec i e cogni ion in a s. Lea n. Mo i . 43, 181-191. Doniol-Valc oze, T., Lesage, V., Gia d, J. and Michaud, R. (2011). Op imal o aging heo y p edic s di ing and eeding s a egies o he la ges ma ine p eda o . Beha . Ecol. 22, 880-888. Dunlap, A. S. and S ephens, D. W. (2012). T acking a changing en i onmen : op imal sampling, adap i e memo y and o e nigh e ec s. Beha . P ocess. 89, 86-94. Fais, A., Aguila So o, N., Johnson, M., Pe ez-Gonzalez, C., Mille , P. J. O. and Madsen, P. T. (2015). Spe m whale echoloca ion beha iou e eals a di ec ed, p io -based sea ch s a egy in o med by p ey dis ibu ion. Beha . Ecol. Sociobiol. 69, 663-674. F iedlaende , A. S., Johns on, D. W., Tyson, R. B., Kal enbe g, A., Goldbogen, J. A., S impe , A. K., Cu ice, C., Hazen, E. L., Halpin, P. N., Read, A. J. e al. (2016). Mul iple-s age decisions in a ma ine cen al-place o age . R. Soc. Open. Sci. 3, e160043. Gallon, S. L., Spa ling, C. E., Geo ges, J.-Y., Fedak, M. A., Biuw, M. and Thompson, D. (2007). How as does a seal swim? Va ia ions in swimming beha iou unde di e ing o aging condi ions. J. Exp. Biol. 210, 3285-3294. Hazen, E. L., F iedlaende , A. S. and Goldbogen, J. A. (2015). Blue whales (Balaenop e a musculus) op imize o aging e iciency by balancing oxygen use and ene gy gain as a unc ion o p ey densi y. Sci. Ad . 1, e1500469. Johnson, M. P. and Tyack, P. L. (2003). A digi al acous ic eco ding ag o measu ing he esponse o wild ma ine mammals o sound. IEEE J. Oceanic Eng. 28, 3-12. Johnson, M. P., Madsen, P. T., Zimme , W. M. X., Aguila de So o, N. and Tyack, P. L. (2004). Beaked whales echoloca e on p ey. P oc. R. Soc. Lond. B Biol. Sci. 271 Suppl. 6, S383-S386. Kadane, J. and Penne , R. (1983). Range ambigui y and pulse in e al ji e in he bo lenose dolphin. J. Acous . Soc. Am. 74, 1059-1061. Madsen, P. T., Wilson, M., Johnson, M., Hanlon, R. T., Bocconcelli, A., Aguila de So o, N. and Tyack, P. L. (2007). Clicking o calama i: oo hed whales can echoloca e squid Loligo pealeii.Aqua . Biol. 1, 141-150. Mille , P. J. O., Johnson, M. P. and Tyack, P. L. (2004). Spe m whale beha iou indica es he use o echoloca ion click buzzes ‘c eaks’in p ey cap u e. P oc. R Soc. Lond. B Biol. Sci. 271, 2239-2247. Moline, M. A., Benoi -Bi d, K., O’Go man, D. and Robbins, I. C. (2015). In eg a ion o scien i ic echo sounde s wi h an adap able au onomous ehicle o ex end ou unde s anding o animals om he su ace o he ba hypelagic. J. A mos. Oceanic. Tech. 32, 2173-2186. Mo i, Y. (1998). Op imal choice o o aging dep h in di e s. J. Zool. Lond. 245, 279-283. Mulcahy, N. J. and Call, J. (2006). Apes sa e ools o u u e use. Science 312, 1038-1040. Naqshbandi, M. and Robe s, W. A. (2006). An icipa ion o u u e e en s in squi el monkeys (Saimi i sciu eus) and a s (Ra us no egicus): es s o he Bischo - Kohle hypo hesis. J. Comp. Psychol. 120, 345-357. Os a h, M. and Ma in-O das, G. (2014). The u u e o u u e-o ien ed cogni ion in non-humans: heo y and he empi ical case o he g ea apes. Philos. T ans. R. Soc. Lond. B Biol. Sci. 369, 20130486. Os a h, M. and Os a h, H. (2008). Chimpanzee (Pan oglody es) and o angu an (Pongo abelii) o e hough : sel -con ol and p e-expe ience in he ace o u u e ool use. Anim. Cogn. 11, 661-674. P ei e , B. E. and Fos e , D. J. (2013). Hippocampal place-cell sequences depic u u e pa hs o emembe ed goals. Na u e 497, 74-79. Raby, C. R., Alexis, D. M., Dickinson, A. and Clay on, N. S. (2007). Planning o he u u e by wes e n sc ub-jays. Na u e 445, 919-921. Robe s, W. A., McMillan, N., Musolino, E. and Cole, M. (2012). In o ma ion seeking in animals: me acogni ion? Comp. Cogn. Beha . Re . 7, 85-109. Sa o, K., Shiomi, K., Ma shall, G., Kooyman, G. L. and Ponganis, P. J. (2011). S oke a es and di ing ai olumes o empe o penguins: implica ions o di e pe o mance. J. Exp. Biol. 214, 2854-2863. Sawada, K., Fu usawa, M. and Williamson, N. J. (1993). Condi ions o he p ecise measu emen o ish a ge s eng h in si u. Fish. Sci. 20, 15-21. Saye s, K. and Menzel, C. R. (2012). Memo y and o aging heo y: chimpanzee u iliza ion o op imali y heu is ics in he ank-o de eco e y o hidden oods. Anim. Beha . 84, 795-803. Scalab in, C., Dine , N., Weill, A., Hillion, A. and Moucho , M. C. (1996). Na owband acous ic iden i ica ion o monospeci ic ish shoals. ICES J. Ma . Sci. 53,181-188. Seibe , A.-M., Kobli z, J. C., Denzinge , A. and Schni zle , H.-U. (2013). Scanning beha io in echoloca ing common pipis elle ba s (Pipis ellus pipis ellus). PLoS ONE 8, e60752. She lewo h, S. J. (1998). Cogni ion, E olu ion and Beha iou . New Yo k, US: Ox o d Uni e si y P ess. She lewo h, S. J. (2001). Animal cogni ion and animal beha iou . Anim. Beha . 61, 277-286. She lewo h, S. J. (2007). Animal beha iou : planning o b eak as . Na u e 445, 825-826. Simmons, J. A. (1973). The esolu ion o a ge ange by echoloca ing ba s. J. Acous . Soc. Am. 54, 157-173. Sla e , P. J. B. and Les e , N. P. (1982). Minimising e o s in spli ing beha iou in o bou s. Beha iou 79, 153-161. Sou hall, B. L., Mo e i, D., Ab aham, B., Calambokidis, J., DeRui e , S. L. and Tyack, P. L. (2012). Ma ine mammal beha io al esponse s udies in sou he n Cali o nia: ad ances in echnology and expe imen al me hods. Ma . Technol. Soc. J. 46, 48-59. S eele, J. H. (1976). Pa chiness. Philadelphia, US: Saunde s Company. Suddendo , T. and Co ballis, M. C. (1997). Men al ime a el and he e olu ion o he human mind. Gene . Soc. Gen. Psych. 123, 133-167. Su lykke, A., Ghose, K. and Moss, C. F. (2009). Acous ic scanning o na u al scenes by echoloca ion in he big b own ba , Ep esicus uscus.J. Exp. Biol. 212, 1011-1020. Thompson, D. and Fedak, M. A. (2001). How long should a di e las ? A simple model o o aging decisions by b ea h-hold di e s in a pa chy en i onmen . Anim. Beha . 61, 287-296. Tul ing, E. and Donaldson, W. (1972). Episodic and seman ic memo y. In O ganiza ion o Memo y (ed. E. Tul ing, W. Donaldson and G. H. Bowe ), pp. 381-403. Camb idge, MA, USA: Academic P ess. Vonk, J. and Shackel o d, T. K. (2012). Ox o d Handbook o Compa a i e E olu iona y Psychology. New Yo k, US: Ox o d Uni e si y P ess. Wa wood, S. L., Mille , P. J. O., Johnson, M., Madsen, P. T. and Tyack, P. L. (2006). Deep-di ing o aging beha iou o spe m whales (Physe e mac ocephalus). J. Anim. Ecol. 75, 814-825. Williams, T. M., Haun, J. E. and F iedl, W. A. (1999). The di ing physiology o bo lenose dolphins (Tu siops unca us) I. Balancing he demands o exe cise o ene gy conse a ion a dep h. J. Exp. Biol. 202, 2739-2748. Wilson, R. P. (2003). Penguins p edic hei pe o mance. Ma . Ecol. P og. Se . 249, 305-310. Ydenbe g, R. and Hu d, P. (1998). Simple models o eeding wi h ime and ene gy cons ain s. Beha . Ecol. 9, 49-53. 9 RESEARCH ARTICLE Jou nal o Expe imen al Biology (2018) 221, jeb165209. doi:10.1242/jeb.165209 Jou nal o Expe imen al Biology