BIOMECHANICS OF TURNING MANOEUVRES IN STELLER SEA LIONS {EUMETOPIAS JUBA TUS) by OL IV IER C H E N E V A L B . S c , Univers i te du Q u e b e c a Mont rea l ( U Q A M ) , 1998 A T H E S I S S U B M I T T E D IN P A R T I A L F U L F I L L M E N T O F T H E R E Q U I R E M E N T S FOR T H E D E G R E E O F M A S T E R O F S C I E N C E in T H E F A C U L T Y O F G R A D U A T E S T U D I E S Zoo logy T H E U N I V E R S I T Y O F BRIT ISH C O L U M B I A May 2005 © Olivier Cheneval, 2005 II ABSTRACT Otari ids such as the Stel ler sea lion (Eumetopias jubatus) a re a m o n g the most manoeuvrab le of mar ine m a m m a l s (expressed as a m in imum turning radius and speed dur ing manoeuv res ) . They evo lved in terrestr ial and aquat ic env i ronments that a re structural ly comp lex , and feed on prey that are an order of magn i tude smal ler than themse lves . C o m p a r e d to o ther aquat ic o rgan isms, Stel ler sea lions have an unstable body des ign and are p resumed to invoke s w i m m i n g techn iques that ref lect their need to be highly manoeuvrab le . Deta i led in format ion was exper imenta l ly ob ta ined abou t the turn ing techn iques emp loyed by otar i ids through jo int ly ana lys ing k inemat ic a n d kinet ic parameters m e a s u r e d f rom v ideo record ings o f th ree cap t i ve Ste l ler s e a l ions. Centr ipeta l force and thrust product ion we re de te rmined by examin ing body movemen ts th roughout a ser ies o f turns. Resul ts s h o w e d that mos t o f the thrust w a s p roduced dur ing the power phase of the stroke cyc le of the pectoral f l ippers. As opposed to prev ious f ind ings, very little or no thrust was genera ted dur ing initial abduc t ion o f the pectora l f l ippers and dur ing the f inal d rag -based paddl ing style of the stroke cyc le . Peak of the thrust fo rce w a s reached hal fway th rough the power phase , whi le the centr ipetal force reached its m a x i m u m va lue at the beginning of the power phase . K inemat ic aspec ts of the manoeuv res c h a n g e d wi th the t ightness of the turns and the initial veloc i t ies. T h e degree of dorsa l f lex ion of the body changed wi th the turn ing radius and the degree of f l ipper abduct ion var ied wi th sw imming s p e e d . However , the genera l manoeuvr ing techn ique and turn ing sequence remained the s a m e in all the recorded manoeuvres . Cont ras t ing the turn ing per fo rmance of the Stel ler sea l ion wi th a s imple dynamic mode l of unpowered manoeuv res in aquat ic an imals showed signi f icant depar tu res f rom mode l predict ions d u e to the hydrodynamic effects of body movemen ts . Overa l l , the turn ing sequence of the Stel ler sea lion was found to be very cons is tent , and their manoeuvrab i l i t y w a s found to c o m e f rom their abil i ty to vary the durat ion and intensity of m o v e m e n t s wi th in the turn ing sequence . iii TABLE OF CONTENTS A B S T R A C T II L IST O F T A B L E S IV LIST O F F I G U R E S V A C K N O W L E D G E M E N T S VI I N T R O D U C T I O N 1 M A T E R I A L S A N D M E T H O D S 4 Morpho logy 4 Exper imenta l se t -up 10 V ideo analys is 12 R E S U L T S 16 Morpho logy 16 Kinemat ics 19 Kinet ics 26 D I S C U S S I O N 32 Morpho logy 32 K inemat ic ana lys is 37 Kinet ic analys is 48 C O N C L U S I O N S 54 R E F E R E N C E S 56 iv LIST OF TABLES Tab le 1: Mass of the three female Stel ler sea lions on the day of the f i lmed tr ials 17 Tab le 2 : Morpholog ica l da ta of the three female Stel ler sea l ions 18 Tab le 3 : Sequence of movemen ts per formed by SL3 dur ing a 180 degrees turn 23 Tab le 3 (cont inued) : Sequence of movemen ts per fo rmed by SL1 and SL2 dur ing a 180 degrees turn 24 Tab le 4 : M e a n kinetic parameters for the S L 3 , S L 2 , and SL1 31 V LIST OF FIGURES Fig. 1: 3-D representat ion of the body of SL3 as obta ined f rom girth measu remen ts 5 F ig. 2 : L i fet ime var iat ion of 4 girth measuremen ts as a funct ion of body mass 7 Fig. 3 : On -sc reen morpho log ica l measuremen ts of the left pectoral f l ipper of S L 3 9 Fig. 4 : On-sc reen morpho log ica l measurements of the left pelv ic f l ipper of SL3 9 Fig. 5 : Schemat ic of the exper imenta l se t -up 11 Fig. 6 : Sequence of movemen ts based on turn H128 per fo rmed by SL3 22 F ig . 7 : Relat ionship be tween turn ing radius a n d the degree o f body curva ture o f th ree Stel ler sea l ions per forming a 180 degrees turn 25 Fig. 8 : Tra jec tory and speed profi le of the shou lder , cent re of grav i ty , and hip markers o f a Stel ler sea lion per forming a 180 degrees turn 29 Fig. 9 : Tangent ia l and normal accelerat ion profi les of the shou lder , the cent re of gravi ty, and hip markers of a Stel ler sea lion per forming a 180 degrees turn 30 F ig . 10: Tra jectory of the nose and shou lder o f a sea l ion per forming a 180 deg rees turn 44 Fig. 11 : Compar i son of the speed profi les of the shou lder , cent re of grav i ty , and hip markers wi th the predic t ions of a theoret ica l mode l 45 F ig . 12 : Compar i son of the speed profi les of the shou lder , cent re of grav i ty , and hip markers of a Stel ler sea lion per forming a fast and a s low 180 degrees turn 46 Fig. 13: Relat ive turn ing radius and average turn ing speed of three Stel ler sea l ions in compar i son to Cal i fornia sea l ions 47 vi ACKNOWLEDGEMENTS I wou ld l ike to acknow ledge m y thesis adv isors , Dr. A n d r e w W . Tr i tes a n d Dr. Rober t W . B lake for their suppor t dur ing this project. Dr. Rober t W . B lake w a s ex t remely helpful and suppor t ive dur ing the analys is and prov ided m e wi th lab space . Dr. Dav id A . S. Rosen was very accommoda t ing of m y research needs at the Vancouve r Aqua r i um . I wou ld l ike to acknow ledge the Vancouve r A q u a r i u m for prov id ing research faci l i t ies and the staff for the i r help. I especia l ly wan t to thank the Stel ler t e a m at the Vancouve r A q u a r i u m : Rebecca Barr ick and Chad Nords t rom, as wel l as the sea lion t ra iners : S h a w n Carr ier , And rew Irv ine, Billy Lasby, V a n c e Mercer , T roy Nea le , Nigel Wal le r , and Gwyne th Shepha rd for their precious help and suppor t . Thanks also to all the staff, co l leagues and fe l low s tudents in the Mar ine M a m m a l Research Unit w h o provided help on f requent bas is . Th i s wo rk w a s partial ly funded by N S E R C , and grants to the North Pacif ic Universi t ies Mar ine M a m m a l Research Consor t ium f rom N O A A and the North Pacif ic Mar ine Sc ience Founda t ion . Last ly, I wou ld l ike to acknow ledge my fami ly and f r iends. I a m especia l ly gratefu l to my parents for their cont inual suppor t and encouragement . 1 INTRODUCTION Most studies on the locomot ion of aquat ic an imals have focused o n fast-star t responses or on per fo rmance dur ing s teady sw imming (see B lake, 2004 for an ex tens ive rev iew; Domen ic i and Blake, 1997; Firth and B lake, 1 9 9 1 ; Harper and B lake, 1990 ; Ward le , 1975 ; W e b b , 1977 ; W e b b , 1978; We ihs , 1973) . Compara t ive ly , little research has been done on the manoeuvrab i l i t y of aquat ic o rgan isms — wh ich Norberg and Rayner (1987) de f ined as the abi l i ty to turn in a conf ined space (Gerstner , 1999 ; Schrank et a l . , 1999; Wa lke r , 2 0 0 0 ; W e b b , 1983) . Ye t , the major i ty of aquat ic an imals rarely sw im in a s teady l inear fash ion especia l ly in coasta l env i ronments tha t are of ten structural ly comp lex . Organ isms in coasta l env i ronments constant ly need to manoeuv re and to adjust their t ra jector ies in the face of destabi l is ing currents or to avo id obs tac les . Manoeuvrab i l i ty is a lso a key c o m p o n e n t of predator -prey interact ions in the aquat ic env i ronmen t w h e r e predators are often substant ia l ly larger and faster than their prey (How land , 1974) . In the case of mar ine m a m m a l s , and especia l ly p inn ipeds, the major i ty of kinet ic s tudies have concent ra ted on measur ing d rag , funct ional des ign , and m a x i m u m l inear s p e e d (Domenic i and Blake, 2 0 0 0 ; Eng l ish , 1976 ; Fe ldkamp, 1987b; F ish , 1993 ; Fish et a l . , 1988 ; Pongan is et a l . , 1990 ; Stel le et a l . , 2000 ; Wi l l iams and K o o y m a n , 1985) . Few have cons idered the sw imming manoeuvrabi l i ty of mar ine m a m m a l s (Fish et a l . , 2 0 0 3 ; Gers tner , 1999 ; Maresh et a l . , 2004 ; Schrank et a l . , 1999 ; Wa lke r , 2 0 0 0 ; W e b b , 1983) . In a ser ies of recent papers , We ihs (2002) and Fish (2002) d i scussed the d i rect conf l ict be tween manoeuvrab i l i ty and stabi l i ty in aquat ic locomot ion and its effect on the funct ional des ign of act ive aquat ic o rgan isms . B y def in i t ion, a m a n o e u v r e is a change o f t ra jectory o r ve loc i ty caused by a l inear or rotat ional acce lera t ion . In o ther words , it is a contro l led instabi l i ty dur ing wh ich the s u m of all forces and momen ts of force act ing on the cent re of gravi ty of the an ima l do not equa l zero . Th is creat ion of unba lanced forces is theoret ical ly p romoted by morpho log ica l character is t ics , such as body flexibil i ty or highly mobi le contro l sur faces posi t ioned c lose to the cent re of gravi ty (F ish , 1997 ; Fish et a l . , 2003) . In contrast , adapta t ions such as a rigid body, rigid a p p e n d a g e s , the isolat ion of 2 the thrus t -produc ing unit f rom the rest of the body, or the l imited mobi l i ty of the contro l sur faces are all representat ives of a stable des ign and poor manoeuvrab i l i ty . B lake eta/. (1995) show that the m e a n turn ing radius of yel lowf in tuna (0.47 L) , a rigid thunn i fo rm s w i m m e r (a special is t Body and Cauda l Fin — BCF — per iodic sw immer , W e b b , 1984) , is much h igher than more f lexible te leosts such as the dolphinf ish (0 .13 L) , the smal lmouth bass (0 .13 L) , or the t rout (0 .18 L). However , mat ters are m a d e more comp lex by the fact that s o m e r ig id-bodied s w i m m e r s such as the boxf ish have smal l turn ing radii (B lake, 1977; Walker , 2000) . T h e boxf ish uses comb ined osci l lat ions and undulat ion of the pectora l , dorsa l and ana l f ins (B lake, 1977) . Blake eta/. (1995) therefore proposed that o ther pa ramete rs , such as decoup led propulsors ( i .e. Med ian and Pa i red Fins — M P F — for s low and t ight manoeuv res , and BCF mot ion for fast -s tar ts) , p lay a role in the sw imming and manoeuvr ing per fo rmance o f var ious spec ies . T h e turns of most an imals are m a d e up of two e lements : rotat ion and t rans lat ion (except ions to this rule are some M P F s w i m m e r s , wh ich can genera te rotat ion wi thout l inear s p e e d , e . g . the boxf ish ; or s o m e spec ies that exhib i t t ranslat ion wi thout rotat ional speed due to a di f ferent ial def lect ion of their f ins, e .g . seahorses ) . T h e most successfu l indiv iduals engaged in predator -prey interact ions are typical ly those w h o max im ise both turn ing componen ts , i.e. per form a t ight turn and mainta in a high t ranslat ion speed (How land , 1974). In a recent s tudy, Maresh et al. (2004) s h o w e d that bot t lenose do lph ins (also B C F per iodic sw immers wi th a fair ly rigid body) a re capab le of h igh manoeuvrab i l i ty , as exhib i ted by the p inwheel turn ing techn ique (also in Nowacek , 2002 ) . Th is p inwheel techn ique a l lows the an ima l to m in im ise turn ing radius a n d max im ise turn ing rate by t rans forming its fo rward speed into rotat ional s p e e d . T h e relat ive inflexibil i ty of such rigid o rgan isms as the boxf ish or the bot t lenose do lph in theoret ical ly impairs their manoeuvrab i l i ty even though they are capab le of per forming t ight turns. Th is is because the ang le be tween the body and the incoming f low of a rotat ing rigid body is c lose to 90 degrees a long the ent i re body length, wh ich results in an impor tant pressure d rag a round the body that opposes rotat ion and causes dece lera t ion . A second ou tcome of hav ing an inf lexible body is that 3 counterba lanc ing momen ts of force are c reated on both s ides of the dorso-ven t ra l rotat ional ax is , wh ich a lso resists rotat ion (Walker , 2000) . T o m y know ledge , no s tudy has yet prov ided da ta to test these two theoret ica l assumpt ions in aquat ic an imals . An ima ls mus t genera te a force in the direct ion of the manoeuv re to change thei r t ra jectory using a force p roduced by the body itself, the contro l sur faces , or bo th . In the aquat ic env i ronment , act ive an imals can genera te this force th rough ei ther a l i f t -based or a d rag -based m e c h a n i s m , both of wh ich induce d rag . Th is addi t ional drag results in manoeuvr ing be ing hydrodynamica l l y more cost ly than s teady s w i m m i n g . Th i s conc lus ion is suppor ted by W e b b (1991) , and H u g h e s a n d Kel ly (1996) w h o found that a f ish works harder w h e n sw imming at a g iven ave rage s p e e d in uns teady sw imming than at the s a m e speed in s teady sw imming . Va luab le in format ion on the m in imum amoun t of d rag exper ienced by sea l ions s w i m m i n g at var ious speeds can be ob ta ined by analyz ing decelerat ion dur ing pass ive g l ides. Howeve r , d rag is l ikely to be greater dur ing act ive sw imming — and especia l ly dur ing manoeuv res — due to f l ippers and body movemen ts (Stel le, 1997) . Ana lys ing the detai ls of the pectoral and body m o v e m e n t s and their effects on the an imal 's speed can thus deepen our unders tand ing of the b ioenerget ics of unsteady sw imming and the costs of locomot ion. S o m e k inemat ic da ta have been col lected for otar i ids on sw imming speeds , pectoral f l ipper propuls ion, turn ing radius, turn ing rate, and genera l turn ing techn ique (Eng l i sh , 1976; Fe ldkamp, 1987a ; Fish e t a l . , 2 0 0 3 ; Pongan is e t a l . , 1990) . T h e y sugges t tha t otar i ids have severa l k inemat ic features that enhance thrust product ion as wel l as manoeuvrab i l i ty . First, the i r thrust product ion techn ique is based on both a lift- and a d rag -based m e c h a n i s m , wh i ch a l lows t hem to produce a m a x i m u m a m o u n t of thrust th roughout the st roke cyc le . S e c o n d , their body des ign is ideal for promot ing instabi l i t ies ( i .e. , they have a highly f lexible body that has a round c ross-sec t ion and large mobi le pectoral f l ippers p laced c lose to the cent re of gravi ty) . Cal i fornia sea l ions are a m o n g the most manoeuv rab le of mar ine m a m m a l s w h e n compared to spec ies that sw im at mechanica l ly equiva lent speeds (Fish et a l . , 2003) . Howeve r , knowledge abou t 4 the k inemat ics of o ther otar i ids is l imi ted. T h e Stel ler sea lion is an ideal s tudy an ima l to expand current know ledge abou t the manoeuvrab i l i ty of otar i ids. Be ing the largest of the fami ly , Stel ler sea lions exper ience a high va lue of d rag and important inertial fo rces , both of wh ich are likely to constra in manoeuvrab i l i ty . Drag is a backwards act ing force that opposes fo rward mot ion , and inertial forces tend to mainta in the direct ional i ty and resist a change in t ra jectory whi le sw imming . Study ing deta i led k inemat ics and morpholog ica l adapta t ions will prov ide in format ion on how Stel ler sea lions m a n a g e these constra ints to opt imize turn ing capabi l i t ies and sw imming veloci t ies. T o prov ide data abou t the deta i led b iomechanics of an unstable body des ign dur ing a manoeuv re , I repeatedly f i lmed 3 Stel ler sea lions per forming 180 degree turns. Movemen ts of the f lexible body were t racked digital ly th roughout the manoeuv res using three markers p laced a long the body midl ine. T h e sequence of result ing movemen ts was corre la ted wi th deta i led speed var iat ions and wi th the forces act ing on the cent re of gravi ty th roughout the manoeuv res . T h e eva luat ion of accelerat ion (posi t ive and negat ive) dur ing the turn a lso prov ided pract ical in format ion regard ing the power requ i rements and energet ics of a manoeuvr ing sea l ion. M A T E R I A L S A N D M E T H O D S All p rocedures and protocols involving an imals we re conduc ted under the author i ty o f the Universi ty o f Br i t ish Co lumb ia An ima l Ca re Permi t No . A 0 4 - 0 1 6 9 . Morphology Morpholog ica l measu remen ts we re taken on three female Stel ler sea l ions at the Vancouve r Aquar ium Mar ine Sc ience Cent re (Vancouver , BC , Canada ) . T w o fema les w e r e 3 year old juven i les (F00YA and FOOTS — later referred to as SL1 and SL2 respect ively) and the third female w a s a 6 year old adul t (F97HA — referred to as SL3) . Measuremen ts w e r e ga thered on each an imal prior to the f i lmed exper iments . To ta l length was measu red f rom the t ip of the nose to the t ip of the hind f l ippers. S tandard length w a s taken f rom the t ip of the nose to the base of the ta i l . Fig. 1: 3-D representat ion of the body of SL3 as ob ta ined f rom girth measu remen ts . G 1 - G 8 represent gir th measu remen ts 1 to 8 and form the base and the top of each t runcated cone. Dis tances be tween each gir th measu remen t ( d l - d 8 ) prov ide the height of the t runcated cones . C G indicates the cent re of gravi ty, jus t before the trai l ing edge of the pectoral f l ippers. Body vo lume and wet ted sur face a rea we re ca lcu la ted as a success ion o f t runcated cones . Eight girth measu remen ts were taken at known intervals a long the body ( the ears , the neck, direct ly in f ront of the pectoral f l ippers, direct ly behind the pectoral f l ippers, two p laces a long the t runk region be tween f l ippers and hips, the hips, and the posit ion where the body and the hind f l ippers meet ) , each forming the base of a t runcated cone (F ig . 1). T h e an imals we re we ighed dai ly on a G S E sca le , Mode l 350 (scale accuracy ± 0 .1kg) . Body densi ty was obta ined by d iv id ing the ca lcu lated vo lume ( including the vo lume of the f l ippers, see below) by the mass of the an ima l . 6 T h e locat ion of the cent re of gravi ty was de te rmined using the me thod of Domn ing and De Buffreni l (1991) . In brief, an anaesthet ised sea lion was p laced on a f lat board wi th the pectora l f l ippers lying flat against its f lank. T h e board and sea lion we re then careful ly ba lanced in a s e e s a w fashion on a s teel p ipe. T h e posi t ion o f the equ i l ib r ium point co r responded to t he cen t re o f m a s s o f both the an imal and the f lat board together . T h e exact posi t ion of the an imal 's C G f rom the t ip of the nose was then ca lcu la ted by subtract ing the effect of the board . T h e relat ive posi t ion of the C G was thus : _ , .. ... distance nose -CG . . . Relat ive C G posit ion = 100 . standard length T h e d is tance be tween the nose and the posit ion of m a x i m u m th ickness w a s a lso measu red whi le the an imal was anaes the t ised and the relat ive posit ion o f the m a x i m u m th ickness was ca lcu la ted us ing : „ . ... .... . .... distance nose - max thickness Anr. Relat ive max th ickness posit ion = 100 . standard length As two of the th ree s tudy an imals were still g row ing juven i les , body mass w a s measu red dai ly. Addi t ional measu remen ts of body length and 4 gir ths we re taken at least once a week . Fluctuat ions in mass we re observed dur ing the course of the s tudy (be tween 5 . 6 % and 7 . 6 % of the body weight ) . I de te rmined the possib le morpho log ica l impl icat ions o f this we igh t change by looking at the l i fet ime morpho log ica l da ta o f e a c h an ima l (F ig . 2 ) . B a s e d o n the long- te rm re lat ionships be tween weight and gir ths ( G 3 , G 4 , G 7 and G 8 , see F ig. 2) I de te rm ined that a gir th measuremen t error of 3 % covered the poss ib le morpholog ica l effects of the change in mass that occur red over the course of the s tudy. 160 ,120 ' 80 40 0 SO SL1 90 100 110 120 130 H C 150 160 y - 0.42x + 77.69 y = 0.34x + 7S.17 y • 0.27X + 44.99 y a 0.12x + 39.70 170 R2 a 0.87 R2 = 0.87 R2=0.50 H1 a 0.33 160 ' 80 40 0 E-3 90 100 110 SL2 ! 120 120 140 152 160 y = 0.38x + 84.63 y - 0.33x + 80.47 172 R - 0.80 R J = 0.79 ^ ^ r - ^ y = G.34x + 36.37 ; R*=0.62 y = 0.12x + 41.58 ; RS = 0.40 160 SL3 80 f 1 t 1 1 1 90 100 UO 130 Mass [kg] I GO G4 © 6 7 140 O G 8 •M: y a 0.40x + 80.60 y a 0.35X + 77.32 :0.89 :0.91 y - 0.24x + 52.88 ; RfaO.53 y = O.llx + 44.75 ; R2=0.49 Fig. 2: Lifetime variation of 4 girth measurements as a function of body mass. Each cloud of data represents the girth of one cross-section of the body (see Fig. 1): G3 is taken just in front of the pectoral flippers; G4, just posterior to the pectoral flippers; G7 on the hipbone; and G8 on the base of the tail. The dashed lines indicate the minimum and maximum weight of each animal over the study period. 8 Fl ipper measu remen ts we re obta ined via 2 di f ferent techn iques . T h e pro jected sur face a rea , length of the f l ipper f rom the insert ion to the tip of the appendage and w id th of the f l ipper we re taken f rom sca led still pho tographs ana lysed on a PC wi th Scion Image sof tware (Beta vers ion 4 .0 .2 ) . Th ickness measuremen ts we re ob ta ined wi th a spr ing- joint cal l iper a n d a mi l l imetr ic ruler (measu remen t accuracy : ± 0 . 5 m m ) at 13 locat ions a long the pectoral f l ipper wh i le the an ima l was under anaesthet ics . T h e 13 measu remen ts we re located as fo l low: 4 a long the lead ing e d g e , 4 a long the mid l ine, 4 a long the trai l ing edge , and 1 at the tip (all measu remen ts w e r e approx imate ly 15cm apar t ) . T h e f l ipper A s p e c t Rat io (AR) w a s ca lcu la ted a s : AR= {lengthf {projected surface area)' T h e vo lume of the pectoral f l ipper was calcu lated by assuming that it was equ iva lent to a success ion of t runca ted , squa re -based pyramids ( the th ickness o f the base o f the py ramid is the ave rage th ickness of the leading and trai l ing edges of the f l ipper) . T h e s e measu remen ts we re taken on one a p p e n d a g e on ly , and it w a s a s s u m e d that both f l ippers w e r e ident ica l . T h e v o l u m e of the pelv ic f l ippers was ca lcu la ted as a percentage of the vo lume of the pectora l f l ippers. Th is percentage w a s g iven by the ratio of the sur face areas of the pelv ic and pectora l f l ippers. T h e s e va lues we re used in the calculat ion of the total vo lume of the an ima l . Fig. 3: On-screen morphological measurements of the left pectoral flipper of S L 3 . The white dots indicate points where thickness of the flipper was measured with a spring-joint calliper. Fig. 4: On-screen morphological measurements of the left pelvic flipper of S L 3 . Length was measured from the base of the tail to the tip of the middle digit. Width was measured perpendicular to the flipper longitudinal axis at the base of the digital extensions. 10 Experimental set-up T h e an imals we re kept in an outdoor facil i ty wi th constant access to amb ien t , f i l tered seawater (Rosen and Tr i tes , 2004) . Tes ts we re per formed in a 19m long, 5 m deep pool w i th rock and w o o d e n haul-out a reas . Dur ing the course of the s tudy, the an ima ls we re fed p redominant ly Pacif ic herr ing. In addi t ion to the da i ly we igh t measu remen ts , morpho log ica l da ta ( i .e. length a n d gir ths) w a s ga thered once a week . Data we re col lected f rom Augus t 1 5 t h to December 3 r d , 2003 . Exper iments occu r red over a period of 11 days wi th S L 3 , 50 days wi th S L 1 , and 29 days wi th S L 2 . T h e t ra in ing techn ique used was simi lar to that used by Fish (2003) . Us ing posi t ive re in forcement , the s e a l ions w e r e t ra ined to s w i m back a n d forth be tween t w o t ra iners pos i t ioned a t oppos i te ends of the test poo l . As the an imal app roached Tra iner 2 , Tra iner 1 wou ld per form a recall s ignal ( i .e. hit the sur face of the wa te r wi th a target pole) indicat ing to the an ima l to immedia te ly return to T ra iner 1. T h e an ima l wou ld then execute a 180-degrees turn to change its d i rec t ion. Pre l iminary test ing revealed that the longer the d is tance be tween the t ra iners, the lower the an imal 's sw imming s p e e d . Hav ing one t ra iner si t t ing in a kayak posi t ioned 3-5 met res f rom the f ield of v iew of the v ideo camera consequent ly reduced this d is tance. Even though the d is tance be tween the two trainers var ied (be tween 9 and 12 met res) , I ensured that the an imal had room for at least one comple te f l ipper s t roke before enter ing the f ield of v iew of the c a m e r a . T h e 180-degrees turns we re f i lmed wi th a digital v ideo camera (Canon GL -2 ) a t tached 5m a b o v e the wa te r sur face. T h e f i lming rate was set at 60 f rames per second and the z o o m sett ing was equiva lent to a 3 9 . 5 m m open ing on a 3 5 m m focal length, wh ich co r responded to a d iagona l ang le of 57 .42° (angle fo rmed at the apex of the t r iangle def ined by two oppos i te corners of the f ield of v iew and the focal point of the camera ) . T o reduce f lares and sur face ref lect ions, the camera was moun ted wi th a c i rcular polar is ing fi lter (Hoya circular polar is ing f i l ter, 5 8 m m , p i tch: 0 .75) . A c lear Plexiglas sheet (d imens ions : 2.61 metres by 1.98 metres) f loated on the wa te r sur face in the cent re 11 of the field of view of the camera to eliminate visual distortions produced by the surface waves. Only the turns that occurred directly under the Plexiglas sheet were analysed. Fig. 5: Schematic of the experimental set-up. A number of temporary marks were drawn on the fur of the test animals that could be tracked on the video images. Oil-based pastel crayons were used because the captive animals could not be marked with long-lasting paint. The dots were visible for 48 hours before having to be reapplied. However, an oil-based zinc cream was used (Desitin, Zinc Oxide 37%, Cod Liver Oil 13.5%) for one of the three test animals. This was quicker to apply but it had to be re-applied during the experiments because it faded faster while the animal was in the water. The first point was placed on the shoulder blades to represent the movements of the anterior part of the body during each manoeuvre. The second point was situated at the centre of gravity of the animal lying flat with its pectoral flipper tucked in. The third point was placed on the hipbone to represent the movements of the posterior part of the body during turns. Each visual marking was drawn 3 times around the 12 an imal g i r ths: on the left s ide , on the r ight s ide , and a long the backbone . T h u s a total of 9 dots ( roughly 2 .5 c m in d iamete r ) we re pa in ted o n each test an ima l . Video analysis Dur ing each manoeuv re , a ser ies of 12 events were identi f ied and c lassi f ied in a t ime sequence to i l lustrate the turn ing techn ique of each an ima l . T h e s e 12 componen ts w e r e : 1) m o v e m e n t of the head inside the tu rn ; 2) start of the abduc t ion of the pectoral f l ippers; 3) s tar t o f the roll o f the body ; 4) open ing of the interdigital w e b of the pelvic f l ippers; 5) start of the dorsa l f lex ion ; 6) end of the abduct ion of t he pectoral f l ippers; 7) m a x i m u m rol l ; 8) m in imum radius o f curvature of the f lexed body; 9) start of the adduct ion of the pectoral f l ippers; 10) body back in a straight pos i t ion; 11) end of the adduct ion of the pectoral f l ipper; 12) the pelvic f l ippers return to a pass ive, g l id ing posi t ion. T h e depth at the beginn ing and end of a manoeuvre occas ional ly d i f fered because the mot ion of the an imals was not l imited vert ical ly. W h e n this happened , the turn ing radius as seen in 2 d imens ions by the c a m e r a p laced overhead (with its axis normal to the wa te r sur face) appears smal ler than it really is in 3 d imens ions . Th is was corrected using a scale that l inked the s ize of an object on sc reen to its dep th : i.e. a v isual sca le (a 2 m long ruler wi th 10cm black and wh i te increments) was built and f i lmed in the exper imenta l set -up at var ious known depths . From th is, a mathemat ica l relat ionship was ob ta ined that l inked the two ent i t ies: Depth = -0 .0564 (Size on screen in pixels) + 8 .633 . T h e ob jec t measu red before a n d af ter each manoeuv re w a s the d is tance be tween the hip a n d the shou lder dots , wh ich was measu red digital ly on sca led v ideo images the day of the tr ial . 13 Tu rn ing radi i , ins tantaneous speed at the start and end of a tu rn , ave rage s p e e d , acce lera t ion , dece lera t ion , rol l ing deg ree , and durat ion of the manoeuv re we re all measu red f rom the sca led v ideo cl ips using Lenox So f tworks ' V ideopo in t 2 .5 . Al l manoeuv res we re re ferenced to an on-sc reen or igin that w a s pos i t ioned 16 pixels r ight and 16 pixels up f rom the bo t tom left co rne r o f the image . T h e posi t ions of the three lateral dots (shoulder , C G , and hips) w e r e manua l l y t racked at a sampl ing rate of 30Hz . O n most images , the painted marks covered a sur face of a f ew pixels on screen (typical ly be tween 2 and 9 square pixels). T h e exact posi t ion of each mark had to be de termined subsequen t l y using the s p e e d prof i les (see be low) . Ins tantaneous speed w a s ca lcu la ted a s : .. _ P n+1 ~ Pn-1 tn+1 tn-1 w h e r e U n is the ins tantaneous s p e e d o f point n (in m /s ) , P n + 1 a n d P n _ i a re the posi t ions o f the points direct ly after and before point n respect ively (in m) , and t n+i and t n - i is the t ime code of the points direct ly after and before point n respect ively (in s ) . T h e speed prof i le w a s ob ta ined by plott ing ins tantaneous veloci t ies over t ime. T h e speed curve w a s then s m o o t h e n e d by mov ing the t rack points a f ew pixels left, r ight, up, or down wi th in the onsc reen sur face of the painted dot. Th is de termined the exact posi t ion of the marker onsc reen . T o correct for the d i f ference of dep th before and after the manoeuv re , the speed calculat ion w a s mod i f ied , such a s : Uncorrected) - c Q S a , w h e r e a is the ang le that descr ibes the d i f ference of dep th . Tu rn ing radius (R) w a s ca lcu la ted mathemat ica l ly by f i t t ing a half-c irc le to the curved part of the t ra jector ies o f the shou lder , the cent re o f grav i ty , a n d the hip us ing least s q u a r e d regress ions ( i .e. three turn ing radii we re obta ined per manoeuv re ) . T h e trajector ies ob ta ined in V ideopo in t 2.5 w e r e impor ted into S - P L U S 6 . 1 , whe re the least squared regress ions we re pe r fo rmed . O n c e R w a s m e a s u r e d , the real turn ing radius w a s obta ined by tak ing the dep th d i f ference into account : 14 Rreal ~ ^( Rmeasured ) + \^~2~ J ' where R is turn ing radius, and A D is the d i f ference of dep th be tween the en t rance and exi t of a manoeuvre . Negat ive a n d posi t ive acce lera t ions we re ob ta ined by ca lcu la t ing the s lope o f t he best f i t t ing l ine through the appropr ia te port ion of the t ime-speed g raph ( i .e. one line w a s f i t ted to the decelerat ing sect ion of the g raph and another was f i t ted to the acce lerat ing sect ion) . T h e s e port ions of the data set we re ident i f ied v isual ly . T h e speed at the beg inn ing of the turn was def ined as the ins tantaneous s p e e d of the an imal jus t before the f irst manoeuv r ing movemen t (most of the t ime , a h e a d m o v e m e n t a n d the abduct ion of the pectoral f l ippers) . T h e speed at the end of a manoeuv re was def ined as the ins tantaneous speed of the an imal as soon as the midl ine of the body had regained a stra ight pos i t ion, wi th the pectoral f l ippers adduc ted a long the body f lanks. Rol l ing degree could not be measu red direct ly in degrees or radians using the one camera ang le . Ins tead , an index of roll w a s used to g ive a sense of how much and h o w fast the an ima l tu rned its back into the turn before execu t ing the m a n o e u v r e . Th is ' ro l l ing index ' — t racked at 30Hz th roughout the turn — was def ined as the d is tance , norma l to the body ax is , be tween the shou lder marker and the edge of the body as seen on the t op -down camera v iew. Th is d is tance therefore dec reased wi th an increasing rol l ing degree (up to a m a x i m u m roll ing degree of 180 degrees — wh ich cor responded to a d is tance of Ocm). T h e rol l ing index a lso prov ided informat ion on how fast a sea lion ro l led. It w a s measu red be tween the " t ime of en te r i ng " ( i .e. the t ime at wh ich the ins tantaneous enter ing s p e e d w a s measured) and the t ime of m a x i m u m roll ( i .e. w h e n the roll ing index is min imal or , in o ther wo rds , w h e n the d is tance be tween the shou lder marker and the edge of the body — as seen f rom a b o v e — was min imal ) . T h e durat ion of a manoeuv re was def ined as the t ime e lapsed be tween the " t ime of en te r i ng " and the " t ime o f ex i t ing" . F rom the ins tantaneous speed data of the three markers , I ob ta ined the resul tant acce lera t ion : 15 _ Up+i ~ Up-i dp — i tp+1 ~ tp-l where a p is the ins tantaneous resul tant accelerat ion of point p (in m / s 2 ) , U p + i and U p - i are the ins tantaneous veloci t ies o f the points d i rect ly af ter a n d before point p respect ive ly ( in m /s ) , and t p + 1 and tp-i is the t ime code of the points direct ly after and before point p respect ive ly (in s ) . T h e m e a s u r e m e n t o f acce lera t ion (and therefore force) w a s very sens i t ive to t he posi t ion o f the markers on each f rame. I therefore f i t ted the posi t ion- t ime data wi th a po lynomia l funct ion of the 6 t h degree , and ca lcu la ted the second der ivat ive of this po lynomia l funct ion to obta in acce lera t ion . Th is process prov ided the X and Y componen ts of the acce lerat ion vector , wh i ch w e r e then t ranslated and rotated to t rack the fo rward veloci ty vector ( i .e. the veloci ty vec to r hav ing no X componen t in the new referent ial sys tem) . Calculat ing the X and Y componen ts of the acce lera t ion vec tor in this new referent ial sys tem prov ided one componen t paral lel to the ins tantaneous veloci ty vector ( tangent ial acce lerat ion a t ) and one componen t perpendicu lar to the veloc i ty vec tor (normal accelerat ion a n ) . Tangent ia l force w a s F t = m a t and normal force w a s def ined as F n = m a n . T h e degree of curvature of the f lexing body w a s quant i f ied by ca lcu lat ing the radius of the c i rcumcirc le that passed through each one of the th ree on-sc reen markers . Howeve r , the degree of curvature of the body w a s clear ly in f luenced by the posit ion o f the h e a d dur ing the tu rn , wh ich w a s not cap tured by the markers located on the shou lder , C G and hips of the an ima l . There fo re , the posi t ion of the an ima l ' s nose w a s t racked using a four th marke r o n 31 tu rns ( S L 1 : 6 ; S L 2 : 9 ; S L 3 : 16) to prov ide in format ion on the head 's posi t ion and its impor tance in the k inemat ics of the tu rn . A N O V A tests w e r e pe r fo rmed o n var ious parameters descr ibed a b o v e ( i .e. en te r ing s p e e d , ex i t ing s p e e d , roll ing t ime, turn dura t ion , dece lera t ion , acce lera t ion , and turn ing radius) to de termine inter-an imal d i f ferences and di f ferences be tween var ious body parts (SPSS 8.0) . Resul ts we re cons idered signi f icant at P < 0 .05. 16 RESULTS Morphology T h e morpho log ica l da ta for each of the three sea lions is presented in Tab les 1, 2 , and 3, as wel l as Figs 1, 3 , and 4. T h e data shown in Tab les 2 and 3 we re ob ta ined pr ior to the first tr ial of each an ima l . T h e an ima ls ' we ights f luctuated over the course of the s tudy. T h e m a x i m u m mass change for each an ima l over the s tudy per iod was 7 . 6 % for S L 1 , 5 . 8 % for S L 2 , and 5 . 6 % for S L 3 . Body length showed cons iderab ly less var iat ion wi th an ave rage increase of 0 . 6 5 % ( range 0 .22 -1 .43%) . T h e h ighest va lues of the range we re used to de te rmine the uncer ta int ies of the morpho log ica l measuremen ts direct ly af fected by mass var iat ions (Table 2) . Wi th the except ion of the h ipbone a rea (where the body is sl ight ly dorso-vent ra l ly compressed ) , the body of the Stel ler sea l ions has a rounded cross sect ion w h e n in the wa te r (personal observa t ions) . There fo re , a c i rcu lar c ross-sec t ion w a s a s s u m e d in al l ca lcu la t ions o f v o l u m e , we t ted sur face a r e a , and frontal sur face a rea . Pectoral f l ippers and pelv ic f l ippers represented 5 5 . 7 5 % and 4 4 . 2 5 % of the tota l pro jected f l ipper area respect ively. T h e mean aspec t ratio of the pectoral f l ippers was 3 .23 , and the m e a n aspec t ratio of the pelvic f l ippers was 2 .39. Table 1: Mass of the three female Stel ler sea l ions: S L 1 , S L 2 , and S L 3 on the day of the f i lmed tr ials. Animal Age Birth year Date Mass [yrs] [dd-mm-yyyy] [kg] SL1 3 2000 12-08-2003 124.6 15-08-2003 126.5 18-08-2003 127.0 22-08-2003 ' 128.4 13-09-2003 118.7 15-09-2003 121.2 19-09-2003 121.5 03-10-2003 125.5 SL2 3 2000 30-10-2003 138.8 05-11-2003 142.9 06-11-2003 142.8 07-11-2003 143.6 11-11-2003 144.7 12-11-2003 145.9 25-11-2003 146.4 27-11-2003 147.3 03-12-2003 147.3 SL3 6 1997 03-07-2003 138.2 28-08-2003 145.2 29-08-2003 145.8 30-08-2003 145.6 31-08-2003 145.4 02-09-2003 145.7 07-09-2003 146.4 18 Table 2: Morpholog ica l da ta col lected on three female Stel ler sea l ions. T h e s e measu remen ts we re ob ta ined once for each an ima l before thei r f irst f i lmed tr ia l . To ta l length represents the length o f the an imal f rom the nose to the t ip of the pelvic f l ippers. S tandard length represents length to the base of the ta i l , and frontal sur face a rea cor responds to the sur face o f the largest c ross-sect ion of the an ima l . F ineness ratio was ca lcu la ted as : (max leng th ) / (max breadth) . T h e posi t ion of the centre of gravi ty of the an imal lying s t retched was de termined using the techn ique descr ibed by Domn ing and D e Buffreni l (1991) . To ta l length , s tandard length , f ronta l su r face a r e a , pos i t ion o f m a x i m u m th ickness, as wel l as C G posi t ion were obta ined whi le the an imal w a s under anaes thes ia . T h e vo lume is obta ined f rom a ser ies of t runcated cones (see F ig. 1) def ined by 8 gir th measu remen ts taken at known intervals a long the body o f the an ima l . Al l measu remen ts a n d ca lcu la t ions o f the pectora l a n d pelvic f l ippers we re ob ta ined f rom dig i ta l , sca led pictures (sea Figs 3 and 4) . Pectora l length represents the length f rom the base of the pectoral f l ipper to its t ip. Pelv ic length represents the length of the pelvic f l ipper be tween the base of the tai l and the t ip of t he midd le digi t . F inal ly , pelvic max wid th represents the wid th of the spread-out pelvic f l ipper r ight at the base of the digi tal ex tens ions. Measu remen ts we re taken on one f l ipper only and it was a s s u m e d that the second f l ipper w a s ident ical . Tota l f l ipper area was therefore 2 x (pectora l f l ipper a rea + pelvic f l ipper a rea) . SL1 SL2 SL3 Total length [m] 2.27 ± 0.02 2.29 ± 0.02 2.26 ± 0.02 Standard length (L) [m] 1.83 ± 0.02 1.87 ± 0.02 1.92 ± 0.02 Frontal surface area [m2] 0.149 ± 0.004 0.156 ± 0.004 0.167 ± 0.004 Total wetted surface area [m2] 2.391 ± 0.066 2.551 ± 0.061 2.481 ± 0.059 Volume [1] 137.3 ± 10.4 154.7 ± 8.9 150.8 ± 8.4 Fineness ratio - 5.2 ± 0.2 5.1 ± 0.2 4.9 ± 0.2 Position of max thickness [% of L] 44.3 42.8 45.8 CG position [% of L] 57.4 55.6 51.6 Pectoral flipper area [m2] 0.104 0.115 0.107 Pectoral length [m] 0.58 0.60 0.60 Pectoral max width [m] 0.23 0.24 0.24 Pectoral aspect ratio - 3.23 3.13 3.32 Pelvic flipper area [m2] 0.085 0.082 0.092 Pelvic length [m] 0.45 0.46 0.46 Pelvic max width [m] 0.28 0.25 0.29 Pelvic aspect ratio - 2.38 2.53 2.26 Total flipper area [m2] 0.378 0.394 0.396 19 Kinematics A total of 419 turns we re f i lmed f rom Augus t 1 5 t h to December 3 r d , 2 0 0 3 . Al l turns we re partial ly unpowered manoeuv res per fo rmed wi th a non-zero initial speed (a m a n o e u v r e start ing f rom a rest ing posit ion was never observed) . In all 419 even ts , the th ree an ima ls used the s a m e genera l turn ing techn ique to per form the 180-degrees turns. S o m e k inemat ic parameters f luctuated be tween turns, such as the rol l ing deg rees , the degrees o f abduc t ion ( m o v e m e n t a w a y f rom the mid l ine o f the body — Fish et a l . , 2003) of the pectoral f l ippers, and the dorsa l a rch ing of the backbone , e tc . T h e an imals we re never observed per forming a ventra l ly induced tu rn . T h e turn ing techn ique observed w a s in all regards comparab le to the turn ing techn ique descr ibed by Fish et al. (2003) for the Cal i forn ia sea l ion. Tab le 3 presents the sequence of m o v e m e n t s tak ing p lace dur ing the manoeuv re and div ides the techn ique into 6 ma in events (or ser ies of even ts ) . Before per forming the turn ing sequence , the an imal g l ided hor izontal ly w i th its dorsa l s ide towards the sur face. T h e plantar face of the pectoral f l ippers was app l ied aga ins t the ventro- la tera l s ide of the an imal whi le g l id ing, and the pelvic f l ippers were cont rac ted and held toge ther w i th their p lantar faces in contact . Th is posi t ion min imised the inter-digital w e b of the pelv ic f l ippers. Upon enter ing the tu rn , the an ima l per fo rmed three movemen ts : 1) the head was or ien ted and ex tended towards the inside of the tu rn , 2) the pectoral f l ippers we re abduc ted and 3) the an ima l rol led to or ient its back inside the turn (F ig . 6.1) . T h e sequence of these th ree events var ied f rom turn to turn and they we re of ten per fo rmed s imul taneous ly . T h e m o v e m e n t of the pectoral f l ippers was as fo l lows. From their g l id ing pos i t ion, the sea l ions rotated their f l ippers ou twards and brought t hem a w a y f rom the mid l ine of the i r bodies (abduct ion , Figs 6.2 and 6.3) . A t the end of the abduc t ion , the pectoral f l ippers w e r e approx imate ly perpend icu lar to the mid l ine o f the body a n d rema ined s tat ionary unti l the anter io r par t o f the body (head , neck, and torso) star ted to exi t the turn (F ig . 6.4). Dur ing the cu rved part of the t ra jectory, the body w a s ex tended a n d a rched dorsal ly in a U-shape posi t ion (F igs 6.4 a n d 6.5) . T h e inter-digi tal w e b of the pelv ic f l ippers was then ex tended (thus increas ing their sur face area) wi th the 20 ventra l s ides fac ing the outs ide of the turn. As the anter ior body star ted to regain a stra ight posi t ion, the an imal per formed a pectoral f l ipper s t roke and acce lera ted out of the turn (Figs 6.5 and 6.6) . Final ly, the pectoral f l ippers previously held away f rom the body 's mid l ine w e r e brought back a long the an imal 's ventro- la tera l sur face ( i .e. adduct ion) . T h e s t roke m o v e m e n t o f t he pectora l f l ippers was or ien ted d o w n w a r d s a n d backwards , a n d the f ront edges of the f l ippers we re rotated inwards to reposi t ion the plantar face of the pectoral f l ippers against the body of the an ima l . Th is mot ion was c o m p o s e d of two phases as descr ibed by Fe ldkamp (1987a) : 1) the power phase (forceful dorso-vent ra l adduc t ion , wh ich ends in a ful l ex tens ion of the pectoral f l ippers be low the body) and 2) the paddle phase (f l ippers or ien ted ' b roads ide to the f low ' and brought backwards and upwards towards the body) . T h e midl ine of the body of the an imal regained a stra ight posi t ion before the end of the pectoral f l ipper s t roke. A s the p o w e r phase was per fo rmed, the body rol led back to reor ient the dorsa l sur face up, the head rega ined its s t ra ight forward or ientat ion, and the neck rema ined ex tended . Last ly, the pelv ic f l ippers we re slowly rotated and brought back toge ther in the gl iding posit ion descr ibed above (F ig . 6.6) . Somet imes the an imals d id not roll back comple te ly and remained at an angle as they g l ided out of the turn . T h e t ra jectory of the nose of the an imal di f fered f rom the rest of the body. Wh i le the t ra jectory of the shou lder , C G and hips we re smooth ( i .e. a l inear g l ide into the tu rn , fo l lowed circular or el l ipt ical tu rn , fo l lowed by a l inear ex i t ) , the t rajectory o f the nose w a s m o r e i r regular. First , it was d isp laced and ex tended into the turn as the an imal en tered the manoeuv re (as descr ibed above ) , wh ich resul ted in an angu la r t ra jectory. S e c o n d , whi le the an imal re-acce lera ted at the end of the turn , the t rajectory of the nose d id not fo l low the genera l d i rect ion of the rest of the body. Ins tead the nose appeared to initially fo l low a path leading towards the inside of the turn before chang ing its course and tak ing the d i rect ion fo l lowed by the rest of the body (F ig . 10). Th i s change of t ra jectory co r responded wi th the onse t of the adduct ion of the pectoral f l ippers. 21 T h e radius of the m in imum c i rcumcirc le that passes th rough each body marker , including the nose , was measu red in 31 turns. Th is radius, wh ich measu red the deg ree of dorsa l curva ture dur ing the tu rn , ranged f rom 0.27L ( i .e. expressed relat ive to the an imal 's body length — L) to 0 .39L wi th an average of 0 .32L (which cor responds to a range of 0 .51m to 0 .69m wi th an ave rage of 0 .6m) . T h e radii of the 31 turns in wh ich body curvature was measu red var ied be tween 0 .17L and 0.36L wi th an average of 0 .27L (which cor responds to a range of 0 .32m to 0 .64m wi th an ave rage of 0 .5m) . F ig . 7 i l lustrates the relat ionship be tween the relat ive turn ing radius and the relat ive dorsa l curvature. T h e larger the turn ing radius, the stra ighter the body ( l inear regress ion , F 3 0 = 18 .27 , p<0 .001) . 1. ! Plexiglas sheet 4. Fig. 6: Sequence of movements based on turn H128 performed by SL3. Arrows indicate the principal movements of the animal on each frame. 1. The head is oriented towards the inside of the turn and the pectoral flippers are abducted. 2. The body starts flexing dorsally and rolls. 3. The body flexion continues, the abduction of the pectoral flippers reaches an end, the body rolls, and the interdigital web of the pelvic flippers starts to unfold. 4. The roll has stopped, the body is maximally arched dorsally, the digits of the pelvic flipper open. 5. As the body regains a straight position, the pectoral flippers are adducted, the interdigital web of pelvic flippers is spread out. 6. The body regains a straight position, the pectoral flippers reach the last stage of the power stroke, and the pelvic flippers return to their gliding position. 23 Table 3: Sequence of movements performed by SU3 during a 180 degrees turn. Each row corresponds to a turn. The numbers between 1 and 12 in the tables describe the sequence of movements. When several actions take place at the same time, they get the same number and are highlighted in light or dark grey. "NA" signifies that a movement happened out of the field of view of the camera or that it could not be identified clearly. In that case, it is ignored from the sequence. The vertical blanks delimit shorter sequences (of 2 or 3 movements) that are repeatedly distinct from the rest of the sequence. A dashed line in the header between two sections indicates that these two sections tend to be distinct but not always. H i : movement of the head inside the turn; Ab 0 : start of the abduction of the pectoral flippers; RQ: start of the roll of the body; P s t a r t : opening of the interdigital web of the pelvic flippers; F s t a r t : start of the dorsal flexion; Ab i : end of the abduction of the pectoral flippers; R M A X : maximum roll; F m a x : minimum radius of curvature of the flexed body; A d 0 : start of the adduction of the pectoral flippers; F e n d : body back in a straight position; Ad i : end of the adduction of the pectoral flipper; P e n d : the plantar surfaces of the pelvic flippers are back in contact, and the inter-digital web closes gradually afterwards. Turn Hi A b 0 Ro Pstart Pstart A b , p "max H68 2 1 3 4 5 6 7 H71 1 1 2 3 3 4 5 H73 1 1 2 NA 3 4 5 H74 1 1 2 NA 3 4 5 H75 1 1 2 NA 3 4 5 H77 1 1 2 4 3 5 6 H78 1 1 2 4 3 5 6 H84 1 1 2 4 3 5 6 H89 1 1 2 4 3 5 5 H92 1 1 2 4 3 5 6 H93 1 1 2 NA 3 4 5 H95 1 1 3 3 4 4 H96 1 1 2 NA 2 3 3 H97 1 1 1 3 2 4 5 H98 1 1 mem 2 2 3 3 H99 1 1 2 3 4 5 6 H101 1 1 HHflH 3 2 4 5 H103 1 2 4 3 5 5 H104 1 2 4 5 6 6 H105 1 H9B 2 3 4 4 H106 1 1 2 3 4 5 5 H107 1 3 NA 4 5 5 H108 1 2 3 4 4 5 5 H109 1 NA NA 3 4 H110 1 HSE 2 NA 3 4 5 H113 1 1 2 HH 3 4 5 H114 1 NA 2 NA 3 4 5 H115 N H HSH 2 4 3 5 6 H116 1 3 ^ . 5 6 H117 1 1 2 NA 3 4 5 H128 1 1 2 4 3 5 6 ^max A d 0 Pend A d , Pend 9 8 10 10 11 6 6 7 8 9 6 6 HH 7 8 7 6 8 9 10 7 6 8 8 9 7 7 8 9 10 7 7 8 8 9 7 HH 8 9 10 7 6 8 8 9 7 8 9 10 6 6 7 7 8 6 5 7 7 8 4 5 6 6 7 6 6 7 7 8 5 4 6 6 7 7 7 8 9 10 6 6 7 8 9 6 6 7 8 9 7 7 8 8 9 5 5 6 7 8 6 6 7 8 9 6 6 HH 8 6 6 7 8 9 5 5 6 7 8 6 6 7 8 9 7 6 8 9 10 6 6 7 8 9 7 7 8 9 10 8 7 9 10 11 6 6 7 8 9 7 7 8 9 10 24 Table 3 (continued): Sequence of movements performed by SL2. Turn Hi A b 0 Ro Pstart Fstart A b , "max ^max A d 0 Fend A d , Pend T 0 0 7 1 2 3 N A 4 5 N A 7 6 8 9 10 T 0 1 2 1 N A 2 N A 3 4 5 6 7 8 8 N A T 0 8 0 1 2 2 N A 3 4 4 6 5 7 7 8 T081 1 1 1 N A 2 3 3 5 4 6 6 7 T 1 0 8 2 1 1 N A 3 4 5 7 6 8 8 9 T 1 0 9 2 1 1 N A 3 4 5 7 6 8 8 9 T111 3 1 2 4 5 6 7 9 8 10 10 11 T 1 1 2 2 1 1 N A 3 4 5 7 6 8 N A 9 T 1 i |3 1 1 2 N A 3 4 5 7 6 8 8 9 T 1 1 4 3 1 2 6 4 5 7 9 8 10 10 11 Tints 3 1 2 5 4 6 6 8 7 9 9 10 T t f 6 1 1 2 4 3 5 5 7 6 8 9 10 T 1 1 7 2 1 2 5 3 4 6 8 7 9 10 11 T 1 1 8 1 1 1 3 2 4 4 6 5 7 8 9 T 1 2 0 1 1 1 3 2 4 4 6 5 7 8 9 T 1 2 2 2 1 1 • H i 4 4 6 5 7 8 9 T 1 2 3 2 1 1 4 3 5 5 7 6 8 8 9 T 1 2 4 2 1 1 4 3 5 5 7 6 8 9 10 T 1 2 5 1 1 1 3 2 4 4 6 5 7 8 9 T 1 2 6 2 1 1 4 3 5 6 IHI 7 8 9 10 T 1 2 7 2 1 1 4 3 5 5 6 6 7 8 9 T 1 2 9 2 1 1 4 3 5 5 7 6 8 8 9 T 1 3 0 2 1 1 4 3 5 5 7 6 8 8 9 T131 2 1 1 N A 3 4 4 HBfl 5 6 7 8 T 1 3 2 2 1 1 N A 3 4 5 7 6 8 9 10 T 1 3 3 2 1 1 3 3 4 5 7 6 8 9 10 T 1 3 4 1 1 1 N A 2 3 3 5 4 6 7 8 T 1 3 6 1 1 1 2 2 3 3 5 4 6 7 8 T 1 3 8 3 1 2 4 4 5 5 7 6 8 9 10 T 1 3 9 2 1 1 4 3 5 5 7 6 8 9 10 Table 3 (continued): Sequence of movements performed by SL1. Turn H, A b 0 R 0 P.t. r t F „ . „ A b , R„» , Fm.« A d 0 F . n d A d , P . n d Y 0 0 5 i i 2 3 3 4 5 6 7 8 9 10 Y 0 0 6 1 1 2 N A 3 4 5 6 7 8 8 9 Y 0 6 3 1 1 2 H f l f l H H 4 5 6 7 8 8 9 Y 0 6 4 1 N A 2 3 4 5 6 7 8 10 11 Y 0 6 6 1 1 1 N A 2 3 3 5 4 6 6 7 Y 0 6 7 1 1 2 N A 3 4 5 6 6 7 8 9 Y 0 6 8 1 1 1 ygg/gggg/l 3 4 5 5 6 7 8 Y 0 6 9 2 1 1 4 3 5 6 7 8 9 10 11 Y 0 7 0 1 N A 2 4 3 5 5 7 6 8 8 9 Y 0 7 2 2 1 1 4 3 5 N A B L H H H B 7 7 8 Y 0 7 3 2 1 3 5 4 • ;;. 5 6 7 8 9 10 11 Y 0 7 4 1 1 2 4 3 5 6 7 7 8 9 10 Y 0 7 6 1 2 2 4 3 5 5 6 6 7 8 9 Y 0 9 7 2 1 2 3 3 4 4 5 5 6 7 8 Y 0 9 9 1 N A 1 3 2 4 4 5 5 6 7 8 Y 1 0 0 1 N A 1 N A 2 3 3 4 4 5 6 7 Y101 1 1 1 N A 2 3 3 5 6 7 Y 1 0 2 2 1 3 5 4 6 6 8 7 9 10 11 Y 1 0 3 1 N A 1 N A 2 3 3 4 4 5 6 7 Y 1 0 4 2 1 1 4 3 5 5 6 6 7 7 8 Y 1 1 7 2 1 2 3 3 4 4 5 5 6 N A 7 _ 25 CO «J 3 U > o CO 0.41 0.37 ]• 0.33 l 0.29 h 0.25 0.15 0.20 0.25 0.30 Turning radius [BL] 0.35 0.40 Fig. 7: Relat ionship be tween turn ing radius and the deg ree of body curvature of th ree Ste l ler s e a l ions per forming a 180 deg rees tu rn . 31 tu rns are represented here. Body curvature is expressed as the radius of the c i rcumci rc le that passes th rough the 4 body markers : nose , shou lders , C G , and hips. Al l measu remen ts a re exp ressed relat ive to each an ima l ' s body length to compensa te for s ize d i f ferences. 26 Kinetics Out of the 4 1 9 turns that we re f i lmed with the 3 test an ima ls , 195 occur red direct ly under the Plexiglas sheet and we re kept for fur ther ana lys is . In each ana lysed tu rn , I measu red turn ing radius, s p e e d , angu lar s p e e d , acce lera t ion , s l ippage, roll ing t ime , and manoeuvr ing t ime (64 turns for S L 1 , 70 for S L 2 , and 61 for SL3 ) . T h e turn ing sequence w a s per formed in 1.65±0.17s for S L 1 , 1 .74±0.20s for S L 2 , and 1.32±0.13s for S L 3 . T h e s e d i f ferences in turn ing durat ion we re s igni f icant a m o n g all an ima ls (one-way A N O V A ; F 2 = 80 .2 , p<0 .0001) . In ter -an imal var iat ions we re observed on all measu red parameters , except for the decelerat ion of the C G for wh ich data we re often miss ing because the abduc ted pectoral f l ippers covered the marke r (Table 4) . T h e ins tantaneous speed da ta plotted aga ins t t ime produced a typical V - s h a p e d curve in all turns recorded (see F ig . 8) . T h e speed of the an imals whi le gl id ing before the first m o v e m e n t of the turn was constant or sl ight ly dec reas ing . Th is p lateau w a s fo l lowed by a per iod of dece lera t ion , wh ich co r responded to the start of the turn ing movemen ts ( i .e. m o v e m e n t of the head inside the tu rn ; start of the abduct ion of the pectoral f l ippers; start of the roll of the body) . T h e decelerat ion of the centre of gravi ty s topped at or after the t ime of m in imum rol l , before the start of the f l ipper adduc t ion . Immedia te ly fo l lowing this per iod of dece lera t ion , the speed reached a m in imum for a short durat ion before increasing aga in . T h e an imal s tar ted acce lera t ing jus t prior to or at the onset of the f l ipper adduc t ion . Jus t before the end of the f l ipper s t roke, the speed once aga in at ta ined a p lateau of cons tan t or sl ight ly decreas ing s p e e d . At this point in t ime , the mid l ine of the body had not ye t regained a ful ly straight posi t ion. Wi th in a tu rn , the speed profi les of the di f ferent body parts showed s o m e tempora l var ia t ions. T h e anter ior part of the body ( represented by the shou lder marker ) dece le ra ted faster than the middle and poster ior parts ( represented by C G and the hip markers respect ively) (Table 4 and F ig. 9) . T h e overal l range of decelerat ion of the shou lder marker was f rom - 0 . 2 7 m / s 2 to - 5 . 1 8 m / s 2 wi th an average - 2 . 1 9 m / s 2 ( ± 0 .84) ; the C G ranged f rom - 0 . 3 9 m / s 2 to - 4 . 3 7 m / s 2 w i th an average of -27 1 .48m/s 2 ( ± 0 .89) ; the hips ranged f rom - 0 . 2 5 m / s 2 to - 3 . 6 3 m / s 2 wi th an ave rage of - 1 . 6 4 m / s 2 ( ± 0.79) (Table 4 ) . M in imum s p e e d w a s a t ta ined sl ightly before the middle of t he 180 deg rees t ra jectory o f each marker . Consequent l y , the shoulders were the first to at ta in m in imum s p e e d , c losely fo l lowed by the C G , wh ich w a s fo l lowed by the hips. As the an imal ex i ted the tu rn , the shou lders and the C G re-acce lera ted at a s imi lar rate, whi le the acce lerat ion o f the hips w a s cons ide rab ly h igher . T h e range of accelerat ion of the shou lder marker was f rom 0 . 4 8 m / s 2 to 6 . 6 6 m / s 2 wi th an ave rage 3 . 2 9 m / s 2 ( ± 1.18); the C G ranged f rom 0 . 4 1 m / s 2 to 6 . 5 6 m / s 2 w i th an ave rage of 3 . 0 0 m / s 2 ( ± 1.22); the hips ranged f rom 0 . 5 7 m / s 2 to 1 1 . 9 8 m / s 2 w i th a n ave rage o f 5 . 0 0 m / s 2 ( ± 2 .26) (Tab le 4) . A s noted above , the ins tantaneous speed of the three markers reached a c o m m o n p la teau at the end of the turn . Dur ing the tu rn , one componen t of the accelerat ion vec tor w a s paral lel to the veloci ty vector ( tangent ia l accelerat ion) and the other was perpendicu lar to it (normal acce lera t ion) . Each componen t behaved di f ferent ly (F ig . 9) and fo l lowed the tempora l var ia t ions of the body parts. T h e po lynomia l curve descr ib ing tangent ia l accelerat ion c losely fo l lowed the var ia t ions in the speed prof i le. A s the an imal g l ided before enter ing the manoeuv re , the va lue of tangent ia l accelerat ion was sl ightly negat ive and c lose to zero . As the an imal rol led into the turn and abduc ted its pectoral f l ippers, the tangent ia l accelerat ion of the shou lders and C G fur ther dec reased until it reached a m in imum approx imate ly w h e n the pectoral f l ippers at ta ined their full abduc ted pos i t ion. It then c a m e back to zero w h e n the speeds of both body parts reached their m i n i m u m . A s speed inc reased, the tangent ia l accelerat ion became posit ive and reached a m a x i m u m par tway th rough the pectoral f l ipper s t roke before coming back c lose to a nil — or sl ightly negat ive — va lue as the an ima l g l ided out of the turn . T h e hips s tar ted decelerat ing later, w h e n the body s tar ted bending and the trajectory of the hips dev ia ted f rom the trajectory of the o ther two markers . T h e onset o f the hips decelerat ion co r responded to the movemen ts of the pelvic f l ippers. It rema ined negat ive th rough the first half of the manoeuv re and reaccelerated abrupt ly as the hips t ra jectory " cu t t h r o u g h " the o ther 28 t ra jector ies and the body of the an imal s t ra igh tened. T h e m a x i m u m tangent ia l acce lerat ion of the hips w a s reached jus t be fore the e n d o f the power phase . T h e tangent ia l acce le ra t ion o f al l th ree markers peaked after the curved port ion of their respect ive t ra jectory. T h e normal acce lerat ion had a di f ferent profi le than the tangent ia l acce lera t ion . Dur ing the l inear gl ide preced ing the manoeuv re , the normal acce lerat ion of the shou lder and cent re of gravi ty w a s c lose to zero . It then increased w h e n the an imal or iented its head into the tu rn , s tar ted roll ing and abduct ing the pectoral f l ippers, and reached a m a x i m u m jus t af ter the star t o f the power phase. Th is co r responded to a point in the manoeuv re that was sl ight ly af ter the midd le of the 180 degrees turn . Final ly, normal acce lerat ion of the shou lder and cent re of gravi ty qu ick ly c a m e back to a nil va lue at the end of the power phase . T h e normal accelerat ion of the hips s tar ted at the beginn ing of the dorsal f lex ion of the body and reached its m a x i m u m hal fway th rough the power phase . It c a m e back to a zero va lue as the body regained a straight posi t ion. As wi th tangent ia l acce le ra t ion , normal accelerat ion fo l lowed the tempora l var iat ions of the di f ferent body parts ( i .e. the shou lders reached their m a x i m u m va lue first, fo l lowed by the C G , and then the hips) . T h e m a x i m u m va lue of norma l accelerat ion was systemat ica l ly g rea ter than the m a x i m u m va lue of tangent ia l accelerat ion for all three an imals and all three markers . Fu r the rmore , the m a x i m u m normal acce lerat ion a lways preceded the m a x i m u m of tangent ia l acce lera t ion for each marker (F ig . 9) . A t the beg inn ing a n d e n d o f the prof i les, the acce lera t ion cu rves occas iona l l y " o v e r s h o t " c reat ing seeming ly ano ther m a x i m u m or m in imum. Th is was an ar tefact of the calcu lat ion techn ique and d e p e n d e d o n the s h a p e o f the 6 t h deg ree po lynomia l f i t ted to t he pos i t ion d a t a ( for wh i ch I took the second der ivat ive) . 29 • Trajectory of the shoulder • Trajectory of the centre of gravity • Trajectory of the hips Origin 3.9 3.5 7 * 3 . 1 S CL in 2.7 2.3 1.9 • Shoulder • Centre of gravity • Hips : « • • • •••• • • ••: :•• • • • • • • a • • • 0.4 0.8 Time [s] 1.2 1.6 2.0 F i g . 8 :Typical t ra jectory and speed profi le o f the shou lder , centre of gravi ty, and h ips markers o f a Stel ler sea l ion per forming a 180 degree turn . T h e dashed l ines del imi t the bouts o f cons tan t s p e e d , f rom the bouts of decelerat ion and acce lerat ion o n both char ts . 3.9 3.5 -SU 3.1 E &2.7 2.3 1.9 • Shoulder • Centre of gravity . • Hips • « • •ni ' ••••• 1 • • • • • •••• •• • •••• • • • • • • . • • • • • • • * • • • • • • • Fig. 9: Tangent ia l and normal acce lerat ion prof i les o f the shou lder , the centre o f gravi ty, and hips markers of a Stel ler sea l ion per forming a 180 degree turn in corre lat ion wi th its sw imming speed . T h e dashed l ines represent tangent ia l accelerat ion and the plain l ines are normal acce lera t ion. Tab le 4: Mean kinetic parameters for the SL3, SL2, and SL1. The different letters and grey tones (white, grey, dark grey) represent a significant difference between each animal at alpha=0.05. SL1 SL2 SL3 In speed [m/s] 2.69 a 2.93 b 2.92 b Out speed [m/s] 3.22 b 3.08 b 3.77 a Rolling time [s] 0.76 b 0.74 b 0.64 a Turn duration [s] 1.65 0 1.74 b 1.32 a Turning radii [BLs] Shoulders 0.32 b 0.31 b 0.26 a C G 0.32 b 0.30 b 0.27 a Hips 0.33 b 0.33 b 0.26 a Deceleration [m/s 2] Shoulders -1.75 a -2.40 b -2.43 b C G -1.31 -1.65 -1.48 Hips -1.20 a -1.84 b -1.89 b Acceleration [m/s 2] Shoulders 2.50 c 3.39 b 3.96 a C G 2.70 b 3.01 a b 3.29 a Hips 4.11 b 3.86 b 7.12 a 32 DISCUSSION Morphology A ser ies of recent art ic les have i l lustrated the conf l ict be tween stabi l i ty and manoeuvrab i l i ty in act ive, mobi le indiv iduals (F ish , 1997; F ish , 2 0 0 2 ; Fish et a l . , 2 0 0 3 ; We ihs , 2002 ) . Stabi l i ty, by def in i t ion, is a property of the body that creates forces that restore its or ig inal condi t ion w h e n d is turbed f rom a condi t ion of equi l ib r ium. In contrast , manoeuvrab i l i ty is the capac i ty to rapidly change d i rec t ion, wh ich means quick ly creat ing and mainta in ing highly unba lanced forces. Stabi l i ty is advan tageous w h e n an individual is constant ly and steadi ly mov ing for an ex tended per iod of t ime, e .g . dur ing migrat ion, or dur ing tr ips to forag ing g rounds . In these c i rcumstances , the individual tr ies to opt imise its energet ic expend i tu res in relat ion to d is tance covered in order to reach its dest inat ion as cost-ef fect ively as poss ib le . T o do so , it makes use of s t ructures that dec rease the energet ic cost of s teady locomot ion , such as kee ls , r igid dorsa l f ins, and /o r lateral compress ion of the body. T h e s e morpholog ica l character is t ics resist roll and s ide to s ide movemen ts , wh ich o therwise wou ld have to be contro l led by muscu la r act iv i ty and wou ld induce cos ts . A l so , contro l su r faces located far f r om , a n d poster ior to , the cent re o f grav i ty (CG) possess more leverage to correct for unnecessary and 'was te fu l ' movemen ts . Other adapta t ions that min imise drag and max imize thrust product ion include reduced mot ion of the contro l sur faces , anter ior p lacement of the cent re o f grav i ty , r educed f lexibi l i ty o f the body , isolat ion o f the th rus t p roduc ing unit f rom the rest of the body , and a lunate caudal fin wi th a high aspec t ratio ( large span and relat ively smal l cord) (B lake, 2004 ; F ish , 1997 ; F ish , 2 0 0 2 ; Fish et a l . , 2003) . In cont rast to stabi l i ty, manoeuvrabi l i ty is benef ic ial w h e n escap ing a fast predator or w h e n try ing to capture an e lus ive prey (How land , 1974) . In these s i tuat ions, energy conserva t ion is less crucial than k inemat ic per fo rmance, because a) being caught by a predator is not an opt ion and b) predatory success depends on the predator 's abil i ty to ou t -manoeuv re its prey. 33 Stel ler sea l ions (and otar i ids in genera l ) d isplay morpho log ica l character is t ics of a very unstable body des ign . T h e y have highly mobi le contro l sur faces p laced at their cen t res of gravi ty, rounded cross-sect ions (with the except ion of the heads and the hips that a re both sl ight ly dorso-vent ra l ly compressed ) , and very f lexible bodies. T h e cent re of gravi ty is pos i t ioned past the middle of the body (Table 2) , sl ight ly anter ior to the insert ion of the trai l ing e d g e of the pectoral f l ippers. T h e s e morpho log ica l fea tures contrast marked ly wi th the list of character is t ics assoc ia ted wi th stabil i ty presented above . A n unstable body des ign l ikely prov ides a number o f eco log ica l advan tages for Stel ler sea l ions. Domenic i (2003) a rgues that the body des ign of an o rgan ism is highly in f luenced by their env i ronment and life history traits (e .g . thunn i fo rms, such as tunas , inhabi t pe lag ic mar ine habitat and have one of the most s tab le funct ional des ign — Blake, 2004) . Stel ler sea l ions, on the other hand , a re amph ib ious creatures that spend a substant ia l amoun t of t ime on land (part icular ly dur ing the breeding season) a n d , w h e n in the water , a re most ly found f rom nearshore to the edge of the cont inenta l shel f (Loughl in et a l . , 1987; Nat ional Mar ine Fisher ies Serv ice , 1992) . S t rong and mobi le pectoral f l ippers a re an advan tage on land because they suppor t the an imal 's we igh t and a l low greater mobi l i ty. Otar i ids are capab le of agi le quadrupeda l locomot ion on land (Eng l i sh , 1976) . In compar i son , phoc ids tha t have less deve loped pectora l f l ippers a n d s w i m us ing thei r pelv ic f l ippers and body osci l la t ions, a re awkward and s low on land. Sea l ions use thei r on- land agil i ty to reproduce, rest, and escape f rom mar ine predators . More speci f ical ly , ma le sea l ions are highly terr i torial dur ing the breed ing season and use their terrestr ial mobi l i ty to establ ish terr i tor ies and f ight-off compet i to rs . The i r dorsa l f lexibil i ty a l lows t hem to keep their torso and head in an upright posi t ion to wa tch over terr i tor ies. Final ly, Stel ler sea l ions use thei r mobi l i ty on land to c l imb rocks out of reach of mar ine predators and f ind areas she l tered f rom s to rms. T h e unstab le body des ign of Stel ler sea lions is a lso advan tageous in the aquat ic env i ronment . Stel ler sea l ions are oppor tunis t ic predators that fo rage on a w ide var ie ty of prey spec ies wi thout a c lear pre ference for one part icular kind (R iedman , 1990; Sinclair and Zeppe l i n , 2002) . Prey spec ies 34 range across severa l taxas f rom pelagic to benth ic , and f rom gregar ious to indiv idual ist ic. Merr ick et al. (1997) c lass i f ied the d ie t o f the Ste l ler sea l ions in A laska a s : gad ids ( i .e. wa l l eye pol lock, Paci f ic c o d , and Pacif ic hake) ; Pacif ic s a l m o n ; smal l school ing f ish ( i .e. cape l in , Paci f ic herr ing, eu lachon , and Pacif ic sand lance) ; f latf ish ( i .e. ar rowtooth f lounder and rock so le ) ; o ther demersa l f ish (i .e. scu lp ins, rockf ish, S t ichae idae, skates , sharks , and lamprey) ; A tka macke re l ; and cepha lopods (i.e. squ id and oc topus) . T h e s e prey spec ies use a w ide var iety of ant ipredator techn iques , such as schoo l ing and confus ion techn iques , crypt ic behav iours , use of natural covers ( i .e. kelp beds and rocky ocean f loor) , e tc . For an act ive predator such as the sea l ion, predatory success depends on the abil i ty to c o m e wi th in str ik ing d is tance of the prey and to main ta in this d is tance long enough to launch a str ike (normal ly a rapid neck extens ion fo l lowed by a b i te, wh ich occas iona l ly involves suct ion of wa te r — personal observat ions) . T h e ex tent to wh ich Stel ler sea l ions use col lect ive foraging techn iques is not c lear, but it appears that sea lions typical ly cap ture one prey at a t ime and have to ingest it before captur ing the next one (personal observa t ions) . T h u s the f inal bout of all predatory str ikes is a one -on -one interact ion be tween the sea lion and its prey. It is dur ing this crucia l f inal app roach towards a m u c h smal le r and m o r e manoeuv rab le p rey i tem that manoeuvrab i l i ty becomes an advan tage (How land , 1974) . T h e most impor tant mar ine predators of the Stel ler sea lion are kil ler wha les and great wh i te sharks (R iedman , 1990) . Being signi f icant ly larger, these specia l ised s w i m m e r s can reach h igher sw imming speeds than sea l ions. T h e only chance of surv ival for the sea l ions cons is ts of ou t -manoeuv r ing their predators o r escap ing to a terrestr ia l re fuge. A g a i n , an unstab le body des ign a n d good manoeuv r ing capabi l i t ies a re a def in i te advan tage in these s i tuat ions (How land , 1974) . F rom a k inemat ic perspect ive , the morpho log ica l features of the otar i ids present undeniab le advan tages and contr ibute to their super ior manoeuvr ing capabi l i t ies c o m p a r e d to o ther mar ine m a m m a l s (Fish et a l . , 2003) . T h e rounded cross-sec t ion of the body o f the sea l ion faci l i tates rol l ing, one o f the prepara tory manoeuv res execu ted before per fo rming a tu rn . T h e lack o f lateral compress ion and dorsa l f in , wh ich wou ld resist rol l ing, a l lows the sea lion to genera te a roll ing 35 m o m e n t wi th only a sl ight def lect ion of one of the pectoral f l ippers. Cons ider ing that rol l ing appears to be a necessary manoeuv re preceding a tu rn , it is advan tageous — if no t necessa ry — to reduce its durat ion and its cost in order to turn as quick ly and as economica l l y as poss ib le . Hav ing a f lexible body a l lows the an imal to bend in the direct ion of the turn and thus to reduce cross- f low and addi t ional pressure d rag dur ing the manoeuv re . Bend ing the body whi le en ter ing the manoeuv re wi th a non-zero speed a lso helps generat ing the rotat ional m o m e n t necessary to the turn (Tucker , 2000) . T h e large and very mobi le pectoral f l ippers (average of 2 1 8 0 c m 2 , wh i ch represents 5 6 % of the total f l ipper sur face a rea — Tab le 2) of the Stel ler sea lion are posi t ioned very c lose to the cent re of gravi ty and play a crucia l role dur ing a turn . First, they act as independent def lec tors and provoke the body rol l ; and s e c o n d , they act as hydrofoi ls genera t ing an impor tant lift force towards the inside o f the turn tha t c h a n g e s the t ra jectory o f the cent re o f gravi ty . T h e y a re a lso used to p roduce thrust dur ing a bi lateral s t roke cyc le that makes use of both lift and d rag fo rces . As shown by Fe ldkamp (1987a) for the Cal i fornia sea l ion, the s t roke cyc le of the Stel ler sea lion is m a d e up of a l i f t -based recovery phase and a power phase based on lift a t f irst and end ing up in a d rag-based paddl ing movemen t . T h e recovery and power phases are per fo rmed dur ing the turn . T h e recovery phase (or abduc t ion) takes p lace dur ing the f irst hal f o f t he tu rn a n d the p o w e r phase (or adduct ion) occurs dur ing the second half of the turn . T h e grea ter the mobi l i ty of the pectoral f l ippers, the higher the ampl i tude of the stroke cyc le — and there fore , the grea ter the thrust . T h e pelv ic f l ippers, wi th an average sur face of 1 7 0 0 c m 2 ( 4 4 % of the total f l ipper sur face a rea — Tab le 2) , serve two purposes . First, the interdigital w e b of the pelv ic f l ippers is ful ly ex tended dur ing the turn a n d resists an ou twa rd sl ip o f the pelv ic a rea . S e c o n d , as in m a n y f ly ing ver tebra tes , in wh ich tai ls faci l i tate genera t ing ae rodynamic torques and substant ia l ly enhanc ing the qu ickness of body rotat ion (Dudley , 2002) , they are used as a rudder to genera te rotat ional momen ts . Fish et al. (2003) m e a s u r e d morpho log ica l parameters on the f l ippers of two Cal i forn ia sea l ions, wh ich present s o m e interest ing d i f ferences wi th the f l ippers of Stel ler sea l ions. T h e total f l ipper a rea 36 of the Cal i fornia sea lion used in their s tudy was a fract ion of the ave rage Stel ler sea lion f l ipper a rea (total f l ipper a rea : 0 . 2 2 7 m 2 for the Cal i fornia and 0 . 3 8 9 m 2 for the Stel ler sea l ions) , even though the an imals in both studies were of comparab le s ize (L= 1.89m and m a s s = 1 3 7 . 8 k g for the male Cal i fornia and L=1 .87m and mass=137 .7kg on ave rage for the Stel ler sea l ions). Th is d isc repancy in the pro jected a rea of the f l ippers might reflect an ear ly deve lopmen t of the f l ippers in the Stel ler sea lion preced ing the growth of their body, wh ich is ul t imately much larger and heav ier than the Cal i fornia sea l ion. Stel ler sea lions are the largest otar i ids wi th the ma les reaching up to 1,120kg and females to 350kg (Loughl in et a l . , 1987; Nat ional Mar ine Fisher ies Serv ice , 1992 ; Winsh ip et a l . , 2001) . In compar i son , Cal i fornia sea lions g row up to a m a x i m u m of 390kg for ma les and 110kg for females (R iedman , 1990). T h e d i f ference in f l ipper a rea o f the Ste l ler a n d Cal i forn ia s e a l ions a lso af fects the kinet ics o f s w i m m i n g . Accord ing to the lift equat ion (/_ = APCLU2 ), the lift fo rce (Z.) is direct ly proport ional to the sur face a rea of the lift genera t ing appendage (AP) such that the larger the sur face a rea , the higher the lift va lue (Hoerner and Borst , 1975) . F rom a k inemat ic perspec t i ve , h igher lift forces produced by the f l ippers t rans late into a more sudden change of t ra jectory and a qu icker body rotat ion. Fur thermore , the a m o u n t o f wa te r acce lera ted dur ing the d rag -based padd l ing m o v e m e n t of the pectoral f l ippers is re lated to the sur face a rea of contact be tween the l imb and the water . In this way , large f l ippers m o v e d s lowly p roduce thrust m o r e eff ic ient ly than sma l l f l ippers m o v e d rapidly (Eng l i sh , 1976) . Based only on f l ipper s ize, Stel ler sea l ions wou ld appea r more manoeuv rab le and more ef f ic ient s w i m m e r s than the Cal i forn ia s e a l ions. Howeve r , this apparen t advan tage is probably negated by the fact that the larger Stel ler sea l ions suf fer f rom a substant ia l ly h igher body d r a g . Theoret ica l ly , the relat ively smal l Stel ler sea l ions that I s tud ied shou ld have been more manoeuvrab le than the Cal i fornia sea l ions that Fish et a l . (2003) s tud ied . Howeve r , my results do not suppor t this hypothes is (see F ig. 13), perhaps because other parameters such as mot ivat ion levels p layed a role in the individual sw imming per fo rmance of each a n i m a l . Ano ther possib le 37 exp lanat ion is that the aspec t ratio of the pectoral f l ippers of the Cal i forn ia sea l ions w a s greater than that of the Stel ler sea l ions, wh ich cou ld have resul ted in the f l ippers o f the Cal i forn ia sea l ions genera t ing a greater lift force (Hoerner and Borst , 1975) . T h u s the degree of manoeuvrabi l i ty be tween Cal i fornia and Stel ler sea lions my have been compensa ted by the d i f ference in f l ipper shape be tween the two spec ies . In the case of large endo the rms inhabit ing the cold waters of the North Pacif ic O c e a n and the Bering S e a , g row ing large s w i m m i n g appendages wil l induce high the rmo-energe t i c cos ts . In large adu l t an imals that exper ience cons iderab le inertial forces and require large contro l sur face areas to mainta in a cer ta in degree of manoeuvrabi l i ty , the b iomechan ica l advan tages d rawn f rom these large f l ippers can be p resumed to compensa te for the thermoregu la t ion cos ts . But the fact that younger and smal ler an imals — wh ich exper ience less inert ia — grow large f l ippers at an ear ly age suggests that young an ima ls a lso d raw k inemat ic advan tages f rom en la rged f l ippers, such as h igher lift and thrust fo rces. Kinematic analysis S o m e aspec ts of the turn ing techn ique appear to be c losely re la ted, even though the dif ferent e lements of the turn ing sequence show a certain degree of tempora l var iabi l i ty. Tab le 3 i l lustrates that the f irst th ree m o v e m e n t s o f the sequence (notably the head d i sp lacement ins ide the tu rn , the abduct ion of the pectoral f l ippers and the start of the roll ing m o v e m e n t of the body) d o not fo l low a def ini te sequence and are in terchangeable . I ndeed , t he posi t ion o f t he head is c losely l inked to the an imal 's f ield of v iew, whereby d isplac ing the head inside the turn m e a n s commi t t ing to the manoeuv re . In other wo rds , the an imal moves its head inside of the turn and stops looking a h e a d , thereby creat ing a rotat ional momen t (yaw) that init iates the rest of the manoeuv re . T h e turn is not i rreversible at this point but d e m a n d s ad jus tments to contro l y a w to reverse the movemen t , wh ich is t ime consuming and induces a higher va lue of d rag (see Tucke r , 2000) . Moreove r , costs increase wi th increased s p e e d , g iven that d rag scales wi th the square of s p e e d . 38 In a s i tuat ion w h e r e the sea lion is uncertain abou t the recall s igna l , it wou ld be advan tageous to de lay the head m o v e m e n t sl ight ly be fore commi t t ing to the tu rn . It is wo r th ment ion ing tha t I never observed the rol l ing m o v e m e n t wi th my s tudy an ima ls before the start of the abduct ion of the pectoral f l ippers. Th is suppor ts the hypothes is that rol l ing is contro l led by an asymmet r i ca l def lect ion of the pectoral f l ippers dur ing the ear ly s tage of the abduc t ion . Directly fo l lowing this initial phase (head movemen t , body rol l , and start of the pectoral f l ipper abduc t ion) , the an ima l p roceeds to a rch its body dorsal ly , ro tates its pe lv ic f l ippers ou twards and ex tends the interdigi tal w e b of the pelvic f l ippers. T h e s e events of ten happen s imul taneous ly , but w h e n they do not, the body f lexion precedes most of the t ime (Table 3) . T h e pelvic f l ippers serve two purposes dur ing a manoeuv re : a) they contro l the m o v e m e n t of the poster ior part of the body and prevent an ou tward sl ip and b) they genera te rotat ional momen t . W h e n the an ima l ' s t ra jectory is rect i l inear, the pelv ic f l ippers a re in a posi t ion that min imises exposed sur face and therefore, fr ict ion d rag . Rotat ing the pelvic f l ippers and expos ing a m a x i m u m surface before the body starts manoeuvr ing is thus det r imenta l kinet ical ly because of the undesi red forces thus c rea ted . T h e onset of the pelvic f l ippers ' m o v e m e n t is s imu l taneous to the start of the dorsal a rch . A t that point, the pelvic contro l sur faces start to be at an ang le wi th the rest of the body and serve as a rudder. Fur thermore , the body itself takes part in genera t ing rotat ional momen t w h e n it a rches into the turn (as sugges ted by Tucke r , 2000) . In short , the second phase of the turn ing sequence sees the g rowth of the centr ipetal force, wh ich results in a change of sw imming trajectory. Dur ing Phase 3 , the rol l ing movemen t and the abduct ion of the pectora l f l ippers s top at their m a x i m u m va lue . Both even ts a re c losely related and often happen s imu l taneous ly because the pectoral f l ippers create a large sur face area that resists the rol l ing m o v e m e n t once they are stat ionary and maximal ly abduc ted . W h e n both events a re not s imu l taneous the f l ipper abduct ion a lways ends before the rol l ing m o v e m e n t (Tab le 3) . Th is con f i rms the fact tha t the abduc t ion o f the pectoral f l ippers is respons ib le for the creat ion of the rol l ing movemen t . 39 Roll ar ises w h e n the two contro l sur face a rea genera te di f ferent amoun ts of lift force (Hoerner and Borst , 1975) . In the case of an aircraft, the a i lerons a re used to vary the a m o u n t of lift genera ted by each w ing . In the absence of f l ipper a i lerons, sea l ions rely on o ther m e c h a n i s m s to create this d i f ference in lift force. O n e mechan ism is to vary the ang le of at tack o f each f l ipper, wh ich af fects lift force. T h e second mechan i sm acts not on the force direct ly, but on the d is tance be tween the f l ipper and the midl ine of the body. T h e greater this d is tance, the higher the m o m e n t of fo rce . There fo re , roll ing m o m e n t is c rea ted by abduct ing the outer f l ipper ( i .e. the left f l ipper in a r ight turn) qu icker than the inner f l ipper. Th is exp la ins w h y roll ing cont inues whi le the f l ippers a re apparent ly ful ly abduc ted ( i .e. the inner f l ipper that is away f rom the camera — F ig . 5 — w as presumably still in mot ion) . Howeve r , these hypotheses canno t be ver i f ied wi th the present record ings because the ang le of v iew of the c a m e r a and the resolut ion o f the images d id not a l low deta i led analys is o f the mot ion of the pectoral f l ippers. Dur ing Phase 4 , the body of the an imal is max imal ly a rched and the adduc t ion ( m o v e m e n t towards the midl ine of the body) of the pectoral f l ippers star ts. Wh i le Phase 3 co r responded to a bout of m in imum s p e e d , this phase marks the beg inn ing o f the power s t roke a n d the acce le ra t ion ou t o f the manoeuvre . T h e adduct ion of the pectoral f l ipper is ak in to the descr ip t ion o f Fe ldkamp (1987a) , and starts wi th a dorso-vent ra l power stroke (or power phase in Fe ldkamp 's te rmino logy) . A s noted by Fe ldkamp, such a m o v e m e n t creates a force or iented forward ly and dorsal ly . T h e t iming of the onset of the pectoral adduct ion w i th , or sl ightly before, the m a x i m u m curvature of the body, a l lows the an imal to m a k e opt imal use of the dorsal componen t of the force . A t this point in the manoeuvre , the body is a rched in a U-shape and the centre of mass is in the midd le of the curved part of the trajectory. T h e t imely onse t of the dorsal ly or iented force therefore prov ides a useful centr ipetal componen t and a fo rward componen t . Per fo rmed ear l ier in the manoeuv re , the power stroke wou ld leave the an ima l w i thout contro l over the later s tages of the turn (because the contro l sur faces wou ld then be in a ventra l posi t ion). Shou ld the sea lion produce a late s t roke , it wou ld cause the an imal to loose substant ia l amoun t of speed over the ear l ier s tages of the turn ing manoeuv re dur ing wh ich no thrust is p roduced . 40 Phase 5 is the end of the manoeuvre . T h e body regains a stra ight or ienta t ion, and the adduct ion of the pectora l f l ippers reaches an e n d . T h e end of the body f lexion of ten c o m e s f irst as the an ima l s lowly terminates the paddle phase. O n these occas ions , the end of the padd le phase consists of a passive adduc t ion dur ing wh ich the f l ippers are brought up towards the vent ra l f lanks of the an ima l . No thrust is p roduced dur ing the paddle phase. Dur ing Phase 6, the pelv ic f l ippers exit the turn and regain a g l id ing pos i t ion. T h e interdigital w e b remains ex tended as long as the pelvic f l ippers fo l low the curved t ra jectory. A s soon as the ang le be tween the midl ine of the body and the pelvic f l ippers returns to zero , they begin to rotate and the f l ipper sur face a rea d imin ishes. T h e turn ing sequence var ied be tween the three an imals I s tud ied . First, all th ree d isp layed a s t rong direct ional pre ference. SL1 and SL3 a lways per formed right turns whe reas SL2 per fo rmed left turns. In the case of S L 3 , the s ide preference was potent ial ly af fected by her impai red eyes ight o n the left eye . She m a y have chosen the direct ion of the turn accord ing to her f ie ld of v i ew , wh ich was most ly on her right s ide. For the other two an ima ls , it is diff icult to de te rmine wh ich one of the muscu lar , s t ructural , or behavioura l preferences most ly caused these di rect ional p re ferences . Tab le 3 a lso shows that SL3 tended to start roll ing only after her head and pectoral f l ippers we re in mot ion , whereas SL1 and SL2 per formed all three behav iours in terchangeably . Here aga in , de lay ing the roll for SL3 m a y have been a behavioura l adaptat ion to her impai red left eye . Wh i le the an imals per formed their manoeuv res they tended to keep v isual contact w i th the t ra iners , and thus rol l ing the r ight s ide downwards in a right turn wou ld have reduced SL3 's f ield of v iew of the sur face. For the s a m e reason , al l three an imals rol led their heads less than the rest of the i r body dur ing the tu rn . T h e degree of body rol l , wh ich was l inked to the 3D sw imming path dur ing the tu rn , a lso var ied be tween manoeuv res . If it w a s super ior to 90 deg rees , the an ima l t ended to d i ve , a n d converse ly if it was less than 90 degrees , the an imal c a m e c loser to the sur face. O n top of the var iat ions o f turn ing techn ique and sequence of ac t ions , d i f ferences in t he w a y certain movemen ts we re per fo rmed we re observed . For ins tance, dorsa l curva ture var ied wi th turn ing 41 radius, i.e. the larger the turn ing radius, the stra ighter the body (F ig . 7 ) . In w ide turns, wh ich involved a low rotat ional momen t , reducing the dorsa l f lexion and thus main ta in ing a s t reaml ined body shape was advan tageous because it l imited the decelerat ion due to fo rm d r a g . In t ight turns however , the fast rotat ion of a non-f lex ing body wou ld result in the creat ion of an important lateral pressure d rag over the ent i re body length. In these s i tuat ions, the curved body l imited the lateral d rag . Abduc t ion of the pectora l f l ippers was another var iable that d i f fered be tween an ima ls . T h e genera l st roke techn ique can be d iv ided in three major sect ions as presented by Fe ldkamp (1987a) . Dur ing the recovery phase , the leading edge of the pectoral f l ipper is rotated ou twards and brought forward and dorsal ly to vary ing degrees of sweep ( i .e. ang le be tween the midl ine of the f l ipper and the midl ine of the body) and d ihedra l ( i .e. angle be tween the sur face the f l ipper and the hor izontal p lane wh ich conta ins t he mid l ine o f the body) . It fo l lows the power phase dur ing wh ich the f l ippers are rotated inwards, and brought ventral ly and backward . T h e s t roke ends wi th the padd le phase dur ing wh ich t he f l ippers cont inue to rotate inwards a n d m o v e backwards a n d up , to rest o n the vent ra l f lanks of the an ima l . A s ment ioned above , the s w e e p and d ihedra l of the pectora l f l ippers vary in di f ferent turns. Part icular ly in very s low turns, the pectoral f l ippers a re not ful ly abduc ted , i.e. the sweep does not app roach 90 degrees (perpendicu lar to the midl ine of the body) and the d ihedra l is much less p ronounced than dur ing faster turns (F ig . 12). Accord ing to the ae rodynam ic theory , lift a n d d rag dec rease w i th t he d is tance be tween the lift-generat ing appendages and the body (Hoerner and Borst , 1975) . For sea l ions per forming a s low turn , it is therefore advan tageous to keep the pectoral f l ipper c lose f rom the body because a) it min imises the drag force and decelerat ion rate and b) the sys tem does not requi re a high lift force to make its t ra jectory change . Final ly, an important posi t ive d ihedra l is s y n o n y m o u s to a st roke of high ampl i tude , wh ich has been posit ively corre lated wi th sw imming speed (Fe ldkamp , 1987a) and therefore is unnecessary in a s low manoeuvre . 42 T h e turn ing techn ique of the Cal i fornia sea lions and Stel ler sea l ions is str ik ingly s imi lar (Fish et a l . , 2003) even though the Cal i fornia sea lion tends to per form the manoeuv re relat ively faster (F ig. 13) and have sl ight ly di f ferent morpholog ica l character is t ics (see a lso Eng l ish , 1976 ; Fe ldkamp, 1987a) . Fish et a l . (2003) noted tha t the at t i tude o f the f l ippers in the Cal i forn ia sea l ion is h ighly var iable and that the body of the an imal is very f lexible dorsal ly . It is this comb ina t ion of highly mobi le contro l sur faces a n d body f lexibi l i ty that prov ides both spec ies wi th a n impress ive a r ray o f manoeuvr ing capabi l i t ies w i thout hav ing to change the basic turn ing techn ique itself. For examp le , the incl inat ion of the turn ing p lane is de te rmined by the roll ing deg ree ; the t ightness of the turn var ies wi th the body f lex ion; and the ampl i tude of the s t roke in f luences s p e e d . Th is is w h y the six phases of the genera l turn ing techn ique are qui te cons is tent even though minor var iat ions in the turn ing sequence are obse rved . Sea l ions are capab le of f ine- tuning parts of the turn ing sequence and adapt ing it to var ious s i tuat ions, wh ich g ives the manoeuvr ing techn ique greater versat i l i ty ( i .e. rol l ing w i thout deviat ing f rom a rect i l inear sw imming path before start ing the turn ing m a n o e u v r e ; contro l l ing the roll ing degree of the head to mainta in v isual contact wi th a part icular ob jec t ; modi fy ing the roll ing degree of the body to inf luence the an imal 's posi t ion in the wa te r co lumn at the end of the manoeuv re ; vary ing the pectora l f l ipper s t roke to contro l thrust product ion) . Us ing th is one genera l turn ing techn ique , the an imal can therefore produce an a lmos t infinite number of underwate r manoeuvres . It is probably for th is reason that the turn ing techn ique obse rved in o the r otar i ids, such as the Cal i fornia sea l ion, is vir tual ly identical to the one observed here. In cont rast , o ther mar ine m a m m a l s such as do lph ins a n d wha les d o not have this manoeuv r ing capabi l i ty and have to rely on a range of turn ing techn iques whe the r they w a n t to min imise turn ing radius, or dece le ra t ion , or max imise turn ing rate. T h e bot t lenose do lph in for examp le relies on a " p i n w h e e l " techn ique to min imise the turn ing radius of its mouth a rea and max im ise its turn ing rate. Maresh et a l . (2004) descr ibed the manoeuv re itself in these t e rms : "Du r i ng the p inwhee l , the an imal appea red to keep its rost rum at a f ixed point, and rapidly rotate its body a round that po int . " 43 Such a manoeuv re essent ia l ly t ransforms all the t ranslat ion speed o f the sys tem into rotat ional speed . But the dolph in thus of fer ing its ent i re s ide to an impor tant c ross- f low probably pays a substant ia l pr ice and endures a signif icant decelerat ion due to the creat ion of p ressure drag around its body. Moreover , as thrust canno t be p roduced dur ing the p inwhee l itself, the an ima l has to "wa i t " until the end of the manoeuv re before it can re-acce lera te . In the case of the sea l ions, this de lay is min imised as they rely on their large and mobi le pectoral f l ippers to star t acce lerat ing half way th rough the tu rn . • Trajectory of the nose • Trajectory o f the shoulder •* • • Plexiglas sheet * • . | \ • * Origin Fig. 10: Tra jectory of the nose and shou lder of a sea l ion per forming a 180 degrees turn as f i lmed f rom above th rough a Plexig las sheet . 1. T h e nose of the an imal is d isp laced inside the turn and genera tes a t ra jectory that has an initial h igh curvature. 2. As the an imal per forms a s t roke ou t o f the manoeuvre and s t ra ightens its body, the dorsal componen t of the thrust force has to be cor rected by a movemen t o f the head and neck o f the an imal ( represented here by the t ra jectory o f the nose) . Fig. l l : Compar i son of the speed prof i les of the shou lder , centre o f gravi ty, and hips markers of a Stel ler sea l ion per forming a 180 degrees turn wi th the predict ions o f a theoret ical model o f the speed var iat ion of a Stel ler sea l ion th rough an unpowered turn (Blake and C h a n , in rev iew). T h e black l ines represent the mode l predict ions. Each one o f the three l ines is based o n a di f ferent re ferenced a rea , i.e. total wet ted sur face a rea , f rontal sur face a rea , or v o l u m e 2 7 3 . For a d iscuss ion on the d i f ference between re ference a reas , see A lexander (1990) . 3.9 — Shoulder — Centre Of arsvity - - Wetted surface area 3-5 Frontal surface area • Volume2'3 1.6 2.0 Time [s] Fig. 12: Compar i son of the speed prof i les of the shou lder , centre o f gravi ty, and h ips markers o f a Stel ler sea l ion per forming a fast and a s low 180 degrees turn. Note the d i f ference in decelerat ion and acce lerat ion as wel l as the average speed in both turns. T h e black l ines o n the graphs represent the predict ions o f the theoret ical model o f an unpowered turn (B lake and C h a n , in rev iew) . T h e an imal 's midl ine and f l ipper posi t ion is indicated in the t op -down v iew on the lef t -hand s ide. In turn A. the profi le of the pectoral f l ipper in the midd le o f the turn is very shor t ( indicated by the circ le and the ar row) , wh ich means that the sweep ang le of the pectoral f l ippers is c lose to 90 degrees . Th i s is not the case in turn B. where the prof i le of the pectoral f l ipper is longer and does not move far f rom the body 's midl ine ( indicated by the circ le and the ar row) . 4 7 2 5 1.5 u o 2 0 0.1 0.2 0.3 0.4 Radius (L) Fig. 13: Relative turning radius and average turning speed of three Steller sea lions in comparison to California sea lions (shaded area). Both turning radius and speed are expressed relative to the body length (L). The California sea lion data is taken from Fish et al. (2003). 48 Kinetic analysis T h e speed profi les of mos t turns had a typical V - s h a p e , p receded and fo l lowed by bouts of constant (or sl ight ly decreas ing) speed (F ig . 8 ) , wh ich co r responded to the g l ides at the beg inn ing and at the end of the turns. As the an ima l star ted or ient ing its body inside of the turn and abduc ted its pectoral f l ippers, the dece lera t ion became more p ronounced (F ig. 8 ) . Th is dece lera t ion contrasts wi th the analysis o f the foref l ipper propuls ion of the Cal i forn ia sea lion (Fe ldkamp , 1987a) , wh ich s h o w e d that abduct ion of the pectoral f l ippers ( i .e. the recovery phase) creates thrust . T h e d isc repancy be tween m y f ind ings and those o f Fe ldkamp (1987a) might be because Ste l ler sea l ions and Cal i fornia sea l ions use a di f ferent s t roke techn ique. But the d i f ference is more likely because Fe ldkamp stud ied sea l ions sw imming l inearly aga ins t an incoming current . The re are obv ious advan tages to produc ing thrust dur ing as much of the s t roke cyc le as possib le w h e n sw imming in a straight l ine. However , creat ing a fo rward force whi le prepar ing and adjust ing for a manoeuv re can be destab i l is ing. Pectoral f l ippers serve a dua l purpose of creat ing centr ipeta l force and thrust dur ing the turn . T o per form a tu rn , the an ima l abducts and posit ions its contro l sur faces in a m a n n e r that wil l genera te as much centr ipeta l force as poss ib le . Th is requ i rement is qu i te di f ferent f rom t ry ing to genera te as much forward thrust as possib le ( i .e. the f l ippers have a di f ferent ang le of a t tack dur ing the abduc t ion , the degree of d ihedra l var ies , e tc . ) . There fo re , I suggest that otar i ids f ine- tune the st roke techn ique to preferent ia l ly p roduce thrust dur ing de f ined parts o f t he s t roke cyc le , i.e. by chang ing the ang le of a t tack of the pectoral f l ippers, or the d ihedra l and ampl i tude of the s t roke , or the s w e e p of the pectora l f l ippers, e tc . T h e adduc t ion of the pectora l f l ippers of the Stel ler sea lion is c o m p o s e d of the two phases descr ibed by Fe ldkamp (1987a ) : the power phase fo l lowed by the padd le phase . In m y s tudy, I observed the latter w h e n a pass ive gl ide fo l lowed the st roke but noted that it w a s omi t ted w h e n the an imal per fo rmed two success ive s t rokes. In Cal i fornia sea lions s w i m m i n g rect i l inearly, mos t of the propuls ive force comes f rom the paddle phase (Fe ldkamp, 1987a) . In c o m p a r i s o n , m y results 49 showed that most of the thrust product ion occurs dur ing the power phase . Acce le ra t ion s tops at the end of the power phase , wh ich indicates that the paddle phase p roduces little thrust . T h e fact that the an imal by -passes the padd le phase a l together dur ing a doub le s t roke fur ther suppor ts this f ind ing. A t the onse t of the power phase , the an imal is in the middle of the turn and its body is bent dorsal ly in a U-shape. A t this point, the dorso-vent ra l movemen t of the pectoral f l ippers genera tes a dorsal ly or iented fo rce , wh ich points in the s a m e direct ion as the anter ior part of the body . Th i s force is used to reacce lera te . In a l inear s i tuat ion, a dorsa l fo rce is waste fu l hydrodynamica l l y because of the heave (a dorso-vent ra l t ranslat ion movemen t ) thus c rea ted . In order to correct the heave and mainta in a rect i l inear pa th , the an ima l has to spend energy correct ing its t ra jectory by mov ing its head and neck (as noted at the end of a f ew turns I recorded — see F ig . 10). As previously men t ioned , Cal i fornia sea l ions asked to sw im l inearly aga ins t a current used the padd le phase to p roduce mos t o f the thrust (Fe ldkamp, 1987a) . Dur ing this phase , the main componen t of the thrust force is d i rected fo rward , wh ich l imits ver t ica l movemen ts . In the l imited space of a sw im mi l l , it makes most sense for the an imal to use the paddle phase as much as possib le. A g a i n , this sugges ts that otar i ids modi fy their s t roke cyc le , i.e. emphas is ing the thrust product ion of one phase ove r the other , depend ing on their in tended s w i m m i n g t ra jectory. Final ly, in a manoeuvr ing s i tuat ion, the power phase not on ly p roduces thrust , but a lso plays an impor tant role in choos ing the f inal sw imming d i rec t ion. For examp le , the sea lion can per form more than a 180 degree turn if it de lays the onset of the power phase and main ta ins the dorsa l arch longer. Converse ly , if the an ima l mainta ins a low body curvature and per forms a f l ipper s t roke ear ly o n , it wil l turn less than 180 degrees . Each body marker (shou lders , C G , hips) was seen to fo l low a sl ight ly d i f ferent t ra jectory, wh ich was ref lected in the di f ferent speed profi les (Figs 8 , 12). T h e m in imum speed of each body marker was reached before the middle of their respect ive curved t ra jectory. For the shou lders and the C G , the 50 m in imum w a s reached before the start of the power phase . T h e hips on the o ther hand , reached their m in imum s p e e d later, approx imate ly a t the star t o f the power phase . T h e m in imum speed of the shou lders and the cent re of gravi ty w a s of ten fo l lowed by a bout of constant or even sl ight ly increasing speed ( i .e. a ' f lat ' m in imum) , dur ing wh ich the pectoral f l ippers we re abduc ted and mot ion less. Th is per iod somet imes carr ied th rough to the ear ly s tage of the power phase. Th is impl ies that the lift force genera ted off the st i l l , abduc ted pectoral f l ippers can produce thrust before the start of the power phase . A s the an ima l a rches dorsa l ly , the hips ( representat ive of the poster ior end of the body) m o v e outs ide of the turn and their t ra jectory depar ts f rom the t ra jectory o f the o ther t w o markers . Th is is a resul t o f t he rotat ional m o m e n t c reated by the d isp lacement of the head inside the turn (for mo re o n the effect of head d isp lacement on rotat ion momen t , see Tucker , 2000) . A s the body regains a stra ight posi t ion at the end of the tu rn , the t ra jectory of the hips c rossed the two other t racks (F ig . 8 ) , and the poster ior end of the body therefore acce lera ted faster than the shou lders and C G . T h e dorsa l f lex ion and the ex tended pelvic f l ippers limit the ou tward mot ion of the poster ior end of the body and reduce the overal l dece lera t ion . If the an imal does not take these measures and remains in a stra ight pos i t ion , the hips will rotate ou twards , thus expos ing the body to an important cross f low and pressure drag much like the bot t lenose do lph ins dur ing a p inwhee l manoeuv re (Maresh e t a l . , 2004) . In the c a s e o f a rigid object , the pressure d rag c rea ted ove r t he body 's prof i le reduces both the t ranslat ional and the rotat ional veloci t ies because it ac ts on the ent i re length of the an ima l , on both s ides o f the cent re o f mass . A s the turns b e c o m e t ighter a n d faster , t he rotat ional and t ranslat ional veloc i t ies increase as does the ou tward slip of the poster ior e n d , thus inducing a high pressure d rag , wh ich sca les wi th the square of s p e e d . In the case of the f lexible Stel ler sea l ion, the relat ionship be tween initial s p e e d , turn ing radius, and poster ior slip is not as s t ra ight forward as wi th a rigid body. In fact , such a c lear re lat ionship be tween these th ree var iab les wou ld only be expec ted w h e n the an ima l s w i m s c lose to its m a x i m u m sw imming capabi l i ty , and cannot a rch its back any further. I d id not obse rve this in m y study 51 because the test an ima ls d id not m a k e use of their m a x i m u m dorsa l f lex ion and d id not sw im at their m a x i m u m sw imming capabi l i ty (personal observat ion) . Never the less , it is c lear tha t the dorsa l f lex ion a l lows an imals to take advan tage of the rotat ional m o m e n t genera ted ove r the anter ior part of the body wi thout suf fer ing f rom the added pressure drag over the poster ior e n d . Stel le et a l . (2000) de te rm ined that the average coeff ic ient of hyd rodynamic d rag of 6 Stel ler sea lions pass ive ly gl id ing was 0 .0046 ( re ferenced to total wet ted sur face a rea) , 0 .044 ( re ferenced to v o l u m e 2 7 3 ) , or 0 .080 ( re ferenced to frontal sur face area) at a m e a n Reyno lds n u m b e r of 5 . 5 2 X 1 0 6 . Recent ly , B lake and C h a n (in rev iew) deve loped a s imp le dynamic m o d e l , wh ich predicts the speed of a submerged aquat ic an ima l per forming an unpowered turn as a funct ion of t ime. By combin ing this mode l w i th the da ta o f Ste l le e t a l . (2000) , I pred ic ted a theoret ica l dece lera t ion o f 0 . 4 m / s 2 a t an initial speed of 3 m / s (F ig . 11). However , my exper imenta l da ta for the s a m e initial speed indicates decelerat ions of 1.6 m / s 2 , l . l m / s 2 , and 0 . 8 m / s 2 for the shou lders , cent re of gravi ty and hips respect ive ly , all fo l lowed by an important accelerat ion (for ave rage accelerat ion and decelerat ion va lues , see Tab le 4) . T h e r e a re a n u m b e r o f potent ia l exp lanat ions for the substant ia l d i ve rgence b e t w e e n observat ions and mode l predict ions of decelerat ion rates. First, the observed sea lion turns w e r e only unpowered dur ing the first half of their manoeuvres . Dur ing the second half o f the manoeuv res , the sea l ions m a d e use of the centr ipeta l accelerat ion and per fo rmed a f l ipper s t roke to c reate a posit ive acce lerat ion that the mode l does not account for. S e c o n d , as B lake and C h a n (in rev iew) ment ion , the coeff ic ient of d rag ( C d ) used in the mode l equat ions does not inc lude the ef fect of d rag on the contro l sur faces . G iven the s ize and posit ion of the contro l sur faces of the Stel ler sea lion dur ing a tu rn , the amoun t of d rag they induce is potent ial ly important . Th i r d , the C d used in the mode l w a s ob ta ined f rom ins tantaneous rates of decelerat ion dur ing l inear g l ides. In o ther words , it was a s s u m e d that C d is the s a m e dur ing a manoeuv re and a long a rect i l inear s w i m m i n g path . Dur ing a pass ive g l ide, a sea lion is mot ion less, keeping its pectoral f l ippers adduc ted a long its ventra l f lanks, and reducing its pelvic f l ippers a rea . Th is conf igurat ion min imizes hyd rodynamic drag (Stel le et a l . , 52 2000) because it exposes a m in imum amoun t of sur face a rea to fr ict ion d r a g , and the sea lion mainta ins a s t reaml ined shape . Dur ing a manoeuv re however , an an ima l genera tes a s ide force to dev iate f rom its l inear t ra jectory and this can only be accomp l i shed by a body a n d / o r a fin movemen t . It is therefore likely that C d is greater dur ing a manoeuv re due to these movemen ts (Hughes and Kel ly, 1996 ; Stel le et a l . , 2000 ; W e b b , 1991) . Unfor tunate ly , mos t of the l i terature avai lab le on the coeff ic ient of d rag is re lated to pass ive drag (Bilo and Nacht iga l l , 1980 ; Fe ldkamp, 1987b; Wi l l iams and K o o y m a n , 1985) . T h e d i f ference of d rag exper ienced by a sw imming sea lion in a pass ive gl ide and in a manoeuv re that involves both body a n d f l ipper m o v e m e n t s is i l lustrated in t he d i f ferent dece lera t ion rates o f Figs 11 and 12. T h e body and f l ipper movemen ts change the s t reaml in ing of the an imal (its shape in relat ion to the incoming f low) , thus in f luencing the va lue o f C d , a n d m a k i n g the a v e r a g e dece lera t ion rate at least 3 t imes greater than predicted by the mode l . T h e speed var iat ion of a s low turn and a fast turn both d ive rge f r om the predict ions o f the theoret ica l mode l (F ig . 12) e v e n though the movemen ts of the f l ippers are not as p ronounced dur ing the s lower manoeuv re . Th is sugges ts that body m o v e m e n t s a re most ly respons ib le for the dece lera t ion a t t he beg inn ing o f the tu rn . B lake and Chan (in rev iew) showed that their dynamic mode l predicts the dece lera t ion of Thunnus albacares accurate ly . T. albacares is a spec ia l ized thunn i fo rm cru iser , wh i ch has l imi ted body a n d f in movemen ts , and canno t per form very t ight turns (0 .47L) . In o ther wo rds , their unpowered manoeuv res a r e c loser to a l inear g l ide than the manoeuv res o f t he f lex ib le s e a l ions. Th is exp la ins w h y the dynamic mode l y ie lded a better fit of the thunn i fo rms than it d id for Stel ler sea l ions. T h e normal acce lerat ion (a n ) of the shou lders and the cent re of gravi ty of the Stel ler sea l ions s tar ted increasing dur ing Phase 1 of the manoeuvr ing sequence ( i .e. head d isp lacement , rol l , and f l ipper abduc t ion) . It cont inued to increase until reaching a m a x i m u m at the onset of the power phase of the pectoral f l ippers, and c a m e back c lose to zero dur ing Phase 5 of the sequence , at the end of the power phase. T h e m a x i m u m va lue of normal acce lerat ion never exceeded 2 0 m / s 2 53 (approx imate ly 2g) , a va lue much lower than previously eva luated for otar i ids (notably 5.13g for the Cal i forn ia s e a l ion - Fish et a l . , 2003) . T h e d i f ference be tween these m a x i m u m accelerat ion va lues in di f ferent spec ies of otar i ids might be exp la ined in two ways . First, my exper imenta l des ign may not have forced the Stel ler sea lions to reach their h ighest per fo rmance level — and even the upper 2 0 % of m y data m a y not be representat ive of the ex t reme manoeuvr ing and sw imming capabi l i t ies of the an imals (personal observat ions) . Up to s o m e ex t reme va lues , sw imming speed and turn ing radius are under behavioura l contro l and thus are principal ly in f luenced by mot ivat ion levels. Th i s m a y expla in w h y Fish et a l . (2003) ob ta ined higher relat ive veloci t ies and turn ing radii for Cal i forn ia sea l ions (F ig . 13) even though their exper imenta l setup was simi lar to ours . U2 A second poss ib le exp lanat ion is that Fish et al. (2003) used the equat ion ac = an = — and the average turn ing speed to calculate centr ipetal acce lerat ion (equal to the norma l acce lera t ion , assuming the turn fo l lows a circular t ra jectory) . G iven that the m a x i m u m va lues of a n a re reached dur ing the t ime of m in imum s p e e d , the equat ion wil l tend to overes t imate the m a x i m u m centr ipetal accelerat ion for ave rage va lues of speed . Based on the d i f ference be tween the ave rage speed and the m in imum speed of the 195 turns I obse rved , I assess that the overes t imat ion of m a x i m u m a c was about 3 0 % . T h e m a x i m u m tangent ia l acce lera t ion (a t ) c lose ly fo l lowed the m a x i m u m o f a n , the reby i l lustrat ing the dua l role of the power phase — to reposit ion the body at the end of the turn and acce lera te . T h e hips a lso fo l lowed the s a m e progression after a short de lay (as seen by the peak in a n fo l lowing the peak in a t ) . In all of the turns that I recorded , the m a x i m u m normal acce lera t ion w a s systemat ica l ly greater than the m a x i m u m tangent ia l acce lera t ion. In o ther wo rds , it took a greater force to mainta in a cu rved t ra jectory and resist slip than to reacce lera te . In the a b s e n c e o f dorsa l keels or s t ructures that help to c rea te this impor tant centr ipeta l force, sea l ions (as wel l as penguins - Hu i , 54 1985) depend o n their large pectoral f l ippers, wh ich expla ins w h y they have to roll to appropr ia te ly posi t ion these cont ro l sur faces. CONCLUSIONS Stel ler s e a l ions use the s a m e techn ique to turn as has been repor ted for ano the r o tar i id , the Cal i fornia sea lion (Fish et a l . , 2003) . However , s igni f icant new in format ion was ob ta ined about the techn iques e m p l o y e d by otar i ids by jo int ly ana lys ing both k inemat ic a n d kinet ic parameters o f the turns per fo rmed by Stel ler sea l ions. First, the da ta s h o w tha t otar i ids a re one o f the m o s t manoeuv rab le mar ine m a m m a l s even though they dep loy a cons is tent turn ing techn ique. Changes in initial speed or tu rn ing ang le do not affect the turn ing sequence . Rather , Ste l ler sea l ions vary the durat ion a n d intensi ty o f m o v e m e n t s wi th in the turn ing sequence , and thus have an a lmost infinite number of opt ions to de termine direct ional i ty. S e c o n d , the major i ty of obse rved sea lion turns had a V - s h a p e speed pat te rn , wh ich reflects the part ial ly powered manoeuv r ing s ty le o f otar i ids. Dorsa l f lex ion a n d abduc t i on m o v e m e n t s o f the large pectoral f l ippers inflict more drag compared to a l inear g l ide dur ing the f irst s tages of a manoeuv re . Th is t rans la tes into increased dece lera t ion , wh ich becomes less p ronounced once the pectoral f l ippers a re ful ly abduc ted . T h e m a x i m u m centr ipeta l acce lerat ion is reached sl ight ly af ter the m in imum speed at the beginn ing of the power phase of the pectoral f l ipper s t roke. Final ly, the power phase of the pectoral f l ipper s t roke causes the an ima l to acce lera te at the end of the tu rn . A th ird notab le f ind ing w a s that m y assessmen t o f f l ipper m o v e m e n t o f otar i ids dur ing a turn (pectora l propuls ion) d i f fered signif icant ly f rom the results obta ined by Fe ldkamp (1987a) for l inearly s w i m m i n g an imals . I found that the abduct ion of the pectoral f l ippers (a lso k n o w n as the recovery phase) by an imals prepar ing for a turn did not p roduce thrust as ind icated by the cons iderab le 55 decelerat ion I obse rved . Howeve r , Fe ldkamp (1987a) found that thrust was genera ted dur ing the recovery phase . Dur ing the second part of the tu rn , I found that most of the thrust was p roduced dur ing the power phase of the f l ipper s t roke. Little thrust was created dur ing the padd le p h a s e , to the point that it wou ld be sk ipped a l together dur ing a doub le s t roke. In contrast , Fe ldkamp (1987a) found that the paddle phase p roduced most of the thrust genera ted over the ent i re s t roke cyc le in l inearly sw imming an ima ls . Th is cou ld suggest that Stel ler sea l ions and Cal i forn ia sea l ions have a di f ferent st roke techn ique . However , g iven the or ientat ion of the forces dur ing the s t roke cyc le and the great mobi l i ty of the th rus t -p roduc ing appendages , it is more likely that both spec ies modi fy their s t roke techn ique accord ing to the si tuat ion (i .e. l inear versus curved sw imming t ra jectory) . A f inal notewor thy f ind ing s temming f rom my data is that Stel ler sea l ions are not as manoeuvrab le as Cal i fornia sea l ions in te rms of pure turn ing per fo rmance (Fish et a l . , 2003) . However this conc lus ion may need fur ther scrut iny g iven that the turn ing per fo rmance o f the sea l ions in both exper imenta l setups m a y have depended heavi ly on the mot ivat ion level of the s tudy an ima ls . 56 R E F E R E N C E S Alexander, D. E. (1990) . Drag coeff ic ients of sw imming an ima ls : ef fects of us ing di f ferent reference a reas . Biological Bulletin 179, 186-190 . Bilo, D. and Nachtigall, W. (1980) . A s imple me thod to de te rmine d rag coef f ic ients in aquat ic an imals . Journal of Experimental Biology 87, 357 -359 . Blake, R. W. (1977) . O n ostrac i i form locomot ion. Journal of the Marine Biological Association of the United Kingdom 57, 1047-1055 . Blake, R. W. (2004) . Fish funct ional des ign and sw imming per fo rmance. Journal of Fish Biology65, 1193-1222 . Blake, R. W. and Chan, K. H. S. (in rev iew) . A s imple mode l of the dynamics of turn ing in aquat ic an imals . Journal of Fish Biology. Blake, R. W., Chatters, L. M. and Domenici, P. (1995) . Tu rn ing radius of ye l lowf in tuna (Thunnusalbacares) in uns teady sw imming manoeuv res . Journalof Fish Biology46, 536 -538 . Domenici, P. (2003) . Habi tat , body des ign and the s w i m m i n g per fo rmance o f f i s h . In Experimental Biology Reviews. Vertebrate biomechanics and evolution, eds V . L. Bels A . Cas inos and J . - P . Gasc ) , pp. 137-160 . Ox fo rd : B IOS Scient i f ic Publ ishers Ltd. Domenici, P. and Blake, R. W. (1997) . T h e k inemat ics and pe r fo rmance o f f ish fast-s tar t sw imming . Journal of Experimental Biology 100, 1165-1178 . Domenici, P. and Blake, R. W. (2000) . B iomechan ics in behav iour . In Biomechanics in Animal Behaviour, eds P. Domen ic i and R. W . B lake) , pp . 1-17. O x f o r d : B I O S Scient i f ic Pub l ishers L td . Domning, D. P. and De Buffrenil, V. (1991) . Hydrostas is in the s i ren ia : quant i ta t ive da ta and funct ional interpretat ions. Marine mammalscience!, 331 -368 . Dudley, R. (2002) . Mechan isms and impl icat ions of an imal f l ight maneuverab i l i t y . Integrative and Comparative Biology 42, 135-140 . English, A. W. (1976) . L imb movemen ts a n d locomotor funct ion in t he Cal i fo rn ia s e a l ion (Zalophus californianus). Journal of Zoology, London 178, 341-364 . Feldkamp, S. D. (1987a) . Foref l ipper propuls ion in the Cal i fornia sea l ion, Za lophus cal i forn ianus. Journal of Zoology, London 212, 4 3 - 5 7 . Feldkamp, S. D. (1987b) . Sw imming in the Cal i fornia sea l ion: morphomet r ies , d rag and energet ics . Journal of Experimental Biology 131, 117 -135 . Firth, H. R. and Blake, R. W. (1991) . Mechan ics of the start le response in nor thern pike, Esox lucius. Canadian Journal of Zoology 69, 2831 -2839 . Fish, F. E. (1993) . In f luence of hydrodynamic des ign a n d propuls ive m o d e on m a m m a l i a n sw imming energet ics . Australian Journal of Zoology 42, 7 9 - 1 0 1 . 57 F ish , F. E. (1997) . Biological des igns for enhanced maneuverab i l i t y : ana lys is of mar ine m a m m a l per fo rmance. In Tenth International Symposium on Unmanned Untethered Submersible Technology: Special Session on Bio-engineering Research Related to Autonomous Underwater Vehicles, (ed . A . a . O N R ) . D u r h a m . F ish , F. E. (2002) . Ba lanc ing requi rements for stabi l i ty and maneuverab i l i t y in ce taceans . Integrative and Comparative Biology 4 2 , 8 5 - 9 3 . F ish , F. E., Hur ley , J . a n d C o s t a , D. P. (2003) . Maneuverabi l i ty by the sea lion Zalophus californ/anus. turn ing per fo rmance of an unstable body des ign . Journal of Experimental Biology 206 , 667 -674 . F ish , F. E., I n n e s , S. a n d R o n a l d , K. (1988) . K inemat ics and es t imated thrust product ion of sw imming harp and r inged sea ls . Journal of ExperimentalBiology137, 157 -173 . Gers tne r , C. L. (1999) . Maneuverab i l i ty o f four spec ies o f cora l - ree f f ish tha t d i f fer in body a n d pectoral- f in morpho logy . Canadian Journal of Zoology 77 ', 1102-1110 . Harper , D. G. a n d B l a k e , R. W. (1990) . Prey capture and the fast-star t pe r fo rmance of the ra inbow trout Salmo gairdneriand the northern pike Esoxlucius. Journal of Experimental Biology 150, 321 -342 . Hoerner , S. F. a n d Bors t , H. V . (1975) . Fluid dynamic lift. Brick T o w n , N .J . : Mrs . Liselot te A . Hoerner . H o w l a n d , H. C. (1974) . Opt ima l st rategies for predator avo idance : the relat ive impor tance of speed and maneuverab i l i ty . Journal of Theoretical Biology47, 333 -350 . Hughes , N. F. a n d Ke l l y , L. H. (1996) . A hydrodynamic mode l for es t imat ing the energet ic cost of sw imming maneuvers f rom a descr ipt ion of their geomet ry and dynamics . Canadian Journal of Fisheries and Aquatic Sciences 53 , 2484 -2493 . Hu i , C. A. (1985) . Maneuverab i l i ty of the Humbo ld t penguin (Spheniscus humboldti) dur ing s w i m m i n g . Canadian Journal of Zoology 63, 2165 -2167 . Lough l i n , T. R., Pe rez , M. A. a n d Mer r ick , R. L. (1987) . Eumetopias jubatus. In Mammalian Species Account no.283, pp. 1-7: T h e Amer i can Society of Mammalog is t s . M a r e s h , J . L., F i sh , F. E., N o w a c e k , D. P., N o w a c e k , S. M. a n d W e l l s , R. S. (2004) . High per fo rmance turn ing capabi l i t ies dur ing forag ing by bot t lenose do lph ins (Tursiops truncatus). Marine mammal science 20 , 498 -509 . M e r r i c k , R. L , C h u m b l e y , M . K. a n d B y r d , G . V . (1997) . Diet d ivers i ty o f Ste l ler sea l ions (Eumetop ias juba tus) and their populat ion dec l ine in A l a s k a : a potent ia l re lat ionship. Canadian Journal of Fisheries and Aquatic Sciences 54, 1342-1348 . N a t i o n a l M a r i n e F i s h e r i e s S e r v i c e . (1992) . Recovery P lan for t he Ste l ler S e a L ion (Eumetopias jubatus). (ed . Stel ler Sea Lion Recovery T e a m for the Nat ional Mar ine Fisher ies Serv ice) , pp. 92 . Si lver Spr ing , Mary land . N o r b e r g , U. a n d Rayner , J . M. V . (1987) . Ecological morpho logy and f l ight in bats ( M a m m a l i a : Ch i rop tera) : W i n g adapta t ions , f l ight per fo rmance, forag ing s t ra tegy and echo loca t ion . Philosophical Transactions of the Royal Society B 316, 335 -427 . 58 Nowacek, D. P. (2002). Sequential foraging behaviour of bottlenose dolphins, Tursiops truncatus, in Sarasota Bay, Fl. Behaviour 139, 1125-1145. Ponganis, P. J. , Ponganis, E. P., Ponganis, K. V., Kooyman, G. L., Gentry, R. L. and Trillmich, F. (1990). Swimming velocities in otariids. Canadian Journal of Zoology 6S, 2105-2112. Riedman, M. (1990). The Pinnipeds: Seals, Sea Lions, and Walruses. Berkeley, CA: University of California Press. Rosen, D. A. S. and Trites, A. W. (2004). Satiation and compensation for short-term changes in food quality and availability in young Steller sea lions (Eumetopias jubatus). Canadian Journal of Zoology %1, 1061-1069. Schrank, A. J. , Webb, P. W. and Mayberry, S. (1999). How do body and paired-fin positions affect the ability of three teleost fishes to maneuver around bends? Canadian Journal of Zoology 77, 203-210. Sinclair, E. H. and Zeppelin, T. K. (2002). Seasonal and spatial differences in diet in the western stock of Steller sea lions {Eumetopias jubatus). Journal of Mammalogy S3, 973-990. Stelle, L. L. (1997). Drag and energetics of swimming in Steller sea lions (Eumetopias jubatus). In Department of Zoology, pp. 84. Vancouver: University of British Columbia. Stelle, L. L , Blake, R. W. and Trites, A. W. (2000). Hydrodynamic drag in Steller sea lions (Eumetopias jubatus). The Journal of Experimental Biology 203, 1915-1923. Tucker, V. A. (2000). Gliding flight: drag and torque of a hawk and a falcon with straight and turned heads, and a lower value for the parasite drag coefficient. Journal of Experimental Biology 203, 3733-3744. Walker, J. A. (2000). Does a rigid body limit maneuverability? Journal of Experimental Biology 203, 3391-3396. Wardle, C. S. (1975). Limit of fish swimming speed. Nature255, 725-727. Webb, P. W. (1977). Effects of median-fin amputation on fast-start performance of rainbow trout (Salmo gairdneri). Journal of Experimental Biology 6&, 123-135. Webb, P. W. (1978). Fast-start performance and body form in seven species of teleost fish. Journal of Experimental Biology 74, 211-226. Webb, P. W. (1983). Speed acceleration and maneuverability of 2 teleost fishes. Journal of Experimental Biology 102, 115-122. Webb, P. W. (1984). Body form, locomotion and foraging in aquatic vertebrates. American Zoologist24, 107-120. Webb, P. W. (1991). Composition and mechanics of routine swimming of rainbow trout, Oncorhynchus mykiss. Canadian Journal of Fisheries and Aquatic Sciences 48, 583-590. Weihs, D. (1973). The mechanism of rapid starting of slender fish. Biorheology 10, 343-350. 59 Weihs, D. (2002) . Stabi l i ty versus maneuverab i l i ty in aquat ic locomot ion . Integrative and Comparative Biology 42, 127-134. Williams, T. M. and Kooyman, G. L. (1985) . S w i m m i n g per fo rmance and hydrodynamic character is t ics of harbor seals Phoca vitulina. Physiological Zoology58, 576 -589 . Winship, A. J. , Trites, A. W. and Calkins, D. G. (2001) . Growth in body s ize of the Stel ler sea l ion. Journal of Mammalogy S2, 500 -519 .