Behavior of Adult Arnerican Shad (Alosa sapidissinto) Homingto the Connecticut River from Long Island Sound

JurraN J. DonsoNl a.uo lVtrrtau C. Luccr:rr

Department of Biology, tuIcGill University.2 p.6. Box 6070, Montreal 101, Que.ana

Essex Marine Laboratory, Essex, Conn. 06426, USA

DoDSoN, J. J., .rnn W, C. Lpccrrr. 1973. Behavior of adult American shad (Alosa sapi-dissima) homing to the Connecticut River from l-ong Island Sound. J. Fish. Res.Board Can. 30: 1847-1860.

The migratory behavior of America-n shad (Alosa sapidissima) approaching their natalriver during the final saltwater stage of the spawning migration was studied using ultrasonictracking and conventional tagging procedures. trnitial displacement of most sonic-taggedshad released without dispiacement adjacent to and 10 km west of the Connecticut R-iverwas not in the direction of the home river. These fish, however, homed successfully to theConnecticut River as did dart-tagged shad released in the same areas.

Shad exhibited two major behavior patterns; countercurrent orientation in response to thereversing tidal current and adjustment of swimming speed to changes in tidal velocity. Counter-cuffent orientation was equally significant during daylight and darkness, whereas the adjust-ment of swimming speeds to tidal current velocity was more signiflcant during daylight thandarkness.

Shad tracked to the west exhibited a westerly bias inherent in the basic open water behaviorpatterns. Shad exhibited a greater degree of directed movement rvhen oriented against the ebbtide and adjusted their swimming speeds to exceed the ebb tide velocity and to approximatelyequal the flood tide vetrocity. Shad tracked to the east exhibited the same major behaviorpatterns but with the opposite directional bias.

.A hypothesis is presented suggesting that the location of the home river is achieved bymeans of a nonrandom search. Environmental clues indicative of the Connecticut River act toestablish a prefered direction of displacement while the actual unidirectional displacementis achieved by reference to the rate and direction of tidal clrrrents.

DoDsoN, J. J., aNo W. C. LEccrrr. 1,973. Behavior of adult American shad (Atosa sapi-dissima) homing to the Connecticut River from Long Island Sound. J. Fish. Res.Board Can. 30: 1847-1860.

Les auteurs ont 6tudi6 le comportement d'aloses savoureuses (Alosa sapidissima) aumoment oir elles s'approchent de leur rividre natale dans 1a phase f,nale en mer de leur migra-tion de fraie. Ils ont fait appel au d6pistage aux ultrasons et aux marquages ordinaires.

Le d6placement initial de la majoritd des aloses munies de marques sonores et relAch6esd l'endroit de capture d 10 km dL l'ouest de la rividre Connecticut ne s'est pas produit en di-rection de la rividre natale, Ces poissons ont cependant rdussi d revenir d la rividre Connecticut,tout comme des aloses marqu6es avec des dards et relAch6es aux rnOrnes endroits.

Les aloses manifestent deux types principaux de comportement: une orientation d contre-courant en r6ponse aux changements de direction des courants de mar6e et r6glage de la vitessede nage en fonction des changements de vitesse de ce courant. Cette orientation d contre-courant se produit d la lumidre du jour aussi bien qu'iL I'obscurit6, alors que le r6glage de lavitesse natatoire par rai2port d. la vitesse du courant de rnar6e est plus prononcd d la lumidredu jour qu'd I'obscurit6.

Les aloses suivies vers l'ouest montrent un biais marqu6 vers I'ouest qui est inhdrent aucomportement de base en pleine eau. Les aloses ont un mouvement dirig6 de fagon plus prdciselorsqu'elles sont orient6es contre le reflux et elles rdglent leur nage d une vitesse d6passantcelle du reflux et d peu prds 6ga1e d celle du flux. Les aloses suivies d I'est se comportent g6n6rale-ment de Ia m0me fagon mais avec biais en direction oppos6e,

Received March 29, 7973

rPresent address: Departmenf of ZooTogy, University of British Columbia, Vancouver 8, B.C.2Send reprint requests to this address.

Printed in Canada (J2984)

1847

J. F

ish.

Res

. Bd.

Can

. Dow

nloa

ded

from

ww

w.n

rcre

sear

chpr

ess.

com

by

Uni

vers

ité d

e M

ontr

éal o

n 12

/02/

14Fo

r pe

rson

al u

se o

nly.

1848

AurntcrN shad (Alosa sapidissima) undertake longdistance oceanic migrations (Talbot and Sykes1958; Sykes and Talbot 1958) and return to theirhome rivers to spawn (Walburg and Nichols 1967).Leggett and Whitney (1972) demonstrated that thetiming and pattern of these extensive migrationsare regulated by adherence to speciflc water tem-peratures in the range of 13-18 C. Seasonal changesin the 13-18 C isotherm along the Atlantic coastcoincides with the southward migration of shad inthe winter and the northward migration in theearly spring and summer. The northward migra-tion brings shad to the vicinity of their home riversat times when river conditions are approaching theoptimum for spawning.

Tagging studies also indicate that shad return totheir natal streams to spawn (Truitt 1940; Hollisi948; Vladykov 1950; Talbot and Sykes 1958;Nichols 1961). Leggett and Whitney (1972) reportedthat between 1 9 65 and 19 69, 18,37 4 mature shad weremarked and released in the lower ConnecticutRiver. Over 300 shad were recovered in the Con-necticut River one year or more after tagging whilenone were recovered in other rivers.

Although the overall timing, route, and controlmechanism of the shad migration is now established,the means by which shad locate and recognizespecific natal rivers was heretofore unknown. Thispaper describes the behavior of adult shad, nativeto the Connecticut River, during their approach tothe river from Long Island Sound and commentson possible environmental factors controlling thebehavior.

Materials and Methods

The orientation and migratory behavior of shad wasinvestigated by a combination of conventional taggingand ultrasonic tracking techniques. Conventional tagswere used to assess the frequency of return of shad tothe Connecticut River and the time taken to home.Ultrasonic tracking techniques (Trefethen 1956; Stasko,1971) were used to analyze the minute to minute behaviorof shad in response to various environmental variables.

Ur,rn,qsoNrc Tna.crrNo Pnocrounrs

A total of 43 adult American shad fitted with sonictransmitters were tracked in Long Island Sound (Fig, 1)during April, May, and June of 1970 and 1971. The tags(Smith-Root model 69-B) had a frequency of 70 KHZ,a range of 30-300 m, depending on water conditions,and an effective life of 15 days. They were housed inpolystyrene cylinders 6.4 cm long and 1.3 cm in diame-ter; approximate weight was 14.5 g in air. The receivingunit consisted of a portable battery operated receiver(Smith-Root model TA-60) tunable in the range 60-180KHZ coupled to a hydrophone having a peak frequency

JOURNAL FISHERIES RESEARCH BOARD OF CANADA, VOL. 30, NO. 12, PT. 1, 1973

of 70 KHZ and a cone of reception of approximately15". Swimming depth of tracked flsh was unknown.

All fish were captured with 14-cm (stretch measure)monofilament gillnets and removed from the net im-mediately upon capture, as indicated by the movementof the headrope floats. If unharmed, a sonic tag wasinserted into the stomach via the mouth and each fishwas gently returned to the water.

Tagging was accomplished in two areas of Long IslandSound; (a) Westbrook, Connecticut (72'28t, 41"16')and, (b) the west side of the Saybrook Point breakwater(72'20',40",41"15',502) (Fig. 1). A thifd flshing area,located 10 km east of the Connecticut River, was fishedln 1971, No shad were ever captured in this area.

Movements of these tagged fish were monitored byimmediate and continuous tracking from a 6'4-m boatwhich was maintained behind and in line with the taggedfish. This was possible due to the directivity of thehydrophone which was manually rotated in the water.The distance between the fish and the boat was estimatedby changes in incoming signal strength. The position ofthe fish was obtained at least every 30 min by triangula-tion. Whenever possible, cloud cover, wind strength anddirection, wave condition, tidal condition, depth, tem-perature and salinity (measured aI 2-m increments),and fish swim-speed (expressed as boat speed throughthe water) were recorded coincident with fish position.Temperature and salinity were measured with a HydrolabTC-2 conductivity meter. Boat speed through the waterwas measured with a Gurly flowmeter.

Automatic shore monitors were utilized to detect theupriver passage of sonic-tagged shad which had pre-viously been tracked at sea. In 1970, fwe shore-basedmonitors were located along the Connecticut River atpoints 7, 13, 31., 33, and 34 km upstream of the river'smouth. In 1971, two shore-based monitors were located7 km upstream of the river's mouth. Each monitor con-tained two receivers which were connected to twohydrophones placed on the river bottom. The incomingsignali to each receiver were recorded on separate tracksof a magnetic tape recording system. A time signal wasautomatically recorded on a third track every 6 min.The time of fish passage was obtained by adding the timemarks in tenths of hours to the starting time of thetape. Individual fish were identified by the tag pulse,which was determined for each tag and recorded for eachfish at the time of tagging.

The displacement of shad in Long Island Sound re-sulted from two factors: the speed and direction of tidalcurrents, and the heading and swimming speed of theflsh. Data on tidal rate and direction were available fromour records, and from tidal tables and charts for LongIsland Sound. These data were combined and usedtogether with observations of fish displacement over timeto calculate fish bearing and swimming speed usingtriangulation and vector velocity techniques. Swimmingspeeds so calculated are hereafter referred to as "cal-culated swim-speeds." This represents the minimumspeed at which a given fish must swim to travel a straightline between two points in a given increment of time.As such, it tends to underestimate the actual swimmingspeed, since straight line courses were seldom followed.

J. F

ish.

Res

. Bd.

Can

. Dow

nloa

ded

from

ww

w.n

rcre

sear

chpr

ess.

com

by

Uni

vers

ité d

e M

ontr

éal o

n 12

/02/

14Fo

r pe

rson

al u

se o

nly.

DODSON AND LEGGETT: SIIAD HOMING BEIIAVIOR I 849

1",

Ftc. l Location map of Long Island Sound.

Current velocity values presented in the tidal chartsand tables for Long Island Sound were obtained frommeasurements made to a maximum depth of 6 m (U.S.Coast and Geodetic Survey). As the depth of trackedshad could not be measured, a knowledge of differencesin current velocity throughout the water column wasnecessary to assess the accuracy of swimming speedestimates. Riley (1956) reported that at midtide, whencurrent velocities are maximum, the flood (westerlyflowing) tide velocity remains relatively constant through-out the water column, whereas the ebb (easterly flowing)tide velocity is greater on the surface than on the bottom(30 m). Thus, the absolute magnitudes of swimmingspeeds calculated during ebb tide conditions may beoverestimated if the fish were swimming deeper than 6m. As the majority of shad were tracked within the 9-mcontour line, we do not feel that this is a major source oferror. Nevertheless, to avoid incorporating this possiblesource of error, swimming speeds calculated under eachtidal condition were not directly compared. Rather,swimming speeds were examined relative to tidal velocity.

Swimming speeds were also estimated by recordingthe average speed of the boat through the water as theflsh proceeded from one point to the next along the track.These estimates were biased as it was impossible to ac-curately integrate the observations over the entire *-hrperiod. They therefore tend to represent the swim-speedof the flsh at the times of positioning and data collectionrather than an average value between two points. Theseswim-speeds are hereafter referted to as "observedswim-speeds."

CoNvnNrroual TAGGTNG Pnocnounrs

Coincident with the sonic tracking experiments, a totalof 330 shad were externally marked with nylon dart-typetags in 1970 and 1971 and released in the same areas asthe sonic tagged shad. The numbers of fish recaotured

in the Connecticut River by commercial and sportfishermen were analyzed for the frequency of return.

Analysis of dart-tag recoveries in the ConnecticutRiver provided estimates of mean homing times fromWestbrook and Saybrook. Dart-tagged shad were cap-tured by sport and commercial fishermen from Old Say-brook at the river's mouth to Enfield Dam which islocated 137 km upstream. Mean homing times were cal-culated by subtracting the time spent in the river fromthe total time elapsed between the tagging and subsequentrecapture of shad. The time spent in the river was basedon an estimated upstream migration rate of 6.1 km/day(Glebe and Leggett 1973).

Results

P,lrHs or Mrcn.q.:rroN AND HoMTNG Arttrry

Forty-three sonic-tagged shad were continuouslytracked in Long Island Sound during 1970 and1971 for a total tracking time of 613 hr. No shadwere tracked into the Connecticut River. Further-more, 40 of 43 shad exhibited a westerly displace-ment away from the Connecticut River (Fig. 2,3). However, 7 (39.8%) of 18 dart-tagged shadreleased at Westbrook in 1970 were recovered inthe Connecticut River by commercial fishermenduring the same year. This recovery did not differsignificantly from the recovery of 879 (21 .8%)of 4072 dart-tagged shad released in the lowerConnecticut River and subsequently recovered inthe river during 1970 (Leggett 1970) (chi-square :1.78, df : 1 , P>0.5) . S imi lar ly , 58 (18.6/o) of312 dart-tagged shad released in Long Island Soundin 1971 were recovered in the Connecticut River.This recovery did not differ significantly from the

J. F

ish.

Res

. Bd.

Can

. Dow

nloa

ded

from

ww

w.n

rcre

sear

chpr

ess.

com

by

Uni

vers

ité d

e M

ontr

éal o

n 12

/02/

14Fo

r pe

rson

al u

se o

nly.

1850

recovery of 864 (20.8%) of 4147 dart-tagged shadreleased in the lower Connecticut River and subse-quently recovered in the river in 1971 (I-eggett197i ) (ch i -square: .59, df : 1 , P>.05) . Thesefindings indicated that shad tagged at Westbrook

JOURNAL FISHERIES RESEARCH BOARD OF CANADA, VOL. 30, NO' 12, PT' 7) 1973

and Saybrook were homing to the ConnecticutRiver.

In 1970, 4 sonic tags were recovered by commer-cial and sport fishermen in the Connecticut River.The automatic shore-based monitors were plagued

\.\-( ry ,j;rrrrji " ' i .v \ .-ro"* ' 'l , - - .F-. . - , , . ' , i l i " '

\ \ \N(\ ) ,

.-J' ^i"i:." "r. ;-, , .

: 1 . . , . 6

Frc. 2. Migration routes of 10 shad tracked in 1970. These shad were selected

ur fr"i"g chaiacteristic of the other tracks. Solid line represents shoreline, dotted

line replesonts 9 m contour line, arrows represent direction offish displacement.

"1:."*:::

J. F

ish.

Res

. Bd.

Can

. Dow

nloa

ded

from

ww

w.n

rcre

sear

chpr

ess.

com

by

Uni

vers

ité d

e M

ontr

éal o

n 12

/02/

14Fo

r pe

rson

al u

se o

nly.

DODSON AND LEC'GETT: SI{AD HOMING BEHAVIOR 1851

by repeated malfunction in spite of daily inspectionand testing. Interrogation of the magnetic tapesrevealed 11 sonic tags recorded in 1970 and five in1971. 'tape advance malfunctions and time gapsin the monitoring system eliminated the possibilityof identifying individual tags or their time of passage.However, it was possible to discriminate between

diferent signals and thus to determine the mini-mum number of sonic-tagged shad entering theriver.

Daily inspection of the monitors involved checkingthe tape advance mechanism, assessing the sensitivityof the hydrophones, and recording a short messageon the tape for reference at the time of interroga-

Frc. 3. Migration routes of eight shad tracked irl1971. These shad were select-ed as being iharacteristic of the other tracks. Solid line represents shoreline,dotted tnJ represents 9 m contour line, arrows represent direction of fishdisplacement.

J. F

ish.

Res

. Bd.

Can

. Dow

nloa

ded

from

ww

w.n

rcre

sear

chpr

ess.

com

by

Uni

vers

ité d

e M

ontr

éal o

n 12

/02/

14Fo

r pe

rson

al u

se o

nly.

1852

tion. Over 5ATo of these messages were not detectedduring tape interrogation indicating that the moni-tors also suffered malfunctions of the recordingsystem. It was therefore impossible to estimate themonitors' efficiency of tag detection. The frequencyof detection (377o or 1,6 of 43 sonic tags released)is believed to be compatible with the detection effi-ciency of the shore monitors and, therefore, indica-tive of successful homing to the Connecticut River.

The frequency of recovery of dart-tagged shadreleased at Westbrook in l97l (23 of i06) was notsignificantly different from the frequency ofrecoveryof dart-tagged shad released at Saybrook in 1971(35 of 206) (chi-square : .69, df : I, P>.1).The mean homing time for dart-tagged shad re-leased at Westbrook in 1970 and 7977 was 7.9 days(so : 6. 7, n : 25), and for dart-tagged shad releasedat Saybrook was 5.2 days (so : 4.9, n : 20).

These long mean homing times for fish migratingover relatively short distances indicated that thehoming migration in Long Island Sound was ex-tensive and not direct. This conclusion was supportedby observations of the paths of shad tracked ln 1970and 1971 (Fie. 2 and 3).

In 1970, three shad were tracked east fromWestbrook towards the Connecticut River. Oneshad turned west before arriving at the river, whereasthe remaining two reached the river and remainedin the vicinity of its mouth for approximately 2 hrbefore leaving. One was tracked west beyond West-brook while the second migrated southeast (Fig. a).No fish were tracked east in 1971.

RBspor,{sp To TIDE

That portion of the data associated with openwater migration, i.e. not associated with fish move-ments about breakwaters, islands, and similar ob-structions, was extracted from the main body ofdara and analyzed. The lst hr of individual trackswas excluded from the analysis to avoid incorporat-ing nonsense behavior arising from possible tem-porary disorientation due to handling.

The data included 442 observations on 16 shadtracked in 1970 and 680 observations on 20 shadtracked in 1971. Seven shad were excluded from theanalysis as they were either lost shortly after releaseor were tracked in areas where tidal cuffent readingscould not be obtained. Individual shad were analyzedif the shad were tracked through at least one tidalcycle. Twelve shad tracked in 1970 and 13 shadtracked in 1971 were atalyzed individually.

Fish bearing -Long Island Sound is subjectedto a reversing tide every 6 hr. The incoming (flood)tide flows approximately west and the outgoing(ebb) tide flows approximately east.

JOURNAL I'ISHERIES RESEARCH BOARD OF CANADA, VOL. 30, NO. 12, PT. 1, 1973

In 1970 and 1971, shad exhibited a strong ten-dency to orient into the tidal current.

1970, r : -.479, n : 388, F<.001, grouped data1971, r : -.502, n : 432, P<.001, grouped data

This countercurrent orientation was equally signifi-cant during daylight and darkness. However, theprecision of the countercurrent orientation differeddepending on the direction of the tidal current. Thecountercurrent orientation of 9 of 12 shad trackedin l97O was more precise when bearing west againstthe ebb tide (Fig. 5A). The remaining three shadexhibited greater precision when oriented to theeast. Of these three, two (no. 07 and 08) initiallypursued an easterly course towards the ConnecticutRiver. Four shad, although exhibiting precisecountercurrent orientation when bearing west,exhibited concurrent behavior under the influenceof the westerly flowing tide. In L97l, the counter-current orientation of 12 of 13 shad was moreprecise when bearing west into the ebb tide (Fig.5B). The remaining shad exhibited similar precisionduring both tides. Four shad tracked in 1971exhibited concurrent orientation under the influenceof the flood tide. Three other shad approached thissituation.

The differences in precision of countercurrentorientation between the two tidal conditions wasalso evident in the magnitude of angular changesin bearing exhibited by shad. Turning angles weresmallest when shad were bearing west against theebb tide. In 1970, the frequency of angular changesgreater than 45" exhibited by shad tracked duringebb tide conditions was 33 .3% (65 of 195 observa-tions). The corresponding frequency during floodtide conditions was 57 .3/o $2 of 143 observations).This difference is statistically signiflcant (chi-square== 7 .48, P<.01). Similarly in 7977, the frequencyof angular changes greater than 45' obsen'ed duringebb tide conditions was 24.7lo (46 of 185 observa-tions) and during flood tide conditions was 52.87o(104 of 197 observations). Again this difference isstatistically significant (chi-square : 14.05, df :1 , P< .01 ) .

Nineteen of 25 fish tracked in 1970 and 7971 andanalyzed individually exhibited smaller turningangles when bearing west during ebb tide conditions(Fig. 6A-F). Six shad exhibited smaller turningangles when bearing east during flood tide conditions(Fig. 6G-I). Of these shad, three were initiallytracked east to the mouth of the ConnecticutRiver.

Angular changes in bearing exhibited by shadwere not dependent on the time of day or the avail-ability of celestial clues. There was no difference in

J. F

ish.

Res

. Bd.

Can

. Dow

nloa

ded

from

ww

w.n

rcre

sear

chpr

ess.

com

by

Uni

vers

ité d

e M

ontr

éal o

n 12

/02/

14Fo

r pe

rson

al u

se o

nly.

the frequency of large turning angles (>45") ob-served during day and night tracks (Table 1A).

KilomH

ffi

DODSON AND LEGGETT: SI{AD HOMING BEI{AVIOR

Similarly, the frequency of large turning anglesdid not differ between clear and overcast daytime

.{: >

. ',..1

I 853

\

\\ t , /

:J_^

IhJ

\v<,

:J-^

L\$\Y t l z

:J--

1 ' ' ,

Frc. 4. Migration routes of shad initially tracked east to the ConnecticutRiver. A, Fish no. 07. a-b, flood tide, b-c, ebb tide, c-d, flood tide; B, Fishno. 13. a-b, ebb tide, b-c, flood tide; C, Fish no. 08. a-b, flood tide.Solid line represents shoreline, dotted line represents 9 m contour line, arrowsrepresent direction of fish displacement.

J. F

ish.

Res

. Bd.

Can

. Dow

nloa

ded

from

ww

w.n

rcre

sear

chpr

ess.

com

by

Uni

vers

ité d

e M

ontr

éal o

n 12

/02/

14Fo

r pe

rson

al u

se o

nly.

1854

skiesnight

JOURNAL FISIIERIES RESEARCII BOARD OF CANADA, VOL. 30, NO. 12, PT. I, 1973

Fish swimming speed - The mean calculatedswim-speed of shad tracked in 1970 was 75.3 cm/sec(so : 42.7) and in 1971 was 71 .9 cm/sec (so :36.6). The frequency distribution of calculatedswim-speeds for 1970 and 1.971are presented in Fig.7. No diel periodicity was evident in shad swimmingspeeds.

A strong positive correlation was revealed be-tween swim-speed and tidal current velocity in both1970 and 1971. The correlation was more signiflcantduring daylight than darkness (Table 2). Analysis

(Table 1B) or between clear and overcastskies (Table 1C).

of 25 individual shad revealed that eight shad ex-hibited a signiflcant adjustment of swim-speed totidal velocity, five shad approached significanceand 20 shad exhibited a positive relationship be-tween swimming speeds and tidal current velocity.

The adjustment of swim-speed to tidal velocitydiffered according to tidal condition. Linear regres-sion analysis of swim-speed/tidal velocity datarevealed thatin 1970, under ebb tide conditions, D :0.64, and under flood tide conditions, b :0.25'Thus. increment increases in ebb tide velocityproduced greater increases in swim-speed than didsimilar increases in the flood tide velocity. In 1971,linear regression analysis revealed that, under ebb

A

Fra. 5,1970 and

Individual mean fish bearins and tide direction for shad tracked in1971.

B

J. F

ish.

Res

. Bd.

Can

. Dow

nloa

ded

from

ww

w.n

rcre

sear

chpr

ess.

com

by

Uni

vers

ité d

e M

ontr

éal o

n 12

/02/

14Fo

r pe

rson

al u

se o

nly.

DODSON AND LEGGETT: SHAD HOMING BEIIAVIOR 1855

tide conditions, b : 0.45. The regression was notsignificant for flood tide conditions.

In 1970 and 1971, shad swam west at a rate signi-ficantly greater than the ebb tide velocity (l-test,P<.001, grouped data) but under flood tide condi-tions swam at a rate approximately equal to thetidal velocity (Table 3). This relationship wasobserved during daylight and darkness (Table 4).

A sign test (Siegal 1956) applied to the swim-speed/tidal velocity data for individual shad trackedin 1970 and l97l revealed that the majority ofindividual shad (22 of 25) swam at speeds in excessof the current velocity during ebb tide conditions(P<.01). Under flood tide conditions, the propor-tion of shad swimming at speeds in excess of tidatvelocity (11 of 25) did not exceed the proportionswimming at speeds less than tidal velocity (P<.4).

Respor:qsn ro TrupBnq,ruRE AND Slr-tNtrv

Decreased surface salinity and increased surfacewater temperatures in Long Island Sound areindicative of the presence of Connecticut Riverwater. The results of a correlation analysis carried

out on observed swimming speeds, surface salinity,and surface temperature revealed that in 1970and 1971, 13 of 17 shad exhibited a negative rela-tionship between observed swimming speed andsurface salinity. This proportion was signiflcantlygreater than the proportion of shad exhibiting apositive relationship (binomial test (Siegal 1956)P : .025).

In 1970 and 1971,11 of 20 shad exhibited a posi-tive relationship between observed swim-speedand surface temperature. This proportion was notsignificantly different from the proportion of shadexhibiting a negative relationship (binomial test,( S i e e a l 1 9 5 6 ) P : . 4 1 2 ) .

Discussion

The observed westerly displacement of 40 of 43shad captured west of the Connecticut River sug-gested that these fish entered Long Island Soundfrom the east and migrated beyond their home river.Stevenson (1899) stated that the majority of shadentered Long Island Sound through the Race (Fig.1). He further stated that most of these shad migrat-

1

T

,D i---r++?

+- €

-"- -z;-'7 i -'--z'-

i - , i i r , 1T

Frc. 6. Turning angles of six shad (A*F) initially tracked west in 1970 and 1971 and three shad (G-I) initially trackedeast in 1970. e, Fistr no. 02, 1970; B, Fish no. 0k, teZO; C, Fish no. 03, 1970; D, Fish no.72, 1977; E, Fish no' 08,1971 ; F, Fish no. 13, 1971.; G, Fish no. 08, 1970; H, Fish no. O'7,1970; I, Fish no. 13, 19'70. Solid bar on the timeaxis represents duration of ebb tide (not including slack tide).

l r i l i ri i i

i , i i l i lo ; - - - - - - - - - - t - ' . " - - ' -

i l ii 1 t l

I

I

" l

ll-TJ.

Fis

h. R

es. B

d. C

an. D

ownl

oade

d fr

om w

ww

.nrc

rese

arch

pres

s.co

m b

y U

nive

rsité

de

Mon

tréa

l on

12/0

2/14

For

pers

onal

use

onl

y.

1856

ed up the Connecticut River, but a Targe numberproceeded westward to spawn in the HousatonicRiver and other smaller streams along the Con-necticut shore. No shad were taken in the HousatonicRiver in 1960 and only occasional shad have been

Tasr,n 1. Frequency of large turning angles exhibitedby American shad (Alosa sapidissima) during (A) dayand night, (B) clear and cloudy days, and (C) clear andcloudy nights in Long Island Sound.

Year n /6>45o n 70>45' Chi2

JOURNAL FISI{ERIES RESEARCH BOARD OF CANADA, VOL. 30, NO. 12, PT. 1, 1973

(A)

1970(B)

1971

1970(c)

1971

Day

1970 236 43.2

I97I 260 36.2

Clear days

124 45.2

1 3 5 3 8 . 5

Clear nights

6t 40.9

91 43.9

Nisht

1 3 5 4 s . 1 . 0 5

149 42.3 .66

Cloudy days

112 41 . r .159

125 33.6 .319

Cloudy nights

74 48 .6 .301

5 8 3 9 . 7 . 1 1 0

> . 0 5

> . 0 5

> . 0 5

> . 0 5

> . 0 5

> . 0 5

caught there since that time; dams and pollutionhave virtually eliminated this and other runs inrivers and streams located west of the ConnecticutRiver (Walburg and Nichols 1,967). In the light ofthis evidence and the recovery of dart- and sonic-tagged shad in the Connecticut River, it is con-cluded that shad caught west of the river wereConnecticut River shad which had migrated beyondtheir natal river.

Historically, the most successful shad fishery inConnecticut was located west of the ConnecticutRiver along the north shore of Long Island Sound.Three-pound nets located here caught more shadthan all others combined (Walburg and Nichols1967). In contrast, few shad have ever been caughteast of the Connecticut River. A shad populationnative to the Thames River, located to the east ofthe Connecticut River (Fig. 1), was eliminated by1896 (Stevenson 1899). Only a few shad have beencaught in the Thames River since that time (Wal-burg and Nichols 1967). This evidence and ourown failure to capture shad east of the ConnecticutRiver leads to the conclusion that shad, native tothe Connecticut River entering Long Island Soundvia the Race, do not approach thenorthern shorelineat points east of the Connecticut River. Rather,the main body of Connecticut River shad appearsto concentrate along the north shore of Long Island

REoUE

C

. M F I R S E C

Frc, 7. Frequency distribution sf calculated shad swimming speeds, 1970 and 1971,

J. F

ish.

Res

. Bd.

Can

. Dow

nloa

ded

from

ww

w.n

rcre

sear

chpr

ess.

com

by

Uni

vers

ité d

e M

ontr

éal o

n 12

/02/

14Fo

r pe

rson

al u

se o

nly.

DODSON AND LEGGETT: SI{AD HOMING BEHAVIOR 1857

1970

t97l

AllDayNightA11DayNight

416272144470288182

. J I J

.349

.257

. 1 5 5

.206

.083

Sound adjacent to and west of the ConnecticutRiver prior to entry into the river.

Analysis of the behavior of shad tracked in LongIsland Sound revealed a well-defined behaviorpattern highly correlated to tidal conditions. Shadexhibited countercurrent orientation in response tothe reversing tidal current, but the precision anddirectivity of the orientation was greater whenbearing west against the ebb tide than during flood

Taerp 2. Correlation analysis of calculated swim-speedvs. tidal current velocity in Long Island Sound, 1970,l97l.a

Year Condition

tide conditions. Shad adjusted their swimmingspeed to changes in tidal current velocity in amanner which resulted in swimming speeds inexcess of the ebb tide velocity. Under flood tideconditions, swim-speeds were approximately equalto tidal current velocity. The result of this westerlybias inherent in the open water behavior patternwas a westerly displacement.

Analysis of individual shad revealed that duringflood tide conditions, the behavior of shad variedfrom countercurrent orientation to concurrentorientation. Eight of 25 shad exhibited the lattertendency. Such behavior, however, also resulted ina westerly displacement in Long Island Sound.

To successfully home to the Connecticut River,shad tracked to the west must alter their behaviorpattern to affect the opposite directional displace-ment. Three shad tracked to the east exhibited apartiaT alteration ofthe open water behavior pattemdescribed for shad tracked to the west. Fish no. 07-70, 08-70, and 13-70 exhibited a lower frequency oflarge turning angles during flood tide conditionsimplying a greater degree of directed movementwhen bearing east. However, no directional biaswas evident in the adjustment of swimming speedto tidal velocity. Fish no. 07 swam at speeds inexcess of both tidal velocities, whereas fish no. 08and 13 swam at speeds less than both tidal velocities.In spite of the limited number of observations onshad initially tracked east in Long Island Sound, theresults indicated that shad tracked to the east andto the west exhibited the same major behaviorcomponents of countercurrent orientation and swim-speed adjustment to tidal velocity. Only the uni-directional bias inherent in these behavior com-ponents was altered to produce the initial easterlydisplacement. Thus, at least two orientation mecha-nisms are implicated in the migration. One is in-volved with orienting shad to tidal rate and direc-tion, the other deterrnines the directional biasinherent in the open water behavior pattern.

Several sensory modalities could be involved ingoverning the observed rheotaxic response of shad.

P < . 0 1P< .001P < . 0 1P< .01P < . 0 1P > . 0 5

&All figures cm/sec.

Ta.nr.p 3. Mean shad swim'speed and mean tidalvelocity under ebb and flood tide conditions, 1970 and1,971.

Westerly tide Easterly tide

1970

Tide 68.8 30.2 t73Fish 69.1 42.2 173

63.8 29.1 21383.5 43.6 213

58.5 25.6 21480.2 33.9 214

1971

Tide 64 1Fish 65. 5

3 0 . 6 2 l B35.7 218

T.q,er,r 4. Mean tidal velocity and mean shad swim speed during hours of daylight and darkness, 1970 arrdl9'7t.P - level of signiflcance as determined by t-test.

Westelly flowing tide Easterly flowing tide

FishsD vel sD P

Tide Fishn v e l s D v e l s D P n

Tidevel

1970,Day1970, Night1971, Day1971, Night

2 9 . 5 6 8 . 6 4 2 . 43 t . 7 7 0 . 3 4 2 . 13 0 . 6 7 1 . 9 3 6 . 83 0 . 6 5 4 . 9 3 1 . 5

>.05 1.24 62.7>.05 89 65.4>.05 129 57 .6< .05 85 59 .9

84 .9 40 .2 < .018 l . 4 47 .9 < .0182 .3 34 .4 < . 0176 .7 33 .2 < .01

12350

134R1

67 .07 3 . 265.464 .5

28 .030 . 624.727.O

J. F

ish.

Res

. Bd.

Can

. Dow

nloa

ded

from

ww

w.n

rcre

sear

chpr

ess.

com

by

Uni

vers

ité d

e M

ontr

éal o

n 12

/02/

14Fo

r pe

rson

al u

se o

nly.

I 858

The countercurrent behavior observed in speckledtrout (Elson 1939), salmon and eels (Davidson1949), coho and chum salmon fry (Mackinnon andHoar 1953) and cod, whiting, whiting-pout, smelt,and herring (Harden Jones 1963) appears to havebeen achieved through reference to optical stimuli.However, shad tracked in Long Island Soundexhibited countercurrent orientation at night aswell as day and displayed similar frequencies of largeturning angles under both conditions. Riley andSchurr (1959) observed extinction coefficients inEastern Long Island Sound that varied from .39to .82. High extinction coefficients were believedto be due to silt kept in suspension by strong,scouring tidal currents. Therefore, the maintenanceof countercurrent orientation at night by opticalstimuli would be possible only if shad were migratingvery near the surface. Tactile stimuli may have beenutilized if the fish were migrating close to thebottom.

The tendency of shad to orient in an easterlydirection during flood tide conditions appears tobe ill-adaptive for fish affecting a westerly displace-ment. This phenomenon, however, may be a basicbehavior trait resulting from the oceanic migration.The northern migration of shad in the spring fromNorth Carolina and Virginia is coastal (Talbot andSykes 1958; Leggett and Whitney 1972).In inshorewaters along the middle Atlantic states, the prevailingocean current is southerly. This southerly flow ex-tends well across the continental shelf during thespring and may attain velocities of up to 18 km/day(Bumpus and Lauzier 1965). Thus shad migratingnorth in the spring must maintain colrntercurrentorientation at all times to proceed northwards.Therefore, it appears that the behavior exhibitedby shad in Long Island Sound is a continuation ofa more basic behavior pattern with only the uni-directional bias inherent in the behavior beingaltered.

The partial breakdown of the ability of shad toadjust swimming speeds to changes in tidal currentvelocity during darkness suggested that opticalclues may be involved in this response. Such clueswould be removed or reduced during darkness.Shad, however, did not exhibit any speed reductionat night. Thus, shad did not display a diurnalrhythm of activity such as has been observed inchinook salmon in a river system (Johnson 1960)and sockeye salmon in a costal situation (Madisonet al. 1972). Madison et al. observed a departurefrom the diel rhythm of swimming speeds character-istic of sockeye salmon when one fish was trackedinto an inlet situation. They believed this may havebeen due, in part, to the bidirectional flow in theinlet. This is compatible with our observations of

JOURNAL FISTIERIES RESEARCH BOARD OF CANADA, VOL. 30, NO. 12, P"T. 1, 1973

shad migrating in Long Island Sound which alsoexhibits a strong bidirectional flow.

Although the behavior of shad was most obviouslyrelated to tidal current rate and direction, these twovariables in themselves are not indicative of thelocation of the Connecticut River. Therefore. otherenvironmental variables, indicative of the Connecti-cut River, must be responsible for establishing theunidirectional bias inherent in the open waterbehavior pattern of shad. Celestial clues do notappear to play a role as shad behavior remainedrelatively constant throughout periods of clear andcloudy daytime and nighttime skies.

The observed tendency of shad to increase swim-ming speeds in response to lowered salinities sug-gested that water-borne chemical clues may playa role in the location of the natal river. During theshad spawning season of 1970 and 1971,, the Con-necticut River was responsible for 75-787o of thetotal freshwater inflow into Long Island Sound(Water Resources Division, United States GeologicalSurvey, Hartford, Conn.). Furthermore, ConnecticutRiver water has been detected off Montauk Pt.(Riley 1959), well within the proposed oceanicmigration route of shad. Thus, shad migrating inareas of decreased salinity could obtain clues char-acteristic of the natal river. However, the fresh-water plume discharged from the Connecticut Riverbecomes fragmented and spread to the east and tothe west of the river by the strong bidirectionalcuffent. As a result, the response of shad to loweredsalinities cannot be defined in terms of orientationtowards or away from the Connecticut River. Indeed,if shad do use chemical clues to establish a preferreddirection of movement in Long Island Sound, thedispersal and dilution of Connecticut River waterover much of northeastern Long Island Sound maybe in part responsible for the initial migration ofshad beyond their natal river. As shad proceedwest, chemical clues indicative of the ConnecticutRiver would become less evident and foreignchemical clues originating from other rivers wouldbecome more prevalent. This in turn may evoke theopposite directional bias.

Other clues may be responsible for the reversalin unidirectional displacement that must be affectedby shad migrating west of the Connecticut River.The shallow and enclosed western portion of LongIsland Sound is 3-5 C warmer than the eastern end(Riley 1959). West of the Connecticut River, springfreshening begins earlier and the more sluggish cir-culation patterns of the western end cause anapproximate 5%o reduclion in the level of salinityrelative to the eastern end (Riley 1959). The majoreast-west trench systems running from the deepwaters of the eastern end to the shallow waters of

J. F

ish.

Res

. Bd.

Can

. Dow

nloa

ded

from

ww

w.n

rcre

sear

chpr

ess.

com

by

Uni

vers

ité d

e M

ontr

éal o

n 12

/02/

14Fo

r pe

rson

al u

se o

nly.

DODSON AND LEGGETT: SHAD HOMING BEHAVIOR 1859

the western end could also act as topographicalreference points. Any or all of these clues couldprovide shad with orientational or positional infor-mation.

The location of the Connecticut River by shadmigrating in Long Island Sound appears to dependto a large degree on an extensive but nonrandomsearch. This conclusion is based on the long periodof time required by shad to home to the ConnecticutRiver from points relatively close to the rivermouth and the observed westerly displacementaffected by Connecticut River shad already westof the river. We hypothesize that shad migratingalong the eastern coast of North America in thevicinity of Long Island Sound detect clues indicativeof the Connecticut River that establish a unidirec-tional bias resulting in a westerly movement towardsthe Connecticut River. If the river is not locatedon the first leg of the migration and shad proceedwest of the river, some stimuli, presently unknown,evoke the opposite unidirectional bias resultingin an easterly displacement towards the ConnecticutRiver. Under both conditions, the open waterbehavior pattern continues to be most directlyoriented by the rate and direction of the tidal cur-rent, while the clues responsible for evoking theunidirectional bias cease to be orienting clues oncethe bias is established. This sequence may be repeatedseveral times before the river is located.

The culmination of the behavior sequence exhibit-ed by shad occurs on entry into the river. The resultsof this study have shown that proximity to theConnecticut River does not ensure entry into theriver and, therefore, other factors must be involved.A temporal restriction related to the shad's stateof receptiveness to certain environmental stimulimay be imposed upon the behavioral sequencedescribed.

Acknowledgments

D. Merriman, L. M. Thorpe, R. R. Whitney, D.Pratt, and M. Carriker reviewed the manuscript. A.Dolan, J. Farrow, D. Friendly, B. Grayton, R, Galvin,B. McKercher, C. Henault, and B. Medford aided inthe netting, tagging, and tracking of shad. R. A. Jonesprovided valuable suggestions and field and technicalsupport. Funding was provided by the U.S. NationalMarine Fisheries Service (P.L. 89-304), the ConnecticutRiver Ecological Study, the Connecticut Departrnent ofEnvironmental Protection, and the National ResearchCouncil of Canada (Grant No. A6513). J. Dodson wasalso supported by a NRC Postgraduate Scholarship.

Buurus, D. F., axo L. M. Lauzrrn. 1965. Surfacecirculation on the Continental shelf off eastern NorthAmerica between Newfoundland and Florida. Amer.Geogr. Soc. Ser, Atlas Mar. Environ. Folio 7.

Da.vrosoN, V. M. 1949. Salmon and eel movement inconstant circular current. J. Fish. Res. Board Can.7 :432-448.

Er,soN, P. H. 1939. Effects of current on the movementof speckled trout. J. Fish. Res. Board Can. 4:491-499.

Gr,psr, B. D., .qND W. C. LnccBrr. 1973. Study ofweight loss in American shad during the freshwatermigration. U.S. Nat. Mar. Fish. Serv. Rep. Contrib.AFC-8-1: 61 p.

H.tnorN JoNrs, F. R. 1963. The reaction of fish tomoving backgrounds. J. Exp. Biol. 40: 437-446.

Hor.r,rs, E. H. 1948. The homing tendency of shad.Science 108: 332-333.

JosNsoN, J. 1960. Sonic tracking of adult salmon atBonneville Dam,7957. U.S. Fish Wildl. Serv. Fish.Bu1l.60: 471-485.

LEGGETT, W. C. 1970. Population dynamics andmigratory behavior of American shad (Alosa sapi-dissima), p. 2l-31. In D. Merriman [ed.] TheConnecticut River ecological study llth semi-annualProgress report to the Connecticut Department ofEnvironmental Protection.

1971. Population dynamics and migratory behaviorof American shad (Alosa sapidissima) p. 50-53. InD. Merriman [ed.] Ttre Connecticut River EcologicalStudy 13th semiannual Progress Report to theConnecticut Department of Environmental Protec-t1on.

Lnccprr, W. C., nlu R. R. WnrrNrv 1972, Watertemperatures and the migrations of American shad.U.S. Fish Wildl. Serv. Fish. 8u11. 70: 659-67O.

Mrcrrxxow, D., nru W. S. Ho.qn. 1953. Responsesof coho and chum salmon fry to current. J. Fish.Res. Board Can. l0: 523-538.

Ma,orsou, D. M., R. M. HonnAll, A. B. S:rasro, .aruA. D. Hlsr,rn. 1972. Migratory movement ofadult Sockeye salmon (Oncorhynchus nerka) tn coastalBritish Columbia as revealed by ultrasonic tracking.J. Fish. Res. Board Can.29; 1025-1033.

Nrcnor.s, P. R. 1961. Homing tendency of the Ameri-can shad in the York River, Virginia. ChesapeakeSci. 1(3-4): 200201.

Rrr,nv, G. A. 1956, Oceanography of Long IslandSound, 1952-1954. II. Physical Oceanography. Bull.Bingham Oceanogr. Collect. Yale Univ. 15: 1546.

1959. Oceanography of Long Island Sound. Bull.Bingham Oceanogr. Cotrlect. Yale Univ. 17: 9-30.

Ru,rv, G. A.,.a.No H. M. Scnunn 1959. Transparencyof Long Island Sound water, Bull. Bingham Ocean-ogr. Col1ect. Yale Univ. 17: 66-82.

Srrcrr,, S. 1.956. Nonparametric statistics. McGraw-Hill Book Co., Toronto, Ont. 312 p.

Sra.sro, A. B. 1971. Review of fie1d studies on fishorientation. Ann. N.Y. Acad. Sci. 1.88: 12-29.

SrBvrNsoN. C. H. 1899. The shad fisheries of theAtlantic coast of the United States. U.S. Comm.Fish Fish., Part 24, Rep. Comm. 1898: 101-269.

Svrns, J. E., aNo G. B. Tareor. 1958. Progress inAtlantic coast shad investigations.. . migration. Proc.Gulf Caribbean Fish. Inst. llth Annu. Sess. p. 82-90.

J. F

ish.

Res

. Bd.

Can

. Dow

nloa

ded

from

ww

w.n

rcre

sear

chpr

ess.

com

by

Uni

vers

ité d

e M

ontr

éal o

n 12

/02/

14Fo

r pe

rson

al u

se o

nly.

1 860

TALBor, G. 8., aNo J. E. Svrns. 1958. Atlantic coast. migrations of American shad. U.S. Fish Wildl.

Serv. Fish. 8u11. 58: 473490.TnmnrmN, F. S. 1956. Sonic equipment for tracking

individual fish. U.S. Fish Wildl. Serv. Spec. Sci.Rep. Fish. 179: 1l p.

Tnurrr, R. V. 1940. Report of, the Chesapeake Bio-logical Laboratory. l8th Annu. Rep. Conselv. Dep.Md. p. 38-45.

JOURNAL FISHERIES RESEARCH BOARD OF CANADA, VOL. 30, NO. 12, PT. I, 1973

Vr,aovrov, V. D. 1950. Movements of Quebec shad(Alosa sapidissima) as demonstrated by tagging.Natur. Can. (Que.) 77: 121-135.

WALBURG, C. H., aNp P. R. Nrcnors 1967, Biologyand management of the American shad and status ofthe flsheries, Atlantic coast of the United States, 1960.U.S. Fish Wildl. Serv. Spec. Sci. Rep. Fish. 550:105 p.

J. F

ish.

Res

. Bd.

Can

. Dow

nloa

ded

from

ww

w.n

rcre

sear

chpr

ess.

com

by

Uni

vers

ité d

e M

ontr

éal o

n 12

/02/

14Fo

r pe

rson

al u

se o

nly.