Xcx::X:>R~IDAB

.( \. , . •

DISTRIBUI'IOO AND ABUNDANCE OF FLOISAM, LARVAL FISH AND JUVBNILB FISH OFF

BARBAim WITII PAilfiCULAR REFEimNCB 'IO 'llm EXcx::x:>R~IDAB

Mario Ronmel T. Lao

Institute of Oceanography,

( ~JcGill University, Nontreal

• April, 1989

A thesis submitted to the Facu1 ty of Graduate Studies and Research in partial fulfillment of the requirements for the degree of Master of Science.

composition and seasonal abundance of flotsam was surveyed by neuston

tows at stations 3 nmi, 6 nmi and 9 nmi offshore of Barbados. A total • of 431 tows were conducted between October, 1987 and September, 1988 • ThE~ flotsam was of coastal marine and terrestrial origin. Only 38% of

tows contained flotsam, and only 1% of tows produced flotsam with

flyingfish eggs. Flotsam was most abundant between March and September,

whtm water reaching Barbados originates primarily from the South

American mainland. This period does not correspond with the spawning

season of the commercially exploited flyin.gfish, Hirundichthys affinis,

near Barbados ( December-May) •

Fish larvae of 34 and 24 families were collected in day and night

neuston tows respectively, and larval abundance was higher by day til.al1

night. The larvae ranged from oceanic families (myctophids and

istiophorids), to offshore families (hemiramphids and exocoetids) to

coastal families (mugilids and mullids). Hemiramphids ( 46% of day

catch) and myctophids (51% of night catch) dominated the catches. In

both day and night samples, larvae were n~st abundant between February

and Jw1e. When families were considered separately, larvae were either

most abundant between February and June or least abundant between

February and June. Families in the first category included the

myctophids, hemiramphids 1 exocoetids and dactylopterids; families in

the second category included the istiophorids, scombrids, carangids and

mullids. Seasonal variation in abundance of exocoetid larvae (February

to Jw1e) corresponds well with the seasona.lity of spawning in !J.. affiuis

(December to 1'1ay} . For all larvae combined, and separately for the myctophid.s,

{1emiramphids, dactylopterids, exocoetids and carangids (in night samples), larval abundance differed significantly between Stations 1, 2 and 3 ( 3, 6, and 9 nmi offshore) • Larvae were less abundant at Station

1 than at Station 2 and/or Station 3. The high larval abundance offshore may indicate that larvae are aggregated and retained in eddies downcurrent of the island. For 8 of the 10 coomon families collected, larval size varied significantly between stations. Larvae were largest and most abundant at Station 3 and smallest and least abundant at

Station 1. This is consistent with the suggestion that larvae are retained in downcurrent eddy systems; larval retention leading to both higher larval abundance and larger size through growth.

A total of 2,204 juvenile flyingfish were captured in the 49 nightlight trips made during the 1-year study. Juveniles of

Parexocoetus brachyPterus, Exocoetus voli tans, Hirundichthys affinis and

Oxyporhamphus micropterus dominated the catch. Adults of !:· brachyPterus and Q. micropterus were also caught, suggesting that the sampling procedure is less size-selective for these species. This is the first record of juvenile ~. voli tans and Q. micropterus off

Barbados. In "E· brachypterus, ~. volitans and Q. micropterus, juveniles appeared in February - March, and remained abundant through summer

(t-larch - July for ~· brachYPterus, February - August for !f. volitans,

February - July for Q. mibropterus). Juveniles of H· affinis appeared in

December-January, and abtmdance increased from February to August. The results suggest that all life history stages of H. affinis are sequentially present year-round near Barbados. This does not support the hypothesis of large-scale migration of !!· !!ifl.!lis from and towards the island. Puisqu'on croit que les poissons volants fraient sur des surfaces

flottantes, la composition et 1 'abondance saisonniere des objets

flottants ont ~u; mesure par des filet de neuston a des stations

situes a 3, 6, et 9 mille nautiques au large de la Barbade. En tout,

431 echantillons, utilisant un filet de neuston, ont ete ramasses entre

octobre, 1987 et septembre, 1988. Les objets flottants etait soit

d'origine te:rreatre ou coti~re. Seulement 38% des echantillons

contenait des objets flottanta et seulement 1% des echantillons

conte1mit des objets flottants avec des oeufs de poissons volants.

L'abo~tce des objets flottants etait le plus eleve entt~ les mois de

mars et septenilire, quand l'eau atteignant la ~·bade est principalement

d'origine Sud-Americaine. Cette periode ne correspond pas avec la saison

de fraie de l'espece de poisson volant exploite commercialement,

Hirundichthys affinis pres de 'la Barbade (decembre - mai).

Les larves de 34 et 24 families de poissons ont ete ramassees

respectivement1 dans des filet de neuston de jour et de nuit, et l'abondance des larves etait plus haut le jOU1' que la nuit. Les larves

representaient des families oceaniques (Myctophidae, et Istiophoridae), des familles au large (Hemiramphidae et Exocoetidae), ainsi que des

familles cotiere (Mugilidae et Mullidae). Hemiramphidae (46% de

l'eclmntillon de jour) et Myctophidae (51% de l'echantillon de nuit) dominaient les echantillons. Panni les echantillons de jour ainsi que

CeUX de nuit, les larves etaient les plus nombt~UX entre fevrier et

juin. Quand les familles etaient analysees separemment, les larves etaient soil les plus nombreux entre fevrier et juin ou les moins nombreux entre fevrier et juin. Les familles dans la premiere categorie comptaient en.tre-eux Myctophidae, Hemiramphidae, Exocoetidae, et

Dactylopteridae; et parmi les families de la deuxieme cat~gorie se

trouvaient Istiophoridae, Scombridae, Carangidae, et Mullidae. La

variation saisonniere dans 1 'abondance des larves dans 1 'Exocoetidae (de

fevrier a juin) correspond bien. avec la saisonalite de reproduction dans

!!· affinis (de decembre ~ mai).

Pour 1 'ensemble des 1arves et aussi pour Myctophidae,

Hemiramphidae, Dacty1opi:eridae, et Exocoetidae consideres seuls et pour

Carangidae dans 1es echanti11ons de nui t, 1 'abondance de 1arves demontrait une difference significative entre 1es stations 1, 2, et 3

(3, 6, et 9 milles nautiques au large de la rive). Les larves etaient

mains nombreux a la station 1 qu'a la station 2 et/ou 3. L' abondance

eleve au large peut indiquer que les larves sont groupes et retenues

dans des contre-courants en aval de 1 •isle. Pour 8 des 10 families

communes echantillolmees, la taille des larves variait significativement

entre les stations. Les larves ;;;taient les plus grandes et 1es plus

nombreuses a la station 3, et les plus petites et mains nombreuses a la

station 1. Cela est consistante avec la suggestion que les larves sont

retenues dans des systernes de contre-courant; la retention de larve

resultant en des abondances de larve elevees ainsi qu, une plus grande

taille,suivant la croissance.

// En tout, 2204 juveniles de poissons volants ont ete captures en les

49 sorties de nui t (en utilisant des lumiE~res) pendant la periode d'etude d'un an. Les juveniles de Parexocoetus brachYPterus, Exocoetus voli tans, Hirw1diohthys affinis, et Oxyporhwupl!IJS__!!!ieropterus dominaient

la prise. Les adultes de ~· P_!J:lPhypterus et Q. J!!_.i,cropt~rus ant aussi et~ captures, qui suggere que la methode d'~haiitillonna.ge est rnoins selective selon la taille pour ces especes. Ceci est la premiere documentation de juveniles de ~· volitans et de Q. micropterus pres de la Barbade. Pour f. brachyPterus, ~· volitans, et Q. micropterus, les juveniles apparaissaient en fevrier-mars et restaient abondantes durant l'ete (mars - juillet pour f. brachyPterus, fevrier - ao'Ut pour ~. volitans, fevrier - juillet pour Q. micropterus). Lea juveniles de H· affinis apparaissaient en decembre-janvier, et 1 'abondance montai t de fevrier a aout. Les resultats suggerent que toutes lea ~tapes de cycle vital de H· affinis sont trouvees en sequence le long de l'ann~e pres de la Barbade. Cela ne soutenir pas 1: hypothese de migration de grand.e echelle de H· affinis de et vera l'!sle. The completion of this thesis would not have been possible witl1out

the supervision and support of my adviser, Dr. Wayne Hunte. His comments and suggestions on earlier versions of the thesis are

gratefully acknowledged. His moJral support when Murphy's Law repeatedly

struck is equally appreciated. Dr. Hazel Oxenford made sure that the

thesis was finished before I left Barbados. I thank her for her

COIIIIlents on early versions, pa.t:ience during the most trying times and

help in production of the final version. Dr. Robin l'tahon provided valuable comments on several sections of the thesis and helped in the

survey of the most suitable neuston gear to be used. He also helped in

resolving problems with computer software. Dr. Jolm Lewis gave helpful advice and encouragement during the early stages of the study and his

~sional visits to Bellairs were always eagerly anticipated. He made my stay in Montreal enjoyable. From start to finish, these people have always provided their time and energy in resolving unexpected problems and shared their knowledge and expertise in numerous discussions.

I would like to thank Mr. Somkiat Khokiattiwong for his help dLn·ing

the first few months of sampling 1 particularly when I was absent on a tagging trip for a week. His companionship during the entire study and the discussions we had together were extremely valuable. The fishermen who accompanied me in the field surveys, Emerson and Gabby, eased some of the difficult work at sea. Mr. Victor Small helped in procuring the materials for the neuston gear and in its construction. He also helped in constructing a boom for the original neuston gear. I thank him for his help and friendship. Students who were in Bellairs during the final stages of this study helped in one way or another in providing encouragement, giving me unlimited access to computers, BID in making my departure memorable. I would like to thank in particular Karen, for making my stay in Barbados enjoyable, Carrie for checking the references, Fred for doing some of the figures, Trudi for typing some of the thesis BID Kathy for the French translaticn of the abstract.

In Montreal, I would like to thank Mr. D. Briere and Ms. C.

Cazabon, CIDA coordinators, for their assistance and help in making the necessary arrw-.gement.s when the study failed to finish on time. In the

Philippines, I would like to thank my family for their forebearance when

I became too busy to write them. Also, the understandil-.g and help of

the Chancellor and staff of the University of the Philippines in the

Visayas are greatly appreciated, despite the problems in COillllWlication.

This study was funded ~Y a scholarship within the ASEAN-CIDA

Cooperative Program for Marine Science Bllii by Dr. W. Hunte through his

NSERC Operating Grant A0264. It was a component of a larger program entitled the Eastern Caribbean Flying fish Project funded by the

International Developnent Research Center. The thesis is dedicated to the memory of the late Dr. Elvira Tau who encouraged me to take the scholarship B11d who remained as an inspiration throughout ti1e course of ti1e study. STATBMHNT OF QUGINALITY

This is the first investigation of seasonal variation in abundance of flotsam off Barbados, hence the first to conment on the role of flotsam in influencing spawning seasonality and spawning Sl..K:Cess of flyingfish. It is the first investigation of the composition and seasonal variation in abundance of fish larvae collected by neuston nets off Barbados and the first full-year sttdy of juvenile flyingfish in nightlight samples. By simultaneously investigating the seasonal abundance of all life stages of flyingfish off Barbados, the thesis presents a more comprehensive picture of flying fish distribution and population dynamics than previously possible. Abstract ••• ••••••••••••••••••••••••••• 0 •••••••••••••••••••••••••• .ii

~s~ •••••••••••••••••••••••••••••••••••••••••••••••••••••••••••• i v

Acknowledgements •••••••••• ...... vii Statement of Originality...... ix

List of Figures •• ••••••••••••••••••••••••••••••••••••••••••, • 4- •• .xiii

List of Tables •••.••.••...... •.••...... •.•••••...... •••.. xvi

List of ~rrlices ...... x:x

1. I~IOO ••••••••••••.•••••••.•••••••••••••••••••••.•••••••• 1

1.1. Distribution and abundance of flyingfish •••••• jj •••••••• 1 1.1.1. Adults •••• ...... • •••••••• 1 1.1.2. Juveniles. • • 4 1.1.3. Larvae ••••• • • 5

1. 2. Reproduction and Spawning Beh.avior of Flyingfish .•••••••• 7

1. 3. SJlEl'Wilir.g Su'bs trates • • • • • • • • • • • • • • • • • • • • • • • • • • • . • • • • • •••• 10

1.4. Food and Feeding Rythyms of Flyingfish •••••••••••••••••• l4

1.5. Oceanic Environmental Characteristics in the easter11 Cari booa.ll • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • ••••••• 15

1.6. Economic Importance of Flyingfish in the eastern Caribooarl •••••••••••.••••••••••••••••••••••••••• 20

1. 7. Objectives of the Study .•..•...... ••••••...••• * •••••••• 21

2. ~lA.~IAI..S AND r1EI1-IOI>S •••••••••••••••••••••••••••••••••• , •••••• 23

2.1. Definition of Terms •• • .•• 23 2.1.1. Neuston ••• ... 23 2.1.2. Flotsam •• .23 2.1.3. Flyingfish .••••••• .24 2.1.4 • Larvae and juveniles •. .24

• 2.2. S8.1Dplitli Statiotl.S ...... 24

2.3. Neuston and Flotsam C~llection. • ••••••••••• 25 2.3.1. Neuston samplers ..•..... •• 25 2.3.2. Description of the neuston sampler .. ..27 2.3.3. Operat 1 ~-,n and handl i m! . .30 2.3.4. SmnpJ ing schedule ...... 32 2.3.5. Treatment of srunples ... . • •. 33 2.3.6. Visual estimation of flotsam. .35

2.4. Nightl ight Sampling .••••...... •....••.••..•..•.•.•.•• 35 2.4.1. Equipment and procedure ••••••••••••••••••••••••• 35 2.4.2. Smnpli!lg sche

2.6. Measurement of ~1vironmental Characteristics •••••••••••• 38

2.6. I>a.ta Arlalysis ••••••••••••••••••••••••••••••••••••••••••• 38

3. ~'I'S ••••••••••••••••••••••••••••••••••••••••••••••••••••••• 40

3.1. Occurrence and Abundance of Flotsam ••••••••••••••••••••• 40 3.1.1. Variation between replicate tows •••••••••••••••• 40 3. 1. 2. Variation between day and night tows •••••••••••• 40 3.1.3. Seasonal va.riation •••••••••••••••••••••••••••••• 43 3.1.4. Spatial variation •.••••••••••••••••••••••••••••• 43 3.1.6. Composition of flotsam •••••••••••••••••••••••••• 43 3.1.6. Flotsam utilization as spawning substrates •••••• 67

3.2. Abundance, Composition and Size of Neustonic Fish l.arv"ae ••••••••••••••••••••••••••••••••••••••••••••• 57

3.2.1. Identification of larvae •••••••••••••••••••••••• 57 3.2.1.1. Hemi r8Jll:pll i da.e • • • • • • • • • • • • • • • • • • • • • • • • • 59 3.2.1.2. Mtlllidae ••...•.••••••••••••••••••••••• 59 3.2.1.3. Dactylopterida.e ••••••••••••••••••••••• 59 3.2.1.4. Ex"

3.2.2. Composition of neuston collection •••••••.•••••.• 62

3.2.3. Variation between sampling replicates ••••••••••• 62

3.2.4. Variation between day and night samples ••••••••• 65

3.2.5. Seasonal variation in abw~ce ••••••••••••••••• 73 3.2.5.1. I>a.y sa.rnples ••••••••••••••••••••••••••• 73 3.2.5.2. Night S8111ples ••••••••••••••••••••••••• 78

3.2.6. Seasonal variation in size ••••••••••••••••.•.••• 82 3.2.6.1. I>a.y sa.rnples ••••••••••••••••••••••••••• 82 3.2.6.2. Nigltt S8lllples ••••••••••••••••••••••••• 85

3.2.7. Spatial variation in abundance •••••••••••••••••• 85 3. 2. 7. 1 . I>a.y sBIIIples ••••••••••••••••••••••••••• 85 3 • 2. 7 • 2 • Night samples ..••••.•••••••..••••••••• 90

3.2.8. Spatial variation in size ••.....•....••••..•••.• 95 3.2.8.1. Day samples ....• • •••••••• 9 5 3.2.8.2. Night samples ... • •••••••• 95

3.3. Species Composition, Abundance and Size of Juvenile Flyingfish in Nightlight Samples •••••••••••••••••••••• lOO

3.3.1. Catch as an index of abundance ••••••••••••••••• lOO 3.3.2. Depletion effects during sampling •••••••••••••• lOO 3.3.3. Taxonomic composition of catch ••••••••••••••••• 102 3.3.4. Size composition of catch •.•••••••••••••••••••• 106 3.3.5. Seasonal variation in abundance •••••••••••••••• 109 3.3.6. Seasonal variation in size ••••••••••••.•••••••• 112

4. DISa.JSS!()I\f ••••••••••••••••••••••••••••••••••••••••••••••••••• 121

4.1. Composition and Abundance of Flotsam ••••••••••••••••••• 121

4. 2. Composition, Abundance and Size of Neustonic Fish I..a.rv-ae •••••••••••••••••••••••••••••••••••••••••••• 124

4.2.1. Differences between day and night larvae ••••••• 124 4.2.2. Seasonal variation in abundance •••••••••••••••• 126 4.2.3. Spatial variation in abundance and size •••••••• l28

4.3. Composition, Abundance and Size of Juvenile Flyingfish in Nightlight Samples •••••.•.•••••••••••••••••••••••••• 130

4.3.1. Gear selectivity and catch composition ••••••••• l30 4.3.2. Seasonal variation in ahln~ and size ••••••• 132 4.3.3. Seasonality of Hirundichthys affinis adults near Ba.r1:::8d.os • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • 13 4

5. LI~'11JRE CI'I'ED ••••••••••.••••••••••••••••••••••••••••••••••• 136

Ap~rrlix 1 • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • . • • • • • • • • • • • • • 14 7 LIST OF FIGl.JRE:I

Figure 1 Location of Barbados and of the neuston and nightlighting stations, including the land marks used to determine position ••••.••••••••••••••• 26

Figure 2 a Profile of the modified Sameoto wld Jarozsynski neuston gear used in the study •••••••• 28

b Dimensions of the sampling gear showing mouth opening, upper fin, lower bracket, and

1:xdy ••• Ill •••••••••••••••••••••••••••••••••••••••••••• 28

Figure 3 Towing arrangement of the sampling gear, illustrating the position of the safety line .••••••• 31

Figure 4 An illustration of the procedure used during nightlightiug for flyingfish, showing light positions, effective fishing area and i lllllni.nated zone • . . . . . • • . • ...... • . . . . • . • • • • • • . • • . • • • 3 7

Figure 5 a Box-m1isker plot of the qllWltity of flotsam collected in replicate tows during day trips .•.••... 42

b Box-whisker plot of the quw1tity of flotsam collected in day and night tows •••••••.•••.••••••••• 42

Figure 6 a Seasonal variation in the frequency of occurrence of flotsam expressed as percentage of the number of tows in a month · with flots811l ...... ~~ ...... 44

b Seasonal variation in the qUWltity of flotsam, expressed as monthly wet-weight/tow (gJTl) •••••••••••••••••••••••••••••••••••••••••••••••• 44

Figut'e 7 Seasonal variation in the frequency of occurrence of the different flotsam components: (a) Thalassia testudinum, (b) SyritJgodium filiforme, (c) tar, (d) pine needles, wld (e) .. others" •••.•.•••.••••••.•..••••. 4 9

Figure 8 Seasonal variation in the quw1ti ty of the different flotsam components collected: (a) Thalassia testudinum, (b) Syringodium filiforme, (c) tar, (d) pine needles, and (e)

ft 0 tl1ers U • • t t a I • I I t • t t I t t t I J t I I t t I a I t I t I t I t I t I I t o1 • I t I 52

Figure 9 a Box-whisker plot of the number of the larvae collected in replicate toHs (day samples) •...... 67

b Box-whisker plot of tl1e number of larvae collected in day and night tows ...... •.. £)7 Figure 10 Length frequency histograms for tl1e most CORIDOn families of fish larvae in the neuston tows (night.. samples): (a) All larvae combined, (b) Myctophidae, (c) Leptocephalus larvae, (d) Scombridae, (e) Exocoetidae and (f) -ca.~idae •••••••••••••...•••.••••••••••••••••• 71

Figure 11 Length frequency histograms for the most CORIDOll families of fish larvae caught in the neuston tows (day samples): (a) All larvae combined, (b) Hemiramphidae, (c) Mullidae, (d) Dactylopteridae, (e) Exocoetidae, (f) Mugilidae, (g) Carangidae, (h) Istiophoridae and ( i) Scombridae •••••••••••••••••••••••••••••••• 72

Figure 12 Seasonal variation in abundance of the most c00111.0n families of fish larvae caught in the neuston tows (day samples): (a) total day larvae, (b) Hemiramphidae, (c) Mullidae, (d) Dactylopteridae, (e) Exocoetidae, (f) Mugilidae, (g) Carangidae, (h) Istiophoridae and ( i ) Scombridae ••••••••.•••••••••••••••••••••••• 7 4

Figure 13 Seasonal variation in abundance of the most common families of fish lat~ caught in the neuston tows (night samples): (a) total night larvae, (b) Myctophidae, (c) Leptocephalus larvae, (d) Scombridae, (e) Exocoetidae and (f) Carangidae ••••••••••.•.•••••••••••••••••••••••• 79

Figure 14 Monthly length frequency distributions of all fish larvae caught in the neuston tows (day s8lllples) •••••••••••••••••••••••••••••••••••••••••••• 83

Figure 15 Seasonal variation in size of all fish larvae combined and of the most cOOJllOn families of fish larvae caught in the neuston tows (day samples): (a) total day larvae, (b) Hemiramphidae, (c) Mullidae, (d) Dactylopteridae, (e) Exocoetidae, (f) Mugilidae, (g) ~idae, (h) Istiophoridae md. ( i ) Scanbridae ...... 84

Figure 16 Montl1ly length frequency distributions of all fish larvae caught in the neuston tows (night BBillples) •••••••••••••••••••••••••••••••••••••••••••• 86

Figure 17 Seasonal variation in size of the most common families of fish larvae caught in the neuston tows (night samples): (a) total night larvae, (b) Myctophidae, (c) Leptocephalus larvae, (d) Scombridae, (e) Exocoetidae and (f) Cai""Wlgid.ae ••••••••.•••...••••.••••••••••••••••. 87 Figure 18 a The percentage of fish missed in a sampling trip versus the index of available fish (sum of captures and misses for that trip) •••••••••••••• lOl

b The number of captures in a sampling trip versus the index of available fish (sum of captures a~xi misses) for that trip •••••••••.••••••• lOl

Figw:-e 19 Length frequency distributions (fork length in cm) for the most conmon flyingfish species caught in the nightlight sampling program (Octber, 1987-September, 1988): (a) Parexocoetus brachYPterus, (b) Exocoetus volitans, (c) Hirundichthys affinis and (d) ~rhamphus micropterus ...... ••••...... t08

Figure 20 Length frequency distributions (fork length in cm) of Parexocoetus brachypterus presented at bi-monthly intervals ••••••••••••••••••.••••••••• 114

Figure 21 Length frequency distributions (fork length in cm) of Exocoetus volit811S presented at bi-monthly intervals ...... 115

Figure 22 Length frequency distributiotlS (fork length in cm) of Hirundichthys affinis presented at bi-monthly intervals •.•••••••••••••...••.••..••. 116

Figure 23 Length frequency distributions (fork length in cm) of ()xyp:>t1hamphus micropterus presented at bi-monthly intervals •••.•.••.••....•••.•••..•.•• 117

Figure 24 Bi-monthly variation in median sizes (fork length in cm) of the most cOIIIDOn flyingfish species caught in the nightlight sampling program: (a) Parexocoetus brachYPterus, (b) Exocoetus voli tailS, (c) Hirundichthys affinis and (d) Ox:yporharnphus micropterus. Data points are median sizes for juvenile and ad.ul t cohot... ts ...... 119 LIST OF TABLES

Table 1 Geographic distribution of some conmon flyingfish BJ>ee'ies • •••..•....•••••.•....••••••••••.•• 2

Table 2 Summary of neuston tow scltedule ••••••••••••••••••••• 34

Table 3 Frequency of occurrence and abundance of flotsam in trips and tows (wet weight in gra.JDS) •••••••••••••••••••••••••••••••••••••••••••••• 41

Table 4 Frequency of occurrence of flotsam (%of tows and % of trips) at Stations 1, 2 and 3 over the entire sampling period (October, 1987 - Septetnber, 1988) ...••..•.....•.••...... •.••...... 45

Table 5 Quantity of flotsam collected (wet weight in gm) at Stations 1, 2 and 3 over the entire sampling period (October, 1987 - September, 1988) ...... 46

Table 6 Abundance of the components of the flotsam (wet weight in gm) collected over the entire sampling period (October, 1987 - September, 1988) ...... •...... 48

Table 7 Results of Chi -square tests on the frequency of occurrence of individual flotsam components betloieen months (October, 1987 SepteJDber, 1988) ...... •.. 60

Table 8 Spearman rank correlation coefficients between the monthly variation in frequency of occurrence of the different flotsam c001ponents ...... 51

Table 9 Results of Kruskal-Wallis tests on the abundance of inclividual flotsam components between months (October, 1987 - September, 1988) ...... •...... •.•••.. 53

Table 10 Spea~1 rank correlation coefficients between the monthly variation in abw1da.nce of the different flotsam components •••.•••••••••••••••• 54

Table 11 Chi-squa.re tests on the frequency of occurrence of the different flotsam components between sampling stations .••.•••••...... • 55

Table 12 Kruskal-Wallis tests on the abundance (wet weight in grams) of the different flotsam components between sampling stations .....•••...... •. 56

Table 13 Utilization of flotsam as spawning substrate ..•..... 58 Table 14 Taxonomic comp:>si tion and relative abundance of larval and juvenile fish collected during the neuston sampling program •••••••••••••••••••••••• 63

Table 15 Taxonomic composition and relative abundance of larval types within the Exoooetidae ( flyillltfishes) ...... 64

Table 16 Results of Wilcoxon rank tests for differences in number of larvae caught between replicate tows for tl1e five numerically dominant families (families with > 200 larvae collected), and for all larvae combined (day samples ) •••••••••••••••••••••••••••••• 66

Table 17 Taxonomic composition and relative abundance of larval and juvenile fish caught in day tows of the neuston sampling program •••••••••••••••• 68

Table 18 Taxonomic composition and relative abundance of larval and juvenile fish caught in night tows of the neuston sampling progi'8Jil. • • • • • • • • ...... 6 9

Table 19 Results of Ki~kal Wallis tests for variation in number of larvae between months in day samples • ...... •...•...... •..•... 7 5

Table 20 Spearman rank correlations coefficients of the monthly variation in larval abundance (day samples) between families ••••••••••••••••••••.• 76

Table 21 Results of Kruskal Wallis tests for variation in number of larvae between months in night smnples ...... 80

Table 22 Spearman rank correlations coefficients of the monthly variation in larval abundance (night samples) between families ••••••••••••••.••••• 81

Table 23 Total number of larvae, and m.unbers of tl1e more cormton families, presented separately for Stations 1, 2 and 3 (day samples) ••••.•••••••••• 88

Table 24 Results of Kruskal Wallis tests for variation in m.unber of larvae between stations (day samples) ...... 89

Table 25 Mann Whi tney tests for variation in number of larvae between station pairs (day samples) •..••..•. 91

Table 26 Total m.nnber of larvae, and numbers of the more conlmon families, presented separately for Stations 1, 2 and 3 (night samples) .•.....•..... 92 Table 27 Results of Kruskal Wallis tests for variation in m.unber of larvae between stations (night S811lples) .••...... ••...... •••.•.••.••.••• 9 3

Table 28 f.1ann Whitney tests for variation in m.unber of larvae between stations pairs, Station 1: 3 nmi offshore, Station 2: 6 nmi, Station 3, 9 1mti (night smnples) ...... 94

Table 29 Results of Kruskal Wallis tests for variation in size of larvae between Stations 1, 2 and 3 (night S811lples ) ...•.•...••...••..•••.••••••..••.••.. 96

Table 30 Median larval size (1!1D) of the COIIIJlOn families in day samples at Stations 1, 2 and 3, and results of Mann Whitney tests for comparing larval sizes between stations •••••••••••• 97

Table 31 Results of Kruskal Wallis tests for variation in size of larvae between stations (day srunples ) ...... •...... 98

Table 32 Median larval size (mm) of the conmon families in night samples at Stations 1, 2 and 3, ru1d results of Mann Whitney tests for comparing larval sizes between stations ••••••••••.• 99

Table 33 The number of capture attempts, captures and misses in 10-min intervals, presented separately for. each hour of the sampling trips. Results of ali-square tests for variation between 10-min intervals in captures and misses are presented •...•••••••••••.•• l03

Table 34 The m.nnber of capture attempts, captures and misses presented at hourly intervals for all sampling trips. Results of Chi-square tests for variation between hourly intervals in captures ru1d misses a1~ presented •..•••.••••.•••.•• 104

Table 35 The relative abtmdru1ce of flyingfish species, including the hemiramphid, Qxyporhamphus micropterus, caught in the 49 sampling trips during the one-year sampling period •.•••••••••••••• l05

Table 36 The relative abundance of non-flyingfish species caught in the 49 sampling trips made during the one-year sampling period ..••.•...•••.... 107

Table 37 The number of flyingfish caught in each month of the one-year nightlighting study, presented separately for juveniles and adults of the four conunon speci~s ...... •.•...... 110 Table 38 Results of Kruskal Wall is tests on variation in the number of fish ca:ught per trip between months (October, 1987- September, 1988); presented separately for juveniles wld adults of the four CODIDOn species in the nigbtligbt ca'tch •••••.•••..••••.....••..•••.•••.•.•..••••.•••• 111

Table 39 Spearman rwlk correlation coefficients of the monthly variation in abundance of the most cOII'IIlOn species of juvenile flyingfish caught in the nigh.tlighting program •••••••••••••••••••••• l13 LIST OF APPBNDicmJ

Appendix 1 Mean monthly values for all environmental variables monitored during the study •••.••••••••• 147 1 • 1. DISTRI.BlJI'ION AND ABUNDANCE OF FLYINGFISH

1. 1. 1. Adults

Most species of flyi.ngfish OOt..."'U.l' in the Indian Ocean (Ka.rame.n,

1980) , the Pacific Ocean (Kovalevska.ya, 1978) , and the Atlantic Ocean between 4~ and 14~ (Fischer et. al., 1981). The range of temperatm--e over which flyingfish bave been encountered is 17-28. 8°C, but they are

more c011100n between 21.0 and 27 .5°C (Nesterov and Grudtsev, 1981) o The distribution li.mi. ts of the individual species appear to be primarily determined by the sea surface temperature (Bruun, 1935; Breder, 1938).

In the tropical Atlantic, flyingfish abundance also corresponds with areas of high productivity (Nesterov and Grudtsev, 1980), such as coastal upwellin.gs (e.g. east of the Verde Islands and south eastern

Africa.) and

Flyingfish are also relatively abundant in areas influenced by river

discharge (e.g. the Amazon; Nesterov and Grudtsev, 1981) o Abundance of flyingfish is low in waters of anticyclonic circulation, where subsidence OL.---curs.

The geographical distributions of the more common flyingfish species are sUIIIlla.rized in Table 1. Hirundichthys affinis, a neri tic species, occurs in the tropical and subtropical Atlantic, including the

Caribbean Sea and Gulf of Mexico (Fischer, 1978; Bruun, 1935; Breder,

1938) • Bruun { 1935) suggests that this species is limited by the 25°C north-south isotherm. H· ~:[finis is mainly exploited off lhe eastern

Cari hbean island.'''>, primarily by Barbados (Mahon ~1- ~J., J YH6). Recent work ,.;ugg.c:sts that there is one panmictic stock, with a sl1i ft:u"J.g center Table 1 Geograp1ic distribution of some common flyingfish species.

Species Distribution 1

CJpaelurus coaatus2 A (Southern Gulf of Mexico, Northern Caribbean Seal

CJPsehna CJ&nopterus A {Northeaatern 8oul.h Alerica, Caribbeu Sea, Gulf of Mnico) 3 C1pselurua heterurus A (Western Atlantic)

CJpselurua aelaourua3 A (Gulf of Mexico, Caribbean Sea, Mortaeastern Atlantic, Bra,il)

CJPselurua exiliens4 A{Gulf of Kedco bu,t absent in Caribbean Seal

Cnsel1nus furcatl&s A (Western Morth Atlantic, Gulf of Keiico)

CJpselurus pinnatibarbatus5 A, P, [

Cypselurus ailleri A (Western Africa!

CJpselurus ni&ricans4 A (Western Africa)

Exocoetus volitans A (but present in Gulf of Kerico), P, [

Exocoetus obtusirostris A, P,

HirundichthJs affinis A

Rirundichthys speculiger A (but rare in the Caribbean Sea, Gulf of Mexico) 6 Hirundichtbys rondoletti A (Gulf of Kexico, North Atlantic, Western Africa)

Hirundicbthys coroaandelenais

Parexocoetus brach1pterus5 A, P,

Parerocoetus aento atlanticus A {off Mauritania! Prognichtkys gibbifrons A (West Africa, Western Atlantic)

Fodiator acutus A, p

References: Bruun 11935), Breder {1938), Brown ll94Z), Munro (1954), Hall (1955), Vijayaraghavan (191lj, Fischer (1978), Fischer et al. (1981), loulevskaya (198~), Storey (1983).

1A: Atlantic Ocean, P: Pacific Ocean, I: Indian Ocean. ;cypselurug is called Cbeilopogon by so1e authors. considered to be tbe saae by so1e authors. ,May4 be si•ilar. ~Sub~epecies 1ay be present. 6uirundichthys rondoletti is called ~anichthJs rondoletti by so1e authors. of abundance, in th.e eastern Caribbean (Oxenford, 1988a).

Exocoetus volitans and E· obtusirostris are oceanic species, occurring in the tropical and subtropical Atlantic, but the former is absent from the northwestern Caribbean and the Gulf of Mexico (Fischer,

1978). Reported temperature ranges for ti1e species are 20-29°C and.

17 .6-24.2°C respectively (Gru:ltsev et aL, 1987). Both also occur in the Pacific and. Indian Oceans (Parin and Gorbtmova., 1964; Kova.levskaya,

1978; Fischer, 1978) and in the Mediterranean Sea (Bruun, 1935; Breder,

1938}: These species are not cOillllercially exploited ( Fischer, 1978) •

Parexocoetus brachypterus is colllll.On in the Caribbean Sea and. the

Lesser Antillea.n Islands ( Fisc:her, 1978) , but less abundant in the Gulf of 1'1exico, Gulf Stream and the western Sargasso Sea. The temperature range of !:· brachypterus has been estimated by Sa.uska.n ( 1973, cited in

Neste1~v and Gt~tsev, 1981) to be between 24.8 and 28.5°C. Lewis (1959,

1961) found large schools of ~· bra.chypterus hilianus in Barbados a few hundred meters from shore. The tendency of ~. brachypterus to occm· both offshore and nearshore has been observed in Madras, India where unusually high catches of P. brachyyterus brachypterus were obtained by reach seines in February, 1969 (Ra.o and Basheeruddin, 1973).

Post-larval catches show a strictly tropical distribution for

Oxyporh.amphus micropterus, between latitudes 15°N and 15°8 in the eastern Atlantic and between 20~ and 20°S in the western Atlantic

(John, 1983) . John (1983) placed the lower temperature limit of this species at 23°C, rather than 24°C as suggested by Bruun ( 1935). The small number of samp] es obta.int~ by Bruun ( 1935) and Breder ( 1938) was not suffic:it~t1L to c]assif:v thi:::; Spt::lJies as either nf:riti,· en ucl~a.nic.

.)" 'Ibe ranges of Atlantic Cypselurus species are given by Staiger

( 1965) • Members of the genus occur in both tropical and temperate

Atlantic, Pacific and Indian Oceans, but they occur at different

distances offshore. Q. exilien.s and Q. furcatus occur in mid-ocean

while Q. cyauopterus, Q. comatus, Q. heterurus and Q. pimatiba.rbatus

are fowld close to land (Le. within 400 miles) •

The vertical distribution of flyingfish adults has been studied by

Nesterov and Ba.zanov ( 1986) • Underwater observations made at night

showed that 85.7% of the fish remained at the surface ( 0-2 m) while

only 6.3% occurred at depths greater than 4 m. During the day, it was

estimated that flyingfish were concentrated within the 0-3 m layer, with only about 5% being deeper. These data agree well with the estimate

made by Zuyev and Nikol' skiy ( 1980) that 70% of flyingfish stay in the

0-2 m layer. During spawning, CYJ?Selurus opistophus hiraii is found. in

the 0-1 m depth during the day before going deeper (0-2.5 m) at night

(Shiokawa, 1969).

1.1.2. Juveniles

John (1983) determined the ranges and mean abundance of some beloniform fry in the Atlantic from neuston tows done between 1967 and

1982. Q. micropterus was strictly tropical in distribution, whereas juveniles of most of tlte exocoetid taxa (~. volitans, ~· obtusirostris,

Cypselurus .!ll?• , Prosnichthys gibbifrons, Da.nichthys rondeleletti and

Hirwldichthys .§1?•) extended over the warmest part of the adult ranges and shared polewa.rd boundaries. The ma.in determinants of distribution were surface water temperature and current direction. The distributional patterns agreed well with those of the adults (Nestet'OV and Grudtsev, 1980) . As is the case for adults, larvae and juveniles of Q. micropterus and E· volitans are concentrated in regions of high productivity (Kovalevska.ya, 1978).

Little is known about growth of juvenile flyingfish. Breder (1938) described juveniles of H· affinis rmlgin.g from 2.7 to 8.5 cm. Lewis et al. ( 1962) collected specimens of H· affinis within the size range 1.2 and 15.2 cm. Monthly changes in length frequency of the samples collected by Lewis et al. ( 1962) indicated that, within a six-month period (April to September), growth progressed from 2-4 cm to ·10-15 cm.

Lewis (1961) caught juveniles of P. brachypterus in the 2-3 cm size range in April, 1960, and noted that size had increased to 8-10 cm by

August.

1. 1. 3. Larvae

Richards (1984) conducted two neuston surveys in the Caribbean; the first from July to August, 1972 and the second from February to March,

1973. The most abwldant families in the first survey, ranked in terms of total number of larvae, were: Myctophidae, Exocoetidae, eel leptocephali, Hemiramphidae and Da.ctylopteridae. In terms of frequency of occurrence, the ranking was: Exocoetidae, Carangidae, Balistidae,

Coryphaenidae and Myctophida.e (tied with Pomada.syida.e) • In the second survey, mullids ranked first in terms of number, followed by mugilids, exocoetids, scombrids and myctophids. However, exocoetids again occurred .most frequently, followed by carangids, coryphaenids, myotophids and mullids. The frequent occurrence of the exocoetids is expected as the larvae and juveniles of this group titay mainly within the 0-0.5 m dept~ layer sampled by the neuston net (Nesterov and

Bazanov, 1986) • Fahay (1975) caught 15 species of flyingfish larvae and juveniles off the South Atlantic Bight, with £. bra.chyPterus, £. gibbifrons and Q. heterurus being the most abundant. Only !: . brachYPterus larvae showed a marked seasona.li ty, with the smaller incli vidua.ls becoming more ab.mdan.t between July and October.

Powles ( 1975) conducted a. comprehensive plankton survey off

Barbados but found very few flyingfish larvae (13 larvae, 0.04% of total catch), probably because the bongo sampler he used is not effective for catching neustonic larvae. Powles classified his samples in terms of habitat (inshore or offshore) and taxonomic grouping (higher or lower) •

He considered lower fishes to include the Isospondyli, Iniomi and the order Anguilliformes. All others were classified as higher. The offshore higher and lower fishes were dominated by the families

Myctophidae and Bregma.cerotidae, respectively. nle inshore group was dominated by clupeids and gobiids in the lower and higher groups, respectively. Overall, there were more oceanic than inshore fishes in his samples.

Larval flyingfish are usually more common in the surface waters during the day than at night (Parin, 1967; Parin ~t al., 1972). It is hypothesized that this QL,"'Curs because the motor activity that retains the larvae near the surface is related to their feeding rhythm. Since foraging is visually cued and is effective only during the day, the motor activity weakens or ceases at night causing the larvae to sink out of the near-surface layer. This diurnal pattern is typical of holoepi]Jelagic fishes (0x")'porbamphidae, Exocoetidae, Coryphaenidae,

Tstiophoridae, I~~l~p~U!'Y§ .§1>· and Xighi8§. gladius) . 1. 2. REPRODUCI'ION AND SPAWNING BRHAVIOR OF FLYINGFISH

Most flyingfish eggs have little or no oil globule for buoyancy, but have filaments for attachment to substrates. TI1ey are therefore

considered demersal. The filamer)ts vary interspecifically in length,

mnnber and position in the egg capsule (Collette et al., 1984).

The eggs of !!· affinis bear long (about 4 cm) filaments on one pole of the egg and short ones on the other (Evans, 1961). The longer

filaments hold the eggs in a single mass to at.tach them to substrates

( Evans, 1961) • TI1e function of the short filaments is unknown. Species

in the genus .Exocoetus .§~ill· (e.g. ~. volitans, E· obtusirostris, ~· monocirrhus) produce buoyant eggs without filamentous tendrils (Bruun,

1935; Breder, 1938; Kovalevskaya, 1964; Parin and Gorbunova, 1964).

Cheilopogon nigricans and members of the genus Prognichthys have

planktonic eggs with shortened filaments (Parin and Gorbunova, 1964;

Kovalevskaya, 1982), and the same is true of eggs of Oxyporhamphus

micropterus similis (Breder, 1938; Bruun, 1935). Tilese differences in egg structure, particularly with respect to filaments, indicate divergences in the process of adaptation to life in the open sea

(Kovalevskaya, 1982).

Flyingfish have low fecundity compared to most other fish groups.

Intermittent spawners like Parexocoetus mento and species of Exocoetus and Prognichthys produce between 400 ai1d 1,100 eggs (Parin and

Gorbunova, 1964; Kovalevskaya, 1964, 1965 in Kovalevskaya, 1982). _e. b1~hypte1~ produces between 2,100 a11d 2,600 eggs (Kovalevskaya, 1967, cited in Kovalevslmya, 1982; Imai, 1959). Other species may spawn up to

20,000 eggs (e.g. Cheilopogon heterurus, Okachi, 1959, cited in

Kovalevskaya, 1982; Qheilopogon pinnatibarba.tus japonicus, Imai, 1959; and. Cheilopogon cyanopterus, Parin and Gorbunova, 1964) •

The fecwldity of H· affinis has been estimated at between 4,100 wld

9,200 (Hall, 1955) wxi between 5,700 and 7,100 (Monte, 1965). A more precise estimate was attempted by Storey ( 1983) who fowld 811 average nbatch fecwldity" (the number of mature eggs present in the female ovary at any one time) of 7, 084. He presented the following linear relationship between fecwldity and fish length:

Batch fecundity = 2 x (41.18 x stwxiard length (Rill) - 498·4.68)

Storey ( 1983) also noted that the average gonosomatic index of females

(11.5) was higher than males (6.5) and that there were four distinct size groups of oocytes in the ovaries. He interpreted this as evidence that females spawn fow: times in their lifetime. Hall (1955) fowld only three size classes of eggs in females of H· affinis.

As is typical of most tx:opical fish, flyingfish species spawn throughout the year but with distinct spawning peaks. The aeasonali ty of spawning varies somewhat with the spawning range of the species.

Kovalevskaya ( 1982) classified the spawning ranges of flyingfiah as oceanic-tropical (Exocoetus volitans, g. monocirrhus, Prognichthys sea.lei, Hirwmchthys speculiser and H· albimaculatus), oceanic subtropical (!!. rondoletti) w1d neritic-ocew1ic (Cheilopogon !'m• - Ch. nigricans group). H· speculiger spawns year round, but at higher latitudes spawning occurs primarily during the autumn-winter period.

Similarly, Exocoetus volitwlS spawns year rowld in the Atlantic (Gulf of

Guinea), with peak spawning occurring between October and April

(Grudtsev et al., 1987). By contrast, most spawning of ~· volitans in the Pacific ~ea.n occurs during the wanner periods of the year. E. obtusirostris shares the same reproductive areas as E. voli twlB, but being more eurythermic, may spa'Wil more actively in the U}Melling areas off northwesten1 Africa and the higher latitudes of the Caribbean Sea

where it is found to be abundant (Grudtsev et al. 1 1987). Ji. rondoletti 1 which spa'WilS during the winter-spring period, moves towards the southern warmest part of its range to_ spawn and migrates back to the cooler northern areas afterwa.I'rls. Cheilopogon .ID?• (Ch. nigricans group) lives in open waters but moves towards the coast to spawn. Spawning occurs primarily during the warm months of the year.

H· affinis spawns in large schools (Monte, 1965) and no pairing activity has been observed (Hall, 1955). Off Barbados, the species spawns throughout the year but there are seasonal peaks ( Lewis et al. ,

1962; Storey, 1983). One spawning peak was identified by Hall ( 1955) as being between April and mid-May. In a more extensive study, Lewis et al. ( 1962) reported the spawning peak to run from March to July. Storey

( 1983) found a minor peak in December and a major one in May. A minimum spawning area covering a large zone east of Barbados has been proposed by Hall ( 1955). Data on spawning in !:!· affinis also exist in areas outside Barbados. Brutm (1935) found ripe females in samples taken from the tropical Atlantic between November and February. In northeastern Brazil, the spawning period for !!· affinis is between May and August (Almeida, 1966). H· affinis appears in Tobago between

November and July with spawning peaks occurring in February and March

(Jordan, 1983) •

The spawning period for .E. brachyPterus in Barbados apperu·s to be between September and January (Lewis, 1961). This is slightly earlier thru1 the spawning period for the srune species in Puert.o Rico, which is reported to be from December to April (Erdma.n, 1976). Off t.he Florida. Keys, ~· mesogaster spawns from mid-May to mid-July (Breder, 1938). No data are available on th,e spawning p:::riod of Qxyporhamphus micropterus in the Atlantic.

A long incubation period is typical of most beloniform fish, including .flyingfish {Kovalevskaya, 1982). It ra:nges from 4-5 days for

!!· affinis ( Evans, 1962), 5-6 days for H. coromandelensis

(Vijayaraghavan, 1973} to 9-10 days for ;m. voli tans (Parin and

Gorbunova, 1964).

1. 3. SPAWNING SUBSTRAT.ES

Flyingfish can apparently use a wide range of substrates on which to spawn. Bred.er ( 1938) illustrated a patch of Sargassum weed full of flyingfish eggs which he believed were from the genus Hirtmdicthyg.

Munro (1954) studied the early life stages of Hirundichthyg speculiger from eggs he found attached to an empty bottle floating several miles off a port in Australia. In Naples, flyingfish eggs have been found attached to floating straws used for making hats (Gill, 1903-1904).

Kovalevskaya (1982) includes driftwood, fragments of algae, pieces of wood, bird feathers,· coconuts, drift nets and even some pleustonic orgru1isms as possible spawning substrates for oce&lic flyingfish of the genera Cheilopogon and Hirundichthyg. Near-shore spawning flying fish

(e.g. the genera Fodiator, Parexocoetus, including certain members of the sub-family Cypselurinae) shed their eggs on coastal algae. Off

Barbados, Hirundichthyg affinis has been observed to spawn on floating weed (Sargassum §R.) and brushwood (Brown, 1942; Hall, 1955).

Vijayaraghavan ( 1973) found !!· coromandelensis egg.s attached to

Sargassum shoots off the Coromandel coast in India. Several studies suggest an association between flyingfish and Sar&assl..DJl weeds. Lewis et

. al. ( 1962) found fragments of Sargassum inside the stomachs of some

specimens of J!. affinis in Barbados. Dooley (1972) found young and

adult CyPselurus heterul'US and a few (less than 20) specimens of

Parexocoetus brachypterus, Exocoetus obtusirostris and Hirundichthys

affinis to be associated with the sa.rsassum coumuni ty in the Florida

Current. g. heterurus is known to use the Sa.rgassl..DJl weeds as spawning

substrate while its young use the weeds for shelter (Breder, 1938).

Cypselurus furcatus and Prognichthys gibbifrons are also known to be

associated with Sa.rgassl..DJl weeds (Breder, 1938; Bruun, 1935).

The materials used for making fish attraction devices (FADs) in tl1e

H· affinis fishery in the eastern Caribbean have proven to be good

spawning substrates. Dried coconut branches are used in Grenada ( Steele

and Oxenford., 1986 ) , while banana leaves are used in St. Lucia ( Wal ters

and Oxenfo1~, 1986). In Barbados, fishermen use either sugar cane trash

or coconut branches (Hal~ing, 1986). A combination of hibiscus flowers

and banana leaves are used in Martinique (Guillou and Oxenford, 1986),

while either banana leaf trash or coconut bran.ches are used in Dominica

(Darroux and Oxenford, 1986). Egg deposition on drifting gill nets is

also a common occurrence. However, it is not known whether eggs

deposited on the FADs or on the gill nets are viable. It is also not

clear how much of the effectiveness of the FADs in attracting and

congregating flyingfish is due to their potential as spawning

substrates, as opposed to their attraction as feeding i terns, since tl1e

FADs are usually used in association with various forms of bait (e.g.

macerated fish and oils).

Hall ( 1955) made a qualitative survey of Sargassu_!l! .§12· off

1 1 Barbados to dete1111ine whether variation in its abundance, or in the presence of eggs on the weeds, were correlated with the flyingfish catch. In neither case were the relationships significant. However,

Hall noted that the weeds were not encountered frequently during his sampling period (February to March, 1953) and when present, were either scattered rin small patches or aggregated in riffles. Taylor (1960,

1969) noted that the weeds near Barbados were primarily Sar&assum flui tans, a:nd secondarily §. natans. No other studies are available on the amount or types of materials that can be found floating or beached in Barbado~:~, although the amount of beached materials in the Caribbean may be considerable. For example, a single mass of floating weeds found washed ashore in the port of Santiago, Cube. between June and July, 1957 had a. dry weight of 45 kg. (Michanek, 1975) .

Considerably more information exists on Sa.rgassun !ill• in the North

Atlantic. Recent studies on their abundance will be reviewed since it may directly or indirectly influence the availability of spawning substrates for flyingfish in the eastern Caribbean. Based on the anticyclonic circulation of the North Atlantic sub-tropical gyre, the movement of Sargassum in the Gulf Stream is from the West Indies,

Caribbean Sea and Gulf of Mexico to the North Atlantic (Stoner and.

Greening, 1984). However differences in weed morphology (J?arr, 1937) and associated macrofauna (Fine, 1970; Stoner and Greening, 1984) indicate that the weed carried out by the Gulf Stream and that present in the Sargasso Sea are distinct from each other. Parr ( 1937) believed that the Gulf Stream weed is carried northward where the cold temperature causes it to die, and that the Sargasso Sea populations are self-sustaining. Stoner and Greening (1984) commented that, while

Parr's hypothesis was hard to confirm, their studies tended to corroborate. it, as the rnacrofaunal conmt.mity of the Gulf Stream weed had lower species richness, more large species and less variable species composition than the OOIIIIll..Dlity in the Sargasso Sea. Moreover, there was little influx of littoral fauna from the West lndies through the Gulf

Stream and to the Sargasso Sea, as would be expected if weeds frequently drifted between these areas (Stoner and Greening, 1984).

Between 1933 and 1935, Parr (1937) conducted the first quantitative survey of Sargassum weeds in the western Sargasso Sea, Caribbean Sea and the Gulf of Mexico. He found that the weeds were very scarce in the

Caribbean and Cayman Seas, while the weeds present in the Gulf of Mexico were mostly deteriorating and heavily encrusted. Weeds were most abundant in the Sargasso Sea; the estimated quantity being 2 to 5.5 metric tons (wet weight) per square nautical mile or 529 to 1454 mg m-2 •

Recent surveys (1977 to 1981) by Stoner (1983) indicated a substantially lower weed bioma.ss. Abundance in the Sargasso Sea dropped 6% below that of Parr's estimates. 'l11e only exception was in the Gulf Stream, where mean values increased from 210 to 280 mg m-2 • However, Butler et al.

( 1983) reviewed the same data and concluded that no significant decrease in bioma.ss had occurred, except for an. area northeast of the Antilles

(20-25°N, 62-68~} where measurements made by Stoner were 0.1% of

Parr's estimates. This difference was ascribed to a seasonal or long- term shift of the currents defining the southwestern boundary of the

Sargasso Sea. An examination of Stoner's results in stations close to

Barbados (14° 06'N, 58° 34'W; 13° 10'N, 59° 44'W; and 12° 42'N, 60°

40'W) show zero abundance of Sargassum §.!!· Mean values for the 2 Caribbean Sea were estimated to be 18+34 nJg m- as compared to Parr's

28±28 mg m- 2 • . . . The contradictory results on the presence or absence of ~-gassum off Barbados are difficylt to explain. Sargassum mortality may be caused by increasing levels of marine pollution (Stoner, 1983) , \ fragmentation through wave action and subsequent sinking of encrusted parts, loss of buoyancy due to diseases (Johnson and Richardson, 1977) and entrainment in the downwelling zone of Langmuir cells (Woodcock,

1950).

1. 4. FOOD AND FEEDING RHYTHMS OF FLYINGFISH

Hirnndicht.hys affinis feeds mainly on zooplank:ton. da Cruz and

Araujo (1971) suggest that the diet consists of fish (53%), arthropods

(14%), molluscs (12%), algal fragments (1%) and other unidentified materials (24%). Fish larvae are a consistent if not a dominant

CompOnent, and planktonic crustaceans are often COiml011 (Hall, 1955;

Lewis et al., 1962; Barroso, 1967). Hall ( 1955) estimated. that H· affinis is capable of ingesting fish prey in the size range 4-12 cm.

Barroso (1967) suggested that feeding activity decreases in H· affinis during the spa.wning period but this may simply result from less time available for feeding (Hall, 1955).

Parexocoetus brachrnterus is mainly a zooplankton feeder. The diet is predominated by · copepods such as Undinula vulgaris 1 Candacia pachydactyla and Corycaeus obtusus (Lewis, 1961). Gorelova (1980) studied the diets ·of young flyingfish belonging to the genera Exocoetus,

Prognichthys, Hirnndichthyg, Cheilopogon, and C;ypselurus as well as that of the small-wing flyingfish, Oxyporhamphus micropterus. 'lhe diets consisted primarily of zooplankton, with no substantial differences betwee:n species. Gorelova attributed the slight differences in types of plankton ingested to local availability rather than to preference by the juveniles. Such similar! ties are consistent with the observation that the morphology of the digestive system of most flyingfish is almost

Wliform, differences occurring only in the structure of pharynaeal bones, teeth and the liver (KhachatUI'Ov, 1983).

In contrast with the adults which feE!d at night (Hall, 1965; Lewis et al. , 1962) , juvenile and larval flyingfish feed during the day

(Gorelova, 1980; Parin, 1967).

1. 5. OCEANIC ~AL C!IARACI'ERISTICS IN THE EASTERN CARIBBEAN

'll1e physico-chemical characteristics of the waters off Barbados are influenced by tl1ree major processes; large scale movements of equatorial currents, local formation of eddies due to an "island mass effect" and intrusion of river water from the South American mainland.

Surface waters of the Caribbean Sea are derived from the westward flowing North and South Equatorial Currents (Parr, 1937; Sverdrup et al., 1942). TI1e relative contribution of these water masses may vary between years (Parr, 1937) but water of North Atlantic origin predominates (Sverdrup et al., 1942). TI1is water, which is typically of low temperature and high salinity, and is low in phosphates, has been detected arolUld Barbados between September and February ( Lewis et al. ,

1962). Between March and August, South Atlantic water, which contains higher phosphate and plankton concentrations, mixes with fresh water

from the rivers on the north coast of South America (Lewis et al. 1 1962) ruld enters the Caribberu1 east of the Lesser Antilles as the Guiana

Current. Although its precise influence on Barbados is not yet clear

(Metcalf, 1968; Atwood, 1976; Nahon, 1986), Borstad (1982a, 1982b) has shown that meanders of this current do influence surface waters off the island.

The seasonal variation in surface salinity values at Barbados has been well studied (e.a., Lewis et al, 1962; Steven, 1971; Sander, 1971;

Steven and Brooks, 1972; Kidd snd Ssnder, 1979) • Durina winter, mean surface salinity is usually higher (36.4 o/oo) than in s\..lllller when it drops to about 33.4 o/oo (Kidd and Ssnder, 1979). This decline in surface sa.lini ty has been attributed to the entry of the freshwater lenses from the rivers of South America. The period of maximum discharge from the Amazon snd Orinoco rivera is from June to July and August to

September, respectively. The timing of Amazon run-off corresponds with the gradual decline in salinity at Barbados to a minimum in July' (Mahon,

1986) . The timing of the decline is consistent with the results of a. recent study by Muller-Ka.rger et al. ( 1988) which revealed that from

February to May, during the ~ea.kening of both the North Equatorial countercurrent and retroflection of the North Brazil Current, Amazon water flows northwestward towards the Caribbean Sea. Borstad (1982b) estimated that intera1mual variation in the Amazon River discharge accounts for 64% of the rumual variation in minimlDll surface salinity at

Barbados. Froelich et al. ( 1978) estimated that 60% of the freshwater entering the Caribbean originates from the Amazon ruld Orinoco Rivers ruld

40% comes from rainwater. Borstad ( 1979) believed that the effect of the discharge from the Orinoco River (Gade, 1961; Meade et al., 1983), may be masked by the much larger discharge of the Amazon (Meade et al.,

1979). The sizes of freshwater lenses from the Amazon have been estimated by Ryther et !!1· ( 1967) to be 200 to 300 miles in diameter.

Aside from their low salinity, the lenses are also rich in silicates but low in nitrates ruld phosphates (Ryther et al., 1967; Froelich et al., 1978). Furthenoore, tl1e color of this water tends to be green (Borstad, ( . 1979). a condition also observed by Hall ( 1955) and Brown (1942) and

which fishermen in Barbados believe to be related to flyingfish

abundance.

The impact of the low salinity water on the seasonal productivity

of tl1e surface waters in Barbados is not clear. During tl1e onset of low

salinity water, plankton bioma.ss has generally been observed to

increase (Calef and Grice, 1967; Beers et al., 1968; Lewis et al., 1962;

Lewis and Fish, 1969; Kidd and Sander, 1979; Borstad, 1982a). In

contrast, Steven (1971), Sander ( 1971) and Sander and Steven (1973)

found no seasonal variation in productivity near Barbados, despite the

recurrence of low salinity water during their sampling period (July,

1968 to June, 1970). 'l11is suggests tl1at the influx of low salinity

water had a negligible impact on local productivity. Kidd and Sander

(1979) offered a plausible explanation for the different results

obtained by tlte different workers. 'l11ey attributed the increase in

plankton biomass to a complex interaction between the lenses and

turbulence caused by the piling up of water in the island chain. 'l11is

causes the lenses to fragment and generate upwelling activities along

their periphery, where high production rates have been detected; with

lower rates characterizing the nutrient-poor center of the lenses.

'l11e impact of the low salinity water on fish larval abundance in

Barbados is not clear. Seasonal variation in fish larval abundance may

follow a similar pattern to that of plankton (Lewis et al., 1962), but

night surface tows done by Lewis and Fish ( 1969) showed marked

fluctuations in abundance with no clear seasonal pattern. Powles (1975)

observed no seasonality in larval abwldance of offshore fish species but

17 inshore fish species showed peak larval abundance from March-May and

August-october. This pattern complements the seasonal pattern of spawning reported for reef fishes in Jamaica (Munro et al. , 1973) • No data are available on the effects of these conditions on the abundance of neustonic fish larvae.

In the months when the North Equatorial Current is predominant

(September - February) it piles up water east of the Lesser Antilles causing isobaric surfaces to slope downward, from west to east (Mazeika,

1973). 'I11e North Equatorial Current is diverted southwards and the bsckflow creates a vortex type circulation which is observed between

Barbados and the Lesser Antilles chain (f>tazeika et al., 1980), on the windward side of Tobago (Febres-ortega, 1976) and between Barbados and

Tobago (Brucks, 1971; Johannessen, 1971). The formation of these vortex circulations delays passage of water to the Caribbean Sea and may serve as retaining mechanisms for planktonic populations. It is also possible that floating materials may be aggregated in the vortices. In the Atlantic, aggregation of Sarga.ssum into clumps occurs through the formation of windrows (Langmuir, 1938). Windrows (also called riffles or slicks) are wind-driven circulation patterns composed of alternating left and right vortices whose axes are oriented to the wind direction.

They form at wind speeds greater than Beaufort 2 (over 3 m sec-1 ) and the spacing between rows (L) and wind speed (W) is defined by the linear equation formulated by Faller and Woodcock (1950) (L = W x 4.8 sec ) •

At lower wind speeds, Sarga.ssum occurs in patches rather than in rows.

Windrows are known to occur off Barbados but no data are available on their frequency. However, it is possible that they are more connnon from

Nay to August, the period during which the Northeast Tradewinds are at their hiibest.

~le formation of eddies downstream of islands situated across the

path of ooeanic currents is known as the "island mass effect" (Doty and

Oguri, 1956; Sander, 1971). Emery (1964; 1972) postulated the presence

of a stable eddy system west of Barbados and emphasized its potential

for retaining local planktonic populations. The system is believed to

consist of a pair of eddies, one in the northwest and the other in the • southwest of the island, rotating counter-clockwise and clockwise, respectively. Shedding is estimated to be between 55 hrs and 1 week.

Sander ( 1971) suggested that the system operates 1 or 2 km from shore.

However, it is not yet known whether the system is composed of eddies

that periodically break off as suggested by Emery (1964) or is a

sta.nditi.it pair-eddy system as suggested by Powles ( 1975). The latter

would be effective as a retaining mechanism for fish larvae. It is also

possible that tl1e eddy sys'te!n off Barbados may be more complex than

these two alternatives (Powles, 1975). Eddies were also found on the

western sides of St. Vincent and St. Lucia by Leming (1971) and Febres-

Ortega (1976). These appear to be formed by horizontal shear

instability (Heburn. et al., 1982).

Tagging studies on adults of H· affin.is (Mulloney, 1961; Lewis,

1964) and fish larval surveys (Powles, 1975) support the presence of the

eddy system to the west of Barbados. T~ed flyill.ltfish adults have

exhibited circular travel patterns west of Barbados (Lewis, 1964). This

may reflect preference to stay within the zone, rather than forced

retention of adults by the eddies (Oxenford, 1988). Areas where larval

catches are high correspond with areas where eddies would be expected,

given the constant westward flow of the current (Powles, 1975). 1.6. EOON<:MIC IMroRTANCE OF FLYINGFISH IN THE EASTERN CARIBBEAN

Flyingfishes are economically i111p0rtant in the eastern Caribbean,

particularly in Tobago, St. Lucia, Barbados, Martinique, Dominica and

Grenada (Mahon et al., 1986). In Barbados, 51% of the conmercial catch

is composed of flyingfish, specifically Hi1~chtl1YB affinis, with tlle dolphinfish, Corypba.ena hippurus contributing a further 19% to total catch. Historically, the flyingfish contribution is about 48% of the

total marine catch and about 58% of the pelagic catch (Jones and

Oxenford, 1986}. The relative illlpOrtance of the different fisheries in

Barbados, and their seasonality, have been reviewed by Mahon et al.

(1982).

The habit of flyingfish of spawning on floating material increases their catchability, since they can be aggregated by floating fish attracting devices. Barbado~ presently has the highest flyingfish harvest in tlle eastern Caribbean, but most countries in the region are incr'easing, or intend to increase their exploitation of flyingfish. The introduction of long range and better equipped fishing boats, like the ice-boats in Barbados (Hunte and Oxenford, 1986), have already increased annual catches in Barbados by 50% (Jones and Oxenford, 1986).

Preliminary simulations by Mahon ( 1986) suggest that increasing the exploitation rate on spawners by 50% may cause abrupt decreases in flyingfish abundance. Since most countries in the eastern Caribbean area are probably exploiting a single flyingfish stock (Oxenford, 1988), a sharp increase in fishing effort by any country may have severe effects regiotmlly.

An adequate Jmowledge of the biology of flyingfish, particularly those aspects relevant to the distribution, abundance, availability and

response of the stock to increased fishing effort, is necessary for

• proper management of this shared resource.

1. 7 • OBJECTIVES OF 'IHE S'IUDY

'n1e tendency of flyingfish, particularly H· affinis, to use flotsam

as spawning substrates during reproductive periods has been known for

some time. However, previous studies have not systematically identified

and quantified potential spawning substrates, nor investigated variation

in their availability. It is therefore not known whether seasonal

variation in availability of spawning material may influence seasonal ' variation in spawning of flyingfish, nor whether the availability of spawning material may limit population size. One objective of the

present sttdy is tl1erefore to identify flyingfish spawning material near

Barbados 1 and investigate seasonal variation in its abundance and use.

While there have been several studies of larval fish populations in

Barbados 1 the neustonic component has not been studied anywhere in the

easten1 Caribbean. Consequently, little is known about tl1e distribution

and abundance of flyin.gfish larvae, since these are primarily neustonic.

A second objective of the study is . therefore to investigate the

composition, distribution and seasonal variation in' abundance of fish

larvae in the neuston near Barbados, with particular emphasis on

flyingfish larvae. 'n1is information will' be used to cOIIIJlent on ( 1)

whether flyingfish lat~ae are present year-round in Barbados or whether,

as is true for adults, their presence near the island is seasonal; and

(2) whether the distribution suggests that downcurrent eddy systems may

be important in retaining larvae and hence in influencing stock

structure and fish JX>pulation dynamics in Barbados. Most flyingfish in the eastern Caribbean are captured with

ccmnercial gill nets with mesh sizes between 1.5 and 1.6 inches. These

are specifically designed for capture of !!· affinis. The consequence is

that little is known about what other species of flyingfish may be

present in the region and at what abundance. The exception is

Parexocoetus bra.chypterus, which is known to be relatively abundant in

near-shore waters (Lewis, 1961). A third major objective of the study

is to une the results of the nightlighting collections made by dipnet,

to investigate the presence and seasonal variation in abundance of

juveniles of all flyingfish species taken near Barbados. Since dipnets

typically catch a wider size-range of individua.ls than is true for the other sampling gears, size-frequency analysis may provide preliminary

information on growth rates of juvenile flyingfish.

In overview, it is hoped. that by simultaneously investigating the presence and temporal variation in abundance of spawning substrates, neustonic larvae, and juvenile flyingfish, a more comprehensive picture of the population dynamics, patterns of movement and availability of flyingfish near Barbados will emerge than has been possible in the past. 2. MATBRIALS AND MB'1li)OO

2.1. DEFINITION OF ~5

2 .1. 1. Neuston

Neuston and pleuston are tenus C0111110nly used to describe organisms ! associated witJt tlte uppei1DOSt surface layer of the sea. The pleuston may be defined as those organisms which are exposed to wind drift because they are restricted to the surface by tlteir own buoyancy or are attached to floating particles. The neuston may be defined as those organisms whose occurrence at the surface is more temporary and variable

(Hempel and Weikert, 1972). Cheng (1975) further defined the pleuston as those animals which live exclusively on the air-sea interface.

However, the distinction between the two groups is not clear in practice since it is possible to find species which have been classified into the two different categories co-C?CCurring in the same field sample. For example, Zaitsev (1968 cited in Hempel and Weikert, 1972) included

Janthina and Glaucus, both pleustonic forms, as components of the neuston. Ch.eng (1975) also fowld it difficult to treat each group as completely separate from the other. For purposes of this stldy, neuston will be used in a limited sense to refer mainly to fish larvae found at the water surface layer (0-20 cm). It therefore corresponds with tl1e terms icthyoneuston or neustonic fish larvae as used by Hartmalm ( 1970).

2. 1. 2 • Flotsam

Titis term will be u.sed to refer to any floating or semi-submerged material vulnerable to the neuston gear, whether it be of marine or terrestrial origin. It therefore includes, inter alia, seaweeds, seagrasses, tar balls, garbage ruld marine bird feathers. 2 .1. 3. Flyingfisb

In this study, flyingfish is used to include all members of the

fmhily Exocoetida.e. The hemirampid, Qxyporhamphus micropterus, is not

included except when indicated otherwise. The inclusion of Q.

micropterus in the category flyingfish is a c011100n practice (see Brtnm,

1935,; Breder, 1938) as this species eo-occurs with most flyingfish.

2 .1. 4. Larvae and juveniles

The early life history of fishes is usually described by end point

events which mark· transitional developnents from embryonic, larval,

juvenile to adult stages. The description of these stages varies

markedly with the perspective of the discipline (e.g. embryology,

ontogeny, systematics, population dynamics) (Blaxter, 1975) and the

species being stuiied (see review by Kendall et al. 1 1984) • Larval

(particularly post-larval) and juvenile stages are usually separated by

size and behavior, since morphometric, meristic and pigmentation

transformations used to distinguish earlier stages are no longer

evident. For beloniform. fishes, John (1983) considered those smaller

than 50 11111 to be fry. The length range considered as juveniles is 3-9

. . cm in Parexocoetus bra.chypterus hilliwius (Lewis, 1961) and 2-15 cm in

Hirundichthyg affinis (Lewis et al. 1 1962). In this study, the size

ranges used to classsify flyingfish ·samples collected by the. neuston

gear as either larvae or juveniles were' determined from length-frequency distributions of each species.

2. 2. SAMPLING STATICNS

A line-transect was established northwest of the island. Three

?A neuston.. stations located 3, 6, and 9 nautical miles (nmi) (Stations 1, 2 and 3, respectively) from shore were established on the transect (Figure

1). 'Ihe 6 nmi limit (Station 2) was used as the starting p:>int for the

drift nightlight stations. The positions of the stations on each

sampling trip were determined by compass readings and landmarks,

specifically the cement plant in Checker Hall, St. LAlcy (Figure 1).

Distance from shore was initially approximated by estimating the speed

of the boat and time elapsed. After one month, a trailing log was used

for greater accuracy.

2. 3. NEUSTON AND FI.DTSAM OOILECTION

2 • 3 . 1. Neuston samplers

Sampling for neustonic organisms and floating material requires a

gear which can perform well under a range of sea conditions and a range

of towing speeds. 'Ihe gear· should be towed out of the turbulence

created by the towing craft (i.e. bow wave and wake) so as to sample 1l1e

• undisturbed sea surface. In addition, the gear should not itself create

any disturbance or shadows which may affect the distribution of target

organisms. All of these characteristics are necessary to obtain

quantitative results. As an added consideration, the operation and

handling of the gear should require minimum effort.

David ( 1965) designed a neuston sampler which is the forerwmer of

today's surface samplers. Essentially 1 the gear was a rectSllgula.r frame

mow1ted on and floated by a pair of wooden skis. It was designed to

sample the upper 10 cm of sea surface. Prior to this, Parr (1937) used

a simpler device to collect floating weeds in the Sargasso Sea, and

Russian workers (see review by Zai tsev, 1971) have designed various Figure 1

Location of Barbados and of the neuston and nightlightin.g stations, including the land marks used to determine _position. N Ost. Lucla 1\

0 "Barbados I . .. • tl

0Grenada

~Tobago

1J 1 0° 3

e neuston

) night light BARBADOS

0 cement plant

I I I 3 nautical milet gears that are still in use (e.g. the pleuston trawl). Many swnplers, modified from the desi,en of David ( 1965), have been created and reconmended by different workers. These include Booby II (Bieri and

Newbury, 1966), otter surface samplers (Sameoto and Jaroszcyn.ski, 1969), side-tracking neuston nets (Ben-Yami et al., 1970), a modified David neuston catamaran (Hempel and Weikert, 1972), the "push net" by Miller

(1973), Norwegian multi-layered samplers (Ellertsen, 1977), integrated samplers (Hinton and Boney, 1979), Boothbay neuston nets (Hettler, 1979) and the manta net (Brown, 1979; Brown and Cheng, 1981). A 100re recent design is the frame trawl (Methot, 1986) which is intended to sample surface and mid-water larvae and juveniles that usually avoid conventional nets (see Ha.rtrnrum (1970) for a review of the performance of neuston gears made prior to 1970) •

In this study, available designs were surveyed with consideration given to ease of constt~tion, its use from a small research boat, and the possibility of modifying it to swnple, not only fish larvae and juveniles, but also floating material. The otter surface sampler

(Sameoto and Jarosczynski, 1969) was selected on the basis of these criteria. Moreover, the gear has been used effectively from small 100tor boats (Sameoto, pers. corn.).

2.3.2. Description of the neuston sampler

The gear was constructed of 1/8 " aluminium sheets cut to the desired dimensions, bent and welded to a rectangular frame (Figure 2a).

A 6 m long, 1. 27 nun mesh net was used, with a mouth opening measuring

1. 0 x 0. 5 m (Figure 2b). In its original design this gear haq a square mouth. 'I11e alteration was made to allow the width of the mouth to Figure 2 4 a). Profile of the modified Sameoto and Jarozsynski neuston gear used in the sttdy.

b) • Dimensions of the sampling gear showing (clockwise) mouth opening, upper fin, lower bracket (not drawn to scale), and body. All measurements are in metres.

•

I conform with the dimensions of the gear used by the Cooperative

Investigation of the Caribbean for plankton surveys (Smith and Richardson, 1977), the .Caribbean Pollution Research and Monitoring Program for collecting floating tar (Atwood et al., 1987/1988), and the \ quantitative sampling for Sargassum weeds by Stoner ( 1983). In

addition, the increased width aids in preventing the sampler from

capsirlng aud diving under steep waves (John, 1975). " On each side of the frame are adjustable fins which keep the

sampler up during tows and which determine the depth of water sampled.

The fins were bolted to a supporting 1Jar and positioned such that the

underside of the gear which skims the sea surface does not create the

turbulence that might otherwise be caused by the lJar, bolts and latches.

The latches control the angle of the wings in relation to the sea

surface, and there were four alternative positions for the latches.

During preliminary trials of the gear, it was observed that the third

latch position gave the best results in terms of sampling a water depth

between 20 and 30 cm. However, the ability of neuston gears, including

the present gear, to sample a consistent water depth is strongly

influenced by sea conditions.

The lower fin in the original model was eventually removed in this

study, since during gear trials, manipulation of the length of the

towing line and the bridles was sufficient to keep the sampler close to

the sea surfabe even during choppy conditions. However, the lower fin

brackets were retained as they provided the necessary sid~tracking and

continous flow performance of the gear. The. side-tracking capability,

which steers the gear away from the boat, is achieved by attaching the

towing bridle a short distance back from the leading edge of one of the

?0 a

float -- ~net

towing bridle

1 0.5 1 t----1.0------t 1------1.7 ------4

l 0.5 1 ,__ 0.25--t t----1.0 __...... conform with the dimensions of the gear used by the Cooperative

Investigation of the Caribbean for plankton surveys (Smith and

Richardson, 1977), the Caribbean Pollution Research and Monitoring • Program for collecting floating tar (Abrood et al., 1987/1988), and the \ quantitative sampling for Sargassum weeds by Stoner ( 1983). In addition, the increased width aids in preventing the sampler from capsizing and diving under steep waves (John, 1975).

On each side of the frame are adjustable fins which keep the sampler up during tows and which determine the depth of water sampled.

The fins were bolted to a supporting bar and positioned such that the underside of the gear which skims the sea surface does not create the turbulence that might otherwise be caused by the bar, bolts and latches.

The latches control the angle of the wings in relation to the sea surface, and there were four alternative positions for the latches.

During preliminary trials of the gear, it was observed that the third latch position gave the best results in terms of sampling a water depth

between 20 and 30 cm. However 1 the ability of neuston gears, including the present gear, to sample a consistent water depth is strongly influenced by sea conditions.

The lower fin in the original model was eventually removed in this study, since during gear trials, manipulation of the length of the towing line and the bridles was sufficient to keep the sampler close to the sea surfabe even during choppy conditions. However, the lower fin brackets were retained as they provided the necessary side-::-tracking and continous flow performance of the gear. The side-tracking capability, which steers the gear away from the boat, is achieved by attaching the towing bridle a short distance back from the leading edge of one of the

29 .sides ( Sameoto and Jarosczynski, 1969) • '111e length of the bridles was

adjusted using a wire clamp. The lower and upper bridles were usually

kept at 1.2 m and 1.15 m repectively, while the towing line was usually

11 m. '111e gear tends to sink when stationary, so a styrofoam float

measuring 0.7 x 0.7 x 0.15 m was placed on top of tl1e gear as suggested

by Sameoto and Jarostr;ynski (1969). As an added precaution, a safety line

was attached to tlte net in such a way that no disturbance was intr

near its mouth (Figure 3). This line also facilitated manual retrieval

of tl1e net after each tow.

A smaller version of the gear which follows the dimensions of the t net used by Sameoto and Jaroscynski (1969) was made and was used in

January, 1988. 'l11e purpose of this was to determine whether the low

catch rates (larval counts) being experienced resulted from the low

towing speed, rather than from temporal and/or spatial variation in

larval abundance. Mean catch ,rate of the bigger gear was significantly

greater than that of the smaller gear (T-test, t = -2.371, P < 0.05).

These res;mlts suggest that the lower towing speed for the larger gear

(mean = 2.88 kn, n = 11 for tl1e larger gear; mean = 4.11 kn, n = 24 for the smaller gear) did not markedly reduce larval catch rates.

2.3.3. Operation and handling

The gear was set on the windward side of the boat. The tow began

as soon as the towing and safety lines were out, the safety line had

been positioned outside the mouth of the gear, and any snagged lines had

been cleared. Initially, the towing speed was roughly measured by

throwing a float in at the bow and timing its passage to the stern. TI1e

tows were standardized in tenns of time ( 10 min) which was measured

using a stopwatch. After a month, a trailing log was used to measure Figure 3

Towing arrangement of the sampling gear, illustrating the position of the safety line.

•

• sampler

safety line

boat the towing speed ( 3 kn) and the distance towed ( 0. 6 nmi) • Since it was not possible to maintain consistent towing speeds in tile small 1~search boat, even during calm seas, tows were standardized by distance towed

(0.6 nmi) as measured by the trailing log. Towing speed was nevertheless recorded, based on the modal speed given by the log meter.

The gear was manually retrieved on the windward side of the boat with the lower portion of the gear, including the net, left banging over the side. The net was then ti1oroughly rinsed from the outside, with seawater from a bilge pump, to concentrate all specimens into the codend bucket. The bucket was ti1en detached and the samples transferred to specimen jars •

2. 3. 4. Sampling schedule

Both day and night tows were made at least once a week from

November, 1987 to September, ·1988 at ti1e sampling Stations 1, 2 and 3

(Section 2.2). Replicate tows were made during day trips. Day tows during October, 1987 and in the first week of November, 1987 were conducted in the morning (from 8:30 to 11:00 hr} for Stations 1 and 2 aud in the afternoon (14:30 to 17:00 hr) for the other Station. All other day tows were conducted in the afternoon (between 13:00 and 17:00

hr) 1 except for those on December 26, 1987 which were conducted in the morning (8:50 to 12:50 hr). The sequence of tows was offshore to onshore, with tow direction being landward. A total of 49 daytime sampling trips were made, resulting in 288 tows.

Night tows were conducted immediately after the nightlight stations

(see Section 2. 4. 2. ) . All night tows were made between 23: 00 and 03: 00 hr. TI1e sequence and direction of tow was similar to day tows, A total of 49 night trips were made, producing 143 tows.

Overall, a total of 98 trips were made and 431 tows were conducted

(Table 2). Five trips (3 day, 2 night) were missed in September, 1988, due to bad weather conditions and unavailability of the research boat.

The missing tows were due to bad sea conditions and/or gear problems.

Other reasons for tow ca.ucellation included lack of time before sunset, and in one case, net clogging when a patch of garbage was encountered

(Table 2).

2. 3. 5. Treatment of samples

All samples were preserved in 5% buffered (sodium tetraborate)

Fonnalin solution. All fish larvae and juveniles were sorted from the plankton and counted. An attempt was made to identify flyingfish larvae to species. Non-flyingfish larvae were identified only to family level. Total length of each identified larvae was measured to generate length-frequency distributions. References used for identifying flyingfish larvae, juveniles and adults (from both nightlight and neustoncollections) wereBruun (1935), Breder (1938), Imai (1959,

1960), Bvans (1961), Staiger (1965), Parin and Gorbunova (1964),

Gibbs and Staiger (1970), Fischer (1978), and Kovalevskaya (1978).

References used for identifying larvae other than flyingfish were

Ahlst1~ (1965), Powles (1975), (Miller -.-et al. (1979), Leis and Rennis (1983), Fahay (1975, 1983), Moser et al. (1984), and Nishikawa and

Ri.Jmler ( 1987).

TI1e flotsam collected was identified into type or species, where appropriate and possible, and wet weights (to the nearest 0.01 gm) were taken. The material was later classified as either marine or Table 2

8\.lll'llal'y of neustoo tow schedule.

Time Period Day Missed Night Missed Trips Tows Tows* Trips Tows Tows*

1987 October 5 28 la, lb 4 10 2a November 4 21 3a 4 12 Deoelnber 5 30 5 15

1988 January 4 23 lb 4 11 la February 4 24 3 9 March 4 24 5 15 April 5 30 4 12 May 4 24 5 15 June 4 24 4 12 July 5 30 4 12 August 4 24 5 15 September 1 6 2 5 le

TOTAL 49 288 6 49 143 4

Total Ntanber of Tows: 431 Nunber of Missed Tows: 10

* reasons for missing tows: a :: net/gear (also net clogging by garbage) or boat problems, b = time limitations (after stUlset) and c = rough sea conditions.

N.B. One day trip (6 tows) where a bigger mesh was used is exclooed from the analyses of fish larvae but included in analyses of flotsam. terrestrial.

' 2. 3. 6. Visual esti.lation of flotsam

An attempt was made to quantify floating material by sightings

while the boat was travelling to the sampling stations during day trips.

The objective was to detennine patch size and location of flotsam in

relation to distance from shore. However, during the entire period of

samplir~, the quantity of flotsam observed was negligible, except for a

patch of garbage encowttered in November, 1987.

2. 4. NIGHTLIGHT SAMPLING

Various approaches have been used to study juvenile and adult

flyingfishes. 'I11ese include visual colmts (Nesterov and Grudtsev, 1980;

Karaman, 1980) underwater observations (Nesterov and Bazanov, 1986) and

nightlighting using dipnets (~wis, 1959; Lewis et al., 1962) or lift

and cast nets (Grudtsev et al., 1987). Other techniques such as

gilln.ettit~ (Hall, 1955; Storey, 1983), hook and line (Storey, 1983) and

tagging (Lewis, 1964; Oxenford, 1988) are normally used to study only a

limited size range of specimens. Many workers therefore use a

combination of these approaches to ensure an adequate sample size and a

good coverage by size of specimen. Nightlighting using dipnets was used

in this study to sample juvenile and adult flyingfish.

2.4.1. Equipuent and pr

Two pairs of 50 watt 12.8 volt car headlights were used for fish

attraction. 'I11e lights were positioned on the leeward side of the boat

where wave action was least and therefore catching was facilitated.

Power was provided by a portable AC/DC generator. A long handled-dip net. and an aquarium dip net were used to capture adult and jlNenile flyinafish respectively. There was no difficulty in choosing which net to use, as the size of the flyingfish approaching the lighted area could be visually ascertained prior to any attempt at capture. The effective fishing range of the big dip1.et was about 2 m (Figure 4) while that for the smaller net was about 0.5 m. For this reason, a lift net is unlikely to be effective. The technique used. was to catch the fish as they swam towards the lights and before they moved under the boat.

2.4.2. Sampling scbedule

The nightlighting was conducted at least once a week from October,

1987 to September, 1988. A total of 49 trips were made instead of the expected 52, either because of bad weather and sea conditions or of boat or gear problems. Initially (November, 1987), nightlighting was conducted only at Station 2 . (Section 2.2) for the full three hours.

However, catch rates tended to decrease in the last hour. To minimize this apparent "depletion effect", in all nightlighting trips after

November, 1987 1 the boat was moved 0. 5 nmi every hour. Nightlighting was conducted between 18:00 and 23:00 hr.

2 • 4 • 3 • Treatment of SSDIJ>les

All samples were frozen ashore. Flyiugfish were sorted to species whenever possible. Standard, fork and total lengths (to the nearest

lllll) 1 and body weight (to the nearest 0.01 gm) were recorded for all specimens; sex and gonad weight were recorded for mature specimens.

2. 4. 4. Quantifying sampling effort

At each station, the number of successful and unsuccessful (

Fiaure 4

An illustration of the procedure used during nightlightina for flyingfish, showing light positions, effective fishing area and illuminated zone (not drawn to scale). (

•

• ILLUMINATED ZONE

EFFECTIVE FISHING AREA

2m CAR ,...--.--..,..__,.._L I G H TS , 1 1 i r,.---,....,.___;_7 2

BOAT (missed) capture attempts were recorded to provide an index of sampling effort. A successful dip was defined as one in which the target fish was captured. A missed dip was one in which the fish escaped. Only one miss was recorded for each fish regardless of the number of UllSuccessful attempts made to captm-e it. Sampling effort was not recorded for non flyingfish catches (e.g. cephalopods, myctophids and other hemiramphids ) •

2. 5. MEASURFl1ENT OF ENVIOONMENTAL CHARACTERISTICS

Environmental characteristics were measured before and after eac:h neuston tow. Temperature was measured using a reversing thermometer, water transparency using a Secchi disk and water color using a Forel-Ule color scale. Wind direction was determined using land markers. Sea state and wind speed were graded as weak/calm, moderate, or strong. Wave height was recorded in 0. 5 m intervals, and clou:l cover rated as either clear, slightly cloudy, partly cloudy, cloudy or overcast. Mean monthly values of ti1e environmental characteristics are presented in Appendix 1.

2. 6. DATA ANALYSIS

Several neuston tows contained no flotsam and no larvae of many of ti1e fish families investigated. Consequently, data on flotsam were not normally distributed and could not be normalized by transformations.

The Statistical Graphics System (STATGRAPHICS) procedure for non parametric analysis was therefore used with all data (STSC, 1986; see also Zar, 1984) . Wilcoxon paired-sample tests were used to investigate variation between replicate tows. Manu Whi tney tests were used to compare flotsam and larval abundance between day and night samples, and Kolmogorov-Smirnov two-sample tests to compare length- frequency distributions of day and night samples. Kruskal-Wallis one way analysis of variance was used to investigate seasonal and spatial variation in abundance of flotsam, larvae and juvenile fish; and Mann

Whitney tests for paired comparisons between stations. Spearman rank correlation analysis was used to investigate correlations between seasonal variation in a.btmdance of individual flotsam components and between fish larvae from different families. 3. RBSULTS

3 • 1. Ot'.'DJRRENCB AND ABUNDANCE OF FIDI'SAM

The neuston samples contained flotsam in every month of the study

(Table 3). It was collected on 68.4% of all trips made, but in only

38.5% of all tows made (Table 3). '111e median weight of flotsam per trip was only 0.112 gm, the median weight of flotsam per tow was less than

0.01 gm (Table 3).

3. 1. 1. Variation between replicate towl:J

Replicate tows were conducted during day trips (Section 2.3.4.).

The frequency of occurrence of flotsam did not differ significantly between replicate tows (observed occurrence in second tows not different from expected occurrence based on first tows; Chi -square test, x2 =

1. 58, p > 0. 05 ) • Consequently, replicate tows were pooled for all subsequent analyses of frequency of occurrrence of flotsam. Similarly, the quantity of flotsam collected did not differ significantly between replicate tows (Wilcoxon rank test, T = 0.902, P > 0.05 Figure 5a).

Consequently, replicate tows were pooled for all subsequent ru1alyses of flotsam a~1ce.

3. 1. 2. Variation between day and night tows

TI1e frequency of occurrence of flotsam did not differ significantly between day ru1d night tows (Chi-square test, x2 = 0.187 1 P > 0.05).

Consequently, day ruld night tows were pooled for all subsequent ru1alyses of frequency of occurrence of flotsam. Similarly, the quantity of flotsam collected did not differ significantly between day Wld night tows (Hrum Mlitney test, U = 0.204, P > 0.05; Figure 5b). Consequently, Table 3

Frequency of occurrence and abundance (wet weight in gm) of flotsam in trips and tows.

TRIPS TOllS

Month No. of trips No. of trips I Occurreace Mo. of tows Ro. of ton I Occurrence lleigkt of with flotsa1 with Clotsa1 Flotsu

198T October 9 4 44.4 ·38 1 18.4 3.52 Rotetber 8 5 6Z.5 33 9 Z1.3 37.38 Decetber 10 8 80.0 45 14 40.0 8.99

1988 Januarf 1 4 57.1 34 1 20.6 6.41 Febrnarf 4 5?.1 33 8 24.2 3.81 March '9 8 88.9 39 u 53.8 32.53 April 9 8 88.9 42 21 50.0 15.89 May 9 7 11.8 39 19 48.7 30.13 June 8 8 100.0 36 31 86.0 31.29 July 9 5 55.6 42 15 35.7 24.99 August 9 4 44.4 39 10 25.& 21.34 Septuber 3 z 66.? 11 4 36.4 53.42

Total 98 n 68.4 431 166 38.5 270.11

Median weight for all trips: 0.112 ga Median weight for all tows: <0.01 ,. Mean weight for trips with flotsaa: 3.86 11 Kean weight for tows with flotsaa: 1.63 &• Figure 5

a) • Box-whisker plot of the quantity of flotsam collected ( wet-,1-i'eight in gm) in replicate tows during day trips. The box covers the uni.ddle

50% of the data values, between the upper and lower quartiles. The vertical lines represent the range of values within 1.5 times the interquartile range. Individual points outside this range are shown. • b). Box-~1isker plot of the quantity of flotsam collected (wet-weight in gm) in day and night tows. Box legends are explained in 5a. •

•

• """ e a ""'J: ....[J •3 I ""' 6 •3 e ....,[J

I: ([ 4 Ill.... 0 ..J 11. 11. 0 2 ....>- H.... z <[ a~ e 6

1 2

REPLICATE NUM8ER

b 60 .j,.l'"' J: ....Ill Ill 3 60 I .j,.l •3 E 40 Cl

I: <[ 30 ....(I) 0 ..J 11. 20 11. 0 > .... 10 H.... z

DAY NIGHT

SAMPLING PERIOD day and night tows were pooled for all subsequent analyses of flotsam abundance.

3. 1 • 3. Seasonal variation

TI1e percentage of tows with flotsam differed significantly between months (October, 1987 to. September, 1988; Chi-square test, x2 = 79.27, P < 0.001, Table 3, Figure 6a). TI1ese data suggest that flotsam occurred most frequently between February and July (Figure 6a) • 'lhe quantity of flotsam collected across: the sampling period also varied significantly between months (Kruskal Wallis test, H = 48.66, P < 0.001). As was true for frequency of occurrence, the abundance of flotsam tended to be lower between October and February than between March and September (Table 3,

Figure Gb).

3.1.4. Spatial variation

The frequency of occurrence of flotsam over the full sampling period did not differ significantly between stations, whether the frequency was calculated as percentage of tows with flotsam (Chi-square test, x2 = 0. 198, P > 0. 05, Table 4) or as percentage of trips with flotsam ( x2 = 0. 257, P > 0. 05, Table 4) • Similarly, the quantity of flotsam collected did not differ significantly between stations (Kruskal Wallis test, H = 0.463, P > 0.05, Table 5).

3. 1 • 5. Composition of flotsam

Fourty-four percent of the flotsam collected was of marine origin and 56% of terrestrial origin. TI1e marine flotsam was composed of the seagrasses, Syringodium filifonne (25%) and Thalassia testudinum ( 19%).

TI1e flotsam of terrestrial origin consisted primarily of tar and pine Figure 6

a). Seasonal variation in the frequency of occurrence of flotsam expressed as percentage of the number of tows in a month with flotsam.

TI1e data are presented in original form ( • ) and median smoothed (-) across three month intervals.

b). Seasonal variation in the quantity of flotsam, expressed as monthly wet-weight/tow (gm) • Tile data are presented in original form ( • ) and median smoothed ( .....-- ) across three month intervals. ,... 1 a ,.. X e. e w u zw er er ::l 0.6 (.) (.) 0 11. 0 0.4 >- (.)z UJ ::l aw 0.2 • er 11.

0 N 0 .J F M A M .J .J A s

MONTH

E b 1.11 6 c: •-I +I l: s Ill ·-I 11 3 I 4 +I •3

:E: 3 - 1 ...H z ------.. ~--· .. ~--

MONTH Table 4

Frequency of occurrence of flotsam (% of tows and % of trips) at Stations 1, 2 and 3 over the entire sampling period (

Station 1 Station 2 Station 3 (3 nmi) (6 nmi) (9 nmi)

No. of tows 139 144 148 No. with flotsam 54 58 5·l % with flotsam 38.8 40.3 36.5

No. of trips 98 98 98 No. with flotsam 36 39 39 % with flotsam 36.7 39.8 39.8 Table 6

Quantity of flotsam collected (wet weight in gm) at Stations 1, 2 and 3 over the entire sampling period (October, 1987 - September, 1988).

Station 1 Station 2 Station 3 (3 nmi) (6 nmi) (9 nmi)

No. of tows 139 144 140 Total weight of flotsam 121.60* 74.95 73.63 % of all flotsam 45.0 27.8 27.2

*42.67 gm (35%) was collected in 2 tows during one trip. If the two tows are excluded, Station 1 accounts for 29 % of total flotsam. needles; secondarily, of wood, bird feathers, cane trash and plant

leaves (Table 6). 'Ibe frequency of occurrence of individual flotsam

components differed significantly between months (Fiaure 7, Table 7;

note P = 0.09 for pine needles). As was true for all flotsam, the

marine components ('Ibalassia testt.di.num and Syringodil.lll filiforme), pine

needles and the category "others" occurred more frequently between March

and September (Figure 7). '!he occurrence of tar showed no discernible

seasonal trend. The monthly variation in frequency of occurrence of I·

testidinum, §. filiforme and pine needles were significantly correlated

with each other, but that for tar and "others" were not correlated with

the other components (Table 8) •

The abundance of the different flotsam components also differed

significantly between months (Figure 8, Table 9; note P = 0.06 for pine

needles) • As was true for all flotsam, the general pattern was that the

abundance of ind.ividual flotsam components tended to be higher between

March and September than between October and February (Figure 8) • 'Ibe

monthly variation in abundance of the marine components (Tbalassia

testudinum and Syringodium filiforme) and the category "others11 were

significantly correlated with each other, that for pine needles with T·

testudinum and "others", that for tar, only with "others" (Table 10).

Considering the flotsam components separately, only pine needles

differed significantly in frequency of occurrence between Stations 1, 2

and 3 (Table 11; but note P = 0.09 for Syringodium filiforme and tar).

Moreover, only pine needles and §. filiforme differed significantly in

abundance between stations (Table 12; but note P = 0.06 for tar). For

§. filiforme and pine needles, occurrence and abundance were highest at

Station 1 (near-shore); for tar, they were highest at Station 2. Table 6

Quantity (wet weight in em) of individual flotsam OCEpOnents collected over the entire sampling period (October, 1987 - September, 1988).

Flotsam Type Weight %

Marine • Thalassia testwlim.111 51.59 (19.0) Syringodiliii filiforme 67.57 (25.0)

Total 119.16 (44.1)

Terrestrial

Tar 11.30 ( 4.2) Pine needles 7.17 ( 2.6) Others 132.48 (49.0)

Total 150.75 (55.9} Figure 7

Seasonal Variation in the frequency of occurrence of the different flotsam components: (a) Thalassia testl.dinum, (b) Syrinsodiun filifonne,

(c) tar, (d) pine needles, and (e) "others... The data are expressed as percen~e of the number of tows in a month with the particular flotsam component, and are presented in original fonn ( • } and median smoothed

( _._ ) across three month intervals. - -

F~EQUENCY OF OCCU~RENCE ~ReQVENCY OF OCCURRENCE (X tow•' .. "'.. .. • .. .. • • • ~ ;; ... • ... .. 01 ~ "' "' ~-·~·~·-·-·-~ "'...-~--*•.--·- -·- .-w-~~- ·---~---- • l> -1 I • -1 -~j X 3 • •a. X :1 • .... ~ ... ,.• IJ • ... • t •~ a. l> l> ..----· iI t/1 I. [11 i CL "' ~

FREQUENCY OF 0CCURRENCE (X towe> FREQUENCY OF OCCURRENCE (X tow•) .. ., :"' !" . !" • ,.• !" .. • ... tu ...... , --- " •-•· - • - -•v·•~ •-• -~ ·- - "' - "' '" ···- . -.. -"' 1(."' 0 0 .' z • z l ID 0 0 0 X :I" :X: :r ~~. u • u ti • • G • .... r I> • l> • t/1 . i rr • f1l "' I~•

FREQUENCY OF OCCURRENCE •, :X: :1 u .... • l> :f "' f () Table 7

Results of Chi-square tests on the variation in the frequency of occurrence of individual flotsam components between months (October, 1987- September, 1988). Frequency of occurrence is calculated as percentage of tows with flotsam.

Flotsam Type x2 p

Marine 'lbalassia testulinun 26.91 *** Syringodium filiforme 35.95 *** Combined 56.34 *** Terrestrial Tar 19.93 Pine 17.64 0.09* Others 39.34 *** Combined 37.47 *** Total 55.260 *** *** level of significance: p < 0.001. * level of significance: p < 0.05.

50 Table 8

Spearma:n rank correlation coefficients between the 110nthly variation in frequency of occurrence of the different flotsam COIRponenta. Frequency of occurrence is calculated as percentage of tows with flotsam.

Thalasaia teatudiaua 8Jrincodiua filiforae Tar Pine Needles

Thalasaia teatudiaua

BJrin&odiua filiforae r .8982 p .0029

Tar r •• 1613 .0232 p .5926 .938&

Piae Needles r .1257 .681f .0887 p .0161 .oua .1681

Others r .357% .4941 .1895 .3816 p .Z361 .1008 .5291 .~056

Significant values ( P< 0.05) are underlined.

1-\1 Figure 8

Seasonal variation in tlte quantity of tlle different flotsam oomponents collected, expressed as monthly wet-weight/tow (gm), for (a) 'lbalassia testu:linllll, (b) Syringod.illll filifo:rme, (c) tar, (d) pine needles, and

(e) "others". The data are presented in original form ( • ) and median smoothed ( -+-- ) across tl1ree montll intervals.

f.? Tn•l••••• teat.XS.L,.,t..t~r~ Sw~LhQOd~~ ~~!~~O~me Tar

1. 2. a 3 t •. 12 c • • ~ 1 t & 2.1!1 & l.l c I I c c ~ ' ...... , t I ., e.e > 2 "' e.ee • ...i .f... ! • I I f I ., e.s i ., 1.$ 1 e.ee • I • ..• ! I ! ! • Ill Ill e. 14 I (.J ... • % •I ~

ONO.JFMAM.J.JAS CNO.JFM"d"'-,:0,)..:.0$ -ONC..."!Fr'i~!"''.;:.J;..S

~NTH MONT M MONTH

Pina Nttedl•• otne,...•

e.ee l· 2 '• e • l' • e ! tJ tJ 1 . ( c ' • & • ee · ...' ... r "".r:: e. e · D ...tJ ) •l • I ~ e. &4 ~ .. e. s' • •l • • ! ___.... Ill . :i ....

• e.2 ~·. . Ill~ •

ONO.JFMAM.) JA$

MONTH Table 9

Results of Kruskal-Wallis tests on variation in the abundance of individual flotsam cauponents between 11100ths (October, 1987 - September, 1988).

Flotsam Type H p

Marine 1balassia testu:li.m.m 2.167 *** Syrilllodi\.ID filiforme 4.740 *** Combined 5.170 *** Terrestrial Tar 5.762 *** Pine 18.85 0.06 Others 5.031 *** Combined 0.694 *** Total 8.648 ***

***level of significance: p < 0.001. Table 10

Spearman rank correlation coefficients between the monthly variation in abundance of the different flotsam components.

Thalassia testudinut Szrin&odiut filitorte Tar Pine

Tialasaia testudinu•

SzrinJodiut filiforae r 0.243 p 0.001

Tar r 0.019 0.068 p 0.690 0.150

Pine r O.Zll 0.031 O.OZT p 0.001 O.HO 0.670

Others r 0.194 0.194 0.100 0.138 p 0.001 0.001 0.040 0.004

Si,nificant values (P < 0.051 are underlined. Table 11

Chi-square tests on variation in the frequency of occurrence of the different flotsam components between sampling stations. Frequency of ocxrurrenoe is calculated as percentage of tows with flotsam.

Flotsam Component

Marine 'lbalassia testuii.num 0.809 0.67 Syringodium filifo:r:me 4.913 0.09 Combined 2.442 0.29

Terrestrial Tar 4.908 0.09 Pine 7.951 * others 2.481 0.29 Combined 1.081 0.40

Total 0.198 0.91

* level of significance: P < 0.06. (d.f. = 2)

fifi Table 12

Kruskal-Wallis tests on variation in the abundance (wet weiiCht in Pl) of the different flotsam canponents between sampling stations.

Flotsam Type H p

Marine Thalassia testuiinliD 3.443 0.18 Syringod.i\ID filiforme 6.103 * Combined 1.603 0.62

Terrestrial Tar 5.770 0.06 Pine 12.051 ** Others 3.698 0.16 Combined 1.593 0.11

Total 0.204 0.84

***level of significance: P < 0.01. * level of significance: P < 0.05. 3. 1. 6. Flotsaa utilization as spasming substrates

Eggs of Hirundichth.ys affinis were found on only five occasions

during the sttdy (Table 13). These were in December, 1987 and between

April and Jtme, 1988. The flotsam with eggs was foWld at all three

stations ( 3-9 nmi offshore) , but the eggs could have been spawned

elsewhere. '111e eggs were attached to either Thalassia testudinum. or cane trash in all cases. The substrates on which the eggs were found were present in small fragments and presliDB.bly had been parts of larger chmps. The cane trash fragments were probably remnants of the fish aggregating devices used by fishermen in Barbados.

Apart from the a'bove egg samples collected at the standard sampling sites, a cli..Dilp of Syringodium. filiforme which was fully laden witl1 !I· affinis eggs was collected 5-nmi off the north of the island

16, 1988. 'The fact that this seagrass, and other types of flotsam, were not found with eggs during routine sampling does not therefore indicate that they are not used as spawning substrates by H· affinis.

3. 2. ABUNDANCE, cx:l1f03ITION AND SIZE OF NEUS'IONIC FISH LARVAE

3. 2 • 1 . Identification of larvae

Larvae were identified using morphological, and when necessary, meristic characteristics. Most specimens caught by tl1e neuston net were post-larvae or early juveniles, but a few were adults. The occurrence of larger or "older" larvae in the uppermost water layer is common in neuston samples (Hempel and Weikert, 1972), and the probability of capturing older larvae may have been increased in the present study by the coarse mesh size of the neuston ~ear.

Identification of larvae to families was therefore relatively easy. Table 13

The occassions on which flotsam was observed to be used as spa.wnillll substrata.

Date Distance f~ Flotsam Flotsam Eggs shore (nmi) type wei.ght(gm) (no,)

1987

December 26 9 cane trash 0.7 17

1988

April 19 3 I. testudim.a 1.4 123 May 3 6 cane trash 17.4 78 June 3 3 I· testudinum 0.2 66 June 6 9 I· testudin1.11 <0.1 12

58 Identification to genus and species was attempted only for the family

Exocoetidae.

3.2.1.1. Hemiramphidae

Hemiramphids (halfbeaks) are characterized by an elongated body,

protruding lower jaw, oblique mouth and an anus situated posterior to

the middle of the body (Vatanachai, 1974). Most hemiramphi.ds collected

were either post-larvae or juveniles, and were therefore easy to

identify.

3.2.1.2. Mullidae

Mullids (goatfishes) are characterized by a slender body, short gut, triangular pigment pattern in the head, no head spination, and

pigments on the notochord and lateral and ventral mid-line of the body

(Miller et al., 1979). Juvenile mullids found in the western Atlantic

IDB.Y be confused with juvenile IIIU!tilids. However, the latter have only

four spines in the first dorsal fin (Caldwell, 1962),

3.2.1.3. Dactylopteridae

Dactylopterids (flying gurnards) can be easily identified by their broad round head, short truncate snout, a head armor with supraoccipital, post-temporal and preopercular spines, and heavy pigmentation (Leis and Rennis, 1983; Washington et al. , 1984) • They are very si.nri.lar to some istiophorid larvae but the latter possess a pointed snout and lack the supraoccipi tal spine.

3.2.1.4. Exocoetidae

Exocoetids { flyingfish) can be distinguished from most other larvae

59 by the early formation of their fins (Leis and Rennis, 1983). The cau::ial fin is already fanned at hatching and the pectoral and pelvic fin rays are very long. Flyingfish larvae in the genus Hirundicht.hys could be distinguished but it was not possible to determine whether these were

H· affinis or H· speculiser. 'lhe large stellate melanophore& on the postero-dorsal side and ventral surface of H· aff:inis larvae (Evans,

1961) are similar to those described by Imai (1960) for H· speculiger.

These two species were therefore combined under the heading

Hirundichtbys §.m!• The problem of separating larvae of these two species has aiso been experienced by Jolm ( 1983), but juveniles and adults may be distinguished by pigment patterns on the pectoral and pelvic fins (Bruun, 1935). Given the considerable numerical dominance of H· affinis over H· speculiger adults in the region (Khokiattiwong,

1989) , it seems likely that most of the Hirundichthys larvae in this study were H· affinis. Larval Exocoetus volitans were identified following the descriptions by Kovalevskaya (1964). Larvae of

Parexocoetus brachypterus were relatively easy to separate from those of other flyingfish because of the presence of barbels (Bruun, 1935;

Breder, 1938; Imai, 1959) • Specimens within the genus Cypselurus are not easy to identify to species (John, 1983). These larvae were therefore combined within the category CYpselurus !!BJ?• following the broad descriptions for this genus provided by Staiger (1965).

Unidentified members of the family Exocoetidae were placed under the general beading Exocoetidae-1.

Given the small m..mber of specimens collected within each species, and the uncertainty in the identification of many larvae within the family, flyingfish larvae were categorised as Exocoetidae for the purposes of subsequent analyses.

3.2.1.5. Mugilidae

Mugilids (mullets) are characterized by a short body, a head with \ no spina.tion, two sepa.ra.te dorsal fins, an anus located in the midline of the body, and heavy body pigmentation (Vatanachai, 1974).

Identification of members of this family is usually based on calor, pigmentation pattern, number of anal fin elements and scales, scale morphology, pyloric caeca and gill rakers (de Sylva, 1984).

3.2.1.6. Istiophoridae

Istiophorids (billfish) have well developed serrate preopercular and pterotic spines and an elongate upper jaw (Fahay, 1983; Nishikawa and Rimmer, 1987) • Identification to family is not difficult as similar larvae, such as dactylopterids and holocentrids, have paired operclular spines and a rostra! spine respectively.

3.2.1. 7. Scombridae

All scombrid (tuna, mackerel) larvae are characterized by a relatively large head, relatively short pointed snout and body, a triangular abdominal cavity, sparse body pigments and the presence of preopercular and post-temporal spines (Nishika.wa and Rimmer, 1987; see also Collette et al. , 1984) • Supraoccipital spines are absent except in

Euthynus .!m• Powles ( 1975) was able to identify Auxis !!ll.Q•, Katsuwonus pelamis, Thunnus mm·, Scomberomorus cavalla and Acanthocybiun solandri from his bongo net collections near Barbados.

61 3. 2. 1 • 8. Myctophidae

'Ibe presence of light organs or photophores is a distinguishing characteristic of ~tophids (lantern fish) (Maser et al., 1984). The relatively lat.·ge ~tophids caught in this stt.dy were easy to separate from other larvae. Larvae of Lampanyctus, Diaphus, Lampadena,

Lampa.nyctus, Myctophum and Hygophum are believed to be present off

Barbados (Powles, 1975).

3.2.1.9. Leptocephalus larvae

The leptocephali collected in this study belong to the

Notacanthiformes (spiny eels), Anguilliformes (true eels) and

Elopiformes (tenpounders, tarpons, bonefishes), the only groups with this type of larval phase. Leptocephalus larvae are characterized by an elongate, highly compressed and nearly transparent body, which mostly consists of V- or W-shaped myomeres (Smith, 1979). The head is disproportionately small for the body and rudimentary dorsal, anal and pectoral fins are present. No attempt was made to identify the specimens collected in this study to family.

3.2.2. Composition of neuston collection

A total of 8, 084 larval and juvenile fish were collected during the neuston sampling program. The catch was comprised of 41 families.

These are listed, in order of decreasing abundance, in Table 14.

Taxonomic c001p0si tion and relative abundance within the flying fishes

( Exocoetidae) is shown in Table 15 .

3. 2. 3. Variation between sampling replicates

Replicate tows were conducted during day trips, but did not differ Table 14

, Taxonomic composition and relative abundance of larval and juvenile fish collected during the neuston sampling program.

1987 1988 Total ~ Kontb Oct Hov Dec Ju leb Mar Apr Hay Jun Jul At.!& Sep (if ) 1) Tows 38 33 45 34 33 39 u 39 36 (2 39 11

Heair&apbidae 35 64 88 109 301 434 550 651 306 190 n 35 2861 35.4 Kyctophidae 17 22 37 n 107 100 198 .116 153 51 113 8 954 11.8 llullidae 79 105 n 35 165 G 3t 240 25 49 1 769 9.5 Dactylopteridae 2 10 Z1 34 T5 152 181 56 7& 89 10 z ?13 8.8 Bxocoetidae 4 5 5 26 53 ll6 91 139 58 30 26 5 559 6.9 Carangidae 1( 35 T1 25 IT 1 19 1 30 9 t8 2 252 3.1 Scoabridae 44 13 20 13 9 11 14 8 u 18 50 5 229 2.8 Kugilidae 2 2 174 u 16 2 1 1 1 220 1.7 Leptocephali 1 4 15 4 12 3 3 37 66 4 u 20 191 2.4 Istiophoridae 31 33 24 1 1 3 12 3 13 9 19 4 165 2.0 Coryphaellidae 1 2 4 6 14 43 ~ 14 ~ 2 4 126 1.6 Noaeidae 2 l 5 10 13 u G 4 8 li2 Serranidae 3 4 15 5 10 5 l 1 3 ( 52 Bothidae 2 1 10 8 10 13 4 48 Balistidae 1 z z 3 12 6 15 1 2 2 47 Holocentridae 13 2 5 3 6 3 34 Gerreidae 6 14 zo Gonostontidae 5 6 i 2 3 19 i.yplaosidae 1 2 13 19 Clupeidae 4 4 4 3 3 19 9yngnatbidae 3 z l 2 4 1 14 Priacanthidae z 5 z 3 l4 'l.'e t r aodon t i dae 1 2 3 12 Geapylidae 2 1 7 Chioroptbalaidae 5 6 Paraiepidae 4 2 6 iiphidae 4 Ostracidae z 3 Poaacentridae 1 2 3 Bregucerotidae 3 3 Sphyraenidae 2 3 Engra.ulidae 2 " Scorpaenidae 1 "z Haeaulidae 1 Apogonidae 1 Acanthuridae 1 Bercida.e 1 Bcbeinidae 1 Ophidiidae 1 I Dauged 26 zz 25 35 sa 30 71 9 5 10 8 16 309 3.8 Unidentified 18 27 34 37 70 2~ 16 10 26 24 ~8 8 330 4.1

Total 31~ 375 uz 40Z 1124 1030 1235 !119-1028 m 446 l21 ~OB4 I 3.9 4.6 5.2 4.8 13.9 12.1 15.3 l:U lU 5.8 5.5 J.5

63 Table 16

• Taxonoudc composition and relative abundance of larval types within the Bxocoetidae ( flyingfishes) •

Taxono1ic Cate1orr Nu1bera Cau1ht

1987 1988 Oct lot Dec Jan Feb Mar Apr Kar Jta Jul Aul Sep Total I

Dirundic~t~Js Spp. 0 1 2 0 16 5 z 55 30 3 1 lU 21.8 lxocoetus volitaas 1 0 0 0 10 u zo 15 10 '0 0 0 16 13.6 Parexocoetus brac~Jpteras 1 0 0 0 z 14 1 11 5 0 0 z 36 6.4 Crpael11rua Spp. 0 0 0 0 0 z 0 0 1 0 0 0 3 0.5 Prognic&tbJs libbifroas 0 0 0 l 0 0 0 0 1 0 0 0 2 0.4 lxocoetidae-1 z 4 3 Z5 25 75 u 58 11 23 u z no 51.4

Total 4 6 5 26 53 116 92 139 58 30 Z& 5 559

(Vote: daaaged b11t still ideatiriable apeeiteas are iDcluded) significantly in nliDber of l8J."'V8e collected (Wilcoxon rank test, T = 0. 77 4, n = 137, P > 0. 06; Table 16; Figure 9a) • Considering the five

nuoerically dominant families in the day tows separately, the number of

larvae collected did not differ significantly between replicate tows

(Table 16). Consequently, replicate tows have been pooled for all

subsequent analyses of larval abundance in day tows.

3. 2. 4. Variation between day and night samples

Day tows collected 6,211 larvae; night tows collected 1,873 larvae.

'11le number of larvae collected per tow during day samples (mean = 22) was significantly greater than that collected per tow during night samples (mean = 13; Figure 9b, Mann Whitney test, U = -3.235, P < 0.01).

A difference may be expected since neustonic larval composition typically differs between day and night samples. The consequence is that day and night samples have been treated separately for all subsequent analyses of larval abundance.

The neuston collected during the day was composed of 34 families

(Table 17) • In order of decreasing abundance, the numerically dominant families were the Hemiramphidae (half-beaks), Mullidae (goatfish),

Dactylopteridae (flying gurna.rds) , Exocoetidae (flying fish) , Carangidae

(jacks) and Istiophoridae (billfish). Note that the Hemiramphida.e strongly dolllina.ted the neuston collections by day. 'l11e composition of the neuston collected in night samples was markedly different from that in the day (Table 18). It consisted of only 24 families, the numerically dominant ones in order of decreasing abundance being the

Myctophidae (lanternfishes), leptocephali (eel larvae), ·scombridae

(tuna), and Exocoetidae (flyingfish). The difference in species • Table 16

Results of Wilcoxon rank tests for differences in nllllber of larvae caught between replicate tows for the five numerically dominant families (families with > 200 larvae collected), and for all larvae combined (day samples).

Family T p TP*

Dactylopteridae 0.395 0.69 43 Exocoetidae 1.210 0.23 68 Hemil'811qilidae 0.348 0.73 25 Mugilidae 0.942 0.35 113 Mullidae 0.248 0.80 88

All larvae 0.774 0.44 8

*TP refers to number of tied pairs. (

Figure 9

a). Box-whisker plot of the mmaber of the larvae collected in replicate

tows (day samples) • The box co'vers the middle 60% of the data values, between tJte upper and lower quartiles. The vertical lines represent tJ1e

range of values within 1. 5 times the interquartile range. Individual points outside this range are shown.

b) • Box-whisker plot of the m.unber of larvae collected in day and night tows. Box legends are explained in Figure 9a.

•

• 400 a

"'3 0 .jJ 300 1.. m Q. "' IIJ ([ :> a:: 200 ([ ..J 11. 0 a:: 111 100 ro I: ::1z

0 ~

1 2

REPLICATE NUMBER

400 b

"'3 0 .jJ 300 1.. Ill D.

IIJ ([ :> a:: 200

0 ~·

DAY NIGHT

SAMPLING PERIOD Table 17

Taxonomic canposition and relative abundance of larval and juvenile fish caught in day tows of the neuston sampling program.

198? 1988 Total 1 Month Oct Kov Dec Jan Feb Kar Apr May Jun Jal Aug Sep (if ) ll Tows 28 Zl 30 23 u 24 30 24 24 30 24 6

Reairhaaphidae 34 64 88 109 301 483 550 649 306 188 41 36 2855 U.D Hullidae 79 105 31 35 165 3Z 240 25 49 1 169 12.4 Dactylopteridae 2 10 u 34 75 151 180' 53 TO 89 10 2 f04 11.3 Hxocoetidae 4 4 4 26 41 93 80 119 54 25 23 3 482 ?.8 Kugilidae 2 2 114 11 16 & 1 1 U9 3.5 Car&ngidae u 33 3S 13 ? 1 18 1 Z9 8 24 2 183 2.9 Istiophoridae 31 33 u 1 1 3 u 3 13 9 19 4 165 2.7 Scoabridae 36 1 7 2 1 3 6 18 34 2 116 1.9 Coryphaenidae 2 1 3 13 19 23 5 2 2 3 13 1.2 Serranidae 3 4 ll 4 4 3 1 3 3 36 Gerreidae 6 14 20 n Koteidae ~ l 8 20 Kypbosidae 2 13 18 Balistidae 3 2 1 15 Clupeidae 4 4 1 3 H Tetraodontidae 1 2 2 1 9 Geapylidae 2 1 l 1 Chloroptaalaidae 5 1 6 Kyctophidae 1 1 1 4 liphi4ae l 1 4 Poaacentridae ~ 3 Bregaa.cerotidae 3 A Spbyraenidae ~ 3 Rolocentridae 2 3 Rngraulidae 2 2 Ostracidae 2 z Scorpaenidae 1 1 Syngnathidae l 2 Apogonidae 1 Bothidae 1 Bcheinidae 1 iaet11lidae 1 Leptocephali 1 Opiidiidae 1 1 Dauged 19 17 14 lZ 40 22 67 10 1 4 217 3.5 Unidentified 16 Z3 30 3& 54 16 9 IS 2.2 24 3 249 4.0

Total Z52 317 2?8 297 916 8Z4 976 875 m 402 270 57 6211 ~ 4. I §.1 4.5 4.8 14.7 13.3 15.? 14.1 12.0 6.5 u 0.9

,--~--~~------.

68 Table 18

Taxonani.c cauposition and relative al:ulda.nce of larval and juvenile fish caught in night tows of the neuston sampling program.

1987 1988 Total l Month Oct Hov Dec Jan Feb liar Apr ll&J Ju Jul Aug Sep lif ) 11 Tows 10 1% 15 ll 9 15 12 15 12 lZ 15 5

KJctopbidae 11 u 36 32 10? 99 197 ll6 153 51 113 8 95Q 50.? Leptocepiali 1 4 15 3 1Z 3 3 31 6' 4 22 zo 190 10.1 Scoabridae 8 6 13 11 9 10 11 8 l8 16 3 113 6.0 Krocoetidae 1 1 6 u 12 20 4 3 2 1'1 4.1 Carangidae 2 z 36 12 10 1 1 4 69 3.? CorJphaenidae 1 3 3 1 24 ll 9 l 53 2.8 Botbidae z 1 9 8 10 13 4 47 2.5 No•eidae 4 10 13 1 6 l 42 z.z Balistidae 2 2 5 4 14 1 2 32 1.1 Ho locen tridae 13 5 3 6 3 31 l.6 Gonostoutidae 5 6 2 z 1 3 19 1.0 Serranidae 4 1 6 & 1 16 Priacanthidae 1 2 5 2 3 l4 81f!gnatbidae 3 2 2 4 1 12 DactJlopteridae 2 6 9 Paralepidae 4 2 6 Reaira1pbidae 1 2 z 6 Clupeidae 3 2 5 Te traodon tidae 1 3 Acan tbur idae 1 Bercidae 1 iypbosidae l Mugilidae I Ostracidae l l Daaaged 1 5 11 23 lZ 8 4 9 1 12 92 L9 Unidentified 2 4 14 5 lG 6 1 5 11 2 4 81 4.3

Total 60 58 144 105 208 106 259 ZH 281 68 116 64 1813 I 3.2 3.1 ?.7 5.6 11.1 ll. 0 13.& 13.0 15.0 3.6 9.4 3.4

69 composition observed is further justification for treating day and night

samples separately when considering seasonal and spatial variation in

larval abundance.

'lbe length frequency distributions of larvae caught during day and

night sampling differed significantly (Kolmogorov Smimov ·two-sample

test, D = 0. 662, P < 0. 001) • 'lbe length frequency distribution for

l.a.rvae caught at night was bimodal (Figure lOa), that for day larvae was

unimodal (Figure lla). Moreover, l.a.rvae caught at night were bigger

than those caught in the day (Figures lOa and lla).

Length frequency distributions for the numerically dominant

families collected in the night and day samples are shown in Figures

lOb-f and llb-i respectively. The figures suggest one explanation for

the difference in size structure of night and day samples. Species which are m.1nerically dominant at night differ from, and are larger than, the numerically dominant species in the day. For example, myctophid larvae (Figure lOb) are significantly larger than hemiramphid

larvae (Figure llb; Mann Whitney test, U = 2.12, P < 0.05);

leptocephalus larvae, which are particularly large (Figure lOo), are

COIIIIlOn in night samples (Table 18) but virtually absent in the day

(Table 17). Note tllB.t leptocephalus larvae are primarily responsible for the larger of the two size modes observed in the night samples

(Figure lOa). A second possible cause of the difference in size of day and night larvae is that, for families which are caught botl1 by day and night, the larvae caught at night may be larger than those caught in the day. 'Ibis is true for the Exocoetidae (Figures toe and lle; night larvae > day larvae; Mann Whitney test, U = -11.92, P < 0.001).

However, the size of scombrid larvae does not differ by day a.nd night Figure 10

Length frequency distributions for the most COIIIJIOtl families of fish larvae in the neuston tows (night samples): (a) All larvae, (b)

Myotophidae, (c) Leptocephalus larvae, (d) Scoinbridae, (e) Exoooetidae and (f) Carangidae. Note that the x-axis scale is not the same for all graphs and that damaged or curved larvae are not included.

•

• TOTAL MIGHT LARVAE 11YCTOPHIDAE

1• a 1 9.9 9.9 b e.a e.a 8.7 9.'7 G u>- z 0.a z 9.6 UJ UJ ::;) e.5 ::;) e,5 cr cr UJ 8.4 UJ 9.4 0: 0: IL 9.3 IL 9.3 8.2 0.2 8.1 9.1 8 9 a 58 100 158 2e0 258 9 3a ea 98 128 .159 LEHGTH CLASS LEHGTH CLASS

LEPTOCEPHALUS LARVAE SCOHBRIDAE

1 c 1 d 9.9 9.9 e.a 0.8 >- 8.7 >- 9.7 0 z e.6 0z 9.6 UJ w ::;) 8.6 ::;) 9.5 cr cr UJ 8.4 w 9,4 0: 0: 11.. 9.3 IL 9,3 9.2 9.2 9.1 9.1

a ..L. 9 .L e 49 89 129 160 299 9 29 49 69 89 LENGTH CLASS LEHGTH CLASS (mm)

EXOCOETIDAE CARANGIDAE - 9.'7 >- 0.'7 u 0 z 9.6 z 9.6 w w ::;) e.s ::;) 9,6 cr cr UJ 0.4 UJ 9.4 0: 0: IL 9.3 IL 9.3 9.2 9.2 0.1 0.1 e 9 -~~-·- 9 39 50 90 129 150 9 HI 29 30 40 50 69 LEHGTH CLASS

Length frequency disributions for the most common families of fish larvae caught in the neuston tows (day samples): (a) All larvae, (b) Hemiramphida.e, (c) Mullida.e, (d) Dactylopteridae, (e) Exocoetida.e, (f)

Mugilidae, (g) Carangida.e, (h) Istiophoridae and (i) Scombridae. Note that the x-axis scale is not the same for all graphs and that damaged or • curved larvae are not included.

•

• TOTAL DAY LARVAE HEMIRAMPHIDAE MULLIOAE - 8.7 >- CL7 >- 0.7 (.) (.) (.) % e.e % 8.e % 0.6 Ill Ill w :;:) 8.6 :;:) 0.6 :;:) 8,5 G G cr Ill w w a: 0.4 a: 0.4 a: 0.4 u. 0.3 u. 0.3 11. 0.3 0.2 0.2 0.2 8.1 8.1 0.1 8 e 0 8 38 68 98 120 169 8 38 68 90 128 168 8 4 8 12 16 28 24 LENGTH CLASS LENGTH CLASS

OACTYLOPTERIOAE EXOCOETIOAE MUGILIOAE

1 d 1 e 1 f 0.9 0.9 0.9 e.e e.e e.e >- 0.7 0.7 >- 0.7 (.) >-(.) (.) % 0.s % 0.6 z 0.6 Ill w w :;:) 8.6 :;:) 0.6 ::J 0.6 G G cr w 8.4 w 0.4 w 8.4 a: a: 0: 11. 8.3 11. 8.3 11. 8.3 0.2 0.2 0.2 8.1 0.1 0.1 8 0 0 8 20 48 se 80 180 8 20 40 60 80 Hl0120 0 10 20 30 40 LENGTH CLASS

CARANGIOAE ISTIOPHOIUOAE SCOMBIUOAE

1 1 1 g h 1 8.9 0.9 0.9 e.e 0.8 0.8 >- 0.7 >- 0.7 >- 0.7 (.) 0 (.) z 0.6 % 0.6 z 9.6 w w w :;:) :;:) 0.5 ::J 9.5 cr 9.6 cr cr w 0.4 w 0.4 w 0.4 0: a: 0: 11. 9.3 11. 0.3 11. 0.3 9.2 0.2 9.2 0.1 e. 1 0.1 _J.___

L_~ e ~ 9 0 .L.... ~J,_._..._(,,._.__.t,""' ... t ••_ •• l.~_..,f e 28 40 60 80 0 6 10 16 20 25 30 0 2 4 6 8 10 12 LENGTH CLASS LENGTH CLASS (mm) LENGTH CLASS 0.05); and the carangid larvae cau,ght

by day are larger that those caught by night (Figures lOf, lla; day

larvae > night larvae; Mann Whi tney test, U = 3. 66, P < 0. 001 ) •

3. 2. 5. Seasonal variation in alu:Jdance

3. 2. 5. 1. Day samples

The number of larvae collected per tow in day samples varied significantly between months (Kruskal Wallis test, H = 83.49, P < 0.001,

Figure 12a). Larvae were most abundant between February and June

(Figure 12a). All of the CODmOn families considered separately showed significant variation in larval abundance between months (Table 19;

Figures 12b-i). Some of the monthly values of larval abundance showed significant positive correlations between families, others showed significant negative correlations, and others were not correlated (Table

20) • The CODmOn families can· be roughly grouped into three categories; those in which sub-adults and adults are largely coastal (Mugilidae,

Mullidae and most Carangidae) , those with sub-adults and adults farther offshore but not fully oceanic (Hemiramphidae, Dactylopteridae, and most

Exocoetidae; note that dactylopterids have an extended sub-adult pbase that is offshore, but adults are coastal) and those with larger, fully oceanic adults (lstiophoridae, and most Scombridae). The monthly variation in larval abundance was positively correlated between the

Hemiramphidae, Dactylopteridae and Exocoetidae i.e. within category 2

(Table 20); larvae being most abundant between February and June

(Figures 12b, 12d, 12e). These families, but particularly the

Hemiramphidae which constitute 46% of the day catch, are therefore largely responsible for the observation that total larvae in day samples Figure 12

Seasonal variation in abundance of the most COIIIIlOn families of fish larvae caught in the neuston tows (day samples): (a) total larvae, (b)

Hemiramphidae, (c) Mullidae, (d) Dactylopteridae, (e) Exocoetidae, (f)

Mugilidae, (g) Carangidae, (h) Istiophoridae and ( i) Scombridae. The data are expressed as nunber of larvae per tow, and presented in original fonn ( • ) and median smoothed ( - ) across three month intervals. TOTAL DAY LARVAE HEMIRAMPHIDAE MVLLIOAE .... 3 49 a 3 30 b 3 12 c .,0 • ..0 ..0 L L 25 L 10 • .. Gl •a. 38 a. D. 28 e IIJ w UJ ::> er 28 er 15 er::> 6

MONTH MONTH MONTH

DACTYLOPTERIDAE EXOCOETIOAE MVGILIDAE .... 3 e 3 6 e 3 10 f .,0 d .,0 .,0 L !... 5 L OJ m e D.• 6 D. D. • 4 UJ UJ UJ 6 ::> ::> a: 4 0: 3 a:

MONTH MONTH MONTH

CARANGIDAE ISTIOPHORIDAE SCOM6RIOAE

..... ~ 3 2 3 2 3 2 i ..0 g ..0 h ..0 L L !.. 1.6 OJ 81 •D. D. 1.5 D. 1. 5

UJ 1.2 UJ UJ ::> ::> a: n: 1 n: 1

MONTH MONTH MONTH Table 19

Results of Kruskal Wallis tests for variation in number of larvae collected per tow between months in day samples.

Family H p

carangidae 31.31 *** Dactylopteridae 74.98 *** Exocoetidae 71.89 *** Hemiramphidae 84.25 *** Istiophoridae 57.49 *** Mugilidae 65.46 *** Mullidae 62.19 *** Scombridae 47.84 *** All larvae 83.49 *** *** level of si~1ificance: p < 0.001. Table 20

Spearman rank correlations coefficients of the monthly variation in larval abundance (day samples) between families. The correlation analyses were conduoted on a tow by tow basis.

Carucidae Dactrlopteridae lroeoetidae leairaap~idae Ittlop.oridae lu,ilidae lullidae

Caraacidae

Dactrlopteridae r -0.138 ' o.ou lrocoetidae r -0.081 o.us p 0.145 0.001

Heairaaphidae r -o.on 0.286 0.513 p o.tu 0.001 0.801 htiopioridae r o.&n -0.131 -0.063 -0.036 p 0.001 0.038 0.294 0.550

Ka&ilidae r 0.031 o.us 0.107 o.m -0.0'0 p 0.601 0.001 0.074 0.001 0.245 lu.llidae r Q_JT9 -0.063 -0.031 0.037 0.383 0.164 p 0.001 0.293 0.600 0.540 0.001 0.182

Bcoabridae r o.m -0.191 -0.035 -0.151 0.19! -o.ou 0.114 p 0.003 o.ooz 0.558 0.009 D.OOZ o.uo 0.057 8icaificaat Yalaes (P < 0,051 are aaderliaed. ' are most abtmdant between February and J\me (Figure 12a). t Monthly variation in larval abundance was positively correlated

between the Istiopboridae and Scombridae ( i. e • within category 3; Table

20) , with larvae being least abundant between February and June ( Fiaures

12b, 121) • Consequently, monthly variation in larval abundance of these

families is either negatively correlated or not correlated with the

families in category 2 (e.g. Scombridae neaatively correlated with

Daotylopteridae and Hemiraqilidae, not correlated with Bxocoetidae;

lstiophoridae negatively correlated with Dactylopteridae, not correlated

with Rx:ocoetidae and Hemiramphidae; Table 20) •

Seasonal trends in larval abunda.noe of the Mullidae, Mugilidae and

Carangidae were less evident (Figures 12c, 12f, 12g), However,

variation in larval abundance was positively correlated between the

Carangidae and the Mullidae (Table 20). The Mugilidae were not

correlated with either the Carangidae or Mullidae (Table 20). Carangid

larvae were least abundant between February and May, and mullid larvae

were least abundant between March and May (Figures 12g, lZc).

Consequently, variations in larval abundance of the Carangida.e and

Mullidae were positively correlated with those of the Istiop1oridae and

Scombridae (Table 20; but note P = 0.06 between Scombrida.e and

Mullidae) • Moreover, Cara.ngida.e and Mullida.e were either negatively

correlated or not correlated with the families in category 2, i.e. those

with high abw1dance between February and June (e.g. Carangida.e was

negatively correlated with Dactylopteridae, not correlated with

Exoooetidae and Hemiramphidae; Mullidae was not correlated with any

category 2 family; Table 20). In contrast to the Carangidae and

Mullidae, most larvae of the Mugilidae were caught between February and April {Fliure 12f). 'lhe consequence is that monthly variation in their

larval abundance was positively correlated with that for the

Dactylopteridae, Exoooetidae and Hemil'811Qitidae (Table 20; but note P =

0. 07 with Exocoetidae) •

3.2.5.2. Night samples

'lhe nunber of larvae collected per tow in night samples varied

significantly between months (Kruskal Wallis test, H = 35.69, P < 0.001,

Figure 13a). As for day samples, larvae were most a.b..mdant between

February and June. 'lhe consequence is that monthly variation in larval

abundance in day samples was signifiosntly correlated with that in night

samples (Spearman Rank Correlation Coefficient for monthly median

values, r = 0. 72, P < 0. 05 ) • 'lhis occurs in spite of the fsot that the

taxonomic composition of day and night samples differs markedly (Section

3.2.2).

Considered separately, three of the four COIIIIlOl1 families in the night samples showed signifiosnt variation in larval abundance between months (Table 21). The leptocephali, which did not show signifiosnt seasonal variation, is strictly a larval fo1u, and may be comprised of several families.

Variation in larval abwdance of the Myctophidae was positively correlated with that in the Exocoetidae (Table 22), both families being most abtmdant between February and June (Figures 13b, 13c). Those two families, but particularly the ~fyotophidae which constitutes 50.7% of the night catch, are therefore responsible for the observation that total larvae in night samples are most abundant between February and

June (Figure 13a). Monthly variation in larval abundance of the Figure 13

Seasonal variation in abundance of the conmon families of fish larvae oauaht in the neuston tows (night samples). Only groups showing c

siinificant seasonal variation in abuOOance are presented: (a) total larvae; (b) Myotopbidae, (o) Exocoetidae and (d) Carangidae. The data are expressed as number of larvae per tow, and presented in original form ( • ) and median smoothed ( -J across three month intervals.

•

• MONTH MONTH

EXOCOETIDAE CARANQIDAE

•· 2 c d

"' .... 3 3 2.6 .8 0 • t.s s. ""'l 11 •a. • a. 2 >J IIJ IIJ q q ;) ;) 1 0: 1.6 0: q 5 J IL u. 0 0 1 0: 0: IIJ e.s 111 m Ill 1: E: 9.6 :> :>z z • .._..._.. 9 e _i__L._l -·

0 N D .J F M A M .J .J A s 0 N D J F N A N J J A 9

MONTH MONTH Table 21

Results of Kruskal Wallis tests for variation in nuDlber of larvae collected per tow between months in night samples.

Family H p

Myotophidae 34.64 Leptocephali 10.94 0.46*** Scornbridae 15.26 0.17 Exocoetidae 40.61 *** Carangidae 33.13 *** All larvae 35.09 *** *** level of significance: p < o.oot. Table 22

Spearmsn rank correlations coefficients of the monthly variation in larval abm.dance (night samples) between f81Dilies. The correlation analyses were condueted on a tow by tow basis,

MyctoJitidae Exocoetidae

Exocoetidae r 0.284 p 0.001

r -0.074 -0.170 p 0.373 0.042

Significant values (P < 0.05) are underlined. Carangidae was negatively correlated with that of the Exocoetidae (Table

22) , presumably because of the low abundance of oarangids between March and Jwe (Figure 13d).

Three families, the Exocoetidae, the Carangidae and the Scombridae, were relatively COOIIlOil in both day and night samples. Variation in larval abundance of the Exocoetidae and Carangidae in day samples was significantly correlated with their variation in night samples (Spearman rank correlations for monthly median values of Exocoetidae, r = 0. 78, P

< 0. 01 ; Carangidae, r = 0. 80, P < 0. 01 ) • Both day and night samples therefore support the suggestion that Exocoetidae are most abundant between February and June, ru1d Carangidae least abundant between

February and Jwe. This analysis was not performed for Scombridae since they did not show significant seasonal variation in larval abundance in night samples.

3. 2. 6. Seasonal variation in size

3.2.6.1. Day samples

The length frequency distribution for all larvae in da;y samples is shown separately for each month in Figures 14a-l. The distribution is similar in all months, most larvae being between 0 and 20um, with a strong peak in numbers at around lOnm. No consistent modal progression during the year can be detected (Figures 14a-l) , but monthly median larval length appears highest between February and June (Figure 15a).

This peak in larval length results primarily from the seasonal change in taxonomic composition of the catch. Large larvae, such as the hemiramphids and dactylopterids, are most abundant between February ruld

June; small larvae, such as the scombrids and istiophorids, are least Figure 14

Monthly length frequency distributions for all fish larvae caught in the neuston tows (day samples) •

•

• OCTOBER, 1981' JANUARY, 1988

1 a 1 d e.9 0.9 0.e e.e >- 0.1' >- 0.1' 0 0 z 0.6 z e.s w w ::J 0.6 :I 0.6 CJ cr w 8.4 w 0.4 lt 0:: 11. 8.3 11. 0.3 0,2 0.2 8.1 0.1

0 e I'. It ••• I.'. I •• I I! •• ' e 28 48 ee 80 190 120 140 0 20 49 ee 88 108 129 140 LENGTH CLASS (mm) LENGTH CLASS

NOUEMBER, 1981' FEBRUARY, 1988

1 b 1 e 0,9 0.9 e.8 0.8 >- 0 .1' >- 0.7 0 z e.s 0z 0.6 w w ::J 9.5 cr :Icr 0.5 w 0.4 w 0.4 0:: 0:: 11. 0.3 11. 0.3 0.2 0.2 9.1 0.1 0 0 L ~-•-•-•~L. ~ 0 20 40 60 80 100 120 140 0 20 40 60 80 100 120 140 LENGTH CLASS (mm) LENGTH CLASS

DECEMBER, 1981' MARCH, 1988

1 c 1 f e. 9 0.9 e.e 0.8 >- 0.'7 >- 0,7 0 z e.s 0z 0.6 w w :I e.s ::J 0.5 cr cr w 0.4 w 0.4 0:: 0:: u. e. 3 u. 0.3 0.2 0.2 0.1 1).1 0 L 0 e 20 4e 60 a0 100 120 140 0 20 40 60 90 100 120 140 LENGTH CLASS 1 g 1 j 8.9 9.9 e.8 e.8

)- 8.7 9.7 u )- z 8.6 0z 0.6 w w :::> e.5 :::> 0.6 G G w 8.4 w a: 9.4 If e. 3 u.. 8.3 8.2 9.2 8.1 8.1 e 8 L e 28 49 69 8e ue 128 148 e 28 48 68 88 188 128 149 LENGTH CLASS (mm> LENGTH CLASS

MAY, 1988 AUGUST, 1988

1 h 1 k 9.9 0.9 e.8 0.8

)- 8.7 )- 9.7 0 z 9.6 0z 9.6 w w :J 9.6 :::> 9.5 G cr w w a: 9.4 a: e.4 u.. 9.3 u.. 0.3 9.2 IL2 9.1 e. 1 e 9 J. L~. f __ ~o.-'-..J._L~..l-L-.t..-J...i~-L~ 8 29 40 69 80 108 120 140 0 20 40 60 80 100 120 140 LENGTH CLASS (mm> LENGTH CLASS (mm)

JUNE, 1988 SEPTEMBER, 1988

1 i 1 1 9,9 0.9 e.e 0.e

)- 9.7 )- 0.7 0 0 z 0.6 z 0.6 IIJ w :J 0.6 :::> 0.6 cr cr IIJ IIJ 9.4 a: 9.4 n: u.. 0.3 u.. 0.3 0.2 0.2 9.1 e. 1 9 e L<-...... L~ .. __.__...___J ~-~ ,. __.__.._J~~"'--1--~.-..."" 1 ~1i ..___.,__,__~·-' .. ' -• • j e 29 40 69 ae 190 120 140 e 20 40 60 89 100 120 140 LENGTH CLASS (mm) LENGTH CLASS (mm> Figure 16

Seasonal variation in size of all fish larvae and of the most 001111100 fwnilies of larvae caught in the neuston tows (day samples): (a) total larvae, (b) Hemiramphidae, (c) Mullidae, (d) Dactylopteridae, (e)

Exocoetidae, (f) Mugilidae, (g) Ca.rangidae, (h) Istiophoridae and ( i)

Scombridae. The data are expressed as monthly median total lengths

(Dill) •

•

• TOTAL DAY LARVAE HEMIRAMPHIDAE MULLIDAE

14 a 13.0 b 12 c

~ '"'E E 19 I E 10.4 • E 18.6 • • • • . .; ...I ...:X: ...:X: e 1!1 .... C) 7.e C) • z • z z • Ill 7 • IIJ w 6 • J • .J 6.2 .J " z z z 4 4: ([ ([ .... 3.6 H H a 0 2.6 0 1&1 w w 2 t: J: J:

e e I I • I a I I I I I I 0 OND.JFMAN.J.JAS OND.JFMAM.J.JAS OND.JFMAM.J.JAS

MONTH MONTH MONTH

DACTYLOPTERIDAE EXOCOETIDAE MUGILIDAE

18 ., d 9 e 29 f E E E E e E E • 6.76 .. 15 • • • :I • • :X: :I • • 1- ...C) 6 ' ... • Clz • Clz ij IIJ 4.5 IIJ 19 .J 4 J J • z z z ([ ([

e 9 _.L_J_L~l._L~LL-.L_l_I._J--1_ e OND.JFMAM.JJAS ONO.JFMAMJJAS ONDJFMAMJJAS

MONTH MONTH MONTH

CARANGIOAE ISTIOPHORIOAE SCOMSRIOAE

39 12 a g h ~ '"'E 26 E E • E E E 9 6 :I 29 :I :I • 1- 1- 1- C) z Clz . . Clz . . IIJ 16 w 6 IIJ 4 J J J z 19 z z

MONTH MONTH MONTH abundant between February and June. In addition, hemiramphid larvae,

which constitute 46% of the day catch appear to be largest between

February and June ( Fi.lfl,lre 16b) • No seasonal trend in median larval

lena~th is apparent for the remaining COII'IIJ)ll families in the day catah

(Figures 16o-i) •

3.2.6.2. Night samples

'lbe lena~th frequency distribution for all larvae in niaht samples

is shown separately for each month in Figures 16a-l. The distribution

is more variable than for day samples. Most larvae fall within the size

range 0-40um; but a second group with a size range roughly between 80-

lOOum is evident in some months (Figures 16a-l). Within each of the

common families, the length frequency distribution is unimodal,

leptocephalus larvae being responsible for the 80-lOOmm size group

observed in Figures 16a-l. With the possible exception of the

Exocoetidae, monthly median la1~al length for the COII'IIJ)n families did

not show any obvious seasonal trend (Figures 17a-f). For exocoetid

larvae, as was true for hemiramphids in day samples, larvae appear to be

largest between February and June (Figure 17 e) •

3.2. 7. Spatial variatiOD in ab.mdance

3.2.7.1. Day samples

The total number of larvae in day samples, and the mubers of the

c011100n families, are presented separately for Stations 1, 2 and 3 in

Table 23. The number of larvae collected per tow differed significantly between stations (Table 24), total larval abundance decreasing from

Station 3 to Station 1 (Table 23). The differences in total larval abundance between Stations 3 and 1, and between Stations 2 and 1, were Figure 16 • Monthly length frequency distributions for all fish larvae caught in the neuston tows (night samples).

t OC108!11, 111Ul1"7 .)ANUARY, 191111

a d ••• ••• ••• ••• )o '·"' ,. '·"' ~·-6 ~1.6 ~1.5~··· ,...... r .... t.3 •. 3 .... IL .... ~ . t.l ··:~~~lle~~-~---~~~~~~~======:_ • ;u •• 61 •• ltt 126 148 • ... •• 111 128 141 ~!NGTH CLASS <~m) L£NGTH CLASS

NOVEMBER, 198"7 FEBRUARY, 191111

1 r b 1 c e e. 9 ! e. 9 ~ e. e ~ e. e ~ ; •. it •• 7 ~ ,. I ,. I ~·-s r ~··er !le. s ~ Ill t {J • ii'· s r fla.4 ~ .. l ~8.4 t •• 3 ~ 1.3 f e.a ~ ; •. 1 ~ t.2f"1.1 t •~~~~~~====:.._ .~ . • ae 48 se ee 1ee 12e 148 • ae ee 1ee 128 ue LENGTH CLASS <~m) LENGTH CLASS

OECEMBER, 1911"7 MA~CH, !966

1~ c f uf t.e r > •. .., f ~·-6 r Ill;:)e.s '~ i:.:tLG ' •• 1 t ,;.. ---- 6 •• se ee u~e 126 148 6 48 ee ee 1e8 12e 14e LENGTH CLASS (Mml LENGTH CLASS (mm} APRIL, 1888 .JULY, lS.88 (r>•242>

g j

= .. 21 61 81 111 121 141 I 21 41 68 --81 11e 120 • LENGTH"' CLASS (mml LENGTH CLASS

MAY, 1988 AUGUST, 19BB

h 1 r k 8.9 e. 9;. t e.e e. e '- 8.7 e. '7 ~ ~ ,.. ! \!e. 6 ~e. 6 ~ IIJ t ::J8. 5 "'13 ..~e. 4 f:::tr 8.3 8.3 •. 2 •. 2 t l 8.1 J e.:[. •• 8 28 48 61 88 188 121 148 e 28 48 68 ee ue 128 LENGTH CLASS (mml LENGTH CLASS

.JUNE, 1988 SEP7EMBER, 1966

i 1 e. 9 8.9 e.e e.e •. 7 8.7 ~ ,.. \!e. 6 ~8.6 ii w 5'· 5 58· 5 ~e. 4 &! e. 4 .. IL 8.3 8.3 8.2

•. 1

• 21 41 61 81 1ee 121 14e I 21 48 68 ee ue 128 148 LENGTH CLASS

SP..asonal variation in size of all fish larvae and of the most COIIIDOn families of larvae caught in the neuston tows (night samples): (a) total larvae, (b) Myotophidae, (c) Leptocephalus larvae, (d) Scombridae,

(e) Exocoetidae and (f) Carangidae. 'Ibe data are expressed as monthly median total lengths (11111).

•

• TOTAL NIGHT LARVAE MYCTOPHIDAE

188 a ee b ..... E E E ea E "' 69 • X X ... 68 ... :zCJ :zCJ w w 49 J 48 J :z :z 4: 4: H H 29 0 28 0 ., • • w ...... w • • . . . I: I: e 9 0 N 0 J F M A M J J A S 0 N D J F M A M J J A S

MONTH MONTH

LEPTOCEPHALUS LARVAE SCOMBR.IOAE

128 c e d

""E 189 E E •· E • "" 6 e9 • ...X ...J: • :zCJ :zCJ • • . w 69 w 4 • • J J • :z :z « 48 « H H 2 0 0 w 29 w I: I:

9 ~ e L-L,__L._L__j_ 0 N D J F M A H J J A s 0 N 0 J F M A M J J A S

MONTH MONTH

EXOCOETIDAE CARANGIDAE

128 e e f ..... E 199 E e E 6 ' J: ee J: ... 1- :zCJ CJz w 68 w 4 J • • • • J • • z 49 • z ([ ([ H .. H 2 0 • Cl w 29 • w I: I:

# 9 '--~J~--- !.~J ~-J_ __ • .-J_J_-..-J__J_J_~J~ • • t__j-L-J~-l .. ,-~1-L_ 0 N D J F M A M J J A s 0 N D J F M A M J J A S

MONTH MONTH Table 23

Total number of larvae, and numbers of the more common families, presented separately for Stations 1, 2 and 3 (day samples).

Family Station 1 Station 2 Station 3 (3 imni) (6 nmi) (9 nmi)

Hemiramphidae 635 848 1469 Mullidae 2·12 335 192 Dactylopteridae 191 308 268 Exocoetidae 99 158 226 Mugilidae 52 144 23 Cara.ngidae 80 50 52 Istiopboridae 48 62 55 Scombridae 46 44 26 All larvae 1660 2166 2506 Table 24

Results of Kruskal Wallis tests for variation in number of larvae collected per tow between stations (day samples).

Family H p

RemiramP:tidae 16.73 Mullidae 1.64 0.44*** Daotylopteridae 11.33 u Exocoetidae 15.02 U* Mugilidae 0.16 0.92 Carangidae 0.19 0.91 Istiophoridae 3.36 0.18 Scombridae 0.13 0.94 All larvae 12.01 ** *** level of significance: p < 0.001. ** level of significance: p < 0.01. significant;. the difference between Stations 3 and 2 w.s not (Table 26) • Considered separately; only the hemiramphids; the daotylopterids and the

exocoetids, showed significant variation in larva.! abundance between

Stations (Table 24).; , For the hemiramphids and exocoetids, larvae were

most abundant at Station 3 and least abundant at Station 1; for the

daotylopterids; larvae were most abundant at Station 2 and least

abundant at station 1 (Table 23) • As w.s true for all larvae oonabined,

the differences in abundance between Stations 3 and 11 and Stations 2

and 1 were significant for all the families; the differences between

Stations 3 and 2 were not (Table 26).

3.2.7.2. Night samples

The total number of larvae in night samples, and the n\Dbers of the

I._/ OOiliDOn families, are presented separately for Stations 1, 2, and 3 in

Table 26. As for the day .samples; the number of larvae collected

differed significantly between stations (Table 27); total larval

abundance ·decreasing from Station 3 to Station 1 (Table 26) • As was

true for day samples, the differences in total larval abundance between

Stations 3 and 1, and between Stations 2 and 1, were significant; the

difference between Station 3 and 2 was not (Table 28). Considered

separately, only the Myctophid.ae and Carangidae showed significant

variation in larval abundance between Stations (Table 27) • For both

myctophids and oara:ngids, larvae were most abundant at Station 3 (Table

26). For myctophids, larval abundance w.s significantly hi.lher at both

Stations 3 and 2 than Station 1 (Table 28); for carangids, the

differences in larval abundance between stations were not significant in

pair-wise comparisions (Table 28) • Table 25

Mann Wbitney tests for variation in number of larvae collected per tow between station pairs; Station 1: 3 nnd offshore, Station 2: 6 nmi and Station 3: 9 nmi Cday samples) •

Stations

1 vs. 2 1 vs. 3 2 vs. 3 Family u p u p u p

Dactylopteridae 3.18 u 2.61 ** -0.60 0.64 .Exocoetidae 3.02 3.73 0.59 0.66 Hemiramphidae 2.70 **u 4.08 ***u 1.36 0.18 All larvae 2.69 ** 3.23 ** 0.58 0.66 *** level of significance: p < 0.001. U level of significance: p < 0.01. Table 26

Total number of larvae, and numbers of the more common families, presented sepu-ately for Stations 1, 2 and 3 (night samples).

Family Station 1 Station 2 Station 3 (3 nmi) (6 nmi) (9 nm.i)

Myctophidae 144 376 430 Leptocephali 46 50 94 Exocoetidae 15 21 39 Carangidae 13 8 48 Scombridae 36 44 34

All larvae 397 679 805 Table 27

Results of Kruskal Wallis tests for variation in number of larvae collected per tow between stations (night samples).

Family H p

Myotophidae 10.67 Leptocephali 1.29 0.52** Scombridae 1.18 0.55 Exocoetidae 1.97 0.37 Carangidae 7.80 * All larvae 11.68 ** **Level of significance: p < 0.01. *Level of significance: p < 0.05. Table 28

Ma.tm Whi tney tests for variation in number of larvae collected per tow between station pairs; Station 1: 3 nmi offshore, Station 2: 6 nmi, Station 3, 9 nmi (night samples) •

Stations

1 vs. 2 1 vs. 3 2 vs. 3 Family u p u p u p

MyctoJitidae 3.05 2.59 -0.20 0.83 Carangidae 0.33 0.73** 1.07 0.28** 0.76 0.44 All larvae 2.89 ** 2.99 ** 0.45 0.65 ** level of significance: p < 0.01. 3. 2. 8. Spatial variation in size

3.2.8.1. Day samples

All families except the Scombridae showed sianificant variation in larval size between stations (Table 29; but note P = 0.07 and 0.10 for Carangidae and Istiophoridae respectively). When significant differences in larval size occur between pairs of stations, larvae are larger at the offshore than the nearshore station (Table 30). For example, for Mullidae, larvae are larger at Station 2 than 1, at Station

3 than 1, and at Station 3 than 2; for Carangidae, Ex:ocoetidae and

Hemiramphidae, larvae are larger at Station 3 than at 1 and 2; for

Dactylopteridae, larvae are larger at Station 2 and 3 than 1; and for

Istiophoridae, larvae are larger at Station 3 than 1. The sole exception is the Mugilidae in which larvae are again largest at Station

3, but are larger at Station 1 than 2 (Table 30).

3.2.8.2. Night Samples

The Exocoetidae and leptocephalus larvae showed significant variation in larval size between stations, the Myctophidae, Carangidae and Scombridae did not (Table 31). For the leptc:xrephali, the tendency for larvae to be larger at offshore stations was again observed, larvae being larger at Stations 2 and 3 than Station 1 (Table 32; but note P :::

0. 07 between Stations 2 and 1 ) • 'lhe Exocoetidae were the exception to the day sample pattern, larvae being larger at Station 2 than at 1 and 3

(Table 32). Table 29

Results of Kruska.l Wallis tests for variation in size of larvae between Stations 1, 2 and 3 (day samples).

Family H p

Hemii'BIII}ilidae 40.31 Mullidae 101.64 *** Dactylopteridae 34.62 *** Bxocoetidae 25.53 Mugilidae 56.68 *** Carangidae 5.17 0.08*** Istiophoridae 4.59 0.10 Scombridae 1.55 0.46

*** level of significance: p ( 0.001. Table 30

Median larval size (mm) of the common families in day samples at Stations 1, 2 and 3, and results of Ma.nn Whitney tests for comparing larval sizes between stations.

Median larval si£e ~ann Vkitney teats

Station 1 Station 1 Station 3 1 rs. 2 1 vs. 3 2 vs. 3 (3 n1i) (6 nail f9 n•il

no. (••I no. 1••1 no. (11) u p u p u p

Hnirupkidae 488 9.84 ?44 9.48 1280 10.67 -l.U 0.22 4.12 m 5.89 tU Kullidae 21( 5.31 313 &.n 173 7.51 8.89 tU a.u ut 2.13 t Dactylopteridae m 6.00 305 7.11 266 1.50 4. 70 Ut 5.86 m l.3Z 0.18 Eiocoetidae 89 6.3Z 151 5.93 ZD9 1.90 -l.U 0.16 2.64 u 4.9? au Kugi lidae 41 4.98 131 4.66 %3 7.19 -5.33 Ut 3.53 Ut 6.33 m Carangidae 86 6.3Z 151 5.93 Z09 7.90 11.45 0.91 2.01 t &.02 t [stiophoridae 46 4.34 62 5.09 54 5.53 1.49 0.14 2.16 t 0.56 0.57 Scoabridae 46 4.&3 n 4.50 26 4. !l

ttt level of significance: P < 0.001. u level of significance: P < 0.01~ level of significance: P < 0.05. Table 31

Results of Kruskal Wallis tests for variation in size of larvae between Stations 1, 2 and 3 (night samples) •

Family H p

Myotofhidae 0.45 0.79 Leptocephali 6.92 Soombridae 0.46 0.78* Bxocoetidae 6.38 ~idae 1.58 0.45* * level of significance: p < 0.05. Table 32

Median larval size (lllll) of the COIIIDOn f8lllilies in night B8lllples at Stations 1, 2 and 3, and results of Mann Whi tney tests for comparing larval sizse between stations.

Faaib Median !anal sizes Mann Vhihey tests

Station 1 Station Z Station 3 1 vs. 2 1 vs. 3 1 vs. 3 (3 nail (6 nti) ( 9 n•il

no. (Ill no. {11] no. (111 u p u p u p

Hyctopltidae 107 16.98 293 16.00 442 16.00 Leptocephali 21 80.00 23 94.00 144 89.50 1.83 0.01 z. 39 f -1.16 0.25 Bxocoetidae 14 27.50 11 50.00 41 37.00 Z.03 l 1.02 0.31 -2.16 l Carangidae 9 3.31 8 3.91 36 3.15 Sco1bridae Z7 4.34 34 4.10 36 4.14

a level of significance: P < 0.05. •

3. 3. SPBCIBS c:DJEU.CJITI~, ABUNDANCE AND SIZB OF JUVBNILB FLYINGFISH IN

NIGHTLIOHT SAMPLES

3. 3 .1. Catch as an iD:Iex of abJndance

During nightlight sampling, an attempt was made to capture every

fish that entered the effective fishing area of the illuminated zone

(Section 2. 4 .1. ) • The number of captures and misses were recorded, the

sum of which can be considered an index of available fish. A comparison

of the number of fish caught with the number of capture attempts

therefore allows assessment of whether fish caught is an adequate index

of available fish. In particular, it is possible that the percentage of

fish caught might decrease with increasing number of fish present.

During the study, 3, 384 attempts to capture flyingfish were

recorded, of which 7 4% were successful. The percentage of fish missed

in a sampling trip was not correlated with the index of fish available

in that trip (i.e. the sum of catches and misses; Spearman rank

correlation, r = -0.1667, P > 0.05; Figure 18a}. This suggests that the

efficiency of capture is not a function of the availability of fish.

Moreover, the nunber of captures was positively correlated with the

index of fish availability (Spearman rank correlation, r = 0.98, P > 0.001; Figure 1Bb). This suggests that the m.mber of fish caught is an ' adequate index of the number of fish available. Consequently, the

number of captures has been used as an ·index of fish abundance in

subsequent analyses.

3. 3. 2. Depletion effects during sampling

Since an attempt was. made to capture every fish that entered the

effective fishing area of the illuminated zone, it is possible that (

Figure 18

a) The percentage of fish missed in a sampling trip versus the index

of available fish (sum of captures and misses) for that trip. b) The number of captures in a sampling trip versus the index of

available fish (sum of captures and misses) for that trip. I

•

101 • 1

0.e

0.6 ~ (0 (0 H E 0.4

0.2 ...

0 100 200 300 400

INDEX OF AVAILABLE FISH

b 490

(f) w 300 Q: :::l la. <[ 0 IL 200 0 Q: w m E :::l z 100

0 100 200 300 400

INDEX OF AVAILABLE FISH catch rates during a sampling trip ( 3-hour duration) may decrease with time due to depletion of fish in the vicinity of the sampling vessel.

To investigate this, the number of captures and misses (the index of available fish) was recorded for every 10 minutes during tile 3-hour sampling period. TI1is allowed assessment of depletion effects at 10- minute intervals within each hour and at hourly intervals within each sampling trip. Neitl1er the number of captures nor tile number of misses differed significantly between 10-minute sampling intervals for either the first, second or tl1ird hour of the sampling trips (Table 33) • The data could tl1erefore be pooled into 1-hour intervals for subsequent analyses. Moreover, neitl1er the number of captures nor the number of misses differed significantly between 1-hour sampling intervals across all sampling trips (Table 34). However, there is a tendency for captures and misses to be low in the third hour of samplillll trips (Table

34). Hourly data have been pooled into trip data for subsequent analyses of variation in capture rates between taxonomic groups and between months.

3. 3. 3. Taxonomic cauposi tion of catch

A total of 2, 834 flyingfish comprising 13 species ( includil\il the hemira.mphid Q. micropterus) was caught duril}g the nightlighting study.

TI1e species captured and their relative abundance are shown in Table 35.

The most common species were the sailfin flyingfish, Parexocoetus prachYPterus (41.7%), the two-wing flyingfish, Exocoetus volitans

( 36. 7%) , the four-wing f lyingfish, Hirundichthys !iffinis ( 14.6%) , and the hemiramphid, ~rhamphus miQl'Opt~rus ( 4. 3%) • Subsequent analyses of seasonal variation in abundance and size st.ructure have been restricted to these species. The non-flyingfish component of the

1()? Table 33

'file maber of capture attempts, captures and misses in 10-min intervals, presented sepe.ra.tely for each hour of the sampling trips. Results of 01i-square tests for variation between 10-min intervals in captures and misses are presented.

Time No. of No. of No. of x2 p x2 p (10-min intervals) Attempts Captures Misses (captures) (misses)

First Hour 1 197 152 45 2 216 166 60 3 232 179 53 0.310 0.99 1.559 0.91 4 236 181 55 5 273 205 68 6 215 174 41

Second Hour 1 193 140 53 2 215 163 52 3 177 135 42 0.877 0.97 3.614 0.61 4 187 144 43 5 169 117 52 6 175 139 36 Third Hour 1 124 89 35 2 120 88 32 3 161 113 48 1.011 0.96 4.137 0.53 4 151 107 44 5 144 113 31 6 156 125 31

103 Table 34

The maber of capture attempts, captures and misses presented at hourly intervals for all sampling trips. Results of Chi-square tests for variation between hoU1'ly intervals in captures &ld misses are presented.

Time No. of No. of No. of x2 p x2 F Attempts Captures Misses (captures) (misses)

1 1406 1094 312 2 1120 842 278 0.597 0.74 2.701 0.26 3 858 637 221

104 Table 35

The relative abundance of flyingfish species, incl~ the hemir&ll}ilid Oxyporhamphus micropterus, caught in the 49 sampling trips made d.\ll'ing the one-year nightlighting study.

Species Ntunber %Catch

Pa.rexocoetus brachypterus 1181 41.7 Exocoetus volitans 1039 36.7 Hirundichthys affinis 414 14.6 Oxyporhamphus micropterus 123 4.3 Hirundichthys speculiger 43 1.6 Hinmdichthyg ~. 3 0.1 Prognichthys gibbifrons 8 0.3 Cypselurus cyanopterus 8 0.3 Exocoetus obtusirostris 6 0.2 CYJ)Selurus comatus 4 0.1 Cypselurus furcatus 1 <0.1 Cypselurus heterurus 1 <0.1 Cypselurus melanurus 1 <0.1 Cypselurus exiliens 1 <0.1 Cypselurus ~· 1 <0.1 Total 2834 100.00 nightlighting catch consisted of only 234 individuals (7 .6% of total catch) • The taxonomic composition of this component of the catch is shown in Table 36.

3. 3. 4. Size cx.aposition of catch

'Ihe size structure of the total catch (October, 1987 - September

1988) of Parexocoetus brachypterus is shown in Figure 19a. 'llll"ee size modes are evident; the first consists of fish between 2 - 6 cm in fork length, the second between 7 and 9 cm, and the third between 10 and .12 cm. P. brachypterus reaches sexual maturity at about 11-12 cm (Bruun,

1935; Lewis, 1961). Consequently, the largest mode (Figure 19a) was considered adults, and all fish smaller than this were considered juveniles in this study. By this criterion, 726 juveniles (61.6%) and

455 adults (38.5%) of f. brachypterus were collected.

The size structure of the total catch of Exocoetus volitans is shown in Figure 19b. The distribution is essentially Wlimodal, 99.0% of the 1,027 fish collected being juveniles (estimated size at sexual maturity for !1;. volitans is 14 cm; Bruun, 1935).

The size structure of the total catch of Hirundichthys affinis is shown in Fi~ 19c. The distribution is essentially bimodal; one size group ranging from about 2-12 cm, and the second from about 17-22 cm.

!!· affinis reaches sexual maturity at about '19 cm (BruWl, 1936; Lewis et al., 1962). The larger mode (Figure 19c) was therefore considered ' adults, all fish smaller than this juveniles. By this criterion, 388 juveniles (94%) and 24 adults of H. affinis were collected during tl1e study. Table 36

The relative abundance of non-flyingfish species caught in the 49 sampling trips made during the one-year samplina study.

Species Number %of Catch

Myctophida.e 80 34.2 Ceptalopoda 48 20.5 Coryphaenida.e 41 17.5 Trichiuridae 1 0.4

Hemiramphida.e Eule:ptorbamPhus .!m. 7 3.0 Hemiram:phus brasiliensis 5 2.1 Hemiramphus ba.lao 1 0.4 Unidentified hemil'811l}:i1id 14 6.0

Belonidae Tylosorus ~ 10 4.3 Ablennes hians 6 2.6 Strongylura marina 6 2.6 Strongylura notata 4 1.7 Platy\)elone argulus 1 0.4

Unidentified 10 4.3

Total 234 100.0

107 Figure 19

Length frequency distributions (fork length in cm) for the most coomon flyingfish species caught in the nightlight sampling program (October,

1987 - September, 1988): (a) Parexocoetus brachYOterus, (b) Exocoetus ( . volitans, (c) Hirundichthys affinis, and (d) Qxyporhamphus micropterus.

•

1fl0 • P•r•xocoatu• br•chypt•ru•

8.16 a 0.26 - b

8.12 0.2

)- )- 0 e.0s 0.16 z z0 w w :J :J C1 w (Jw 0: 0.es 0: IL IL 0.1

8.83 e.85

e 3 9 12 16 0 4 8 12 16 29

FORK LENGTH FORK LENGTH

Hirundichthu• arrinis Oxyporhamphus micropterus

9.24 c 0.16 d

0.2 0.12

0.16 >- 0 >- 9.99 z 0z w w ::J 0.12 :J (J w w(J II II 9.96 IL 11.. e.ee

CL 03 e.e4

e e

9 4 8 12 16 20 24 9 3 6 9 12 15

FORK LENGTH (cm) FORK LENGTH The size structure of the total catch of Oxyporhampus aicropterus

·is shown in Figure 19d. The distribution is essentially bimodal; one

size group ranging from about 3-9 cm, and the second fran about 11-14

cm. Q. micropterus reaches sexual maturity at about 12 cm (Bruun,

1935). The larger mode (Figure 19d) was therefore considered adults,

all fish smaller than this juveniles. By this criterion, 71 juveniles

(58%) and 52 adults of Q. micropterus were collected during the study.

3. 3. 5. Seasonal variation in abundance

The number of flyingfish caught per trip across the sampling period

varied between months (Tables 37 and 38; but note P = 0.08). When only

juveniles were considered, the variation in abundance between months

attained statistical significance, juvenile flyingfish being most COIJIIl.Oil

between February and August (Tables 37 and 38). The abundance of adult

flyingfish did not vary significantly between months (Table 38).

Seasonal variation in ab1.mdance of the four COlllllOn species can be

considered separately. With juveniles and adults combined, there wa.s no

significant variation in abundance between months for !:· brachyPterus

(Tables 37 and 38) . However, the abundance of juveniles varied between

months, juveniles being most common between March and July (Tables 37

and 38) . The abundance of adults did not vary significantly between

months (Tables 37 and 38).

The number of adults of ~. voli tans caught ( 6) were insufficient to

justify analysis of seasonal variation in abundance. The adults were

collected in January ( 3) , March ( 1 ) , May ( 1 ) and September ( 1 ) (Table

37) . The abundance of juveniles of this species varied significantly

between months, juveniles being most coiiiilon between February and August

109 Table 37

The number of flyingfish caught in each month of the one-year nightligbting stuiy, presented separately for juveniles and adults of the four ooomon species.

bti Trip lirelDmeblllni,ntn ~ wlita lliMfuib atfilil Oopmlldl! l.iclwt.n tial falxi JIM!Di.le Malt JvaileMalt JlmileMalt Jlllllile *it Madleldalt

liT Mm 4 0 31 5 0 5 0 1 I 11 32 bel8!r 4 14 9 3 0 0 2 0 1 17 12 lmiDr 5 • 111 3 0 15 1 2 1 m liM 118 larr 4 15 Url t 1 3 1 t 4 I 113 ~ 3 6 51 94 0 Zfi 1 13 l 138 53 lfardt 5 n 19 m? 3 18 0 17 34 314 56 AJril 4 166 49 149 0 1 1 5 t 11 52 6 rot 50 113 1 40 0 11 4 311 55 Jme 4 105 15 2 0 38 5 5 0 150 m ""July 4 99 11 t 0 I lZ 6 0 139 29 Qlllt 5 1 14 4ZII 0 ID 0 3 0 645 14 ~ z 0 8 Z5 1 1 0 0 4 I 13

fOCal 49 ?09 412 1033 6 391 234 fl 52 1104 553

110 Table 38

Results of Kruskal Wallis tests on variation in the m.~nber of fish caught per trip between months (October, 1987- September, 1988); presented separately for juveniles and adults of the four COIIIOOn species in the nightlight catch.

Species Juveniles Adults Total Catch

H p H p H p

Ps.rexocoetus brachyPterus 21.03 * 4.84 0.94 9.72 0.56 Exocoetus volitans 28.08 ** 28.00 ** Hirundichthys affinis 24.59 * 22.93 * Qxyporamphus micropterus 18.68 0.07 5.34 0.91 13.92 0.24

Total 24.38 ** 4.31 0.96 18.23 0.08 **level of significance: P < 0.01. * level of si~1ificance: P < 0.05.

111 (but note the very low catches in June and July; Tables 37 and 38).

The nllllber of adults of !!· affinis caught ( 23) were insufficient to justify analysis of seasonal variation in aburdance. The adults were collected in November ( 2) , December ( 1 ) , January ( 1 ) , February ( 1 ) ,

April (1), June (5), and July (12) (Table 37). The abundance of juveniles of !! . affinis varied significantly between months, juveniles being most coomon between February and August (but note the low catch in

April ; Tables 37 and 38 ) •

With juveniles and adults combined, there was no significant variation in abundance between months for Q. micropterus (Tables 37 and

38) • The abundance of juveniles varied between months, juveniles being most coomon between February and July (Tables 37 and 38; but note P ::

0.07). The abundance of adults did not vary significantly between months (Table 38) •

In sl..UIIIl8.ry, the data suggest that juveniles of all four species are more abtmda.nt over a similar period. of the year; i.e. between February and August. The consequence is that seasonal variation in abundance of juveniles is significantly correlated between most species pairs (Table

39).

3. 3. 6. Seasonal variation in size

Length frequency distributions for Parexocoetus brachypterus,

Exocoetus voli tans, Hirmilichthys affinis and ()xyporhamphus micropterus, the four common species in the nightlight catch, are shown at two-month intervals in Figures 20, 21, 22 and 23, respectively.

For P. brachyterus, only one cohort was present in December -

112 Table 39

Spearman rank correlation coefficients of the monthly variation in abundance of the most coumon species of juvenile flyingfish caught in the nightlighting program.

Parexocoetua brachypterus Bxocoetus volitans Hirundiclthys affinia

Parexocoetus brachypterua r p

Brocoetus volitans r .0735 p .6104

Hirundicbthys affinis r .1633 .3128 p .2579 .030Z

Oxyporhaaphus aicropterus r .51i2 .2990 p .0003 .0383

Significant values (P < 0.05) are underlined.

lB • Figure 20

Length frequency distributions (fork length in cm) of Parexocoetus t brachYOterus presented at bi-monthly intervals.

•

I OCTOBER - NOVEMBER, 19a'7 APRIL - MAY, 19a8 d 1 a 1

e.a 0.a >- >- e.e 0 0.e zw i :Jcr w 9.4 w 9.4 0:: 0:: .. u. 9.2 9.2 .I e e ••• I I'. I ••• I ••• I I •• I •• It I •• I e 2 4 6 a 19 12 14 e 2 ••4 6 a 19 12 14 FORK LENGTH FORK LENGTH

DECEMBER, 198'7 - JANUARY, 1988 JUNE - JULY, 1988 1 b 1 e

9.8 0.e >- >- 0z 0.e z0 9.5 w w :J cr cr:J w 9.4 w 9.4 0:: 0:: u. u. 9.2 0.2

9 ....L~ ...... ~J- 0 9 2 4 5 6 10 12 14 e 2 4 6 a 10 12 14 FORK LENGTH

FEBRUARY - MARCH, 19aa AUGUST - SEPTEMBER, 1988 1 c 1 f

e.8 e.8 )- >- 0 z0 0.6 z 0.6 w w :I cr:I cr w 0.4 w 0.4 ..0:: ..0: 0.2 9.2

0 0 .t < ...... t ,__._._t_._---~_&...... ~ __._ ..... _ _t_~-~J __ __ ...___..._~t..,. 0 2 4 6 a 10 12 14 e 2 4 6 e 10 12 14 FORK LENGTH FORK LENGTH (cm) Figure 21

Length frequency distributions (fork length in cm) of Exocoetus volitans presented at bi-monthly intervals.

•

• OCTOBER - NOVEMBER, 198'7 APRIL - HAY, 1988 1 a 1 d

e.e 0.e >- 0 >- z 8.6 0 8.6 w zw :l :l C1 C1 w 8.4 w 8.4 0: 0: 11. 11. 8.2 8.2

8 9 9 3 6 9 12 16 18 8 3 6 9 12 16 18 FORK LENGTH

DECEMBER, 198'7 - JANUARY, 1988 JUNE - JULY, 1988 e 1 b 1

0.8 9.8 >- >- 0 0 z 9.6 z 0.6 Ill Ill :l :l C1 C1 Ill 8.4 Ill 0.4 0: 0: 11. 11. 0.2 8.2

8 8 l1U ~~ 9 3 6 9 12 15 18 9 3 6 9 12 15 18 FORK LENGTH

FEBRUARY - MARCH, 1988 AUGUST - SEPTEMBER, 1988

1 c 1 f

0.e 8,8 >- >- 0 0z e.s z 8.6 Ill w :l :l C1 cr Ill 0.4 w 8.4 0: 0: 11. 11. 0.2 0.2

__,.__.....,.__L __ _._____.__ __ _.._""---! ~-~-~-· 8 a -t__ -~,~-·-·I 1-·~ .. l ______1 __ 0 3 6 9 12 15 18 9 3 6 9 12 15 18 FORK LENGTH (cm) FORK LENGTH

Length frequency distributions (fork length in cm) of Hirundichthys affinis presented at bi-monthly intervals. •

•

• • OCTOBER - NOVEMBER, 1981' APRIL - MAY, 1988

e.e e.e >- >- 0 z e.6 0z 9.6 w UJ :;) a :::)a w 9.4 w 9.4 0: 0: IL 11.. 8.2 9.2

a 9 e 4 8 12 16 28 24 e 4 8 12 16 20 24 FORI< LENGTH FORI< LENGTH

OECEMBER, 1987 - .JANUARY, 1988 .JUNE - .JULY, 1988 e • 1 b 1

e.8 9.8 >- >- (J 0 z e.6 z 9.6 UJ w :::;) a:::) cr w 0.4 UJ 0.4 0: 0: 11.. 11.. 9.2 0.2

e 0 --- L. • • e 4 B 12 16 20 24 0 4 8 12 16 20 24 FORK LENGTH FORK LENGTH

FEBRUARY - MARCH, 1986 AUGUST - SEPTEMBER, 1988

e. 8 0.8 >- >- 0 0z e.s z e.s Ill UJ :I :Ia cr UJ 9.4 UJ 0.4 0: 0: 11.. 11.. 0.2 0.2

0 0 .~.1~·__..___ .-.-...1-~.o--~··~....___.,_j~-~---"'.. .. 1-~~..L e 4 8 12 16 20 24 e 4 8 12 16 20 24 FORK LENGTH

Length frequency distributions (fork length in cm) of Oxyporhamphus • micropterus presented at bi-monthly .intervals. •

•

• OCTOBER - NOVEMBER, 19B7 APRIL - MAY, 1988

e.B e.e >- u u>- z e.e z e.s Ill w ::J :I 0' 0' Ill 8.4 w e.4 rl rl IL u. 8.2 8.2

e lJ e e 3 8 9 12 16 e 3 8 9 12 16 FORK LENGTH (cm) FORK LENGTH (cm>

DECEMBER, 1987 - JANUARY, 1988 JUNE - JULY, 1988 b 1 1 e

e.e e.e >- >- u (.) z e.s z e.s w w :I ::J !3 !3 w 9.4 w rl e.4 u. rlu. 9.2 .JJ._ 0.2 e 0 J .. ~~~ .. 1. k•• _~~- e 3 6 9 12 15 e 3 6 9 12 15 FORK LENGTH

FEBRUARY - MARCH, 1966 AUGUST - SEPTEMBER, 1986 Cn=SS> 1 c 1 f

e.8 e. 8 · >- >- (.) (.) z e.s z 9.6 w w :I :I 0' cr w 9.4 w 9.4 0: 0: u. u. 9.2 9.2

. e ~ e __ ,_ .... ~ • ~~' e 3 6 9 12 15 e 3 6 9 12 15 FORK LENGTH 4. DIBClB9IOf

4 .1. ~ITION AND ABUNDANCE OF FI.DTSAM

Flotsam components off Barbados have not previously been

quantitatively surveyed. 'lbe components of the flotsam collected in

I this study were of coastal (marine) and terrestrial origin. This

suggests that the material originated on or arotmd the island, and/or

that it drifted into the area with water from South American rivers • (Borstad, 1979; Mesde et al., 1979, 1983). Note that no Sa.rgassl.ID was collected. This contrasts with previous studies and observations

suggesting its presence (Brown, 1942; Hall, 1955). However, the t observed absence of Sargasst.Dil arotmd Barbados is consistent with the

large-scale surveys conducted by Parr ( 1939) , Stoner ( 1983) 8IKi Butler

et al. ( 1983) • • 'Ibe most striking results of the flotsam sttrly are the lildted

quanti ties of flotsam collected and the rarity of occasions on which

I eggs were found. ~1ly 38% of all tows contained flotsam, perhaps suggesting a patchy distribution. When flotsam was sampled, the

abundance was low and the components were fragmented. The fragmented

nature suggests considerable disintegration. 'lbe low abundance may

partly result from rapid sinking, and flotsam could therefore be more

abundant at greater depths.

t Only 1% of tows contained flyingfish eggs. The scarcity of eggs

may result from an overall scarcity of spawning substrates near

Barbados. Alternatively, the scarcity of eggs and substrates near the

surface, and hence vulnerable to the gear, may be because eggs are laid

profusely on all available substrates causing both to sink. Whether

121 sinking of substrate COIIIIlOnly occurs during flyingfish spawning, and

¥bether this affects hatching success and viability of hatched larvae

is unknown. Whatever the cause of the surface scarcity of flotsam, it

may suggest that substrate availability limits flyin,gfish spawni.ng near

Barbados. 'lbe fact that cane trash used by ooomeroial fishermen as fish

attracting devices is frequently covered with eggs after only a few

hours of soak, is consistent with the hypothesis of limited spawning

substrata. It certainly suggests that ripe fish are present in the

area, can spawn opportl.mistically, and are acutely sensitive to the

presence of potential spe.wnin,g substrate.

To maintain current population sizes in the face of the scarcity of

' surface flotsam observed, either flyin,gfish are spawning on floating

substrates found elsewhere, or spawning success is not dependent on

floating material, i.e. spawning may be occurring on submerged

substrates or on the bottom. The first suggestion is the more

orththodox perspective, but a large-scale survey conducted during the

breeding season in the eastern Caribbean produced negligible quantities

of flotsam and few flyingfish eggs (Oxenford, 1988b). Whether

flyingfish can utilize sul:merged substrates or spawn on the bottom,

perhaps on seagrasses, in shallower areas requires further

investigation.

Seasonal variation in the abundance of flotsam off Barbados has not

previously been investigated. Flotsam was present throughout the year,

but oocurred more frequently and was most abtmdant between March and

September. This corresponds with the period when South Atlantic water

mixes with fresh water from the rivers on the north coast of Souti1

America and enters the area around Barbados as the Guiana Current ( Lewis

122 et al., 1962; Borstad, 1982a, 1982b). The correspondence of high flotsam

~bundance with the presence of this water mass near Barbados is

consistent with the composition of the flotsam observed, i.e. of

terrestrial and coastal rather than oceanic oriain. It sl.IGCests that a

considerable portion of the flotsam may be transported to Barbados fraa

the South American region. The period of high flotsam abundance

overlaps but is out of phase with the period of availability of

flyingfish near Barbados (December-June; Mahon et al., 1986), and with

the spawning period of flyingfish near Barbados (December to May;

Storey, 1983; Khokiattiwong, 1988). • It is of interest that the scarcity of flotsam and the rapidity with which flyingfish spawn on fish aggregating devices may suggest that

spawning substrata is limiting, yet flyingfish do not appear to be

spawning when surface flotsam is most abundant. However, caution is

required in this interpretation. Using seasonal variation in the

abundance of flotsam to co1010ent on whether variation in spawning

substrata influences spawning seasonality in flyingfish is complex. If

spawning oocurs when flotsam is most available, one might predict a

positive correlation between flotsam abundance and spawning effort.

Alternatively, if eggs cause flotsam to sink, one might predict a

negative correlation between flotsam abundance in surface waters and

spawning effort. However, neither a positive nor a negative correlation

between flotsam abundance on surface waters and spawning effort was

detected in this sttdy. 'Ihis may again imply that the availability of

surface spawning substrata off Barbados does not influence the spawning

success of flyingfish. It therefore emphasizes the need to investigate

whether flyingfish can spawn on submerged and/or bottom substrates.

123 4.2.

Larvae of 34 families were identified as beina present in day

neuston tows and 24 as being present in niah.t neuston tows durina the

study. They ranaed from oceanic families such as myctophids,

istiophorids, to offshore but less oceanic families, such as

hemir&m}:itids and exocoetids, to coastal families such as 11l1..1Cilids and

mullids. Hemil"8JD~ilids ( 46% of day catch) and myctoJitids (51% of ni.@t

catch) dominated day and niaht samples respectively. Hemii'81D}ilids are

surface-dwellers (Fischer, 1978). They are not coomeroially important J in Barbados, but are used as bait for large pelagic fish (e.g. dol,Ptin,

tuna) elsewhere in the Caribbean (Fischer, 1978). Myctophids are

primarily oceanic. They are known to migrate to the upper surface

layers at night (Fischer, 1978, 1981), which explains their abundance in

night samples in this stujy. Myotophids are not exploited coomeroially

in Barbados or elsewhere in the Caribbean, but the potential of the

group as a protein source is being assessed in South Africa and Russia

(Fischer, 1978, 1981). Commercially exploited families which were

collected in the neuston tows include the exocoetids (7% of total

catch), carangids (3%), scombrids (3%), istiophorids (2%) and

corypha.enids (2X).

In an earlier extensive icthyoplankton survey, Powles (1975)

collected no hemiramphids off Barbados. '111is may be because he used

bongo nets, whereas neuston nets were used in this sttrly. Richards

(1984) used both types of nets in his survey in the Caribbean, and also

found no hemiramphids in his bongo net collections. This difference is

a strong reminder of the selectivity of gear used in icthyopla.nkton and

124 n.euston surveys. Given this selectivity, it is inappropriate to compere

'the n.euston composition in the present study with catches made by

Richards ( 1984) and Fa.hay ( 1975) , since the three studies used different

gears. However, preliminary data on the composition of larvae ca.ught in

a recent eastern Caribbean survey, which used the same gear as the

present study, show a predominance of hemiramphids and myctophids for

day and night samples respectively (Oxenford, 1988b).

Apart from differing markedly in taxonomic COIII(X)Sition, day and

night samples differed in larval size. Day larvae ranged from 0-20nm,

with a peak around lOom. Night la.l"Vae were la.l'ger, one size group ' ranging from 0-40mm and a set...""'Ond group ranging from 80-lOOnm. The size differences between day and night larvae result primarily from

differences in taxonomic composition of the neuston by day and night.

t-tyctophid and leptocephalus larvae, which were abundant at night, are

larger than. the lat"Vae conmonly caught by day (e.g. hemiramphids aud

exocoetids). The size differences may also be influenced by day and

night differences in population size structure within taxonomic groups

and/or by variation in the size selectivity of the gea.1' by day and

night. Larger larvae, which may be able to visually detect and avoid the

net by day, may be less able to do so by night. In this context, it is

of interest that exocoetid lat"Vae caught in the night were larger than

than those in the day. However, for scombrids, lat"Vs.e caught in the

day were larger than those caught in the night.

Apart from differing in taxonomic composition and larval size, day

and night samples differed significantly in larval abundance, higher

catch rates oc:curring in the day than in the night. This pattern

differ:" from that of most neuston surveys (e.g. Parin, 1972; Fahay,

125 •

1975). It may partly be the consequence of the time of sampling. Over

a 24-hr period, neustonic larvae are usually most abundant at dawn and

after sunset (Parin, 1967; Hempel and Weikert, 1975). Abundance

typically declines at noon and midnight. Most day samples in the

present study were collected between 13:00 and 17:00 hr, i.e. after the

period of minimum day abundance. Night tows were taken between 23: 00

and 0: 300 hr, which overlaps with the period of minimum night abundance.

However, it should be noted that for some of the families caught by

night and day, catch rates are expected to be higher in the day. For

example, the higher catch rates of exocoetids in the day were expected,

since flyingfish larvae are known to concentrate at the surface by day

and move to slightly deeper water at night (Hempel and Weikert 1975).

Since the timing of day and night sampling was similar throughout the

year, diurnal variation in larval abu:ndru"l.ce at the surface should not

confound patterns of seasonal variation in larval abundance.

4. 2. 2. Seasonal variation in ahmdance

Since only one season of data was collected in this study, the

assumption that any seasonal pattern observed will be repeated through

years should be treated with caution. Seasonal changes in fish larval

abundance may result primarily from seasonal changes in spawning

activities of adults (e.g. Robertson et al., 1988; Hunte and Cote,

1989). Since many tropical marine fish spawn throughout the year, but

with a seasonal peak in activity (e. g . Munro et al. , 197 3 , Tupper,

1988) , fish larval abundance may be expected to vary seasonally.

However, using plankton nets other that neuston nets, seasonal variation

in larval abunda.nce was not detected off Barbados by Lewis t"J:c al.

(1962)' Lewis and Fish (1969) or Sander (1971). By contrast, using

126 bongo nets, Powles ( 1975) fmm.d seasonal variation in larval abundance

·of several nearshore species (e.g. Labridae, Scaridae, Carangidae,

Poma.centridae, Apogonidae and some Serranidae). For these families,

there were typically two periods of high abundance, March to May and

August to September. Note that these periods fall within the time

(March to September) that the less transparent, greener and more filotsam

rich water mass from the South American mainland is present near

Barbados (Borsta.d, 1982a, 1982b; Section 4.1). By contrast, Powles

(1975) found no seasonal variation in abundance of offshore fish larvae.

The present study is the first investigation of seasonal variation ' in fish larvae from neuston tows off Barbados. Significant seasonal variation in abundance was observed in both day and night samples for

all larvae combined and separately for most of the cOlllllOn families. In

both day and night samples, larvae were most abundant between February

and June. Two basic patterns emerged when families were considered

separately. Larvae were either most abundant between February and Ju:ne,

or least abu:ndan.t between February and June. This invites the

speculation that, in terms of larval release, advantages may accrue

from partitioning the environment on a seasonal time scale. Families

collected primarily between February and Jtute were the m.vctophids (51%

of the night catch), hemiramphids (46% of the day catch), exocoetids and

dactylopterids, which are largely offshore but not fully or exclusively

oceanic, and the llllJgilids , which are primarily coastal. Families which

were least abundant between February and Jnne were the larger oceanic

scombrids and istiophorids, and the coastal carangids and mullids. It

is of interest that the myctophids and hemira.mphids, which together

comprise almost 50% of the total larval catch and which are both most

127 abundant between February and June, may be partitioning the upper

.surface environment by day and night.

TI1e sea.sonality of spawning for most of the fish families whose larvae were collected in this study is unknown. Consequently, it is not possible to correlate the timing of peak la1~al abm~ce with that of adult spawning. The exception is adult exocoetids, particularly

Hirundichthys affinis, in which peak spawning activity is known to occur primarily between December and May (I..ewis et al., 1962; Storey, 1983;

Kh.okiattiwong, 1988). The observed seasonal va1·iation in abundance of exocoetid larvae {February to June) therefore corresponds well with the seasonality of spawning reported for adults.

4. 2. 3. Spatial variation in abundance and size

For all larvae combined, and separately for the myctophids, hemiramphids, dactylopterids, exocoetids ( in day samples) , and carangids

(in night samples) , there was significant variation in larval abundance between stations. In all cases where significant variation was detected, larvae were less abundant at Station 1 (3 nmi offshore) than at Station 2 and/or Station 3 (6 and 9 nmi offshore respectively). For myctophids and exocoetids, which are found relatively far offshore throughout their life cycle, tl1is pattern of larval distribution may be expected on the basis of adult distribution. However, it is less expected on this basis for hemiramphids and dactylopterids in which adults are found closer to shore, and not expected for ea1·angids, in which a.dul ts are often coastal. TI1e high larval abundance of these families offshore may therefore largely result from current patterns, perhaps augmented by larval fooo distribution. Fish larvae and larval food niB.y be aggregated and retained in eddies created in the downcurrent

128 wake of the island by the east to west transport of water across

' t Barbados. Note that Powles ( 1975) found no evidence of larval transport from the Antillean islands to Bar'ba.dos via the easterly countercurrents

described by Ma.zeika ( 1973), and hence transport from the Antillean

islands is an unlikely explanation for the high larval abundance

offshore of Barbados. Based on island-specific pa.tterns of population

recovery in the sea urchin Diadema antillarium, Hunte and Younglao

( 1988) have suggested that larvae are retained downcurrent ·of islands

and primarily recruit back to their natal populations. Moreover, Hunte

and c8te (1989), Hunt von Herbing ( 1988) and 'fupper (1988) suggest that

the retention of larvae near Barbados, with subsequent recruitment to

natal populations, may explain why reef fish populations in Barbados are

not "recruitment-limited", but limited by space availabiltiy on reefs.

For 8 of the 10 common families collected (exceptions the scombrids

and myctophids) , larval size varied significantly between stations.

This was detected in spite of the prestuned size selectivity of the

sampling gear. For species with nearshore or coastal adults, a gradient

of increasing larval size offshore might be expected, since larvae must

be released near shore. However, older/larger post-larvae and juveniles

of coastal species must migrate back towards shore, thereby disrupting

a simple offshore size-gradient. For offshore and oceanic species, an

onshore/offshore gradient in larval size might not be expected.

However, the present results suggest that, regardless of where adults

reside, larvae tend to be larger at offshore than nearshore stations.

Interestingly, the pattern of horizontal distribution of larval size

closely reflects that of larval abundance. In moat cases, larvae are

largest and most abundant at Station 3, smallest and least abundant at

129 Station 1; with differences between Stations 3 and 2 being less marked.

!he horizontal distribution of larval size is therefore consistent with the hypothesis that larvae are retained by eddy systems with a strong effect some 6-9 nautical miles downcurrent of the island (see POwles,

1975). Retention of larvae in that vicinity would lead to both higher larval abundance and larger larval size through growth.

4.3. COMPOSITION, ABUNDANCE AND SIZE OF JUVENILE FLYINGFISH IN

NIGIITLIGin' SAMPLES

4. 3. 1. Gear selectivity and catch cauposition

The total number of flyingfish captured in the 1-year ni.ghtl:ighting study was 2,834. This was 74% of the flyingfish observed to enter the effective fishing area of the illuminated zone. The efficiency of capture was not a function of the availabilty of fish in tJ1e zone.

Consequently, the nlDD.ber of fish caught was considered an adequate index of fish availability in the zone.

The m.111ber of fish in the illuminated zone appeared to be lowest in tJ1e third hour of sampling trips. However, given the variation in fish availability between sampling trips, the variation between hourly intervals across trips was not statistically significant. Declining fish availabilty in the third sampling hour might indicate local depletion of fish in tJte vicinity of the sampling vessel, but could also result from nocturnal variation in the tendency of fish to be attracted by the swnpling procedure. Given that the swnpling vessel was moved 0. 5 nmi between sampling hours, the latter may be the more plausible explanation for lower catch rates in the third hour. It is therefore imperative tltat nighlight studies standardize tlte time of night over

130 which sampling occurs, if comparisons of catch oomposi tion and abundance

Qf catch across months are to be meaningful. In this study, sampling

was always conducted between 18:00 and 23: 00 hours.

Although the nlllnber of fish captured may be an appropriate index of

the nlllnber of fish in the illlllninated zone, the fish in the zone are

tmlikely to bE; a representative subsample of all fish present in the

vicinity of the vessel. Specifically, the sampling procedure is likely

to be seleotive by species, and within species, by size. Gulland ( 1983)

has emphasized the need to consider these aspects of seleotivi ty when

sampling fish by light attraction, since phototactic behavior may vary

with species and size.

Parexoooetus brachypterus, Exocoetus voli tans, Hirundichth:ys

affinis and Oxyporhamphus micropterus dominated the nightlight catch.

The presence of P. brachypterus and R· affinis was expected, since

juveniles of these species have previously been reported off Barbados

(Lewis, 1959, 1961; Lewis et al., 1962). However, ~· volitans and Q. micropterus have not previously been reported in nightlight catches from

this area.. The absence of ~. vol i tans and Q. micropterus from the

rep:>rts of Lewis ( 1959, 1961) and Lewis et al. (1962) may indicate

either that these species were captured but not recorded, or that samples were taken at times of night when these species are not attracted by the sampling procedure. The former may be the more plausible explanation.

Both juveniles (60% of catch) and adults of £. brachypterus were caught. The size range collected (2.3-12 cm) was similar to that re~Jrted by Lewis (1959, 1961; 3-12 cm). Similarly, for Q. micropterus, both juveniles (58% of catch) and adults were vulnerable to the sampling

1~1 procedure. By contrast, very few adults of !!· affinis, and almost no . adults of ~· volitans, were collected. This suggests strong size selectivity of the sampling procedure in these species. Adults of H· affinis support a commercial fishery in the eastern Caribbean, and consequently are known to be present in the area. Moreover, both Storey

(1983) and Khokiattiwong (1988) report adults of ~. volitans off

Barbados. The absence of H. affinis adults from the nightli.ght samples obtained by l.ewis et al. (1962), and from those found in the present study, support the suggestion of Nesterov and Bazanov (1986) that adults of this species avoid nightlights. The size range of H· affinis collected by l.ewis et al. (1962; 2-15 cm) was similar to the size range collected in this study ( 1 - 22 cm).

4. 3, 2, Seasonal. variation in a.btmdance and size

Length frequency distributions indicate that recruitment of juveniles to the population began in February - March in f· brachyPterus, ~. voli tans aud Q. micropterus; and December - January in

H. affinis. The monthly median size of fish in the juvenile cohort did not increase consistently with time for either R· affinis or ~. volitans. This suggests extended recruitment of juveniles and/or size selectivity of the sampling procedure against larger juveniles in these species. Interestingly, these are the species in which catches of adults are negligible. By contrast, monthly median size of fish in the juvenile cohort increased more clearly with time for both f· brachypterus and Q. micropterus. This may suggest a shorter recruitraent period and less size selectivity of the sampling procedure in tl1ese species. The suggestion of less selectivity is consistent with the observation that adults of these species are caught in relatively large

132 nl..lllbers •

'lbe abundance of adults of f. brachypterus and Q. lli.cropterus did

not vary significantly between months, but juvenile abundance varied

significantly in all four species. In f. brachypterus, i· volitans and

Q. mict•opterus, recruitment of juveniles began at the same time

(February - March); and juveniles remained abundant over a similar

period (March to July for f· braohypterus, February to August for i·

volitans, and February to July for Q. micropterus). 'Ibis suggests a

similar pattern of seasonal reproduction in these species. 'lbe spawning

season of f. brachyPterus has been suggested to be September to January

by Lewis (1961), March to August by Rhokiattiwong (1988). The present

results suggest that juveniles of f. brachypterus may grow at about 1 cm

per month, and juveniles are nearly 4 cm long at recruitment to the

nightlight catches in February - March (Section 3.3.6.; Figure 24a).

This suggests that the February - March recruits may have been spawned

in the previous October - November. The results are therefore more

consistent with the September to January spawning period of Lewis ( 1961)

than the f-fa.rch to August period of Rhokiattiwong (1988). It is of

course possible that the spawning seasonality of ~· brachypterus varies

between years, and likely that it varies between locations. For

example, Erdman ( 1976) suggested that the spawning of ~. brachypterus in

Puerto Rico occurs between December and April; slightly later than the

September to March proposed by Lewis (1961) for Barbados. Falmy (1975)

reported that larval and juvenile f. brachyPterus were most COIIIDOn in

the South Atlantic Bight between July and October; a period which would

be more consistent with March to August spawning as proposed by

Khokiattiwong ( 1988). There have been fewer studies of spawning

133 seasonality in g. volitans. However, Grudtsev et al. ( 1987) reported

that spawning occurred in the Atlantic primarily between <:Ctober ard

April. This would correspond, with about a 4-month lag, to the

occurrence of juveniles between February ard August observed in this

study.

In H· affinis, some recruitment of juveniles b!gan in December -

January, but juveniles became increasingly abundant from February

through to August. This corresponds well with th~e December to May

spawning season proposed for H· affinis in Barbados (e.g. Storey, 1983;

Khokiattiwong, 1988) , and with the observation that larval exocoetids

were most abundant between February and June (Section 3. 2. 5) •

4.3.3. Seasonality of Hinmdicht.bys affinis adults near lla.rbados

The conmercial fishery for H· affinis in Barbados is seasonal,

adults being captured between December and May. Current hypotheses for

the non-availability of adult H· affinis between June and November are

described by I010kiattiwong ( 1988) and are summarized below. One

hypothesis is that adults are present year-round in Barbados, but are

only available to the gill nets of the CODJJlercial fishery during the

spawning season when aggregated in schools. A second is that the period

of low abundance is an interval between adult cohorts; adult mortality

increasing shat·ply towards the end of the spawning season. Two observations deserve conment in the context of post-spawning mortality.

First, Barroso (1967) suggests that feeding of H· affinis decreases during spawning. Second, a traditional Barbadian phrase is "You look like a June fish!". It refers to a person looking jaded and in poor health. The third hypothesis for the seasonal non-availability of adult

H· affinis is that there is large-scale migration away from Barbados

1 ~A aJ."'tnld May 1 leading to the period of low abunr\a:nce.

n1e present results suggest that all life stages of H• affu1is are

sequentially present year-rotnld near Barb:u:los, and are therefore not

consistent with the hypothesis of large-scale migration away from and

towards the island. Peak s:.:awning of H· affinis occurs between December and May, peak larval abundance (of exocoetids) is between February and

June, juvenile abtn1da.nce rises from February to a peak near August, and small adults begin to be taken by the commercial fishery in November. nte results support the suggestion that the period of low abundance of adults (July to November) is primarily caused by an interval between adult cohorts. ID1okiattiwong ( 1988) reached a similar conclusion based on results obtained from fishing year-round with gill nets of varying mesh size. However, it should be noted that at least some fish of adult size are present in the traditional off-season for the fishery. Twelve of the 23 adults of H· affinis· captured in this study were collected in

July. lliese fish could either be fast growing individuals spawned early in the preceding December - May spawning period, or the remnants of the adult population which had spawned in December - May. Given the size of the adults (19 - 20 cm), the former suggestion seems more likely. The most plausible explanation for the seasonality of ti1e commercial fishery for H· affinis may therefore be that few adults are present between June and November, and that the few present are not spawning and are not therefore vulnerable to the commercial fishery. llie scarcity of adults results from post-spawning mortality and from the fact that few fish, spawned in the December - May period, have yet attained adult size.

1::15 5. LITBRATUI:m CITliD

Ahlstrom, E.H. 1965. Kinds and abundance of fishes in the California Current region based on egg and larval surveys. CALOOFI Report 10: 31-52.

Almeida, Nadja Urt M. De. 1966. Estudos sabre a ma.turidade do peixe voador ( Hirundichthys affinis - Gunther) na Costa Nordestina do Brazil. B. Est. Pesca. 6(3): 31-41. (English SUIIIDB.I'Y)

Atwood, D.K. 1976. Regional oceanography as it relates to present and future pollution problems on living resources - Caribbean. Paper for the IOC/FAO/UNEP International workshop on marine pollution in the Caribbean Sea and adjacent waters, Port of Spain, Trinidad, 13- 18 December, 1976.

Atwood, D.K., F.J. Burton, J.E. Corredor, G.R. Ha.rvey, A.J. Mata Jimenez, A. Va.squez-Botello and B.A. Wade. 1987/1988. Petroleum pollution in the Caribbean. Ocea.nus 30 ( 4) : 25-32.

Barroso, L.M. 1967. Biologica e pesca de peixe-volador (Hirundi.chthys affinis Gunther) no estado do rio Grande do Norte. B. est. Pesca 7: 9-37. (English summary)

Beers, J .R., D.M. Steve and J .B. Lewis. 1968. Primary productivity in the Caribbean Sea off Jamiaca and the tropical Atlantic off Barbados. Bull. Mar. Sci. 18(1): 86-104.

Ben-Yami, M., A. Herzberg, S. Pisanty and A. Lourie. 1970. A side tracking neuston net. Mar. Biol. 6: 312-316.

Bieri, R. and T.K Newbury, 1966. Booby II, a quantitative neuston sampler for use from small boats. Publ. Seta mar. Biol. Lab. 13(5): 405-410.

Bla.xter, J.H.S. (ed.), 1975. The ea1·ly life history of fish. Springer Verlag, Berlin.

Borsta.d, G. 1979. ASTP at Barbados: Mesosca.le pools of Amazon River water in the western tropical Atlantic. In: F. El Ba.z and D.M. Wa1ner ( eds. ) • Appollo-Soyuz Test Project Sl..liiiilery Science Report Vol.II, NASA SP 412, Washington, D.C.

Borstad, G. 1982a. The influence of the meandering Guia.na Current on the surface conditions near Barbados - temporal variations of Trichodesmium (Cyanophyta) and other plankton. J. Mar. Res. 40(2): 435-452.

Borstad, G. 1982b. The influence of the mearrlerin.g Guiana Current and Amazon River discharge on surface salinity near Barbados. J. Ha.r. Res. 40(2): 421-434.

Breder, C.H. 1938. A contribution to the life historiex of Atlantic Ocean flying fish. ]3ull. Bingham Oceanogr. Coll. G( 5) : 1-126.

1:1f) Brown, D.M., 1979. 'nle ma:nta net. Quantitative neuston sampler. Marine Technician's Handbook, Institute of marine Resources, Scripps Institution of Oceanography Refet'ence Series, San Diego.

Brown, D.M. and L. Cheng. 1981. New net for sampling the ocean surface. Mar. Ecol. Prog. Ser. 5: 225-227.

Brown, H.H. 1942. The sea fisheries of Barbados. BulL Develop. Welf. W. I. 1 : 1-42.

Brucks, J.T. 1971. Currents of the Caribbean and adjacent regions a.s deduced from drift-bottle studies. BulL Mar. Sci. 21: 455-465.

Bruun, A. F. 1935. Flying-fish (Exocoetidae) of the Atlantic: systematic and biological studies. Dana.-Report 6: 1-106.

Butler, J .N., B.F. Morris, J. Cadwallader and A. W. Stoner. 1983. Studies of Sarsassum and the Sargassum community. Bermuda Biological Station Special Publication No. 1, 307pp.

Caldwell, M.C. 1962. Development and distribution of larval and juvenile fishes of the family Mullidae of the western North Atlantic. U.S. Fish Wildl. Serv. Fish. Bull. 62(213): 403-457.

Calef, G. W. and G. D. Grice. 1967. Influence of the Amazon river outflow in the ecology of the western tropical Atlantic II. Zooplankton abundance, copepod distribution with remarks on the fauna of low salinity areas. J. Mar. Res. 25: 84-98.

Cheng, L. 1975. ltta.rine pleuston - animals at the sea-air interface. Ocew1ogr. mar. biol. A. Rev. 13: 181-212.

Collette, B.B., G.E. McGowen, N.V. Parin, and s. Mito. 1984. Beloniformes: Development w1d relationships. pp. 335-354. : r-toser, H. G. , W. J. Richards, D. M. Cohen, M. P. Fahay, A. W. Kendall, Jr, wld S.L. Richardson (Eds). Ontogeny and systematics of fishes. Amer. Soc. Ichthyol. Herpetol. Spec. Pub!. No. 1.

da Cruz, F. J. w1d M. Araujo. 1971. Notas sobre a alimentacao de Hirundichthys affinis (Gunther 1866), ao largo do Nordeste Bra.sileiro. Bol. Instit. Biol. Mar. Univ. F .R.G. Norte. 5: 31- 37. (English S\llliBary).

Da.rroux, P. and H. Oxenford. 1986. Dominica. pp. 35-38. In: Mahon, R., H. Oxenfrod, wld W. Hunte (eds.). Developnent Strategies for Flyingfish Fisheries of the Eastern Caribbean. International Develop.nent Research Agency, Ottawa, Cw18da.. IDRC-MR128e.

Brown, D.M. w1d L. Cheng. 1981. New net for sampling the ocean surface. Mar. Ecol. Prog. Ser. 5: 225-227.

David, P.H. 1965. The neuston net: a device for sampling the surfact~ fauna of the ocean. J. Har. Biol. Ass. U.K. 45:313- 320. de Sylva, D.P. 1984. Mugiloidei: develor:ment wld relationships. pp.

137 530-533. In: Maser, H.G., W.J. Richards, D.M. Cohen, M.P. Fahay, A.W. Kendall Jr., and S.L. Richardson (eds.). Ontogeny and systematics of fishes. Amer. Soc. lcthol. Herpetol. Spec. Pub!. 1.

Dooley, J.K. 1972. Fishes associated with the pelagic sargassum complex, with a discussion of the sargassum community. Contr. Mar. Sci. 16: 1-32.

Doty, M.S. and Oguri, M. 1956. The islard mass effect. J. Cons. pe1~. int. Explor. Mer 22: 33-37.

Ellertsen, B. 1977. A new apparatus for sampling surface fauna. Sal~ia: 113-114.

Emery, A.R. 1964. The ecological effects of current patterns around islands with special reference to Barbados. M. Se. 'l,.tesis, McGill University, 145p.

Emery, A.R. 1972. Eddy formation from an oceanic island: Ecological effects. Carib. J. Sci. 12(3-4): 121-128.

Erdman, D.J. 1976. Spawning patterns of fishes from the northeastern Caribbean. Conmonwealth of Puerto Rico, Dept. Airic. and Fish. Contr. 8(2).

Evans, J. W. 1961. Normal stages of the early developnent of the flyingfish, Hirundichthys affinis (Gi.inther). Bull. Mar. Sci. 11 ( 4) : 483-502.

Fahay, M.P. 1975. An annotated list of larvae and juvenile fishes captured with surface towed meter net in the South Atlantic Bight during four "Dolphin" cruises between May 1967 and February 1968. NOAA Tech. Rep. NMFS SSRF 685, 39p.

Fahay, M.P. 1983. Guide to the early stages of marine fishes occurring in the northwestern Atlantic Ocean, Cape Hattera.s to the southern Scotian Shelf. J. Northwest. Atl. Fish. Sci. 4.

Febres-Drtega, G. and L.E. Hen-era. 1976, Caribbean Sea circulation and water mass transports near the Lesser Antilles. Bol. Inst. Oceanogr. Uni v. Oriente 15: 83-96.

Fine, M.L. 1970. Fauna! variation in pelagic Sargassum. Biol. 7(2): 112-122.

Fischer, w. (ed.). 1978. FAO species identification sheets for fishery purposes; Western Central Atlantic (Fishing area 31). Vols. 1-7. Food and Agricultural Organization, Rome. pag. var.

Fischer, W., G. Bianchi and W.B. Scott (eds). 1981. FAO species identification sheets for fishery purposes: Eastern Central Atlantic (Fishing area 34 and 47-in part). Vols. 1-7. Food and Agricultural Organization, Rome. pag. var.

Froelich, P.N. Jr., D.K. Atwood, and G.S. Giese. 1978. Influence of Amazon River discharge on surface salinity and dissolved silicate concentration in the Caribbean See.. Deep-sea Res. 25: 735-744 •

. Gade, H.G. 1961. On some oceanographic observations in the southeastern Caribbean Sea and the adjacent Atlantic ocean with special reference to the influence of the Orinoco River. Bol. Inst. OcewlOSl'• Univ. Oriente 1: 289-342.

Gibbs, R.H. and J.C. Staiger. 1970. Eastern tropical Atlantic flyingfishes of the genus -cypselurus (Exocoetidae). Stud. Trop. Ooeanogr. 4(2): 432-466.

Gill, T. 1903-1904. Flyingfishes and their habits. P• 495-516. Smithsonian Inst. Annual Report 1903-1904.

Gorelova, T.A. 1980. The feeding of young flyingfishes of the family Exoooetidae and of the smallwing flyingfish, OxyPorhamphus mic1~pterus, of the family Hemirhamphidae. J. Ichth. 20(4): 60- 71.

Grudtsev, M.E., L.P. Salekhova and V.G. Lushohina. 1987. Distribution, ecology wld intre.specific variability of flyingfish of the genus Exocoetus of the Atlantic Ocean. J. Ichth. 27 ( 1 ) : 39-50.

Guillou, A. wld H. Oxenford. 1986. Ma.rtinique. pp. 31-34. In: Ma.hon, R. , H. Oxenford and W. Hunte ( eds. ) • Developnent strategies for flyingfish fisheries of the eastern Caribbean. International Developnent Research Agency, Ottawa, Canada. IDRC-MR128e.

Gulland, J.A. 1983. Fish stock assessment; A manual of basic methods. Wiley wld Sons, New York: 223pp.

Hall, D.N.F. 1955. Recent developments in the Barbadian flyingfish fishery and contributions to the biology of the flyingfish, Hirun.dichthys affinis ( Gi.inther 1866) . Eng. Colonial Office, Fish. Publ. No. 7. 41 pp.

Harding, A. 1986. Past developnent and present techniques in the Barbados flyingfish fishery. pp. 27-28. !!!: Mahon, R., H. Oxenford and w. Hunte (eds.). Development strategies for flyingfish fisheries of the eastern Caribbean. International Developnent Research Agency, Ottawa, Canada.. IDRC-MR128e.

Hartmwm, J. 1970. Verteilung und Nahrung des Ichthyoneuston in subtropischen Nordostatlantik. 'Meteor' Forsch. Ergebn., Ser. D. No. 8. 1-60. (English SUilllla.rY) •

Hebut'll, G.W., T.H. Killder, J .E. Allender and H.E. Hulburt. 1982. A numerical model of eddy generation in the southeastern Ce.ribbew1 Sea. p. 299-328. In: J. C. J. Nihoul ( ed. ) Hydrodynamics of Semi Enclosed Seas. Elsevier, Amsterdam.

Hempel, G. and Weikert, H. 1972. The neuston of the subtropical wld boreal Northeastern Atlantic Ocean. Alm. review. Har. Biol. 13: 70-88.

Hettler, W.F. 1979. Nodified neuston net for collecting live and

139 larval and juvenile fish. Progye. Fish Cult. 41(1): 32-33.

,Hinton, G.C.F. and Boney, A.D. 1979. An integrating sampler for quantitative ecological studies on surface waters. Meeresfosch. 27: 48-51.

Houde, E.D. and N. Chitty. 1976. Seasonal abundance and distribution of zooplankton, fish eggs and fish larvae in the eastern Gulf of Mexico, 1972-1974. NOAA Tech. Rep. NMFS SSRF-701: 18p.

Hunt von Herbing, I. 1988. Reproduction and recruitment in the bluehead wra.sse Thalassoma bifa.sciatum in Barbados. M. Se. Thesis, McGill University, 141pp.

Hunte, W. and I. Cote. 1989. Recruitment in the redlip blenny Qphiobletmius atla.nticus: is space limiting? Coral Reefs (in press).

Hunte, W. and H. Oxenford. 1986. The economics of host size in the Barbados pelagic fishery. Proc. Gulf Ca.rib. Fish. Instit. 39.

Hunte, W. and D. Younglao, 1988. Recruitment and population recovery of Diadema antillarium (Echinodermata; Echinoidea) in Barbados. Mar. Ecol. Prog. Ser. 45: 109-119.

Imai, S. 1959. Studies on the life histories of the flyingfishes found in the adjacent wates of Japan. -I. Mem. Fac. Fish. Kagoshima Uni v. 7 : 1-85 , (English sUliillary)

Imai, S. 1960. Studies on the life histories of the flyingfishes found in the adjacent waters of Japan. -II. Mem. Fa.c. Fish. Kagoshima Uni v. 8: 8-45 • (English sunmary)

Joha.tmessen, O.M. 1971. Preliminary results of some OL"'ea.tlogra.phica.l obset~tions carried out between Barbados a.tld Tobago, ~~h/April, 1968. Sym)Xlsium on Investigations a.t1d Resom~s of the Caribbea.tl Sea and adjacent waters. pp. 95-112.

Johnson, D.L. and P.L. Richardson. 1977. On the wind-induced sinking of Sarga.ssum. J, exp, mar. Biol. Ecol. 28: 255-267.

John, H-Ch. 1975. Description of a two-net neuston-sampler after the principle of Sa.meoto and Jaroszynski. Ber. dt. wiss. Kommn. Meeresforsch. 24: 342-344. (English SU1111l8.ry).

John, H-Ch. 1983. Quantitative distribution of fry of Beloniform fishes in the Atlantic Ocean. 'Meteor' Forsch. -Ergebn. D. 36: 21-33.

Jones, R. and H.A. Oxenford. 1986. Barbados. pp. 23-26. In: Ma.hon, R., H. Oxenfrod, a.tld W. Htmte (eds.). Developnent Strategies for Flyingfish Fisheries of the Eastern Caribbean. International Developnent Research Agency, Ottawa, Canada IDRC-t-1R128e.

Jordan, C.!'-1. 1983. The flyingfish industry (Tobago): A continuing case study. Proc. Gulf Caribb. Fish. Instit. 36: llpp.

140 Karaman, I. I. 1980. On the quantitative distribution of flyingfish of the family Exocoetidae in the Indian O:Jean. J, Icthy, 20(5): 155-157.

Kendall, A.W. Jr., E.H. Ahlstrorn and H.G. Maser. 1984. Early life history stages of fishes and their characters. W• 11-22. In: Maser, H. G. , W. J . Richards, D. M. Cohen, M. P. Fahay, A. W. Kendall , Jr, and S.L. Richardson (Eds). Ontogeny and systematics of fishes. Amer. Soc. Icthyol. Herpetol. Spec. Pub!. No. 1.

Khachaturov, V .A. 1983. The digestive system of flyingfishes (Exocoetidae, Beloniformes). J. Icthyol. 23(2): 106-113.

Khokiattiwong, S. 1988. Seasonality and abundance of Hinmdichthys affinis and Parexocoetus brachypterus. M. Se. Thesis, McGill University, 152pp.

Kidd, R. and F. Sander. 1979. Influence of Amazon River discharge on the marine production system off Barbados, West indies. J. Mar. Res. 37(4): 669-681.

Kovalevskaya, N. V. 1964. Study of developnent of the "two-winged" flyingfishes of the genus Exocoetus ( Exocoetidae, Pisces) • Tr. Inst. Okeanol. Akad. Nauk SSSR 73: 204-223. (English SUIIID8.I'y)

Kovalevskaya, N. V. 1978. The distribution characteristics of larvae and fingerlings of the smallwing flyingfish Oxyporhamphus micropterus (Val.) and the flyingfihs Exocoetus volitan.s Linne in the region of equatorial currents of the Pacific O:Jean. O:Jeanol. 18(1): 88-90.

Kovalevskaya, N.V. 1982. Superfluous reproduction and developnent of flying fish of the family Exocoetidae. J. Ichth. 22 ( 4) : 48-54.

Langmuir, I. 1938. Surface motion of water induced by wind. Science 87: 119-123.

Leis, J.N. and Rennis, D.S. 1983. The larvae of Indo-Pacific coral reef fishes. Australia, N. S. W. , University Press.

Leming, T.D. 1971. Eddies west of the southern Lesser Antilles. Symposium on Investigations and Resources of the Caribbean Sea and Adjacent Regions. pp. 113-130.

Lewis, J .B. 1959. Note on a teclmique for catching flyingfish at night. West Indies Fish. Bull. 6: 9-id.

Lewis, J.B. 1961. The growth, breeding cycle and food of the flyingfish Parexocoetus brachYPterus hillianus (Gosse), Bull. Mar. Sci. 11(2): 258-266.

Lewis, J.B. 1964. Tagging experiments on the flying fish Hirundichthys affinis (Gunther). Bull. Har. Sci. Gulf Carib. 14(3): 381-386.

Lewis, J.B. and A.G. Fish. 1969. Seasonal variation of the zooplankton fauna of surface waters entering the Caribbean Sea at Barbados. Carib. J. Sci. 9(1-2): 1-24.

I:.ewis, J .B., J .K. Brundritt, and A.G. Fish. 1962. The biology of the flyingfish, Hirundichthys affinis (Gunther). Bull. Mar. Sci. 12(1): 73-94.

Mahon, R. 1986. Seasonal and interannual variability in abwxiance of flyingfish. PP• 89-123. In: Mahon, R., H. Oxenford, and W. Hunte (eels. ) • Developnent Strategies for Flyingfish Fisheries of the Eastern Caribbean. International Development Research Agency, ottawa, Canada IDRC-MR128e.

Mahon, R., H. Oxenford, and W. Hunte (eds.). 1986. Development Strategies for Flyingfish Fisheries of the Eastern Caribbean. International Developnent Research Agency, ottawa, Canada IDRC MR128e.

Mahon, R., W. Hw1te, H. Oxenford, K. Storey, and R.E. Hastings. 1982. Seasona.li ty in the COITillercia.l marine fisheries of Barbados • Proc. Gulf. Caribb. Fish. Instit. 34: 28-37.

~fa.tarese, A.C. and E.l-1. Sandknop. 1984. Identification of fish eggs. p. 27-31. In H.G. Maser et al. (ed.) Ontogeny and systematics of fishes. Amer. Soc. Ichtyol. Herpetol. Spec. Publ. No. 1. f>Jazeika, P.A. 1973. Circulation and water masses east of the Lesser Antilles. Deutche. Hydrogr. Zeit. 26: 49-73.

Mazeika, P.A., D.A. Burns, and T.H. Kinder. 1980. Mesoscale circulation east of the.southern. Lesser Antilles. J. Geophys. Res. 85(5): 2743-2758.

Meade, R.H., C.F. Nordin, F.M. Costas Rodroguez, C.M. DoVale and J.M. Edmond. 1979. Sedimant loads in the Amazon River. Nature 278: 161-163.

Meade, R.H., C.F. Nordin, Jr., D. Perez Hernandez, B. Abel Mejia and J .r-1. Perez Godoy. 1983. Sediment and water discharge in Rio Orinoco Venezuela and Columbia. pp. 1134-1144. In: Proc. Sec. Int. Symp. River Sedimentation, Water Resources: and Electric Power Press, Beijing, China.

Methot, R.D. 1986. Frame trawl for sampling pelagic juvenile fish. CALCOFI Rep. 27: 267-278.

Michanek, G. 1975. Seaweed resources of the ocean. FAO Fish. Tech. Pap. 138: 127p.

Miller, .J.N. 1973. A quantitative push-net system for transect studies of larval fish and macrozooplankton. Linu1ol. Oceanogr. 18: 175-178. f>filler, J.I1L, W. Watson and J.t-1. Leis. 1979. Atlas of conunon nearshore marine fish larvae of the Hawaiian Islands. University of Hawaii, 179p. Monte, S. 1965. A contribution to the study of the flyingfish reproduction (Hirundichthys affinis) (Gunther 1866). Bol. Instit. Bioi. Mar. Univ. R.G. Norte. 2: 1-18. ( English sunmary)

Maser, H. G. , W. J. Richards, D. M. Cohen, M. P. Fahay, A. W. Kendall, Jr, and S.L. Richardson (eds). Ontogeny and systematics of fishes. Amer. Soc. Icthyel, Herpetol. Spec. Publ. No. 1. 760p.

Muller-Karger, F.E., C.R. McClain, and P.L. Richardson. 1988. The dispersal of the Amazon's water. Nature. 333(6168): 56-59.

Mulloney, B.C. 1961. a preliminary report on the results of tagging experiments on the flyingfish, Hirundichthys affinis Gunther. West Indies Fish. Bull. May/June: 1-4.

Munro, I.S.R. 1954. Eggs and larvae of the four-winged flying fish, Hirundichthys specuriger (Valenciennes). Aust. J. Mar. Freshwater Res. 5(1): 64-69.

Munro, J.L., V.C. Gaut, R. Thompson and P.H. Reeson. 1973. The spawning seasons of Caribbean reef fishes. J. Fish. Bioi. 5: 69- 84.

Nesterov, A.A. and M.Ye. Grudtsev. 1980. The quantitative distribution the flyingfish of the family Exocoetidae (Beloniformes) in the tropical Atlantic. J. Ichtb. 20(3): 137-140.

Nesterov, A.A. and S.I. Bazanov. 1986. Vertical distribution and behavior of flyingfish ( Exocoetidae) . J. Ichtch. 26 ( 5) : 159-162.

Nishikawa, Y. and D.W. Rinmer. 1987. Identifiaction of larval ttmas, billfishes and other Scombroid fishes (suborder Scombroidei} : an illustrated guide. CSIRO Mar. Lab. Rep. 186, 37pp.

Oxenford, H.A. 1988a. Preliminary tagging results on the four- wing flyingfish Hirundichthys !!ffinis in the Easthern Caribbean. Proc. Gulf Carib. Fish. In.stit. 41. (In press)

Oxenford, H.A. 1988b. Eastern Caribbean flyingfish project - abundance survey. Cruise Report, 16pp.

Parin, N. V. 1967. Diurnal variations in the larval occurrence of some oceanic fishes near the ocean surface.

Parin 1 N. V. 1 N. N. Gorbunova and V. M. Chuba.sov. 1972. Quantitative distribution of the near-surface ichthyoplankton in the eastern Pacific Ocean (based on trawls with a towed neuston otter trawl). Oceanology 12(6): 894-899.

Parin, N.V. and N.N. Gorbunova. 1964. On the reproduction and developnent of some synentognaths (Pisces, Beloniformes) from the Indian 0f;ean (based on material from the R/S "Vityaz"). Tr. Inst. Okean. Akad. Nauk SSR 73: 224-234.

tA..., Parr, A. E. 1937. A contribution to the hyd.rographiy of the Caribbean and Cayman Seas. Bull. Binsh.am Oceanogr. Coll. 5 ( 4 ) : 1-110.

Powles, H. 1975. Abundance, seasonality, distribution and aspects of the biology of some larval fishes off Barbados. Ph.D. Thesis, t-k:!Gill Univ., Montreal, 227p.

Richards, W.J. 1984. Kinds and abundance of fish larvae in the Caribbean Sea and adjacent waters. NOOA Tech. Rep. NMFS SSRF-776, 54p.

Roa, K.S. and S. Basheeruddin. 1973. Unusal catches of the flyingfish, Parexocoetus brachYPterus brachypterus (Richardson) in inshore waters at Madras. Indian J. Fish. 20: 629-634.

Robertson, D.R., D.G. Green and B.C. Victor. 1988. Temporal coupling of production and recr·ui tment of larvae of a Caribbean reef fish. Ecology 69(2): 370-381.

Ryther, J.H., D.W. f-1enzel, and N. Corwin. 1967. Influence of the Amazon River outflow on the ecology of the Western Tropical Atlantic. Part 1: Hydrography and nutrient chemistry. J. Mar. Res. 25(1): 69-83.

Sameoto, D.D. and L.O. Jaroszynski. 1969. Otter surface sampler. A new neuston net. J. Fish. Res. Board Can. 26(8): 2240-2244.

Sander, F. 1971. Organic production of inshore waters of Barbados: A study of the island-mass effect and it causes. Ph.D. Thesis. t-k:!Gill University, Montreal, Ca:nada. 151 pp.

Sander, F. and D.M. Steven. 1973. Organic production of inshore and offshore waters of Barbados: a study of the island mass effect. Bull. Mar. Sci. 23: 771-792.

Shiokawa, T. 1969. Study of the swimming habits of flyingfish in the spawning season. Proceeding of FAO Conference on Fish Behaviour in Relation to Fishing Teclmique and Tactics, Bergen, Norway, 19- 27th Oct., 1967, Vol. 3; 1969 FAO Fish. Rept. 62: 579-589.

Smith, D.G. 1979. Guide to the leptocephali (Elopiformes, Anguilliformes and Notacanthiformes). U.S. Dep. Commer. NOAA Tech. Rep. NMFS Circ. 424, 39p.

Smith, P.E. and S. Richardson.. 1977. Standard techniques for pelagic fish eggs and larval surveys. FAO Fish. Tech. Pap. (175): lOOp.

Staiger, J .C. 1965. Atlantic flyingfishes of the genus Cypselurus, with description of the juveniles. Bull. Mar. Sci. 15(3): 672- 725.

Steele, R. and H.A. Oxenford. 1985. Grenada. p. 11-18. ln: Ma.hon, R., H. Oxenford and W. Hw1te (eds.). Developnent strategies for flyingfish fisheries of the eastern Caribbean. International

-t A ;a Developnent Research Agency, Ottawa, Canada. IDRC-MR128e.

STSC, 1986. Statgraphics: statistical graphics system. User's guide. Statistical GraJitics Corporation, pag. var.

Steven., D.M. 1971. Primary productivity of the tropical western Atlantic Ocean near Barbados. Mar. Biol. 10: 261-264.

Steven, D.M. and A.L. Brooks. 1972. Identification of Amazon River water at Barbados, West Indies, by salinity and silicate measurements. Mar. ~iol. 14: 345-348.

Stoner, A.W. 1983. Pelagic Sargassum: an evidence for a major decrease in biomass. Deep Sea Res. 30(4A): 469-474.

Stoner, A.W. and W.S. Greening. 1984. Geographic variation in the macrofaunal associates of pelagic Sarsassum and some biogeographic implications. Mar. Ecol. Pros. Ser. 20: 185- 192.

Storey, K.W. 1983. Aspects of the biology and fishery of the flyingfish, Hirun.dichthys affinis, at Barbados. M.Pb. Thesis. University of West Indies, Barbados. 161 pp.

Sverdrup, H.U., Johnson, M.W., and Fleming, R.H. 1942. The Oceans: their physics, chemistry and general biology. Prentice-Hall Inc., New York. 1087 pp.

Taylor, W.R. 1960. Marine algae of the eastern tropical and sub-tropical coasts of the Americas. University of Michigan Press, Arm Arbor, Michigan, 870p.

Taylor, W.R. 1969. Notes on the distribution of west Indian marine algae in the Lesser Anti 11 es. Contributions from the University of Michigan Herbarium 9(2): 125-203.

Tupper, M. 1988. Recruitment and assemblage structure of reef fish in Barbados, W.I., M. Se. Thesis, McGill ~1iversity, 120pp.

Vatanachai, S. 1974. The identification of fish eggs and larvae obtained from the survey cruises in the South China Sea. IPFC Proceedings, 15th Session, October, 1972. Wellington, New Zealand. p. 111-130.

Vijayaraghavan, P. 1973. Studies on fish eggs and larvae from Indian waters. I. Development of egg and larvae of Hirun.dichthys coromandelensis (Hornell). Indian J. Fish. 20(1): 108-137.

Wal ters, H. and H. Oxenford. 1986. pp. 19-22. fu: Mahon, R. , H. Oxenford and W. Hunte ( eds. ) . Developnen.t strategies for flyingfish fisheries of the eastern Caribbean. International Developnent Research Agency, Ottawa, Canada. IDRC-MR128e.

Washington, H.B., H.G. Hoser, W.A. Laroche and W.J. Richards. 1984. Scorpaeniformes. pp. 405-428. In: Moser, H.G., W.J. Richards, D.M. Cohen, M.P. Fahay, A.W. Kendall, Jr, and S.L. Richardson (eds). Ontogeny and systematics of fishes. Amer. Soc. Ict.byol. Herpetol. Spec. Pub!. No. 1.

Woodcock, A.E. 1950. Subsurface pelagic Sargassum. J. mar. Res. 9: 77-92.

Zar, J .H. 1984. Biostatistical Analysis. Prentice-Hall, Englewood Cliffs, N.J. 781pp.

Zaitsev, Y.P. 1971. Marine neustonology. Naukov Dumka, Kiew. Eng. Translation IPST, Jerusalem: ~07p.

Zuyev, G.V. and V.N. Nikol'skiy. 1980. Procedure for the quantitative recording of flyingfish (Exocoetidae). J. Ichth. 20(6): 147-149.

'1/IC AppeOOix 1

Mean monthly values for all environmental variables monitored durine the study.

'l\lpJmlme Vial .. .. beijt .. state lib!r tlallpll!mJ i.ler oollr 111111 CXMr 1111 directb 11 rtJ 11--31 (•J (1--3) 1•1 (1--3) (l-6) (1-61

1 s.d. 1 s.d. I a.IL 1 a.d. I s.d. 1 s.d. 1 a.d. I a,d,

11'1 atdler 28.9'1 0.09 1.96 O.fl 1.40 0.50 1.69 0.56 21.91 3.56 &.00 0.00 3.33 u; 3.33 0.66 ~MEr •33 28.'13 9.&4 2.19 0.'12 us G.36 1.75 0.50 m.s& 1.40 Ul 0.51 3.36 0.92 3.8) 0.'18 ~ 45 28.1110.32 U60.54 1.40 0.42 1.98 0.61 23.39 3.86 1.27 0.4& 3.11 0.80 1.14 0.34 1988JIIIUf 31 %1.16 0.21 Lll 0.46 1.46 0.44 1.96 Ul 22.36 2.66 2.00 o.oo 3.23 0.1 2.1M18.28 rem., 33 26.'18 Ul U3 0.'18 1.65 0.51 1.99 0.59 m.331.63 z.oo o.oo 2.900.1 z.oo 0.00 !lard 39 1U5 O.ZS z.?A 0.64 1.40 0.45 1.82 0.60 Z0.33 1.68 2.00 0.00 LU 0.86 Z.IYI U6 Ajri.l 42 %1.29 o.l8 U2 0.62 1.32 0.2? 1.?3 0.49 ZO.ll Z.'lli 2.00 o.oo 1.34 0.1 2.19 0.39 lfa,f 39 28.00 0.36 2.28 0.60 1.15 8.23 1.53 0.64 18.91 3.42 2.25 0.60 2.61 0.88 2.46 Q.64 Jtlle 36 28.29 0.13 2.54 0.49 1.27 us 1.90 0.60 m.sa 2.33 2.33 0.48 3.381.10 2.61 Ul JulJ 42 28.48 0.28 2.?.1 0.5'1 1.21 0.23 l.f6 0.60 19.40 1.13 2.66 0.4'1 U'l0.14 2.04 O.Zl AICIBt 39 28.66 0.23 2.52 0.57 1.32 0.?.5 1.91 0.66 ~.~ %.41 1.91 0.28 2.33 0.98 2.35 0.70 ~ 11 28.'1& 0.19 &.36 0.50 1.34 0.46 1.90 D.TO 25.33 U1 1.83 O.Z5 2.00 0.00 3.25 0.45

IDlialB usal for uinal data: rm Clrullials hi Di:rldioo hi~ &a sta.m ilt.er lhlor

l c1s ratherly ek calJ bill! t sll;tt oorhterlJ mrate mlerate bloo creen 3 partly easb:!rly st:roiC atz