• Aucun résultat trouvé

it rti

N/A
N/A
Protected

Academic year: 2022

Partager "it rti"

Copied!
225
0
0

Texte intégral

(1)
(2)
(3)
(4)
(5)

SEASONALPAT T ERNS OF OIVERSITYIN MARINE FlS Il COMMUNI'nES

~1AUGUST INEJAPIENI MUNGKAJE , 13.Sc.[HOllS.J

At.nests submit tedto the Schoo l of ured uctcSludieHIn pa rti a l fulfil men t of the requi r e men t s for theflcgl'ceuL

Maste r of Science

Departmentof Biology MemorialUniversi tyof Newfoundla nd

June199 5

St . John's NQwf ou l lfJl 'll llJ

(6)

...

NallonalofCanadaLibrary

AcQuislhonsand DirectiOndes acquisitionset BibliographicServices Branch desservices bibliographiques

:I'Y..o W<'l4lol(jl()nSl'C<)l 395.rucW~arl

?:1~'(;~~O'1lnr., ?,IX~Onla"O)

The author has granted an irrevocable non-exclusive licence allowing the National Library of Canada to reproduce, loan, distribute or sell copies of his/herthesisbyany meansand in any form or format, making this thesis availab le to interest ed persons.

The author retainsownership of the copyright in his/herthesis.

Nei ther the thesisnor substantial extracts fromit may be printedor otherwise reproduced wIthout his/herpermission .

L'auteuraaccordeune licence Irrevocable et non exclusive permettant it la Blbllotheque nalionale du Canada de reprodulre,preter, dlstribuer ou vendre des copies desa these de quelque manlere et sous quelque forme que ce soit pour mettre des exemplaires de cette these

a

la disposition des personnes

rnteres sees.

L'auteurconservela proprletedu droit d'auteur qui protege sa these. NI la theseni des extraits substantiels de celle-cl ne doivent etre tmprlmes ou autrement reprodultssans son autor isation.

ISBN 0-6 12-13930_1

Canada

(7)

ABSTRAC T

This study used exploratory analysis to discover and describe seasonal patterns of diversity in marine fish communities.Manysmal l -scalestudios have sho wnthat seasonal patternsof diversitydo exist in fish co mmun i t ie s ; there have beennocomparisionstoin ve s t i g a t e la r g e-s c a l e patterns.

Thespecific objectivesof thiscompara t i v e study were to:(i) usepublisheddata setsto discoverseasonal pa t t e rn s of di.ve rsityin marinefishcommuni ties; (ii)describeth e s e seasonal patterns; (iii) separate general se a s o na l patterns fr o msp e c ific ones: (iv) develop further testable hypotheses about thepossiblecauses and mechanisms ofth e s epat t e r n s of diversity ; and (v ) di sc u s s these seasonal patterns and any relationshi psamongri c hn e ss, heterogene ityandeq ui ta b i l ity compone nts of diversity 1n order to en abl e research ers in applied fields such as fisheries man a gem ent and marine pollu t i on , where dive r stty stat i s tics are often used, to de s l gn and executemore sensitivetests.

Publishedfish catchdata setswere compiledan danalysed fo r seasonalpetterne in the di v e r s i t y ind ices S, H' and E which measu r e richness, heterogen eity an d eq uitabil ity, res pe c t i ve l y. Threeprincipal patternswere observed.

A br-ea d geogra p hica l patt e rn was that twomaj or seaso nal pea ksin the th ree indice s occurredin fi s h co mmunit i es from

11

(8)

higher lati t udes (> 4l~l O ' Nl wht Ie two to three suchpenk s were frequent at lower latitudes ($ 4l"l O' N) in the Nort h Ame ri c a n Eastcoe ee•The timeof occu.rr-enc e ofthefinlt peaks in H' and E sho wed a latitudi nal trend in thi s regi on; the fir stpe ak s in H' and Eoccurredearlierat lower Let t t.ud ea thanathigherlat i tudes .

Two major patternsof seaso nal track:l.m) in S, II' ilm l E observ edwe rean . in-p h a s e'(peaksinspecies numb er cofncfdu wi th pe ak s in H' and E) and an 'o ut- o f - phase ' pa t te r n (pe ak s do not coincide) . The •in-phase' pattern was mor epr-evaLont thanthe 'out-of- p hase ' patte rn,which typicallyro s uit.n Lr-om the influ x of lar ge numbe r s of one or two species. All addi tio nal pat t ern observedwasthat the mean s of S uno 11' withi n studi es were hig her inda t a sets wi t h ma rc spectos whil e thatof Ewa s highe rin those wi t hfe we r SP(1(:iCB.

Aut o c orre l a t ion ana lys is of tho tempo ral re sol u tio n of the peak s in the se a sonalpatte rn s snowed th ilt for ill] datil set stha t ha d "smcot h co rre.toqr ams',tnu timeInt.ervntbctweo n peaks was longe r tafive months)atla titude s;::: 33"37'Nthil ll atla t i tu des e33~37 ' N.

Two hy po t hese s were fo rmu l ate d from th ese ro su ttur t tl

"Plan kt o n and Fi sh Pheno log y " hypot h esis ; and (jJJ

"Instabi1i ty-Oo mi n a nc e" hypothesi s .Thepoten t ia l ap plt cot tcn oftbeue patterns and hypo the ses in fisheriesmana gement and marine poll ution moni tor i n garedi scussed .

i ii

(9)

ACKNOWLEDGEMENTS

I am greatly indebted to my supervisor, Professor David C. Schneider for his constant advice and encouragement. His expertise in quantitative methods enabled the easy discovery of patterns among large volumes of data. He is also thanked for proof-reading the manuscripts.'The advice and suggestions of Professors John M. Green ur,d Don Steele,as members of the supervisory committee. are gratefully eppz-ecLa t.ed ,

Associate Professor John c. Pe.rnetrte, formerly of the Biology Department, University of Papua New Guinea (UPNG) is thanked for hisencoureqement;s during the earlier stages of my pursuit for graduate studies and indoctrinating me into academia.

My sincere thai k you to Mr. Dan Genge of the Marine Science Research Laboratory, Logy Bdy for his computer skills in producing the graphs of diversity indices for the 22 data sets from which the in1 tial exploratory analysis and description of seasonal patterns were done.

The response of Or. Staffan Thorman of zoology Department, Uppsala University, Sweden to my initial requests for original data in sending three data sets from his co- authored work in arca Iven Estuary. and reprints of related publications was an encouraging sign at the start of this work. All authors listed in Table 81, Appendix 8 are

iv

(10)

acknowl edgedfo rtheir data used inthi sstudy ,obtai ne dtrom the ircee pect ive publi cati on s listedin the Refere nc es.

On behalf of my wife Martha and daughter crecu who accompani ed me in St. •10hn ' s for six months , and myse lf, I would like to thank th e followi n g families who orrerod ns Ln v e IuebLeassistance an d he lpdu r ingthisperi o d: Pro f ess or Sc h nei d e r , wife Bobbyandchil d r e n ; Mr.Bi l l Goo d ridge,wife Jan e t and chi ld r e n ; Mrs. DoreenWestera and fa mily; and Mr.

Cliff Br o wn, wife Janic e and children . Afterth ei r return to Pa p u a NewGui ne a, my fa mily we r e inthe re l ia ble car eof my brcrtbar Pi u s Mungkaje, wi f e Pat r i c ia and t:hi ld r e n , andnjcc c MariaMurk! and nephewJer ry Be i kum. My cou s inFranc i sDall18111 , wifeKeyan dch i l d re n as s isteduswi thtra velandaccomoetatton on a number of occa si on s. 'r he y ar e al l t hcnk e d [01- the,11 kindnessand hosp Lt aLf.tiy,

I have benefi ted sociall y , cul t u r a l l y and ecedentcejtv fo rminterac ti on swithmany Biologygra duate stud e nts du r In g my two years at Memorial University; espec ially John Christian, Daryl Jones , Rakesh Kuma r ,Hamz a r Sunu.si ,Jcn:ljfor Bates,ChristineCampb ell , Jo hn Horne, Prasad K.S.and Manu e l Gomes. 1 thank themfor the companions hip,

This study was undertaken attheNe\~foundl and Inst itute ofCo l doc e e » Science (NIC OS) andBiologyDepertrne rrt,Memorial University of Ne\olfou nd land. It was made po s sible wi t h a scho la rship from the Interna t i onal Centr e for oc ean

(11)

Dev e l opmen t trccn r. Wi lla Mageeand Lis eLa mon t a g n e fromrcon are th a nked for han d li ng the adminis trati ve aspects of the scholars hi p . Ian Bur row s and Simon Saulei who successiv e l yser ve d ashe a d softhe Biol og y De pa r tm e n t, UPNG aretha nked for their approv a l for theuse of de pa rtme n tal faci l i ties to comp l ete this thesis.The Staff Devel opmentUni t of UPNG pr ov i ded financ ial assistance to my fam ily and cont i nue dmy sal ary while 1 wasonst u d y leave.

In con clu s i o n. I tha n k my wife Ma r t ha an d da ugh t e rs, Gr ace and Ca t heri ne, for the i r understandi n g an d to l erence throug ho utttunst udy,especia l lyfor ma k ing 'n o i se-f r e e' time ava il ab le athome wh e ne ver1needed i t.

vi

(12)

ABSTRACT ACKNOWLEDGEM ENTS TABLEOFCONTENTS LISTOFTABLES LISTOFFIGURES 1 INTRODUCTI ON

TABLE OFCONTENTS

II Lv vii ix xl

2 MATERI ALS ANDMETHODS 11

2.1Dataacq uis i t i o n 1 I

2.2 Da t a processing 17-

2. 3Dataanalysi s 12

2.3.1Ca l c ul atio nof divers ity indi c e s ifl 2.3. 2Samp l i ng biasonse aso n a l patter n s 18 2.3•.')Approxima terando mi za t i o n tes t 7.7- 2,3.4Ana lys i s of se ason a l pat t erns of

diversi t y fromsi gma-pl o t gra p hs 23 2.3.5 Visualanalysi sof sig ma-p lotgr u phs 24 2.3 . 6Spl ine ane LysLs ofseasona l patte rn s

inS,H'andE 26

2.3.7Patternsof seasonal tr ac k ing

ofS, with H' an d E ?:lJ

2.3 .BRanks of varian ce s and me ans of

S,H' andE 7.'J

2.3.9Aut o c o r r elationanalysi s

ofS, lf ' an d E 30

3 RESULTS 35

3.1Data set sco mpdLerd for thisstu d y 35 3. 2 Effec t of ver-reet.cn incat c h a bility of gea r 35 3.:~ Visualanal y si s of sea s on a l patter ns

inS ,H'andE: 31J

3.4 Spline analysisof se a s o nal pat tern s

inS, H' andE srJ

3.5Patterns of seasonal tracking

of SwithH' an d E JO]

3.6Variance s and meansof SI II' andE ]{J{, 3.7Autocor r elationanalysisofS, H' and E 121

3.8Summary of res ults 13 1

vii

(13)

TABLE OF C:ONTENTS Co n t inue d

Page

4 DISCUSSION 137

4. 1 Data sets used in this study 137

4. 2Effect of bias by sampling~ear 138 4.3 Seasonal patterns of diversityin fish

communi ties 14 1

4.4Processes regulati ng seasonalpatterns

ofdi v e r s i t y 145

4.4.1Re c r u i t men t variabili t y 146 4.4. 2 Sea water temperaturetoler a nc e 15 0 4.4.3 Avoidance of competition and predation 155 4.5 Hypotheses onseasonal patternsofdi v e r s i t y 158 4.5.1PlanktonandFish Phenology hypothesis 158 4.5.2Instability-Dominancehypothesis 164 4.6Applicationsfor describedpatterns in

S, H' and E 167

4.6.1Application in fish eriesmanagement 169 4.6.2Application in environmental

mani taring and impact assessment 17 1

5 REFERENCES 179

6APPENDI CES 192

6.1 AppendixA. Illustrativefigures 193

6.1.1Figure AI . Patterns of tracking of

5with H' andE 195

6.1.2Figure1\2 .Comparisionsof visual

and splineplots 19 7

6.2AppendixB.Summar yof data sets 19 8 6.2.1TableB1. Localities and sources

of data sets 19 9

6.2.2Table82. Habitat,gear typesand

durationofeach study 202

viii

(14)

LIST OF TABLES

Page

Ta b le 1. Visualanal ysis ofpeaks1n S withlati tude 48 Table 2.Vi sua l analysis of pea ksin H' wi t h latitude 53 Tab le3.Visualanalysisof peaks in Ewith lat i tud e 56 Table4.Visualana lys isof peaksill S withhabitat 60 Tab le5. Visua l ana l y sis of peaksinH' withhabitat 62 Table 6.Visual ana lysisof peaks inE wIthhabita t 64 Tabl e 't,Visual ana l ysisof peaksin S withgea r type 65 Tabl e8. Visual an alysisofpea k s inH' withgeartype 68 Table 9. Visual ana l y s i s of peaks in E wi th gear type 70 Table 10. Splin e analy sis (If peaksin S withlatitude 77 Table 11 . Spli ne analys i s ofpe a ks in H' withlati t ude 79 Table 12. Spli neanaly s is of pe aks inE !'i'ithlatit u de 81 Tab l e 13. Sp line anal y si s ofpeaks inS withha b i t a t 91 Ta b le 14. Sp linean a l ysis ofpe ak s inH' with habi tat 93 Table15. Sp line ana l ysisof pea ksinE with ha bi tat 95 Tab l e 16. Sp l ine analysisofpe ak s inS with gear type 97 Tab le 17. Sp li n e anal ys isof peaksin H' withgear type 99 Table 18. Sp l ine an a l ysisof pe ak s in Ewithgear type 101 Table 19 . Types ofse as o nal trackinginS.H' andE 10 4 Table 20. Trendin ranks ofmeans andvaria nc eswi t h

test varia b l e s 107

Table21. Rank s inveeaence eof Swi th res pe c t

tonumbe r of species and lati tu d e 10 8

(15)

LIS TOF TABLES Con ti n ue d

Page 'fable 22. Ranksin variances ofH' with respect to

number of species and latitud: 110

Ta ble 23. Ranks in variances of E with respect

to numberof spec iesand latitude 112 Ta bl e 24. Ranksin the means of S with respect

to numberof species andla t i t u d e 115 Table25 . Ranksin the means of HI with respect

to numberof speciesand lati t ude 11 7 Table 26. Ranks inthe means of E with respect

to number of speciesandlatitude 119 Table 27. Summary of autocorre la tionanalysisof S 12 2 Table 28. Summary of autocorrelationanalysisofH' 124 Table29. Summaryof autocorrelationanalysis of E 126 Table30. Dat a sets wi th smoo thcorrelograms'

in S, HI and E an dtest variables 129 Table31. Data sets with. spiky correlograms'

inS, H' and E and testva r i a b l e s 13 0 Table 32. Data sets with both •smooth' and 'spiky'

corre1ogramsin S.H' and E

and test variables 132

Table33. Timelapse of predictionof pattern in S, H' and E for data sets wi th

. smoothcor relograms' 13 3

(16)

LISTOFFIGURES

Page Fig u r e1.Geographicaldis tribu t i on of da tilsets 37 Figure 2.Comp<!lrisonof di spersionof H' rrom

simulatedand real data -10

Figure 3.Comparison ofdis pe rs ion ofE:from

simulated and real data 42

Figure 4.Comparison of seasonal patte rnofH'

fro m simulated and realdata 44.

Figure 5.Comparison of seasonal pa tte rn of E

fromsimulated and real data 46

Figure 6. Vis u a l analys is of peaksinS

with latitude 50

Figure 7. Visual analysisof peaks in H'

withla t i t u d e 55

Figure B. Vi s u al analysisof peaks in E

wi th lati t u d e 58

Figure 9.Spline analysis of pe a k s in S

wi thla t i t u d e '/2

Figure10. Splineana lysisof peaksinH'

withla t i t ud e 74

Figure 11. Spline analysisof peaksin E:

wi thla t i t ud e 76

Figure 12 . Regression of date of first peak in

S wit hlat! tude f!4

Figu re13. Reg ressionof date offi r s t peakin

H' wi th lat!tude 86

Figure 14.Regress ion of date of firstpeakin

E withla t! tude 88

xi

(17)

1INTRODUCTI ON

Seasonal patterns of diversi ty in animal communi ties reflect cha nges in community structure withtime. Temporal variation in community structure results from changes in species compositio n and the i r re l a t i ve abundances in the communityovertime.In trying tounderstand the structu reof communities andthe pa t t e r n s of changein the structure with time, ecologists ha ve traditionall y looked at measures of diversity ofcommunities to try to understand such change in structure. Re v ie ws by Pianka (1966). Pe e t (1974), and Wils hing ton (1984) covered the develop ments inthe concept of diversity and the diversity indices that ha ve bee n used to measure diversity in communitiesof organisms. Pielou (1966) described and cl assifiedtypes of bi ologic a l collectionsone can get by sampling any community. A key for the ...::lassificationof the type of col l ectionandwhi ch index of diversity is suitablefor the coll ection wa s also presented. Pielou (1969)and Magurran(1988) provide a good coverage of indices of dive rsitywhich are in use and also discuss the ir re recLveme r i t s.

Diversitymeasures have two co mpo ne n t s ; the measureof the variety and richness of communities is one whilst the measure of the relati ve abunda nc e s of thedi f f ere n t speciesin communitiesis theothe r . In di c e s suchasth e tota l numbers of species present,S, measures the forme rwhileindices suchas

(18)

the Shannon -Wienerin d e x of diversityH', attempts tomeasure both(Pe et , 1974;Magurran, 19 BB1 . Furthermo re, indices such as the speciesevennessin d e x E, which 13a ratioof observed dive r s ityH',to the maximumdive r s i tyoftheconaaumty(II...,.1, where H....-1 nS (Pe e t, 197 4), ar e of t e n us e d to moasuro th e re l ati v e mag01tude of the two components of divers tty measured,comparedto themaximumpossible(Ma g u rr a n, 19 8 8 ).

Thatis , Emeasures therelat i veabundance of each spe cie s in th eco mmu ni t y ; an evennes sval ue of l.0meansallspeoLn sar-e equally ab u n d a n t . For th e purpos esofconsiste nc y andcl.:lrit y theterms"r-I chne es", "he'te r-oqeneLt.y'", and"t.·,fui t.a b i l i ty " will be used thro u g h o ut the te x t in asso ciation with th e three dive r s i tyindices;S, H'an d E respectiv ely.Thi s term ino logy followsthat of reee (19741,

Seasonal pecteeoeof divers ityin mari ne fi sh co uuuur ut tos ace the pri maryinteres tof thisthesis. I fseason a l patte r n s of va r iationin diversitycanbe di sc e r ned, the n the seshou ld be useful for purposes of predicting changes in community structure.If seasonalpat t erns ofdi v~ r sityca nbepredi cted , this should be useful for applied purposes suchas fr sberIes management and aquatic ec osystems management. In pollut i o n managementstudiesin aquatic ecosyst e ms, fishdatu have been used to esaeae pollutionandhab itatdegradat i on(Bech tel en d Copeland, 19 7 0 ; McErleant lg1., 197 3; HaedrichandlIaedrich, 1974;Ha e d r i c h, 1975;Livi n g ston , 1975; Hilman~t131., 1977;

(19)

Potter et al.~ 19861. In such studies, fish diversit~· st ati s t i c s are often used to describeseasonal patterns of divers i tyAnd cOlMlunity st r uc tu r e .

It isilllpo r t a nt inenvironmental managementprob l ems such asimpa c t assessmentth a t studiesbeapp r opri a t ely designed to solve the specific problem(s) in question. Because natural variations withi necologicalsystemscan easil ycon fou ndwi th varia t ionsincur r ed by anth ropo ge niccauaea, imprope rdesigns and irre levant tempo r a l and spati al eceies sere c ted for a stud yca n fa i l to detectimpacts(Gr e en , 1989). Stres singthe importance of proper des ign , Eberha r dt and Thoma s (1991) sta t e d thatunlike controlled exper i me nt a t i o n,ecolog icaland environmental re s e a r ch often do not meet the criteria of modernexp erime n t a l design. Inimp a c t as sessmentstudies the samp l i ng designis oftenplannedtoaccomp l ish two principal goals (Und e rw oo d and Peterson, 1988) . The first goal 1s to formallydetect and confirm tha t a change 1n the system has occurred. Secondly, the sampling design sho u l d al low the assessment ofwh e t her theobs e rvedchange is dueto theimpact (e.g. pollut i o n ) or whether it is due to Borne other natural processes. To ecnreve the firs t goal often require s the oxco u t.tcn of pil ot stud i es or the utilization of existing ba sel ine ecol og ical data (Cl a rke and Gree n, 1988; Green, 19 8 9 ).Insituationswhe resuc h ba s elineda t a areun a v a il ab le or financ i a l resou rces for pilot stud ies are li mi t e d,

(20)

est a b lis hed pa t tern sin community sucucturoC<111 become useful in thisrespect.

In studyingseasonalpatternsof diversit yand commun ity structure, the choiceof what tempora l sc ales \:0 uscdepends on the nature of the problem being investigated and scare kn o wl e d g e of thebi o l og y and ecologyof the speciesconcerned. Quarterly and monthly timesc a l es are commonly used in the study of seasonal patterns of diversity and cmuunnuty st ructu re in fish communit ies. The fine r re s ol uti o ns of seasona l patter ns are only adequately sho wn over mcmthI y intervals. This has been assessed qu ant rta u tvaLj- for benthic invertebrateand fis hda t a (Livingston,19 87).

The ear.etence of seasonal pat ternsin diversityin fish communities is evident fromvarious scudLes, Mos tof those earl i.er studies whichlo ok e d at seasonalpat terns in di ver s i ty and communitystructure in fishcowo u n re r e e , have be en on the east coast of Un ited States (Me rr iman and Wa r f e l , 1948;

McFarland, 196 3 ; Richards, 1963; Dahlberg and OdUlII, 2970 ; Ty l e r , 197 1 ; Oviat tandNixon, 1973 ; Ha ed ri ch and Haedric h, 19 74 ; Sub ra hman y a mand Drake, 197 5 ; Hilma n ~J;<(1,.,1977;Hoff and Iba r a, 1977). Besidesst ud i e s on the ea st coast of the United States other studies have been done in cc r r rornre (Allen an d Horn, 197 5 ), Mexico (Warburto n, 1978) , aevern Estu ary (U .K.) (C l a r idg e !l!; 1;l1 . , 1'::186), Swe d en (Tho r man, 1986 ) , Australia(Quinn , 1980:Rainer,198 4),Kuwait(Wrigh t,

(21)

19 8 9J, southorn Iraq (AI-Oa ham andYou s i f, 1990 ) , and Cape Coast ,South Africa (Bennet t , 19 8 9).

Publishedstudieshave shownconside rableseasona lchange in diversity. ececree richness, which is measured in this studyby theind e lC S, has bee n reported to vary seasonallyby severa l studies.Itwasgenerally reported tobelow inco l d e r months an d high during warmer months (Mc F a r l a n d, 1963;

Richards, 19 6 3; Shuntov , 1971; Oviattand Nixon, 1973; Hoff andIba r-a , 1977; Warbur ton , 1978 ; Quinn , 1980 ; Rainer,1984;

Thorman, 1986 ) . Variationsdoexastin te r ms of the speci f ic t iming ofthe occurrenc eof these seasonalhighsandlows in sp e cie s rich nessbutno t thegeneraltrendof more speci es in warmermo n t hsandade c line in species richness in thecolder mont hs. Studies by McFarland(19 6 3) in 'rexe s and Rich a r d s (1 963 )inNewYork,both showedthat the richness componentof diversitywas lo w in the wintermonths andhigh inthesu mme r morrt bs. Speciesdch nesswas reported to behi g h in October an dlow in Jan uaryin a stu dyinNa r r a ga n s e t t Bay(Ov i at t and Nixon, 1973). Twomore studiesfromthe North American East Coast, al s o reported speci esrich ness to reach a maximumin thesumme r.Thesewere reported fromFloridatsubrenmenyeaand Dr a ke, 19 7 5 )an d Massachusetts(Hoffand Ibara, 19 7 7) . Th i s s: eesona I tren d in species rich n esswas also reported from California (Allenan d Horn , 1975; All en , 1982) and Mexico (Warburton,197 81. Furthe r studieswhich r-epcr tred thi s pa t tern

(22)

in area s othe r than No~th Amer i c a in Swed e n (ThOr man, 1 9 86); northern Au s t ra li a (Shu ntov, 1971; Qui n n , 19 80 :Ra i n e r , 198 4 ) ;Ku wa i t (Wright , 1989) sout he rnIra q(Al- Daha mand Yo us if , 1990); SouthAfrica (Be nnet t , 19 8 9 ). One studyre por t e dare verseseasona l pa tte r n inspec iesrichness; spec i esric h nesswashighe r in the cold ermont hs, Thiswas sh ow n by a stu dy done in Se vern Es tu a r y , U.K. (Cl<lr i d go at gl., 19 6 6) ,

The se ason al patter nof the he t e r o geneity compo n entof diversityshowed that pe a ks oc c ur r e d ineither wa r m and col d seasons of th eye a r ; usuallyoneor the otheroccurred in an y one study. Maxi mum valu e s in the he t ero g e n e i t y componen tof diversit y in wa r mer month s we r e reported by a num ber or studies(Ovi att and Nixon,1973; Subrah manyaman dDrake, Itj'l~;

Allen and Horn, 1975; War bu r ton, 19 7 8 : Al - Da ham and You sif , 1990).Otherstudies re p or t ed maxi mumvalu esinthisco mpo nent of di ve r s i ty duri ng thecolder months (Ho f fand jbar-zi , i977; Al len , 1982; Cla ridge~~., 19 8 6 ) ,

Th eeq ui tab i l i tycomponentofdiv e r s i tysh o we d a aoeso uot pat t er n si mil arto th at for the he t e r og eneit y component of di versi t y.Tha t is, maximum va l ue s we reobserved inboth werur seas ons (Da h l be r g and Odurn, 1970 ; Wa rb u r t on , 19 7 8 ; A1-Daha lll and You s i f, 19 9 0) as well as cold se asons of the year (Clar idge~gl" 1986).

The abo ve studies ind i c a t e one pet t.er n tobecommon.Tha l

(23)

is therichness compo n e ntof dive rsity rea c hed maxim umvalues in thewa rmer mon t hs . Thisranged fromMarch to abo u t Octo b er in theno r t he r n hemi spher estudi es (McFarl a nd, :L9 63 ;Richa rd s , 196 3 ;Wa rbur t on, 1978; Tho r man, 1986; Wr i g ht, 1989 ; AI - Dah a m andYousif,199 0 )and fr omDecember to Marchin the south e r n hem i s phere stud i es(Qui nn,1980; Raine r ,1984 ; Bennett,198 9 ) . La tespring in t o earl y autumn is generally regarded aswa r m months whilelate aut u mn to ea rly sprin gcan be gene ral ly rega r dedas co l d mo n th s,withinan eonuei cycle. The specific months for these wa rmer end colde r con di t i o ns wil.1 be diffe rent inthe twohemi s phe r e s he nc e th i sdi f feren cesho uld bebor ne inmin dwhe n re f e rring to seaso nalpatt e rns re por t ed by seasons(i.e spri n g,su mmer, autumn andwinter). That is th e summe rin th e northern he mi s phe r e is from about June to la teea rl ySep t e mber while inthe sou t h ern hemi sphe re this period is wint e r.Unl i kespeciesri chness~Ihichappe ared to showa fairlycons i stentwarm- cold trend, the he t.ez-cqeneL'ty and equitabilit ycomponent sofdiv e rs ityseem e d tosho w ei the r pa ttern in di ffe ren t studie s. Thi s impl ies tha t al l thr e e compon e n tsof dive r sity can either track eachother,or th e.t theheterogen e i ty and equi tabU!tycompone nts ofdiversitycan showan oppos ite seas o n al patternto thatofspecies r ichn e s s . Wheneve r the peaks and troughs in thethreedive rsityindic es, r, H' an dE, close ly fol lo weach other, thi s repre sent s an equ itahl e influx and ef fl ux of species. Aneq uita b l ein f l u x

(24)

occurswhen 5, H' andE increasetoget~,er. If5 shows a peak whileIi'an dEdec r eaee,this tre nd re pre s en t san ine q uitable in f l u x and efflux 0)£ species due to one or two dom i na n t speci es.Aschematicillustrationof thetwotype s of trackin g ofth e threeindices is giveninFigu r e AI. Appendi x A.

Fr o m the above de s c r i p t i o ns of seasonal varIation In diversity. i.t is evid e ntthatseasonal patt erns illdiversit y do exi s t in fish co mmuni t:l es. Species di s t r ibu ti on s an d compositions,11S well as environmentalfactors whichareor eeu foun dto influence them, vary geograp hical l y .Thus,oneasks wh etheronecan di f f e r e n t i a t e betwee ngeneralpett.c sn s (tho ~c occurri ngoverlarge geo9raphicalscales) andthoseepectrtc topa r t i cu l artyp e s of fishcommuni ti e s (thoseocc ur r1n g over local scales) . Thi sques t i on de al s withlargespatia l SOul es and can on ly be sufficiently addresse d by a compc rntrv e appr o a ch . If large scalepa t t e rn s do exis t. then thisshou ld be usefu l in app lied fields such as impa c t andpollutJo n assessme nt as we l l as fisheries management.

From thisreviewof theli t e r atu r e on stu dieson seasonal pa t t er ns of diversity and community structure of fish commu n itie s.i t isevi d e n t th atal t hou gh numerou s smal l scoLo or single stud y investi ga t i o ns on the subject have been pub l i she d .no majo r largesca l ecompa ra tiv e studiesnave been re p or ted . One recent stu d yused pub lisheddata setsCor fish capture s on po wer station inta ke scree ns from twctve sites

(25)

throughoutEnglandand Wa lesto study the seasonal pa t t e r n sof diversityan dcommunity structureof inshore fishcommunities (He nderson, 1989 ). This study demonstrated both tempo ral (s e as ona l ) as well as spatial (l a ti t udin a l) trends in diversity and community structure. 1 aim to integrate both local and la r ge scale seasonal patterns of diverl?ity and communi tyst ructureof fi s hcommunitiesin thisstudy, which i f accomplished, will be us e f u l for applied purposes.

The information reviewed above can be summarized as a worki nghypothesis that seasonal patterns in diversity in ma r ine fish communities exist. and these patterns are z-e qulatied by processes influ e nced by the interaction of a range of biotic and abi o t i c variablesopera ting in concert.

The princi pal objective of this study is to analyse this hy po th e s i san d propose further testableones. The study will specifically involve firstly, using published data sets to discover seasonal pa t t ern s of dive rsit y in ma rine fish commu n i t ies. Any appa r e nt patterns discovered wil l be des c ri bed , to allow gen eral seasonal patter ns (patterns occurring overlarg e geo graphicalsca les)tobe separate d from ones that are specific to lo c a l scales. These described pa t t e rn s wi l l.beus e d topr opo s e furt her hy p o t hes e s aboutthe pr oce s s e s an d mechanisms regulating seasonal pat t e rn s of di vl' r s i t y in marine fi sh communi ties. Final l y,these seasonal pa t t e rn s, any rela t i on ships among the thre e compone nts of

(26)

10 diversity(richness, he t e r o g e ne i t y . and equit abfLLt.y } , and tho proposedhyp o t h e s e s will be used to illustrate exan\plesofho w theseinformation canhe usedby researchers in apr11edfIelds that use diversity stat is-cicsde rrved flam fishcatch dot.n, such as £isheri.es managementand marine pollution, to design an d exec utemore se ns itive tests.

To ectueve these objectives, two analytical techniques will be used to identify seasonal patterns of diversity. These are: (1) a graphicaltechnique:and(2 ) tho use of statistics (means, variancesand autocorrelation coefficients) of the indicesS, If' andE. The graphical techniquewill allow forol visual di s pl a y of any seasonal pa t t e r ns as weLj, as ,"'Illy possible trends in these patterns. In the use of sample stat istics, the ranks of the means, variances, and Lhe autocorre lationfunctions of the tinlc seriesoCeecn Index [or different data sets will be used. These ranks wi!t he analyse d with re s pe c t to major explanatory variablessuch us habitat ty pe , type ofgear, and latitude,wit hi ngeogrilp hico'll regions with similar climaticregimes, to determi newhether any relationships exist among the ranks of the dU f e r e n l statisticsand the explanatoryvariables.

(27)

2 MATERIALSANDMETHODS

2.1 Dataacquis i tion

Dataused in thisst u d y wereobta i n ed throughpu blis h e d sou rc e s . Literature search for releva ntpubl ications was done by usi ng comp u t e r search methods, searching through th e contents of every vo lume of key jou rna ls, and fo l l o wing up relevan t refere nce s frOllleveryeppro pr Lete article. The ke y jou r nals searc hedwer'et Ma r i ne Bi ol ogy; Bul l e t i n ofMarine Sci e nc e; Marin e Ecolog yProg r ess Seri es ; Es tuarineCoasta l and Shel f Sc ience; Estuaries; .rcurner of th e Mar i ne Biol ogic a l Associat io n of the United Ki ngdo m: Jou r na l of the Ma r i ne Biological As soci ationof Indi a; Env iro nment a l Biolog y of Fishes; Canadi a nTech n i cal Re por ts of Fishe r i e s and Aquatic Sciences. Additional jour n als searched to some exte ntwere : Fishery Bulle t i n (U.S.); Aus t ral i a nJournal of Ma rine and Freshwat erResearc h; CanadianJournal of FisheriesandAquatic:

Sc i e nc e s; Journal of Fish Bi o l ogy.

Thro ughout the se a r ch, the selecti o n criteria for articl e s were that: (lJthe studyshou ldinvo lve mont hl y or fortnightly sampli ng fo r s1xmon thsor more; (2) all da t a pres e nte dshou ldbe sample d by asi ng lesamplin g gearor whe r e mor e thanonegeartype is used, dat afromeachtype of gear sh o u l d be se pa rab l e; (3) taxonomi c in format i o n sho uld be compl e'teto speciesleve l or i fspeciesareun i d e ntif i ed,the y dho u ld be di sting ui she d from othe r related specie s and be

(28)

12 consistentlyrecorded by atentativeidentification; and (4) the sampl ing procedures uaert should be sUfficient ly an d accurately described to allow for thecalculationof catchper unit effort (CPU !::)statistics. Based on these criteria, a tota lof 22 data setswerese lected for this study.

Data sets were eithe r obtained directly from the publications wheneverthese werepresentedor were req uested from the authors. From an initial attempt of thirteen requests,only CIne eucfrorre s p o nde d by sending original data.

This poor response resulted 1n abandoning this method of obtainlng da t a. The twenty-two data sets used inth i s study were largely compiled from published sources; they e r e avai lable forus e upon request.

2.2 Dat aproces sing

Allda t a setsacquired were stored as ASCII files using MemorialUn i ve r s i t y'sma i nf r a me (VAX!VMS) coerpu tier.The data sets were separated by sampling stations ilnd by type of sampl ing gearus ed. Thesedata fileswere the parent f.lles fromwh i c h subsequ entmanipul ationsand sp ecificanalyseswere done.

2.3 Data analysis

Data analysiswas done with three computer software pa c k age s. Gr a ph i c a l analyses were performed~JithSIGMAPLOT

(29)

13 lin d MATHCAD.Stat is t ic lj l analyseswer~perforraedw.i: t hHI NITAB.

which \lias available on the Un iversity ' Sma inframecomputer.

Twoan a l yt i c a l techn iqueswe re used In data Anal y s i s;the use of gra phi c al ana l ysi s was one technique and th e use of stat i s tics (means ,variancesandautocorrelationCoefficients) wasthe other.

Graph i ca l methods areof t en us e f ul in exploratory ':Illta analys i s asthey dis play any pat ter n s ui sting inthe da t a. This enables the re s earcherto ext ra c tvi s uall y any po s sibl e rela tions h i ps be t we en the pat t ern In the data and the variable s concer ned .Italsoall ows the researche rto de c i d e onap p ropriate methods ofdat a analysi s tous e for tes ti ng these relationsh i p ssta tistica lly.

The statistical technique used wasthe ana l ys i s of the aut ocor re l atio ncoe f fi cients, andth eran ksof thellleansand variances of S. H'and E. Thesest a t i stic s wereanalysedwith re s pect to a number of explanatory variables.Habitat type, gear type, and latitud e wit h inmajor geograph1.cal reg i ons gro up edaccording to si milarities inclimaticpat terns. WQc e themain explana to ryvari able s used. The ran ks ofmea nsan d va riances (o r each ind l1x were chec ked aga i nst each of the axpfe.rtatcr-y variables to uncover any re l a tionships that exr se ed,S1.milarl y , th e aut ocorrelati oncoeffi cient s ofthe thre e indi c esof divers i ty were analysed bytest i n g the types ofccc-r eroq raes of dif fe rent da ta setswiththe eJlplanatory

(30)

var i ab les for anyrelat io nships .

Several maj or indices of divers i ty are availabl e and their descr i p t i o nsand merit s have bee n di s cu s sed in anumbe r of pUbl i ca ti on s (Pi e l ou, 1969 ; Le gendre and Le gendre , 19 83 : Washin gton ,19 8 4: I-la g urra n, 1988). Statistica l propertie s of these major indices have bee n comparedusing data from a copepod communityby Hei pandEngels(19 74).They showe d that the Shannon -Wiene rinde xH', wa s thein dexth at £lave a oet t.er measu re ofth e diversityof theco p e po d community in te rms of th e var ia bil i t y an d confo r mity of valu es calc u l a te d fOi indi v idu a l samples and tha t of al l samp l es pooled . No conse nsus ex i s tsas towhich ind icesare bet t e r andwhi c horu not(Wa shi ng t o n, 198 4; Magurr an, 1988 ) . Besid esthi s lack or conse nsus , the r e are also limi t ati o n s to the bi olog icn! inter preta ti ons among values of th e diff e r e nt diver s t y indices. Thi s isbecaus e man yindices areratiosof mea ningf Ul biol o gica l va r i ables such asspecies ri ch ne s s end relat i v e abundance ,and ther e foresh arein fo rma tion {P o o le,197tl}.Tho probl em arises when these twotypesofinf o r llwti o fl about <1 communi t y are compounded int o a si ngle numer i c al muoauro (P o ole , 1974) as in the Shannon-W i en er inde x H'. Man y such indic e s are in us e today(Washington, 1984: Magurran . 198 8). To gua r d aga ins t th i s pro blem one mustcar efully conside r tho suitabi l i ty of sa mp li ng methods and des i gns pla n n ed fo r collecti n gspecies diversity datafroma partiLo u jar co mmu n i ty

(31)

15 (Gr een, 1979). and ee ie ct.,::p!"roprlateindi ces for thesamples col lected respect ively(Pielou, 196 6; Pool e , 1974).

De s pite thi s limi t a t i on , payi ng critical atte nti o n to samplingandstati stica l as pe c t s of selecte dindices can give in f o r ma ti o n about th e bi ologyofth e sys tem studi ed. Thisha s been shown with data from bird , bentho s and plankton co mmu n i ti e s (We b b, 19 74). My reasons fo r using the above indi ces are as follows. (1) Speci esri chnes s th a t isme a s u r e d bythe ind e xS, isba sed on the tota l countsof the numberof species in a col l ection. It is an absol ut e an d fun damen t a l me as u r e of ep ecs.ee diversity (Mag ur r a n, 1988) since i t measure s dire ct ly the species co n t e n t of a sa mple. (2) Th e Shannon-Wiener in d e xha s beenwidelyused in diversitystudies in fish co mmu n ities an d so it is used he r e to enable comp a ris on s of seasona l pat t e rnsof thi s index in this study wi th previouson e s ifnecessary . (3)The Shanno n-Wienerindax combi nes boththe numberof specie s pres ent in thesa mpl eas wel l as th e apportionmentof the number of indi viduals in t o the different species (Peet , 1974 ; Magurran, 198 8 ) . (4) A C0ll11110n observation in an i ma l and plant communi t ies is that oftenmost individual s in .3 communitybelongto a sma l l number of do mi n a n t spe c ie s while a large number of spe cies are representedby a small number ofind i v i d u a l s (Margalef, 1958, Hughes, 1986; Lawton, 1990) . Th i s generallymeans tha t the number-of individuals perspec ies in most ani ma l an d plant

(32)

16 communitiestend tobea logar i thmic function of th e rank in abundance of the co ns ti t ue nt speci es of th e conununtty . Accordi ngto in f o r mati o nth e ory (Shannonand Weaver , 19-19) the Shannon-Wiener index H'. has properties that app ro x imate propertiesof theoretical mod e lsof speciesdi s tribution such as the logseriesandlog - n o r ma ldi strib ut 1 o ns .The r efo r e, the Shannon- Wi enerindex is chosenfor useinthis st udy. (5)Tim evennessindex E, whichmeasu res th e equitab i l ity compo n e n tof diversi ty . is al sosele c t ed be c au s e it is a mea s u r e DC the ra tio of the heterogenei tyH' of the samp l e to themax illll nll hete r ogene i ty po s sible if all the spec ies were eqnotty rep re sen t e d (Peet , 1974; Ma g u r r a n, 198 6 ). Therefore, th is ind e x givesus a handleon whether or not thespeciesinth e community from whi ch th e sa mp les were obt aine d, nu me ri c a l l y equallyrepresen ted.(6) Becauseeachof th e th r ee iOl.Uces lo ck s at a differen t aspec t of dive r s i ty, the ywere al lused he r e in order to desc r ibe theove rall str uc t u r e of the community.

2.3.1caacu r.a-tIon ofdi v e rsi t y indice s

The divers ity indicescalcula tedwe re: the totni numhnr- of spe cies S, recorded fora st ati onon eachsamp lin g{jilt £!;

theShannon-Wi e ner in d e xof dive r si tyH';th especf e s cvc un us s, ind e x of diversity E.The formulae for th e s e Lndfcun we r e af>

follo~s:

(33)

17 S.. total number of species recorded at a site

sampli ng date;

I Equation1)

where Pi • the proportion of th e itil. species in the sample (PI" OliN );

OJis thetotal number of individua lsof the ith species;

N is thetota l numberof individua lsof al l species in the sample.

E..H'11nS (Equation 2)

where In S is equal to the maximumdi v e r s ity (H...lof the community (Peet, 1974).

H' and E were both calculated using the MINITABstatistical package.Computationrouti nessavedas macros(a saved set of exec u t a ble MINITAB commands) were execu ted seq uentia l ly to obtaineach andex for eachda t a set.

The sequence of the compu ta tions to obtain H' was as follows: (1) the pr o p o r t i o n of each species(PI' Pl••••PI) for each samplingdate was calcu lated; (2) the natura l logarithm (In) of each pwa s cefcnt ated and stored; (3) the pz-oducf of the values in (1 ) and (2) respectively were thencomp ut e d,

(34)

18 gi ving avarve corresponding to each speciesforeachsa mpling dateinth edata set; and (4)summing up the valuesobtained in (J) fo r each sa mpli ngdatefina llyge.ve there s pec t i v e H' values(see equation1).

To calculate the species eve nne ss values E, for each sam pli ngdate, the na t ura l logar it hm of thenumbe r ofspe c i es S. for eac hde.tewas compu n e d,Th is was thendividedintoth('

corr es pondingvalueof Htobt a i ned In(4)above,to gi vethe re spe c tive evennes s value (se eeq ua ti on2). All val ues of5, Ht and Eforeachsampli ngdate for each st ud y werethe nsaved for the subsequent analyses of seasona l pet .te rn e.

2.3.2 Effect ofsampl i n gb:f.a s on se a sona l patterns In anysamp li ngprogranufte,reg a rd l es softhetypeofqecr used, th e catchability of th e gear varies for different species as a result of fa c t ors such as gear efficiency, di stributio n al patte r nsof the different spec i es, and the i r re s pons e tothe gear (Taylor, 19 5 3 ; Kenchlngton, 19 8 0 ; Flyrno t l§l., 198 1;Gullan d, 1983 ) .Catchabil ity is conve ntionally de s i gnated as q in the fis h er ies 11uereecre, and is also termed a"ca tchr.lbil i t y coef f i cient"(Rickel,11)75;Kj el s on and Jo h nso n, 1978; Gulla n d , 1983; Blabe r e~ ~1,., 19 90 ) . ThJs coeffic i e nt q,intheinst anceoftrawling gear, isde fin e d es the rat i oof ind ividua lsof ace r t ainspeciescaug ht by th e t rewlto th eto t alnumbe r of tha t species inthe pathof the

(35)

19 trawl (Ric k e r. 1975; Kjelson and Jo h n son , 1978). Mathematically , thisrelation ship ca n be r epi-euented by the equa ti on:

q '"c/o (Equation3)

whereq _the pr o po r tio nof in d i v i d u a l s of asp e c i e s in the path of the gear. re pr e s e nte d in the catch;

c'* the number ofin d i v i d ua lsof a spec iesin theca t c h;

n ..the totalnumber ofindi v i d ua ls of a sp e ci e s 1n the pathofth e gear.

This stud y used catch data obtained fr o m published studies. Allda t a setsusedin this st ud y useda single gear ty p e , except one (Allen, 1982). which was collected by multiple gear types. This data set was accepted for usein this study because a common eff ortwas possible to calculate fo r thedifferentmethods used, ena blingthe st andardi za t i on of thecatch in termsof thiscommonef fort. It was impo r t a n t to observe the above criteria in compiling the data sets because the catchability of different gears for di ff e r en t species differ and never have a perfectcatchabili ty ofq • 1.0 (Ta ylo r , 1953 ; Kenching t o n; 1980; Byrne §.1 §.!., 19 8 1 ; Gu11a nd, 1983). Therefore, i t was important to determine whether ke e p i n g anyimperfe c t ca t c ha bi l i t i e s in gear ( q<1.0)

(36)

20 for the different specie s co ns t a nt. among samp leswithin() stud y.wou l d giv e co mparabl eseasona l patternsfo r II' andr.to thosedis cernedunder s1tuatio nsofpe r fe ct catch a bl 1 1ty (q..

1.0) fo r al l species. If the twopatte r ns ar e thesame, then one ca n as s u me th a t as lo ngas the samplin!;1desig n and the sampl ing progr amme were constantwithin astudy, the seasonal pa t tern s di s c erned from the catch da ta ar e not bi a sed by differe n tial catc h a bl11tyof species.

To assessthe above question, the da t a set of Macd o na l d g,.t l l . (1984) (see Tabl e s 81 and 82 . Ap pend i x 1]) was ar b i t ra rilychosen to ana lyse whe t he r imper fectcatcha hility aff e c t edth eLnt e r-pz-etetLc n of theseasonalpatter nsobserved;

q ..1.0 wouldimply that c " n , From the catch (c) inth i s data set 300data sets wer e simul at ed , each to re pr e sent pos si ble total numbe r s (n) for each speciesex posed to tho gear. In eachsimu lati o n diff eren t val ues of q were ra ndomly gene r ated for each species in the da t a set. Withi n each si mulation these va l ue s of q were kept constant for each speci es.Th issimulationproc edu r e was as follows:(1)MIN I' r AB was use d to ge nerat e random val uesbetwe en 0. 2and 1.0 for q from a unifo rm proba b i l i t y distributio n fo r ea ch sp ec iesin the da t a set ; (2) the cat ch c. in theorigi na l dat a setfor each specieswas divided by the corresponding :i t.c give a simulated da t asetwhic hreprese nt ed thesimulate dn for each species(see equa tio n3); (3)H' and Efor thissimulateddata

(37)

21 set werethen calc ulatelJ follo wi ng the computationalsequence forcalcu la tingthetwo indicesdescribed in 2.3. 1above;an d (4) th e va l ue s for H' and Ewere then saved. This rou ti newas re pea t e d 300 times.

The 300 vl!llues of H' and E were then gzaph lcal ly anal ysed. Th is wes done to see if the seasonal patt ern of diversityfro mthe ori g i n a l dataset di ffered sig ni fica ntl y fr om that fo r the 300 data sets sim u ll!lted wi th consta n t ra n d oml y generat ed catcha b i l i tf es foreachspec i e s, throughout allsa mp l i ngdates.To asse s s the di s pe r s i on of th e value s of both H' and E cal c u lated for each samp l ing da te from the simula t e d dat a sets, "aoxandWhi sk ers · plotsweredone for

each Lnd ex , These were pl o t s of the quarti les of the distri bution of the 300 va lues for each inde x fo r each samplingdate. The values for H' and E calc ula t ed frO/llthe origi n a l da t a set were th e n superimposed on the "Box and Whiske rs· plot. Thi s wasdone to displa y the vari a bility in thevaluesof each index fromthe 300 simulateddatasets, in relationto thecorrespondingvalues of the two ind ices from theor i g i n a l data set. Va luesof H' an dE ofthe or i gi naldata foreachsa mpl i ng da t e we r e furthercompa r-e dwith thevaiue e of the two indice s ob t a i ne dfr om data from two of the 30 0 si mu l ati ons. This wa sdone by comparing the graphs of H' and Ecalcula t e d fr om th eorigina l data and theval u e s of the two in d i c e s cercure eedfro m data from two of the three hun dre d

(38)

22 simula t i o ns.The two simul ations usedfo r thecomparison we r e arb i trar i lycho sen . Fromthi s comparison, it was de t e r mined wh ether the seasonalpat te rnof diversitychan ged whe n n was varied at eac h si mul ation by a ne w set of random catcha bilitiesfor ea c h species.

2.3.3ApproJ'limaterandomi:z:ationtest

A cendomtaetrontest wasus ed to deter mineWhether H'and E calculated from these simulated values of n differed sign i ficant l y fro mthe va l ue s of the two indices cnj cujeced from the orig i nalcatch (c Ida t a.The simulateddana setswere obtained by rando mlygene ratedcatcha bilitiesasdescribedin 2.3.2 above. This procedure is called an "approx imate ran do mi zation test" (Noree n, 198 91. It involves r-endomjy genera t i ng samples from a pool of all possible valuesof a variablewhichareobtai n ab le by completepermutatio n andthe n calcul atingthe requi red statisticfr o m these samples.Inth is casenwasth e variablesampledendHIand E werethereq uired statistics. Foreachsimu lateddataset obtainedfrom ran doml y genera ted cat c ha b i l i t i e s , H' and I': were calculated by executingmacrostha t carriedou t the computations descri be d in2.3.1above.The valuesof the twoindices correspondingto each simulation were saved together. To comp ute the de s c rip t ive statist ics for the 300 setsof va lues forH' an d E whi c h were obt a ined fro m simul ated val ue s of n, i t

(39)

23 necessaryto separate each set of valuesfor eachindexfrom th e 300 files in which they were stored. The 300 files containi ng H' and E from each simulationwere retrieved in ba t c he s of fiftyfiles at a time into MINITABandthe values for H' and E werethenresaved separately.Fi f t y fileswere sorted at a time by repeatedly executing a macro which conveniently didt.hesortingbecause randomizationmethods are oftenle n g thy and timecon-surm nq,Repeating thisprocedure six timesre s u l t od in six such files each for H' and E, allwith a data matrix of fifty columns (corresponding to fifty s.unutetrons of random catchabil1 ty) sixteen rows (correspondi ngto sixtee nsamplingdates).Finally, th esesix file s fo r H' and Eres pe c t i vel y , werecombined to gi ve th e fina l 300 x 16 matrix for each Lndex, The descriptive statisticsfo r the 300 values forea c h index for each date obt6ined fromthe simulated datasets were calculeated fro m thesetwo fin al mat r i c e s.These descriptivestatistics were then used to analys e and compare the dispersionof the 300 values ofH' and E forthe simulateddata sets wi threspect to the values for eachind e x for theoriginalda t a set.

2.3 . 4Anal ys i sof sea sonal pat t e r nsofdiv8r:'si t y fr:'omsigma- plotgr:'aph s

Gr ap hi c al anal y si ct for seasonal patter nsin 5,H' and E wa sdon e usi ngSIGMA.PLOT.Visualin spe c ti o nof these grap h s

(40)

24 was do ne to identifymajor peaksand troughs and to lookfor general seasonal patterns in the three divers1 ty indices.

Thoseplots of S. H' and E, were al l pl ottedontoa single pa g e , each indexwith a suitabley-ax isand all indices wi th a common x-e x r.s , Exact dates werepl o t: t e d , or where suchwas not possiblefro mth e data, Julia ndatescorrespo n di ng to the middle of each month were used. This ar-r-enqernenu of graphs allowedfor simultaneous comparisons of seasonal patterns in the three indices.

2.3.5Vi s u al analysi sofsigma-p l o t graph s

The graphs dr a wn with SIGMA PLOTwere orderedby latitude within four ma j o r geographical regions from which the data sets were collected. The four regions were: (1) Europe; (2) No r t h AmericanEast Coast; (3)Mediterranean , South Afr ica, and North Ame r i c a n WestCoast; and (4 ) TropicsjSubtropics.

Marine biogeographic re g i o n s mapped according to the di s tr i b u t i o na l patterns of shore and shallow sea reune (Hedg pet h, 19 5 7 ) were us ed to group the above regio ns. He d gpe t hIs bi o g e o g r a p h i c regions were based on 1itt o r a 1 provinces of the worlel classified by Ekman (Ekman, 19 53op cit.He d g p e t h, 19 5 7 ).'rhose provinces were:lIrctic-Antarctic;

Borea1 - An tiborea1; Wa r m Temperate ; Tropic. They reflected temperatu re re gi me s associated with la t i t ud e . Ac c o r d ing to this sch eme , theappropriatebiogeogra phicprovinces for the

(41)

25 respec t i v e regions we re: Europe - Bore al ; North America n Ea s t Coast -Ar c t i c ,Bo r e a l, Tempe ra te;Mediterreneen,South Africa and North Ame r i c an West Coast Wa r m Temperate,

Bo r e a l / Ant i bore a l ; Tropic.:s!Subtropics - Tropic. Wi th i n ea ch re g i on thedata sets were orderedlat!tudinal l yfromno r th to southandth eseasonalpatte r nsof 5, HI and E from datasets fr o m the respecti ve regions were inspectedfor patternswith tho variOUS exp lanatory variables. The main explanatory variables were latitude, habitat type and gear type.

To carry outthe analysis, a form was drawnup to record the location of each stud y. post ticn (lat! tude/longitude) and monthsof sampling.By visual inspectioneachma j o r cy c l e of inc rease and decrea se in each ind ex , S , H' and E, was ide n tified.Monthsof highes t and lowest valuesof each index in eech major cycle were then recorded on the form. This info r matio non the timing of the peaks and troughsof each in d ex were then plotted . To ensure consistency in the ana lysis,definitionsfor "peek ' and. trough,wexenecessary. l\ "peek' is define d he r e as the hig h est point which is ap p r oach e d by a seriesof progr ess ivel yinc reasingdata poin ts and departedfr o m i t bya series of progressivelydecreasi ng points, ina time series plot of any of the three in di c e s . A changewi thin either an overallincreasingor decreasingtrend bya minimum of oneda t a pointwas no t cons idereda ch a ngein the ge ne r altrend.A.trou g h' is define d as thelowest point

(42)

26 between any twoadj a c e n t peaks in a tim e ser iesplotof anyof th e th r e e indices. Pe aks and troug hs as define d he r e are grap h ical l y il l us tr a te d in Fig u r e /1,2 (Appe n di K A). These de f ini ti on s excl u dedany pe ak s and troughs at the start and end of any time seriesin the ana lysis be c au s e those were often no t represe ntedin a comple tecycl e.

Grap hs of all data sets from the thre e geograp hical regio ns were la t i t ud i n al l y ordered and each index was separatelyplotted.Visualinspectionsof these plotsen abl ed comparisonof major seasonal aswe l l as latitudinalpatte r ns in each index wi thin each of the major geographical ar-aas , Pa t te rn s relatedto po s s i ble influenceof sampling gear and typ e ofha b i t at wereals oexpl ore d.

2.3. 6 Spli neanalysisofseasonal patternsin S, H' and E Spl inean a l y s i s for the di ve r s i ty indices5, II' and E, of al l da t a setswere donewith the Ma t h CAo software pa cka ge . Thismetho dwas alsoused be s i d e s the visua lmethod because it ismo re quanti tat iveandexplicit. That is, differentpeop l e usingth etec hn i q u ewouldob t a i nsimila rresults for any da t il set. On theother hand , the visual method nla y give slig htl y differ en t results in thisres pect, it has the ad va nt a g e of permi t t i ng intelligen t judgements about "c ut. Lt er s' in the data, allow ing one to appropriate l y plot seasonal trends in the da ta. Spli ne ana lysis works best whenserial da t a are

(43)

27 collecte d at eq ualinterv alsof time or space. Splineswe re fou nd to be hig h l y sensitive to serial data collected at une qu al interva ls. Hence,using both methods serves to cross chec k theme tho d s of analysisandth e ir resul t s.

The spline anal ys is invo lved fiTting linear splines be tween knots.A$pl'.neis a lineofbe s t- fitthroug htheda t a points withinapre -determined interval (1.e. theresolutionl while a "knot ' in a spline plot is an imagina r y l ine pe rpend i c u l a r to the x-axis, th e time axis in thisca s e , that marks the po i n t at which splineson a1 ther side of i tare con nected. When adja c e n t lines are connected at successive kno t s,an overall li ne ora curve results;whetherth i s spline is straig htor curved dependson thepo we r of the function that describes the re lationshi p bet we e n th e two var i a b l e s plotte d.Detai led de s cr i p ti o n s and examplesofspli neanalysis is givenin Wold (1974).Figu r e A2 (Appe ndi x A) shows pl o t s of both vi sual and spline ana lysiswith respect to theorigin al data, for the timeseries fo r eac hof the threeindices.

The re s o l u ti on (i.e. the int e r va l between success ive kn o t s ) in my analysis was pre-deter mi ned to span two dat a points, giving resol ut io nsof: sixtydayson ave ra ge for all data sets colle cte d over month l y inte r va ls; th irty days on average for tho s e collectedat va ri a ble time intervalsth a t rang ed betwe e nwe e klyto mont h lyinte r vals; and fiftee ndays for datasetscollectedataweekl ytimein t e rval orle s s.The

(44)

2B ave ra ge resolution in ea ch dataset was computed using th o formula:

( Eq uatian -I)

where Rnis the resolu tionat knot111; kisth e nu mbe r ofdata poin tsspanned (2 ) ; tjis ti meat samplin ginJulian da ysof thei th sample:

tl is time at sampling in Julian days of the firs t sa mp l e;

n is the sample size.

Using the abov eformula:sev e n t e e n dataset sharla resojut Lon of sixty days on aver age: fourdata setshad a re s olution of thirty day s; and one data set had a resoluti on of [ifte en da y s,inthesp lin e analysi sforeach indexinthedata se ts. The an alysis re sultedin spli n e pl o t s at app rOilria te l y sele c t e d resolu t i ons foreach Ln d e x in each data set . 'fhl.!

peaks and troughs we r e analysedin the same manner as wasdone in sectio n2. 3.5 , accordingto their re s pect ive de finit Io ns. For plots of each ind e x of all da t a sets, the months and corresponding ,Jul iandates for each pe a k and trough in the time serie s were tabulated. Thi s data was compa red with lati tude, habitat ty pe and sampl i ng ge a r , tode t er mine an y relationshipsin the peaks andtr o u g h s wi t h thes e vorteb res •

(45)

29 2.3.7satt erns of seasonal tracking of S with H' and E

In order to de t er mi n e the differences 1n the type of seaso nal pa t t e r ns 9£species ri c h n e s s S, the heteroge neity ind ex H'and the equ ita bilityinde xE in differe ntdata sets, the major peaks an dtr ou g h svi s i ble from theoriginalsigma- pl ot grap hs ofea ch ind ex for ea ch da t a setwereused. The peaks and troughs of ea c h index fo r each dat a set were examinedto display the pattern of trackingof H' and E with res pe ct to thepe a k s in and troughs in species richness, S.

Themajorpattern(s) of trac king in each data set observed fr om this analysis were tabulated.

2.3.8Ranks of variances and means ofS, H' and E Variancesand means of thethre edive rsi tyindiceswithin each of the datasetswereanal y s e d by rankingand checkingto see whether these statistics were rel<.ted to any of the variab lesused in the analysis. Usin gMINITAB, .;h e variances an dthe meansof 5, H' and E of each studywere calculated. The meansandvariances for each indexforea c hdat a set were then ranke dand tab u latedseparate ly, resultingintwo setsof tables .One set wasfo r th evar ia ncesof eac h index whilethe otherwas fo r the means . In eachset of tables, the typeof ha b i tat for each study; la ti tude for each study; the total numberofspe c i e s rec o r d ed in the da ta; thetype of sampling gear( s) used; th e sampling inter v al; and the du r a ti o n of

(46)

30 sampling, were recorded. Th e s e tables were the n us e d to determinewh e t h er the ra n k s In the variances and means for each in d e xin eachdata se t ahowndany relationships witilany of the var iableslisted above. For example ,was hi gh var i ance in speciesnumb er associatedwitha particularhabitat , suc h as the nearshore ?

Thevariables thatappeared tobere late dona ran ksca l e to the three indiceswer e thenanalysedin1I10re det ai l. ~allks ofthevarian c es and meansof5, H' and Ewer ealso cxneuncc in relation to latitude andthetot a l nu mber ofspec i esin the dat a se ts, to furt h er describe th e pattern with res pec t to thesevariables.

2.3.9Aut o c o r r el a t i on ana l ysisof 5,H' and E

Autocorrelat ionfunctionsmeasu re th ecorrelati o n of .:lily pair of observa t ions tha t are k lags apar t (Montg ome ry ilnd Johnson , 197 6 ) in a se r i e s of ti me - o rien t ed cba ervet Lo na (Montgomery, 1991 ), called a time se r i es (Mon tgome ry find Johnson, 197 6 ). The seasona l values c., diversi ty indi c e s analysed in this study is an ex a mp l e of suc h il se rr.as • Autocorre l a tioncoeffi c ient s , also call e dsc rt arcorr e l a t io n coefficients (Pool e, 19 74 ) , mea sure th e re l at ive stre n g t hof associa t i onof pairs of observations formed bythe or ig inilJ time series sl idingbyoneob s ervati o n each time nea tdeacopy of itself (P l a t t and Denman, 1975) unt il the lug k,._1 is

(47)

31 reache d.Gen e r a lly , whenobs erva tio nskla gsapa r tha ve val ues of similarmagn i t ud e s thei r aut oc orrelat i o n coefficientswi l l be close to 1.O. while those witha larg e val ue succeeded by asmal l valueresults inthe coefficientha vi ng a value close to -1.0. Howe v e r, if such observations have very little re l ati o nshi p , the coefficient would be approximately zer o (Mo n t gome r y and Johnson, 1976).

The formulafor the autocorrelation fun c t i o ns according to Legendre and Legendre (1983)is:

(Equation5)

• autoca va ria nceatk I variance at zero lag ; kis the lag (k, ••• •.•k~_I );

ru{k) is the auto correlatio n co e f f i c i e nt s at k and ranges be tw e e n -1 and +1. The signs indicate the di r e c t i on of the relationship .

Autocorrel ationfunctions fo r the valuescfea ch indexin each da t a set were calc ul a tedwiththeMINI TABstatistical pa ck age by exec u t ing the autocorrelati on funct i on comma n d. Autocor relationcoefficientsforthe first 169 end th l". hi ghes t val ue for eachseries, fo r S, H'andEfor allda t a sets, wez-e tab u lated.Fiveexpla natory var i abl e swhich incl uded lat itude,

!lab!tat type, type of gear used, the resolution of the t ime

(48)

31.

series fo r each in d e ll,and the durat i on of sampling,were also incl ude d 1n the tab l e. Besides comp uti n g autocorre l ation coefficie nts fo r each lagof ea c h series,MINITABals oplo t t e d co r re l o g r a ms of th ecomputedvalues.

A correlogram is a plot of the values of the autocorre lationcoeffici entsve rsuslags intheseries(Poole, 1974 ; Ball an d Jenk i ns , 1976 ; Legendr e and Legendre, 1983). Whe nthe osci l lationsin a correlogram ra pidlydampensto zero ask increases, th isindicates that the oscillationsin the indices with time are not occurri ng in a "regular and periodic "manner (Poole, 1974 ). Iftheyare si n us o i d a l this indica tes that a regular per i od i c ptra ncme non exLata (Stephe ns on. 1978)•

'rtir e emai n patte r nsin cor re logramsfor S, H' andE were identified. A 's inus o ida l pattern' with each succes sive coefficient being sequentia l lyrelated both by ma gnitu deand direction ,was the firstki nd. Inthispattern, con nec tingthe sequ enceof coefficients resu l t e d in asi nu s o i d al curve. Th e secon d pa t tern was one wi th eachsuccessive coefficients being gene r a l ly of a similar mag ni tUde , and with a consta nt di r e c t ion . Thi s pa t t e r n resuLt e d in a "un i f o r ml y rtn oa r pattern'whe neach coeffic ientinthe sequence wasconne c t e d.

Fo r th e purposes of testing for rel a tionsh ips of thesetwo pa t t ernsa.n the co rrelo grams wit h the five var i ab l e s l isted above, they were classified tog e t her as 's mo ot h patt e r n '.

(49)

33

Th i r dl y . there wasthe typewith the magnitu deand direc tion of the coeffi cients being highly variable, giving II 'spiky pa t t er n'. That is, ll. plot of the coeff1cient s in the correlogralllappeared es spikes projecting In an irreg u lar ma nn erei ther in II positiveorneg a ti v e di rec tionInrel ati o n to lags plo tted on thex-exrs.

The correlogra ms of S. H' andE for each data set were inspected to de t ermi ne th e type of pa t t e rn as well as clas sify whether the au toco rrelati onpa t t e r n wa s .smoot h, or . spiky ' • Subseq uen tly. the appr -cp r-Lat. e patte r n of autoco r r elat ionfor each ind e x for eachdata set wa s rec ord e d wi theac hof the ftveexp l a natoryvariables.Forthepur po s e s of clar ityInreporting th ese re sults, par ticula r lywhen not al lind icesof a dat8 set were cla s s i f i e d together under one or the other of the two classes of eucccorrefe ctcn patterns, thedata sets were divided into three categories. The first category was 'a ll ind i c e s with slIlOOth correlogra ms '. This catego ry contai nedal l the da ta sets in wh ich thecorre l ograms for all three ind i c e s had a "s moot hpat tern '.Secon d l y, there was the ca t eg ory , 'all ind i ce s wi t h spiky corr elog r ams'. It conta ined the data sets whose c..;nrrel o g r ams for all thr e e ind i ce s had a ' spiky patter n'.Thethirdone wasthe"mixed' caf:p''1ory;itco n t e.ined data sets wit hone ortwoof the thre e Lndt ces belongingto eithe r the fi rs t or seco n dcat egory.

A. smooth correlogram' foranyindexme ansthatthe ind ex

(50)

34 iscor r ela t e d witha se ries of previ ou sva l ue s . The si nuso idal pattern that is ch a ract e rist i c of .smoot h cor re lograms ' ind i c a t e s that there is strong sea s on a lity In the indices (Stephenson , 19 78). •Spiky correlogra ms' ind i c a t e that the valuesof the indexconcernedarecorrelatedwithfewof the previou s value s.Thisillpl i eslitt l eor no seasonali tyIn the ind ex.

(51)

3 RESULTS

3.1 Data setscompiled for this study

All suit a b ledata sets acqu iredforthisstu dy arelis t ed in Ta ble81 and Tabl e82 (Ap pendix 8).The ordering in thetwo ta b lesisby latitude from north to south . Figu r eIshowsthe geogra ph ica llocationsofth evariousstu d iesfrom wh i c h th e da t a sets were compiled. Tab l e 81 (Appe n dix 8) liststhe differe nt da t a sets,theirgeographic alpo s i ti on in terms of latitude and long i t ud e, and the source reference for ea ch stu d y. The type of hab i t at andthe samplinggear(s)usedfor eac hstu d y ar e pr ese n t e d in Table 82 (Ap pe nd ix B). Most of these stu di es are fro meit h e r estuarine or inshore ma rin e hab ! tats. and mos t l yus ed sei ne andtraw l gear. The 22 dat a se t s compiled fulfiled all four crit e ri a: monthly or fortnightlysamples were col l ecte d fo r aperiod of si x mont hs or mor e; gear us ed for collecti ng da t a were speci fi e d; sampling pr oc edure s we re accurat e ly de s c r i bed to per mi t ca lcu la t ion of CPUE; taxonomic in for mat i o n was comp lete to spec ies lev e l. It is appa r en t from Figu re 1 tha t th e Nort h Ame ri c an Eas t Co a s t as a si ng l e geograph i c al region, is rel a tivelywell represe nted.

3.2 Effect of variation in catchabil.ity of gear

Theseaso na l pat tern of di ve r s i tyfrom the catch datawas fou ndnottobedi s t o rtedby bia ses incat c h ab i l i ty. aslong

(52)

Figure1.Geographical di s tribu tion of da t i) sets comp iled from th e literature for this study. 't'he numbers represent in gthe di f fere ntdata set correspondto the numbersgiven in Tables 81 an d82(Appendixm.

(53)
(54)

3.

as th ebia s was cons tant withinstudy. The dis pe rs ionof 300 val u esfor the snenec n -ar en e rind exofdive r s ity H' . and the spec i es evenn e ss ind e x E, calculated from JOD dat a sets si mulat e dwithrando ml y gen eratedcatchabll1t ycoeffi cien tsq, wer e plot tedIn Figure 2and Figu re 3, re s pe ctive l y. AJI tho val u esforthe twoind icesrepres en ti ng the original da ta Coli wi t hi nthemidd lebcseethat enco mp as sed 26'-75\of theva lue s in the res pec tive fr eq uen cy dis t ribut i on s for each 300 si mula t i o n for each sa mpli ng date.Most of tho s evalue s fo r the original data In both figur e s were very clo se to Che med ian.

The seasonal pat t ernfor H' andEareshown in f"iyu re,., and Fi gure5,re s pe ct ivel y.The se figure s comparethe seasona l

pat tern for ea c h index calc ulat e d from the original data or Mac do nal d i ! Al. (19 84' wit h thos e of two of the 300 si mu lations. FrORlth e comp arison of the two graphs it was app arent 'that 'the seasonalpa tt ern s ofH' and E in theelltc h dat adid not di ff erfro m tha t of the two si mu la ti ons .

3.3 Visua l ana l ysis of grap hsof seaso n alpatterns inS. H' and E

Visualan a l ysisofsi g ma-plo t gr aphsoCth o.!thr e e lndice!i sh o we d gene ralseas ona l and latitu di nelpatt e rns to exi s t In fish commu ni t ies . Th es e seasonal patter ns weremore appa rent for the North AmericanEastCoast; a geo graphical regio ntha t

(55)

Fig u r e 2.Graph for H' ca lculat ed from th e orig i n al data (Macdo nald g,t 9.1., 19 8 4) ( ---+-+- ) and "BoK and Whi ske r s " plots (I--CI:J-/) repres enting que r -eaLes for the disp ersi on of 300 va l ues of H' calcul a t ed fo r eachmo nthfro mda tasimulated witha const ant catc habi lity for each epecte e • The dash ed l ine s (-- ---) indi c ate no mont hlysamp les being tak en.

(56)

~

II

z

0(fl4:

J

r--

r-;

(J)

:;;

4:

:;; (f)

I

W-

e-

J

Z

J

0

J

::2

~

0

/ Z

~

-:

0

(fl <D r-, 4: (J)

N ~ ~ ~~ ",! ~ CO tD'<t" N

N N 0 0 0 0

':r:

(57)

Figu re 3. Gr a ph for E calcu lated fr om the origi na l data (Macdo nald ~ tl. , 1984 ) (- . - . -) and "Bo x and Whisk er s " plo tsO- -C D -i) repr esentin gqu a rtiles for th e dispersion of 30 0 values of Ecalc ulate d fo reach mon t h from data simul a t e dwit ha consta nt ca tcha b ility for eachspe cies. The dashedlines (---- -) indica teno mo nthlysa mples bei ng taken.

(58)

~

II

z a

(j)<l:

I

r-, r-,

rn

::;

<l:

::; (f)

1.L I

, J-

-, , , Z 0

~

0

::;

I

,-

Z

,-I

a

(j) <D r-,

rn

<l:

::;

L-

~

rn

00

I':

<D io

-e- n

N

~

a

0

a a o a a a 0

w

(59)

Figure4. Seasonalpa t t e r n ofil' versustime (months)for H' of originaldata (Macdonaldt l ~.. 198 4 ) and for H' of two othe r of the 300 simulations. The two simulations used for the comparison ar'bttr-e r Ljy chosen; the comparision was don e to show thatthe seasonal pa t ternof H' from simulated data is comparableto that of the originaldata.

(60)

H'

0.5

0.0

o

if o

;

0

~

/ .

v

o ·

o

H'GRIO

H' 1s t . v

H' 2

nd

(61)

Fi g u re5.Seasona l pat t e rn ofE:versus time(mon t hs) for E:of origi naldata (Macdona ld§!.,2!.., 1984 )and for Eof two othe r of th e 300 simulati ons . The two simulations used for the comparison arb!traril y chosen ; thecompa r ison wa sdo ne to show that th e seaso na l pattern of Efr om simu lated da ta is compa ra bleto tha tofthe originaldata .

(62)

1.0

0.8

v0

~

/

t-

0. 6 E

0.4 . h

0

E E

ORIG.1st

0.2 E 2

nd

0.0

I ! I I I I

MJ J A S O N O J FMAMJJASON

19 7 6 1977

Références

Documents relatifs

Give examples of complete, redundant and indeterminate systems in Whittaker classification

Intraspecific phenotypic variation and, thus, probably also genotypic variation among seed sources of salal is limited. Consistent differences existed only between each of the

functions reuches the extreme bound of 9 possible simplification.. function belongs to

We prove that the strong detonation travelling waves for a viscous combustion model are nonlinearly stable, that is, for a given perturbation of a travelling wave, the

These past habits are often translated as 'used to, or 'would.' Distinguish between the use of 'would' for habitual past actions (imparfait) and the use of 'would' for

 « this » est utilisé pour un objet/une chose/une personne qui est PRES de l’énonciateur (celui qui parle).. Exemple: This book is the best book

Parental involvement in ch ildren's reading : Results of a survey of Bre nt primary School Headte achers... The impa ct of the home literary env ironment on read

A longstand i ng set of presumpt ions have considered real folk music to be &#34;of the people&#34; and free of certain t y p e s of external contamination in the form of