This article was downloaded by: [University of Sydney]On: 19 March 2013, At: 12:59Publisher: Taylor & FrancisInforma Ltd Registered in England and Wales Registered Number: 1072954 Registeredoffice: Mortimer House, 37-41 Mortimer Street, London W1T 3JH, UK

New Zealand Journal of Marine andFreshwater ResearchPublication details, including instructions for authors andsubscription information:http://www.tandfonline.com/loi/tnzm20

Ecology of rocky reef fish ofnorth‐eastern New Zealand: A reviewGeoffrey P. Jones aa Department of Zoology, University of Auckland, Private Bag,Auckland, New ZealandVersion of record first published: 30 Mar 2010.

To cite this article: Geoffrey P. Jones (1988): Ecology of rocky reef fish of north‐eastern NewZealand: A review, New Zealand Journal of Marine and Freshwater Research, 22:3, 445-462

To link to this article: http://dx.doi.org/10.1080/00288330.1988.9516315

PLEASE SCROLL DOWN FOR ARTICLE

Full terms and conditions of use: http://www.tandfonline.com/page/terms-and-conditions

This article may be used for research, teaching, and private study purposes. Anysubstantial or systematic reproduction, redistribution, reselling, loan, sub-licensing,systematic supply, or distribution in any form to anyone is expressly forbidden.

The publisher does not give any warranty express or implied or make anyrepresentation that the contents will be complete or accurate or up to date. Theaccuracy of any instructions, formulae, and drug doses should be independently verifiedwith primary sources. The publisher shall not be liable for any loss, actions, claims,proceedings, demand, or costs or damages whatsoever or howsoever caused arisingdirectly or indirectly in connection with or arising out of the use of this material.

New Zealand Journal of Marine and Freshwater Research, 1988, Vol. 22: 445-4620028-8330/88/2203-O445$2.50/0 © Crown copyright 1988

445

Ecology of rocky reef fish of north-eastern New Zealand:a review

GEOFFREY P. JONES

Department of ZoologyUniversity of AucklandPrivate Bag, AucklandNew Zealand

Ab str act This review discusses the major factorsproducing spatial and temporal patterns inabundance and the structure of reef fish populationsin north-eastern New Zealand. It also examines thepotential impact of fish feeding on prey populationsand habitat structure. Major biological features ofthe habitat, such as the distribution of macroalgaeand echinoids, appear to affect fish populations at avariety of spatial scales. These spatial patternsremain coherent over time, despite temporalvariation in population densities, which can beindependent of habitat structure. Demographicstudies have not yet resolved whether pre- or post-settlement processes are most important in struc-turing reef fish populations. Feeding categories, anddetails of diet, prey selection, and the use ofmicrohabitats are discussed as necessary stepstoward examining the impact of fish as predators.Experimental studies have not demonstrated thateither carnivores or herbivores play a major role indetermining the biological structure of habitats, or inmodifying prey communities. A picture is emergingwhich suggests that the habitat has a far greaterimpact on fish populations than vice versa. It isstressed that the factors influencing patterns ofdistribution and abundance are likely to be species-specific; there is unlikely to be a general effectattributable to fish in shallow reef communities.

Keywords marine reef fish; distribution;abundance; habitat; recruitment; post-recruitmentprocesses; trophic organisation; prey selection;predation; herbivory

Received 29 June 1987; accepted 11 December 1987

INTRODUCTION

Subtidal rocky reefs in temperate waters harbour avariety of fish species which derive food and/orshelter from the reef substratum. Until ten years ago,these "reef fishes" had attracted only a little attentionfrom marineecologists (Clarke 1970; Stephensetal.1970; Bray & Ebeling 1975; Russell 1975; Hobson& Chess 1976). The term was virtually synonymouswith the diverse coral reef fish communities, whichhave been a topic of considerable interest (seereviews by Ehrlich 1975; Sale 1980,1984;Munro&Williams 1985). The number of studies on theecology of temperate reef fish has increasedexponentially over the past decade. Although thebody of information does not yet match that for theirtropical counterparts, a consideration of the status oftemperate reef fish ecology has been necessary forsome time.

Studies on the fishes inhabiting reefs of thenorth-eastern coast of New Zealand have made asignificant contribution to our knowledge oftemperate, subtidal ecology. The aims of this revieware to assemble this information, to describe theconclusions which have emerged, and to highlightareas in need of further work. The present literaturecomes mainly from studies carried out at theUniversity of Auckland's Marine ResearchLaboratory at Leigh. The original souce of much ofthis information can be traced to MSc and PhDtheses, but these will be cited only when the datahave not yet appeared in published form. In drawingthis information together, the ultimate goal is to seewhether any generalisations can be made for this onegeographic area.

Ecological studies on the reef fishes of north-eastern New Zealand have been concerned withrelationships between fish populations and benthiccomponents of reef communities, such as kelpforests and grazing invertebrates. These interactionshave been approached from two major perspectives:Many have attempted to assess the major features ofthe benthic habitat influencing the distribution and

Dow

nloa

ded

by [

Uni

vers

ity o

f Sy

dney

] at

12:

59 1

9 M

arch

201

3

446 New Zealand Journal of Marine and Freshwater Research 1988, Vol. 22

abundance of fishes inhabiting shallow reefenvironments. A second set of questions haveconcerned fish feeding activities, and the impact offishes on the structure of benthic algal andinvertebrate communities on reefs. These two broadtopics provide the framework for this review.However, this dichotomy is more for conveniencethan real. The effect of fish predation uponcommunity structure is likely to be closely linked totheir abundance patterns (Cheat 1982). Likewise,any depression in the abundance of prey species can,potentially, have strong feedback effects on theabundance patterns of fish predators. This reviewdiscusses the importance of fish-benthosinteractions, from the points of view of bothparticipants.

A brief historyThe first detailed study on the ecology of reef fish inNew Zealand was carried out between 1969 and1971 (Russell 1971), which included a generalaccount of the colonisation of artificial reefs(Russell 1975), the abundance and biomass of reeffishes (Russell 1977), and their feeding habits(Russell 1983). However, mostof the workfollowedthe establishment, in 1975, of New Zealand's firstmarine reserve between Cape Rodney and OkakariPoint, near Leigh. The baseline survey of the marinereserve, carried out by A. M. Ayling (1978),provided much information on the abundance andsize-structure of reef fish populations in different,biologically defined, habitats. Data on spatialpattern and temporal change in reef fish abundances,accumulated between 1975 and 1986, have recentlybeen published (Choat & Ayling 1987; Choat et al.in press). A study on subtidal encrustingcommunities, carried out between 1975 and 1976;was the first to examine the influence fishes mighthave on the structure of these communities (A. M.Ayling 1981). The first detailed investigation intothe population dynamics and demography of a reeffish species was on the ubiquitous paketi,Notolabrus (= Pseudolabrus) celidotus, and wasconducted between 1976 and 1981 (Jones 1980,1981a, 1981b, 1983,1984a, 1984b, 1984c; Jones &Thompson 1980; Thompson & Jones 1983). Fromthen on the number of studies has proliferated.Twelve theses dealing with aspects of the populationdynamics of reef fishes have been completed at theUniversity of Auckland over the past decade, with afurther five dealing with the potential effects thatfish may have on prey species.

DISTRIBUTION, ABUNDANCE, ANDSTRUCTURE OF REEF FISHPOPULATIONS

The fundamental questions in population ecology ofreef fishes concern the factors limiting or inducingchange in population numbers. The answers aredependent on the spatial scale at which thepopulation is defined, and the time scale over whichit is monitored. Before appropriate scales can bechosen, the variation in numbers and age/sizestructure in space and time need to be considered.The first stage is getting an accurate description ofthe changes in abundance in these two dimensions.During this phase, observations are made uponwhich to erect models about the processesunderlying these patterns. The second phaseinvolves establishing which processes are of majorimportance,bymakingpredictionsfromthemodels,and testing these predictions. Thestudy of temperatereef fish populations is essentially at the first stage.A great deal of information has accumulated onspatial and temporal patterns of abundance andpopulation size-structure, but few of themechanisms underlying these patterns have beenwell established. In this section I consider:(1) spatial patterns in population size and

structure;(2) temporal patterns in population size and

structure;(3) the demographic parameters (e.g., recruitment,

mortality) involved in establishing thesepatterns; and

(4) speculation about the ultimate factors limitingpopulation size and structuring reef fishcommunities.

Spatial patternsPatterns of density change have been considered ata number of spatial scales, ranging from differentgeographic locations throughout the North Island ofNew Zealand, down to differences in patch usewithin sites less than 0.01 km2 in size. To avoidconfusion due to the common use of differentterminologies, these scales of focus have beendivided into a number of categories: comparisonsamong different geographic "locations" (>5 kmapart), comparisons among "sites" within a geo-graphic location scale (<5 km apart) and a within-site (0.1 km2 or less) scale (Table 1). The area ofinterest, regardless of scale, can be stratified intodifferent types of habitat, defined on the basis ofphysical characteristics or dominant benthic

Dow

nloa

ded

by [

Uni

vers

ity o

f Sy

dney

] at

12:

59 1

9 M

arch

201

3

Jones—Ecology of rocky reef fish 447

organisms. Comparisons among habitats, andamong sites within habitats, can be made at anyspatial scale. For reef fishes, this has commonlybeen done at a within-location scale, and to a lesserextent, within sites (for the latter the term "micro-habitats" is usually applied). Such comparisonshave led to most of the insights into the factorsresponsible for different density patterns.

Density estimates have generally been madevisually along transects of different dimensions. It isbeyond the scope of this paper to assess critically thebiases and sources of error often inherent in thistechnique (see discussions by Leum & Choat 1980;Sale & Sharp 1983; Bortone etal. 1986). Importantis that the technique has been extremely useful formaking quantitative comparisons of speciesabundances among areas and over time. Triedescription of pattern is dependent partially upon mesize of the transect employed. In the past, the choiceof the dimensions has been based upon a variety ofcriteria, such as the size of habitat patches, the actualdensities recorded, and a variety of logistic con-straints. Traditional "standards", such as the 50 X 10m transect commonly used for large reef fish, may

have some comparative value, but are likely to beless appropriate for detecting pattern in somespecies. Formal statistical techniques are useful fordetermining the transect size and number whichoffer the greatest possible precision, within theconstraints of the time and effort available(Underwood 1981; Andrew & Mapstone 1987).These techniques have only recently been applied totemperate reef fishes (McCormick & Choat 1987),and their further use is to be encouraged.

Almost without exception, significantdifferences in density have been recorded at allspatial scales examined. However, it is themagnitude of the variation at each scale which isbiologically important, as this knowledgedetermines the scale at which it will be profitable toask questions about pattern and process. Theselevels of variation can be directly measured whenfully hierarchical sampling programs are employed(Underwood 1981; Choat & Bellwood 1985). Suchsampling designs have been used to some extent forreef fishes in New Zealand (e.g., Sylvester 1986;Choat & Ayling 1987), but not to a level where thevariance present at all scales of interest has beenpartitioned.

Table 1 Spatial scales examined in studies on the distribution and abundance of rocky reef fishesof north-eastern New Zealand. Spatial scale: Large, among locations; Medium, among sites withinlocations (B, between habitats; W, within habitats); Small, within sites. For explanation of eachscale see text.

Reference

Ayling 1978Choat & Ayling 1987Handford 1979Jones 1980Jones 1983Jones 1984aJones 1984bJones 1984cKingettfe Choat 1981Kingsford 1980Leum & Choat 1980MacDiarmid 1981MacDiarmid 1981McCormick 1986Meekan 1986Milicich 1986Mutch 1983Poynter 1980Russell 1977Sylvester 1986Thompson 1979

Spatial scale

Large Medium Small

Species B W

Large reef fish • •Large reef fish • • •Blennioids • •Notolabrus celidotus • •N. celidotus • •N. celidotus •N. celidotus, juveniles • •N. celidotus, adults • •Chrysophrys auratus «Chromis dispilus • <Cheilodactylus spectabilis • <Scorpis violaceus <Pempheris adspersus <Cheilodactylus spectabilis • •Odaxpullus «Parika scaber <Parapercis colias <Parika scaber <All reef fish «

•i •

t m

Large carnivores & Blennioids • • •Blennioids • •

Dow

nloa

ded

by [

Uni

vers

ity o

f Sy

dney

] at

12:

59 1

9 M

arch

201

3

448 New Zealand Journal of Marine and Freshwater Research 1988, Vol. 22w

Zea

land

.

z1V

BCA

O

1CA

n-sc

ale

•3to

B

I

ates

:

so13

bii

§

13

CIS

•s

:wee

n r<

£3oagoire

o

gnit

to

H

cto

fici

8• a

Q

oC A"is

a.

8

Q .

a

>Ou

o&ofletj

u

CA

'o

a

£22

ooooooooooodoo

S3 &

8 4to S S g

al> o8 .a

o oo g«i c n SJ h c niiP Q " § "i S 8

§

1.5 §j P If I

g 00 cS 60^0 -

3 O 3 O xl °

g<a t4C a g ,Q j£ 5/) 50

^ £2 •'S ̂ 3 £9 ̂ i« i5OJ *3 <U Hi ' s U *3 "3

S a b .-§ 8 S,2 2 9to

O

I

-3to 5? O

e ti .&.* M i s g g£^J3JJ3lS

I IITi

PATTERNS AMONG LOCATIONS: Not surprisingly,there are substantial changes in species compositionand abundance of reef fish faunas on a broadgeographic scale encompassing the North Island ofNew Zealand, and associated offshore islands to thenorth (Choat & Ayling 1987). Factors which haveproduced and currently maintain thesebiogeographic patterns are unknown. Choat andAyling note the distinct differences among offshoreislands in faunal composition, and among differentlocations along the mainland. When grouped interms of feeding categories, variation in densities atthis scale appeared to relate to habitat, with smallwrasses, regardless of species, dominating areas ofhigh algal cover. Larger carnivores were moreabundant in areas characterised by coralline algalflats grazed by urchins. Distinct i sland and mainlandfish communities have been linked to differences inhabitat-structure in a northern temperate region(Ebeling et al. 1980a, 1980b).

Interestingly, those studies comparingabundances of individual species among differentgeographic locations, and among different sitesbetween and within habitats, have found a similarrange of mean densities at both scales (Jones 1984c;Choat & Ayling 1987). For example, adults of thewrasse N. celidotus exhibit the same range ofdensities, when sites within the marine reserve atLeigh are compared, as they do for comparisonsamong locations sampled throughout the NorthIsland (Jones 1984c). Since marked changes indensity occur over a relatively small spatial scale, itis an important (and logistically feasible) scale uponwhich to focus questions about patterns ofabundance and their causes.

PATTERNS WITHIN LOCATIONS: Most descriptions ofspatial pattern in density have worked at anintermediate or "within-location" scale. Densitieshave been compared among randomly chosen sites(0.5-5km apart). Alternatively, locations have beenstratified in terms of habitats, and densities havebeen compared both among habitat types, andamong sites within habitats (Table 1). Significantdifferences among habitats and sites are the rule,rather than the exception. A. M. Ayling (1978)described the patterns of abundance and speciescomposition characteristic of seven differenthabitats in the marine reserve at Leigh. Species-specific patterns of abundance at this scale havebeen correlated with a variety of physical andbiological variables (Table 2). Biological factors,such as the amount of macroalgae versus turf-algae,

Dow

nloa

ded

by [

Uni

vers

ity o

f Sy

dney

] at

12:

59 1

9 M

arch

201

3

Jones—Ecology of rocky reef fish 449

and physical factors such as the degree oftopographic complexity, appear to have importantconsequences for fish populations and the structureof local communities. Although cause-effect linksbetween these variables have not been established atthis scale, they may be useful as predictors ofrelative abundance. Causation is probably a far morecomplex problem than these simple correlationswould imply.

The relationship between fish faunas and thedisjunct distributions of kelp, coralline-algal turfflats, and areas grazed by urchins, is clearly evidenton this medium scale. The densities of smallwrasses, such as N. celidotus, are positively relatedto the quantity of macroalgae (Jones 1984b; Choat &Ayling 1987), whereas larger carnivores such asChrysophrys auratus, Parapercis colias, and othersare more common in turf areas (Jones 1981a;Kingett & Choat 1981; Mutch 1983; Choat &Ayling 1987).

There is considerable evidence that topographiccomplexity, or the amount of relief, plays anotherfundamental role. Not only do the densities of anumber of individual species correlate with thisvariable (Table 2), the species richness and totaldensities of groups like tripterygiid blennies(Thompson 1979) and planktivorous fishes(Kingsford 1980) are also related in the same way.Topographic complexity may partially reflect thetype of rock substratum. Meekan (1986) found thatthe herbivorous fishes Odax pullus and Girellatricuspidata were more abundant on the dissectedgreywacke reef areas, than on the flatter sandstonereefs.

PATTERNS WITHIN SITES: The distribution ofindividuals within sites at any time appears to benon-random and related to a host of ecological andbehavioural factors, which probably interact in acomplex way (Jones 1984a; McCormick 1986).Behavioural factors can include intraspecificassociation or aggression (Leum & Choat 1980;Kingett & Choat 1981; Jones 1983,1984a; Mutch1983), interspecific association or aggression(Mutch 1983; Thompson & Jones 1983; Jones1984a), and spawning-site choice (Thompson 1979,1986; Jones 1981b).

Within-site variability is demonstrably due tothe effect of micro-habitat on the use of space. Allthe experi-ments which have linked densities tofeatures of the habitat have been done at this scale.Jones (1984b) showed that kelp removal caused adecrease in the densities ofjuvenile N. celidotus, and

kelp addition (via urchin removal) caused anincrease, presumably owing to an effect on localrecruitment patterns. Choat & Ayling (1987)showed that patch use by larger carnivores,including C. auratus, decreased when kelp forestsbecame established following urchin removal.Kingett & Choat (1981) showed by experimentalclearances, that coralline algal turf had a directeffect on patch use by C. auratus, activity beingreduced when turf cover was low. These studiesprovide clear evidence that these ecological factorsaffect the relative use of space at a within-site scale.Extrapolations of these effects to explain patternsamong locations or among habitats within locationsshould be made with caution.

POPULATION STRUCTURE: Knowledge of spatial andtemporal pattern in age structure is crucial to anunderstanding of the mechanisms causing change inpopulation size, since processes may notuniformally affect all age classes (Jones 1987).Also, constancy of overall abundances may mask anunderlying instability in the age composition.Unfortunately, attempts to compare age distri-butions at different locations have been made onlyfor one species (Jones 1980, 1981a). Due to diffi-culties in obtaining unbiased estimates of agestructure, size-frequency distributions arecommonly presented as an alternative descriptionof the structure of populations. For many questionsthese may actually be more appropriate, since lifehistory stages in fishes (such as entry to adultpopulations) and the ecological processes (such asprey selection) are often more dependent upon sizethan age.

Differences in the size-frequency distributionshave been detected at all spatial scales examined.Choat & Ayling (1987) note differences in the sizestructure of populations occupying macroalgal andurchin-grazed habitats, among locations, amonglocal sites, and among patches within sites. Severalworkers have recorded habitat and depth-relatedpatterns in size-distributions, larger fish often beingfound in deeper water (Leum & Choat 1980; Jones1981b; MacDiarmid 1981; McCormick 1986;Meekan 1986). These trends could have importantimplications with regard to spatial pattern inbreeding population size, and the impact of fishpredation.

Temporal patternsAt present most of the data on temporal variability inpopulation numbers spans the average length of

Dow

nloa

ded

by [

Uni

vers

ity o

f Sy

dney

] at

12:

59 1

9 M

arch

201

3

450 New Zealand Journal of Marine and Freshwater Research 1988, Vol. 22

MSc or PhD studies—which is too short, relative tothe life spans of most fish species, to get anappropriate measure of population stability.Ecological studies on other temperate reef and coralreef systems have suffered from similar constraints,with studies extending greater than 3-4 years theexception rather than the rule (Sale 1984; Ebeling etal. 1980b; but see 12-year study by Stephens et al.(1986) in California).

The single exception, for north-eastern NewZealand, was the 12-year survey of reef fishdensities made at Nursery Cove, Poor KnightsIslands, between 1975 and 1986 (Choat et al. inpress). All species exhibited substantial temporalchanges in density over this period, but some morethan others. Species with a subtropical distributiondeclined from moderate numbers in 1974 to nearlocal extinction in 1979, with one labrid species,Suezichthys aylingi, exhibiting a 15-fold fluctuationin numbers. Labrids which have southerndistributions, including Notolabrusfucicola and N.celidotus, showed much less variation in numbers.The magnitude of temporal change could be relatedto sea temperatures, with storm-induced disturbancealso paying a role (also see Ebeling et al. 1985).

Those studies which have combined adescription of spatial and temporal pattern over anumber of years, have shown that temporal variationis relatively small and spatial patterns remainconsistent over time. For example, Jones (1984c)estimated the abundance of N. celidotus at 10 sites(between and within habitats) in the Leigh MarineReserve, over a five-year period, and found that theranking of locations in terms of abundance remainedunchanged, despite measurable changes inabundance. The same appears to apply for thisspecies at the Poor Knights Islands (Choat et al. inpress), and is consistent with results for other reeffishes in north-eastern New Zealand (Thompson1979; Leum & Choat 1980; MacDiarmid 1981;Mutch 1983; McCormick 1986) and other temperateareas (e.g., Ebeling et al. 1980b).

Within-year sampling has shown that somespecies, such as Chrysophrys auratus, exhibitmarked seasonal changes in abundance (Kingett &Choat 1981). The spatial patterns established duringthe summer peak in abundance, however, areconsistent among years.

Demographic parametersThe distribution and abundance of a reef fish speciespotentially reflects survival through all life stages:egg, larval, juvenile, and adult. At present, opinions

are divided as to the relative importance of eventsoccurring during pre-settlement and reef-associatedphases. Early models considered only thoseprocesses affecting settled juveniles and adults (seeSale 1980). It was assumed that events happeningduring the early pelagic phase, which could affectnumbers settling (e.g., mortality, development,movement) were unimportant, since patternsestablished at the time of settlement were modifiedby interactions among juveniles and adults. Recentstudies on coral reefs have suggested that thedistribution and abundance of many species merelyrelect variation in numbers setting (e.g., Doherty1983; Victor 1986). The relative importance of theprocesses occurring in the two phases has yet to beresolved for any species (Jones 1987). Certainly, theparameters responsible for spatial pattern may betotally different from those bringing about temporalchange. In this section I discuss the information thatis available on recruitment and potentially importantpost-recruitment processes such as mortality,maturation, and movement. Kingsford (1988) hasreviewed the information available on larvalprocesses.

RECRUITMENT: The potential importance ofrecruitment has been a topic of great interest in rockyreef fish studies. All reef fishes in north-eastern NewZealand recruit seasonally, most species during themonths of spring and summer (Thompson 1981;Ayling & Cox 1982). Actual numbers settling haveproved difficult to assess, so the term "recruitment"usually refers to estimates of the densities of recentlysettled juveniles. Recruitment to the reefenvironment will only reflect numbers which havesettled, if the latter is not modified by immediatepost-settlement processes (Connell 1985). Milicich(1986) has come the closest to examining settlementpatterns in a temperate reef fish. Input of pre-settledParika scaber into the reef system were monitoredevery two days, by sampling kelp plants suspendedon ropes near the surface waters. Pre-settledjuveniles of many species are attracted to drift-algalmaterial (see Kingsford 1988). Patterns ofrecruitment to kelp plants attached to the reef werealso monitored. The temporal patterns ofrecruitment partially reflected changes in theavailability of potential settlers.

Recruitment appears to be highly variable bothspatially (within and among locations) andtemporarily (within and among years) for mostspecies (Kingsford 1980; Poynter 1980;MacDiarmid 1981; Mutch 1983; Jones 1984b; but

Dow

nloa

ded

by [

Uni

vers

ity o

f Sy

dney

] at

12:

59 1

9 M

arch

201

3

Jones—Ecology of rocky reef fish 451

see Thompson 1979). This generalisation may applyequally well to other temperate locations (Cowan1985; Stephens et al. 1986) and to coral reefs (Sale1984; Munro & Williams 1985). Such variationprobably makes a major contribution to changes inthe abundance and composition of reef fish faunas innorth-eastern New Zealand, between areas and overtime (Jones 1984b; Choat et al. in press). Wherespatial and temporal patterns in recruitment havebeen considered together, differences among arrashave been greater than differences among times(Jones 1984b). The magnitude and direction ofchange appears to be consistent over a number ofsites within a location. Hence, certain sites receiveconsistently higher recruitment than others.

The causes of spatial variation in recruitmentprobably differ from those bringing about temporalchange. Spatial variation in recruitment has beenshown to relate to habitat, but the relativeimportance of habitat selection, before or aftersettlement, or early differential survival is notknown (Jones 1984b). Temporal variation inrecruitment, within and between seasons, probablyrelates to a variety of factors including spawningpatterns, processes influencing larval survival anddevelopment, processes transporting fullydeveloped larvae back to the reef, and long-termchanges in habitat structure (Kingsford 1980; Jones1984b; Milicich 1986; Choat & Ayling 1987).

POST-RECRUITMENT EVENTS: Obtaining reliableinformation on post-recruitment processes whichmay affect adult population size and age/sizestructure has high priority (Sale et al. 1985). Spatialpatterns in recruitment are positively correlated withadult density patterns in some species, but certainlynot in all. No obvious temporal relationshipsbetween interannual variation in recruitment amdsubsequent changes in adult numbers have beenrecorded (e.g., Jones 1984b, 1984c). However, withthe exception of Choat et al. (in press), the periodencompassed by most studies would probably be tooshort to detect such patterns.

Variation in adult densities may be affected bychanges in post-recruitment mortality rates,maturation times, and systematic movements, noneof which have received a great deal of attention inreef fish studies (but see Doherty 1983; Jones 1984c,1987; Eckert 1987). Differential juvenile mortalityappears to have a substantial effect on adultdistributions in the blenny Forsterygion varium innorth-eastern New Zealand (Thompson 1979). Thisis one of the few species in which juveniles tend to

recruit fairly evenly across a variety of habitat types.However, survival appears to be much higher at siteswith greatest topographic complexity, whichclosely correlates with adult numbers. Thesepatterns are unlikely to be explained by movement.Behavioural studies on F. varium show juveniles tobe extremely site-attached; they will in fact return totheir home ranges when displaced enormousdistances (Thompson 1979, 1983).

Other workers in this area have stressed thepotential importance of post-recruitment mortalityand growth (Jones 1984c; Milicich 1986). Jones(1984c) showed that although spatial patterns inadult density in N. celidotus were correlated withrecruitment, the relationship was not linear. Bothdensity-dependent growth and mortality appear torestrict input to the adult population at highrecruitment levels. Density-dependent growth isimportant, because it means that the time taken toreach maturity is prolonged with increasingdensities (Jones 1984c, 1987). There is someevidence of social control of maturation in female N.celidotus at high densities (Jones & Thompson1980). Jones (1984c) argued that the relativeimportance of recruitment and post-recruitmentevents, shifted in favour of the latter, at sites whichconsistently exhibited high recruitment levels.

The role of movement obviously becomesincreasingly important in determining pattern atsmaller spatial and temporal scales. It is clearly oneof the primary factors operating at a within-site scale(see previous section). Local movements on a diel ortidal basis are common, and known to be associatedwith breeding, foraging, and resting activities (Bray1981; Jones 1981b; McCormick 1986; Meekan1986). It may also be implicated in many of thepatterns in abundance and size-frequency evident ata broader scale. Some species show a tendency torecruit into shallow water, and appear tosystematically shift into deeper water as they grow(e.g., Cheilodactylus spectabilis: Leum & Choat1980; McCormick 1986). The summer influx of 0+and 1+year age classes of C. auratus into the reefsystem would appear to be the result of fairly large-scale movements (Kingett & Choat 1981). Reefpopulations of the parore, G. tricuspidata, may bereplenished by migration from sheltered harbourreefs, where they appear to settle in large numbers.

Limiting processesThe question of what ultimately limits populationsize in the reef fishes of New Zealand has yet to beanswered. A variety of processes are likely to be

Dow

nloa

ded

by [

Uni

vers

ity o

f Sy

dney

] at

12:

59 1

9 M

arch

201

3

452 New Zealand Journal of Marine and Freshwater Research 1988, Vol. 22

operating before, during, and after recruitment; theirrelative importance within any one species is likelyto change, as spatial and temporal scales are altered.There is no reason to expect generalisations amongspecies, since each exhibits distinct patterns ofecology and behaviour, both during larval (seeKingsford 1988) and benthic phases of the lifehistory. The multitude of correlations betweenabundance and habitat variables, such as thepresence of kelp and turf, may offer clues as toimportant processes, but they are just as revealing asattempts to find simple solutions. The small-scaleexperiments linking distributions directly to thesevariables (Kingett & Choat 1981; Jones 1984b;Choat & Ayling 1987) clearly establish theirimportance at one level, but raise more questionsthan they answer. Do these factors determine larger-scale patterns of distribution and abundance? Arethese habitat associations based on the availabilityof preferred food or shelter, or a complex interactionbetween the two? How important are habitat selec-tion by larvae, post-settlement survival, or move-ment, in establishing these larger-scale patterns?

The potential roles of competition and predation,in relation to these population patterns and habitatassociations, have not been fully considered.Thompson (1979) suggested that the site-specificmortality patterns in F. varium may be due todifferential predation pressure in habitats offeringless shelter. Jones (1984a, 1984b) could find noevidence of intra-specific competition betweenadults and small juveniles in N. celidotus, butpresented one interesting scenario to explaindensity-dependent growth and mortality. Hesuggested that juveniles were restricted to kelpheads as a shelter site from predators, and, as a result,were subject to intra-cohort competition for food.These arguments are mere speculation, but serve toremind how complex the relationships may bebetween competition for resources and theavoidance of predators.

The diet of N. celidotus of intermediate sizesoverlaps considerably with that of the blennioid fishF. varium, which behaves aggressively toward thewrasse (Thompson & Jones 1983). There is strongevidence from removal experiments that the twospecies compete, but only while the wrasse passesthrough a restricted size range. There is nosuggestion that this interaction affects patterns ofdistribution and abundance. Large-scale effects ofinterspecific interactions have been recorded insome Californian reefs (Hixon 1980; Larson 1980),but not in New Zealand.

Sorting out the relative importance of processesoccurring during pelagic and reef-associated phaseswill require comprehensive studies, which focus onboth portions of the life cycle at the same time. Weneed to monitor the fate of successive cohortsrunning the gauntlet of the pelagic environment,managing to settle in the reef habitat, and survivingto breeding age (e.g., Stephens et al. 1986).Experiments testing the importance of differentfactors at each stage must not only establish that aprocess is operating, but must establish the degree towhich it contributes to the observed patterns in spaceand time (Weinberg et al. 1986).

FISH FEEDING AND THE STRUCTUREOF REEF COMMUNITIES

A growing generalisation that fish feeding has amajor impact on prey populations and the structureof communities in temperate waters was challengedby Choat (1982). He stressed the need for morefundamental information on predator-preyinteractions and urged a cautious interpretation ofexperiments carried out without this knowledge.The potential impact of a fish species on benthiccommunity structure cannot be assessed withoutbasic information on its diet, use of microhabitats,and feeding behaviour. These may vary betweenareas and over time, in response to changes in thedistribution and abundance of prey species. Anindividual may contribute to different effects withtime, as it undergoes ontogenetic changes in diet,and in its choice of microhabitats. The impact will bedependent on how much fish predation contributesto the mortality in prey populations, and can beassessed, unequivocably, only by the experimentalmanipulation of fish feeding pressure overappropriate spatial and temporal scales. Thenecessary dietary information on fish feeding overrocky reefs in north-eastern New Zealand isaccumulating, and is summarised here, along with adiscussion of the preliminary experiments designedto test the effects of fish predation.

Trophic organisation and dietsWhen considering broad trophic categories, the reeffishes of north-eastern New Zealand are primarilycarnivorous. Of 44 "reef species examined byRussell (1983), 82% were carnivores, 11% wereherbivores and 7% were omnivores (Table 3). Theseproportions appear to be broadly similar in othertemperate reef communities (Hobson 1982), and insome coral reef communities, although herbivores

Dow

nloa

ded

by [

Uni

vers

ity o

f Sy

dney

] at

12:

59 1

9 M

arch

201

3

Jones—Ecology of rocky reef fish 453

Table 3 Trophic organisation of the reef fish community around Goat Island, north-eastern New Zealand. Dietinformation from Russell (1983); abundance information from Russell (1977) includes % of total abundance and totalbiomass. NI = not included in Russell's (1977) survey.

Species Family % Abundance % Biomass

CARNIVORES

Benthic Feeders:Myliobatus tenuicaudatus (Hector)Gymnothorax prasinus (Richardson)Conger wilsoni (Bloch & Schneider)Dellichthys morelandi BriggsTrachelochismus melobesia PhillipsLotella rhacinus (Bloch & Schneider)Pseudophycis breviuscula (Richardson)Zeusfaber LinnaeusScorpaena cardinalis RichardsonEllerkeldia huntii (Hector)Chrysophrys auratus (Bloch & Schneider)Upeneichthys lineatus (Bloch & Schneider)Paristiopterus labiosus (Giinther)Chironemus marmoratus GuntherCheilodactylus spectabilis HuttonNemadactylus douglasii (Hector)Latridopsis ciliaris (Bloch & Schneider)Coris sandageri (Hector)Notolabrus celidotus (Bloch & Schneider)N.fucicola (Richardson)Pseudolabrus miles (Bloch & Schneider)Parapercis colias (Bloch & Schneider)Ruanoho whero HardyBellapiscus medius (Giinther)Notoclinops segmentatus (McCulloch & Phillips)Forsterygion capito (Jenyns)F. variant (Bloch & Schneider)Forsterygion sp. (see sp.b, Thompson 1981)Notoclinops yaldwyni Hardy

Open-water feeders:Optivus elongatus (Gunther)Caesioperca lepidoptera (Bloch & Schneider)Pseudocaranx dentex (Bloch & Schneider)Pempheris adspersus GriffinChromis dispilus GriffinObliquichthys maryannae HardyScorpis lineolatus Richardson

HERBIVORES

Girella tricuspidata (Quoy & Gaimard)Kyphosus sydneyanus (Gunther)Parma alboscapularis Allen & HoeseAplodactylus arctidens RichardsonOdaxpullus (Bloch & Schneider)

OMNIVORES

Parablennius laticlavius (Griffin)Parika scaber (Forster)Scorpis violaceus (Hutton)

MyliobatidaeMuraenidaeCongridaeGobiesocidaeGobiesocidaeMoridaeMoridaeZeidaeScorpaenidaeSerranidaeSparidaeMullidaePentacerotidaeChironemidaeCheilodactylidaeCheilodactylidaeLatridaeLabridaeLabridaeLabridaeLabridaeMugiloididaeTripterygiidaeTripterygiidaeTripterygiidaeTripterygiidaeTripterygiidaeTripterygiidaeTripterygiidae

BerycidaeSerranidaeCarangidaePempheridaePomacentridaeTripterygiidaeKyphosidae

KyphosidaeKyphosidaePomacentridaeAplodactylidaeOdacidae

BlenniidaeMonacanthidaeKyphosidae

93.1

15.4>0.1

0.2>0.1NINI

>0.1>0.1>0.1

1.40.20.61.7

>0.14.20.6

>0.1>0.1>0.1

4 01.20.30.8NININININININI

77.74.50.42.8

20.51.3NI

51.2

4.9

4.7>0.1>0.1

0.1>0.1

1.8

NI1.8NI

57.5

39.9>0.1

4.63.9NINI

>0.1>0.1

1.10.50.21.12.3

>0.17.46.63.4

>0.10.85.34.50.71.8NININININININI

17.60.40.43.31.80.5NI

11.2

38.0

36.0>0.1>0.1

1.8>0.1

3.4

NI3.4NI

Dow

nloa

ded

by [

Uni

vers

ity o

f Sy

dney

] at

12:

59 1

9 M

arch

201

3

454 New Zealand Journal of Marine and Freshwater Research 1988, Vol. 22

reach 20-30% at some tropical locations (Goldman& Talbot 1976; Hobson 1982; Williams & Hatcher1983).

Russell (1983) divided carnivores into two broadcategories. The majority of the species (81 %) are the"benthic feeders", which primarily feed on mobileinvertebrates. The group is represented by relativelylarge families, such as labrids and tripterygiids, andnumerous species, including red moki, C.spectabilis, the goatfish Upeneichthys lineatus andthe commercially important snapper C. auratus(Table 3). The most important prey categories areamphipods, with crabs, gastropods, errantpolychaetes, and small bivalves utilised heavily(Russell 1983). Several species include small reeffish in their diets, but only one, Zeusfaber, is totallypiscivorous.

The other group of carnivores are "open-waterfeeders", which include fish that shelter amongreefs, but feed in the water column on zooplankton(Table 3). Although benthic feeders predominate interms of number of species, Russell's (1977)censuses at Goat Island showed that open-waterfeeders make up 78% of the total number ofindividuals recorded. However, these censuses didnot include the numerous small tripterygiid blenniesand gobiosocids, which are largely benthic feeders.The most abundant open-water feeders are thesweep Scorpis lineolatus and the bigeye Pempherisadspersus. The numerical dominance of open-waterfeeders over large carnivores is reversed whenestimates of biomass are considered (Russell 1977).Benthic carnivores account for 40% of the total reeffish biomass, while open-water feeders make up18% (Table 3).

The herbivorous fish of temperate reefs areprimarily "browsers", feeding selectively on foliosemacroalgae. This contrasts with tropical coral reefswhere the predominant herbivorous families, suchas scarids and acanthurids, are "grazers", feedingnon-selectively on microalgae. In north-easternNew Zealand, herbivorous fish, the most common ofthese being the parore, G. tricuspidata, primarilyconsume red or green algae. The only exception isthe odacid, 0. pullus, in which adults feedexclusively on large brown algae. Numerically,herbivores made up only 5% of the reef fish recordedby Russell (1977), but this figure expanded to 38%of the total reef fish biomass (Table 3).

Inter- and intraspecific differences in dietMore recent detailed studies on the diets of indi-vidual species suggest that the above categorisation

may be an oversimplification, which neglects theconsiderable interspecific differences, andontogenetic, spatial, and temporal variation in preyconsumption within a species. Russell's (1983)paper points to a high degree of partitioning of foodresources, at least among adults of most species. Thelevel of partitioning may be inflated, since hisspecimens came from a variety of locations andhabitat types. However, studies on ecologicallysimilar species collected from within the samehabitat types have also revealed significantdifferences among species in prey composition(Thompson 1979; MacDiarmid 1981; Sylvester1986).

What is not apparent from Russell's (1983)general dietary study is that juveniles and adultshave almost totally different diets in 11 of the 13species which have been examined in detail (Table4). Not only do the relative proportions of differentprey categories change as fish grow, but the meansizes of prey types also increase (MacDiarmid 1981;Thompson & Jones 1983; Jones 1984a). Even inspecies in which the dominant prey types do notchange with growth, there are subtle differences inprey sizes and proportions (Clements 1985;McCormick 1986). Thus, a species may have animpact on a much wider range of prey categories andsizes than would be evident from Russell's (1983)study.

When juveniles are considered, the importanceof gammarid amphipods as a food source is startling(Table 4). Nearly all benthic carnivores, and evensome herbivores (e.g., 0. pullus: Clements 1985)grow through a phase when a large proportion oftheir diet consists of these small crustaceans. Somebenthic carnivores, such as C. spectabilis and U.lineatus, which appear to be relatively non-selectivein their feeding, continue to consume amphipodsthroughout their lives. In general, adult fish appearto feed on a much wider range of prey categories, andthis is evident both within and among species.

In considering the importance of predator-preyinteractions, both spatial and temporal variation indiets may be extremely important. In the fewinstances that spatial variation has been consideredfor temperate reef fish, substantial differencesamong geographic locations have been found(Cowan 1986). In New Zealand, juvenile AT.celidotus have been shown to feed almostexclusively on benthic substrata at the Leigh MarineReserve, but feed on pelagic zooplankton at a high-density site on Takatu Peninsula (Jones 1983,1984a). Kingsford (1980) found that Chromis

Dow

nloa

ded

by [

Uni

vers

ity o

f Sy

dney

] at

12:

59 1

9 M

arch

201

3

Jones—Ecology of rocky reef fish 455

dispilus, at the Poor Knights Islands, consume muchlarger quantities of euphausiids and copepods thanthey do near Leigh, on the mainland. Local-scaledifferences among habitats have been recorded bySylvester (1986). He found that the diets of somespecies (e.g., N. celidotus and P. colias) differedmarkedly between shallow turf flats and deep reefs,while others (e.g., U. lineatus, C. auratus, and mostblennioids) did not. Differences among locationswithin habitats may also be important, but thesehave rarely been considered.

Temporal variation in diet can also bepronounced. Several species, such as P. scaber, N.celidotus (Jones unpubl. data), and Pseudocaranxdentex (Russell 1983) are known to switch betweenbenthic feeding and open-water feeding at irregularintervals. Juvenile P. adspersus feed diurnally onsmall copepods near the substratum, but adults feedon zooplankton in the water column at night(MacDiarmid 1981). On a larger time scale, thereappear to be seasonal changes in the diets of most

benthic carnivores (Thompson 1979), open waterfeeders (Kingsford 1980; MacDiarmid 1981), andherbivores (Clements 1985), which have beenexamined.

Foraging patterns within habitatsBroad-scale and among-habitat differences in theeffects of fishes may reflect changes in fish density.Within-habitat variation may reflect preferences forforaging in certain microhabitats. Microhabitats areconsidered to be physically, or biologically,distinguishable feeding substrata, which occur on aspatial scale such that individuals may encounter theentire range present at a site within their normalhome range. As with diet, detailed knowledge offeeding substrata is essential when examining therole of fish predation, since fish feeding will bepatchy at a within-site scale.

There appears to be a high degree of partitioningof microhabitats among co-occurring species intemperate waters (e.g., Bray & Ebeling 1975;

Table 4 Dominant prey categories taken by juveniles (<100 mm SL) and adults of severalreef fishes.

Species

Aplodactylus arctidensCheilodactylus spectabilis

Chromis dispilus

Chrysophrys auratus

Fosterygion varium

Notolabrus celidotus

N.fucicola

Odax pullusParapercis colias

Parika scaber

Pempheris adspersus

Scorpis violaceus

Upeneichthys lineatus

Juveniles

red algaeamphipods

copepods

amphipods

amphipodsgastropods

amphipods

amphipods

red algaeamphipods

amphipods

copepods

copepods

amphipods

Adults

red algaeamphipods

appendiculariafish eggscopepodscrabsurchinshermit crabscrabsgastropodsmolluscscrabscrabshermit crabsmolluscsbrown algaefishurchinsmolluscsspongesascidiansred algaeamphipods

appendiculariafish eggsred algaecrabsamphipods

Reference

Clements (1985)Leum & Choat (1980);McCormick (1986)Kingsford (1980)

Choat & Kingett (1982);Russell (1983)Thompson (1979)

Jones (1984a)

Jones, unpublished

Clements (1985)Mutch (1983)

Poynter (1980);Russell (1983)

MacDiaimid (1981)

MacDiarmid (1981)

Choat & Kingett (1982);Russell (1983)

Dow

nloa

ded

by [

Uni

vers

ity o

f Sy

dney

] at

12:

59 1

9 M

arch

201

3

456 New Zealand Journal of Marine and Freshwater Research 1988, Vol. 22

Ebeling & Laur 1986). Sylvester (1986) studied twogroups of fishes, large carnivores and blennioids,which feed over open rocky areas. He foundsignificant differences among species in the use ofmicrohabitats within both of these groups. Acomparison of two habitats, rock flats and deepbroken rock, indicated that large carnivores differedin their use of microhabitats between the twohabitats. However, it is not certain whether this isdue to changes in microhabitat preference or micro-habitat availability.

Foraging preferences within a species oftenchange as individuals grow larger, both in north-eastern New Zealand (Poynter 1980; Mutch 1983;Jones 1984a; McCormick 1986; Sylvester 1986)and other temperate regions (Holbrook & Schmitt1984; Schmitt & Holbrook 1984). For example, N.celidotus and P. scaber switch from feeding onamphipods in the macroalgal canopy when small, tofeeding on a range of invertebrates from rockysubstrata as adults (Poynter 1980; Jones 1984a).Jones established that both adults and juveniles werefeeding non-randomly with respect to theproportional cover of these microhabitats. Adult P.celidotus avoided foraging on rock surfaces beneaththe algal canopy where prey numbers were low.Other species, such as the generalist carnivoreCheilodactylus spectabilis, primarly forage amongturf algae throughout their lives, although there isgreater emphasis on foraging in crevices onophiuroids as they grow larger (McCormick 1986).Planktivores can also exhibit changes in foraginglocation with size, with larger fish tending to foragehigher in the water column (Kingsford 1980).

Prey abundance, availability, andselectionA description of spatial and temporal pattern in preyabundance, within appropriate microhabitats, iscritical to interpreting experiments and assessing theimportance of fish predation. The above patterns indiet and foraging may reflect changes in preyabundance, prey availability, and/or prey selection.Prey abundance does not equate directly withavailability in most circumstances, since the lattercan be affected by a variety of factors other thanabundance, such as prey mobility, visibility, andbehaviour (Main 1985). However, the relativeavailabilities of different prey types is difficult toquantify. Diet and prey abundance measures can becompared to obtain the degree of "apparent" preyselection, which can indicate the prey species mostlikely to be affected by predation.

There have been a few attempts to quantify preyabundance and examine apparent feedingselectivity in benthic carnivores. Jones measureddifferences in the abundance of prey items taken byboth juveniles and adult N. celidotus, both amongmicrohabitats (Jones 1984a) and habitats (Jones1984b, 1984c). Juveniles, restricted to foragingwithin macroalgae, take microcrustaceans inproportion to their abundance. Adult wrasses aremuch more selective. This pattern of greaterapparent feeding selectivity in adults is also true forP. colias (Mutch 1983) and C. spectabilis(McCormick 1986), although the role of availabilityneeds to be established. Cheilodactylids have beendescribed as essentially "functional grazers", takingbenthic organisms non-selectively (Choat 1982).However, McCormick's study on C. spectabilisshows that adults consume prey in proportionswhich differ from those present in turf, suggestingsome degree of selectivity. Sylvester (1986) is theonly study which has distinguished amongamphipod species when comparing diets and preyabundance. He found no evidence that either largecarnivores, such as C. spectabilis and U. lineatus, orblennioids, were consuming some amphipods inpreference to others. The most abundant speciespresent in the turf, such as Gammaropsis typica,Podocerus karu, and Ischyocerus longimanus, wereconsumed in proportion to their abundance.

Prey selection has also been examined inplanktivorous fishes (Kingsford 1980; MacDiarmid1981). The large degree of temporal variation in thediets of planktivores reflects changes in thecomposition of the zooplankton. Planktivoresappear to be very selective, but a category beingselected at one time may be rejected at another. Forexample, Scorpis violaceus and Chromis dispilusselect appendicularia during the winter months, butin spring-early summer, these are rejected in favouroffish eggs which are present at this time (Kingsford1980; MacDiarmid 1981).

The impact of carnivoresThe large biomass and variety of benthic feedingcarnivorous fish has led to suggestions that they willhave a major influence on the structure anddynamics of benthic invertebrate communities(Bakus 1969; Russell 1983). They may have effectsdirectly attributable to prey consumption, ormechanical disturbance of the habitat when feeding.As an example of the latter, Battershill (1987)suggested that silt-sorting by the goatfish, U.lineatus, may influence the recruitment of sponges

Dow

nloa

ded

by [

Uni

vers

ity o

f Sy

dney

] at

12:

59 1

9 M

arch

201

3

Jones—Ecology of rocky reef fish 457

in some areas. Nocturnal planktivores, such as P.adspersus, which feed on demersal zooplankton,may also produce a feeding-effect on benthiccommunities. Diurnally feeding planktivores maylocally deplete the abundance of zooplankton beingtransported above reefs (Kingsford & MacDiarrnidunpubl. data) and there is a suggestion that this canaffect the recruitment of some benthic organisms(Gaines & Roughgarden 1987). Aggregations ofopen-water feeders in temperate waters may alsohave positive effects through defecation andammonium excretion, which can enrich the habitat,promoting the growth of benthic organisms (Bray etal. 1981,1986).

Studies on the role of carnivorous fish aspredators in north-eastern New Zealand can bedivided into those examining effects on sessileencrusting organisms, large mobile grazers (e.g.,echinoids), and microcrustaceans.

Parika scaber is the only fish in north-easternNew Zealand known to consume encrustinginvertebrates, such as sponges and ascidians(Russell 1983). The data available suggest that itmay have a localised effect on the cover of somepreferred species, but is probably not of generalimportance as a structuring force in thesecommunities. A. M. Ayling (1981) tested theireffect by conducting an exclusion experiment at twolocations supporting low and high densities of P.scaber in the Leigh Marine Reserve. The exclusiontreatments were simple, but ingenious, consisting ofa horizontal shield close to the substratum. Fishcould swim underneath, but could not orientatevertically to feed. The exclusion had a substantialeffect on the cover of a number of preferred sponges,ascidians, and polyzoans at the high-densitylocation, but not at the other. P. scaber only appearto reach those high densities in deep reef habitats,which make up a relatively small proportion of thetotal reef area. Further experiments at other sites inthe Leigh Marine Reserve have found no effects ofpredation by P. scaber on sponges or ascidians (A. L.Ayling 1978; Stacker 1984), but localised effects ofmonacanthid fishes have been detected in temperateAustralian waters (Russ 1980; Keough 1984).

Choat & Ayling's (1987) comparison of fishdistribution with habitat types was conceived toexamine the potential for fish to structure benthichabitats through effects on large invertebrategrazers, such as the echinoid Evechinus chloroticus.As a general pattern, urchins and associated grazinggastropods appear to dominate subtidal reefs at anintermediate depth range, producing relatively

stable coralline crust/turf belts (Choat & Schiel1982; Schiel 1988). Large fish preying on thesegrazers preferentially forage over the urchin-grazed/turf habitats, rather than within thosedominated by macroalgae. Hence, their feedingwould not be responsible for maintaining theintegrity of kelp forests by removing algal grazers.

It is possible that heavy predalion in urchin-grazed areas eventually leads to the establishment ofkelp forests. Andrew & Choat (1982) excludedpredatory fish from plots in a heavily urchin-grazedarea to test what impact fish were having on urchinnumbers. While fish did reduce the densities ofcrevice-bound juveniles, enough survived to anadult size to maintain kelp free areas. Thus, whilefish may be having an effect on certain size classesof important grazers, it may not follow that they areindirectly responsible for general patterns in habitatstructure (see also Andrew 1988). At othertemperate locations, fishes may assume greaterimportance as predators on sea urchins (Cowan1983).

Since amphipods are the most important foodsource consumed by benthic carnivores, an effect offish predation upon them has been considered likely,and to have important consequences. Manyamphipods and other microcrustaceans are grazers,and are thought to play an important role in affectingmacroalgae and seagrass communities in theNorthern Hemisphere (Bernstein & Jung 1979;Zimmerman et al. 1979; Brawley & Adey 1981;Norton & Benson 1983). Indeed there is a general,although largely unsubstantiated, opinion that fishregulate the abundances of microcrustaceansassociated with vegetated habitats (see Choat 1982).

There is no strong support for the latter viewfrom studies carried out in north-eastern NewZealand, although it remains to be unequivocablytested. Gammarid amphipods are a conspicuouscomponent of the fauna associated with corallineturf and macroalgae (Jones 1981a, 1984a, 1984b;Choat & Kingett 1982; Sylvester 1986). A summerpeak in abundance coincides wi th the recruitment ofmost reef fishes, which rely heavily on amphipods asa food source at this time. Choat & Kingett (1982)excluded fish from foraging over 2 rn2 plots of turfin an area frequented by U. lineatus and juvenile C.auratus. This 8-month study did not demonstrate aneffect of fish predation on amphipods, ostracods, orpolychaetes at this scale. However, given the likelymobility of amphipods, which may emerge into theplankton at night (MacDiarmid 1981), any effect ofpredation may only be measurable over a

Dow

nloa

ded

by [

Uni

vers

ity o

f Sy

dney

] at

12:

59 1

9 M

arch

201

3

458 New Zealand Journal of Marine and Freshwater Research 1988, Vol. 22

considerably broader scale. If there really is noeffect of predation, it may be because coralline turfis such a good refuge for amphipods that fish areexploiting only an insignificant proportion of theprey present. The role of turf as a refuge from fishpredation has been examined elsewhere in NewZealand (Coull & Wells 1983).

Jones (1981a) is the only study to examine thepotential impact of small labrids on themicrocrustacean fauna associated with macroalgae.Fish were excluded by placing mesh bags overindividual plants. Although this caused a massiveincrease in gammarid numbers, the predation effectcould not be distinguished from an effect of the bagsthemselves. A preliminary examination of an areadenuded of wrasses suggested that, with appropriatereplication, such clearance techniques would bemuch more effective to examine fish impact.

The general picture emerging is not one of astrong influence offish on invertebrate communitiesor habitat-structure—although species-specific,localised effects clearly occur. Most of theexperiments carried out to date have been donewithout reference to the distribution, dynamics, andmobility of the prey species. They have been carriedout on a small scale, at only one site, and overunjustified time periods. Most have serious designflaws, such as "pseudo-replication" (Jones 1981a;Andrew & Choat 1982) and may confound realpredator effects with cage artefacts (Jones 1981a).

In future, greater consideration must be given todetermining the appropriate scale at which toexamine the effects of predation, based on thedistribution and mobility of the prey. Theappropriate temporal scale will depend on the lifehistory and dynamics of prey species, which is anecessary background for determining theimportance of predator effects, relative to otherprocesses. When cages are necessary, considerableimagination must go into the design and deploymentof cages and cage controls, and collecting directinformation of the relative use of cage control andopen control treatments. Fish removal techniques,which have been used in the past to answer otherquestions (e.g., Thompson & Jones 1983; Jones1984b, 1984c) represent a real alternative whenlooking at larger-scale effects. We can only expectgeneralisations to emerge by repeating experimentsat other places and times (Underwood 1983).

The impact of herbivoresNumerous experiments designed to excludeherbivorous fish have shown a substantial impact offish grazing on tropical reefs (see reviews by Ogden&Lobel 1978; Hixon 1983).These results have been

stated to contrast to temperate reefs, where herbi-vorous fishes were considered to be rare and unim-portant (Ogden & Lobel 1978). While true grazingherbivores, such as scarids and acanthurids, do notpenetrate temperate systems, it has been pointed outthat browsing herbivorous fish are common ontemperate reefs and must therefore be considered aspotentially important (Choat 1982). Althoughherbivores, in north-eastern New Zealand, compriseonly a small proportion of the overall number of reeffish, they make up a substantial proportion of thebiomass (Russell 1977: table 2). In view of this,Russell (1983) argued that the importance of fishherbivory has been largely underestimated. Hestated, "it seems likely that the zone of short algalturf in shallow water on rocky reefs in north-easternNew Zealand is at least partially maintained by theactivities of grazing fishes".

Subsequent work on 0. pullus, the only herbi-vore consuming large brown algae, has notsupported this contention. Meekan (1986) examinedthe distribution of 0. pullus and surveyed plants inhigh-density areas for the characteristic circular bitemarks in the laminae. His quantitative description ofthis record of feeding showed that damage was verypatchy and not extensive enough to have a majorimpact on kelp distribution. Herbivore exclusionexperiments designed to test directly the impact ofherbivores have proved difficult to deploy in theshallow surgy conditions of rocky reefs in north-eastern New Zealand (Schiel 1980). The potentialroles played by browsers on red and green algae,such as the common G. tricuspidata, remain to beexamined.

CONCLUSIONS

In asking precise questions about the ecology ofrocky reef fish in north-eastern New Zealand, thecompiled information provides some answers, andhighlights issues yet to be resolved. Much is knownabout patterns of their distribution and abundance,but little about the underlying causes of thesepatterns. Much is known about their diets andforaging patterns, but further work is required tounequivocally establish the role fish predation playsin structuring the rocky reef system. While the "big"questions remain unanswered, there is a sound basisfor establishing the appropriate techniques to solvethese ecological problems.

Some tentative conclusions are emerging aboutthe interactions between fish and their habitat. Choat

Dow

nloa

ded

by [

Uni

vers

ity o

f Sy

dney

] at

12:

59 1

9 M

arch

201

3

Jones—Ecology of rocky reef fish 459

& Ayling (1987) have suggested "the biologicalstructure of shallow reef habitats, especially thepresence of algae, may determine the character ofthe predatory fish fauna, not vice versa". This modelis largely supported here. However, it is clear thatexperiments testing the effect of the habitat on thefish, and vice versa, need to be repeated on largerspatial and temporal scales.

Spatial and temporal pattern in reef fishpopulations are likely to be caused by totallydifferent sets of processes, which differ in theirimportance as scales of focus are expanded. Theseprocesses involve both pelagic and reef-associatedphases of the life history, and their relativeimportance will only be distinguished by long-termstudies, which monitor what happens to successivecohorts as they pass through each life history stage.

Studies on the ecology of temperate reef fisheshave differed in approach from the early coral reeffish literature, in which questions were asked aboutabundance and structure in the context of largeassemblages, and about the effects of fish "guilds"on prey communities (Ogden & Lobel 1978; Sale1980). Temperate-climate workers haveemphasised ontogenetic and species-specificdifferences in reef fish ecology. These differenceswill be an integral part of any general ecologicalmodels which develop for shallow reefenvironments.

ACKNOWLEDGMENTS

Without the influence of J. H. Choat on the developmentof reef fish ecology at the Leigh Laboratory, there wouldhave been little to review. I thank Howard, along withN. Andrew, A. & A. Ayling, C. Battershill, R. Creese,M. Kingsford, A. MacDiarmid, M. Poynter, D. Schiel, andS. Thompson for making "the early years" interesting,productive, and enjoyable. Thanks are due also to all thosepeople who have recently worked on reef fishes at Leighand allowed me to cite their unpublished material. I amgrateful to M. Cappo, R. Cole, M. Francis, M. Kingsford,and D. Schiel for wading through and makingimprovements to a very rough first draft. I acknowledgethe use of the facilities of the University of Auckland'sMarine Research Laboratory and Department of Zoologyin preparing this review.

REFERENCES

Andrew, N. L. 1988: Ecological aspects of the commonsea urchin, Evechinus chloroticus, in northernNew Zealand: a review. New Zealand journal ofmarine and freshwater research 22: 415-426.

Andrew, N. L ; Choat, J. H. 1982: The influence ofpredation and conspecific adults on the abun-dance of juvenile Evechinus chloroticus(Echinoidea: Echinometridae) Oecologia 54:80-87.

Andrew, N. L.; Mapstone, B. D. 1987: Sampling and thedescription of spatial pattern in marine ecology.Oceanography and marine biology annualreviews 25: 39-90.

Ayling, A. L. 1978: Population biology and competitiveinteractions in subtidal sponge dominatedcommunities in temperate waters. UnpublishedPhD thesis, University of Auckland. 106 p.

Ayling, A. M. 1978: Okakari Point to Cape RodneyMarine Reserve: a biological survey.Unpublished report, University of Auckland.115p.

1981: The role of biological disturbance intemperate subtidal encrusting communities.Ecology 62: 830-847.

Ayling, A. M.; Cox, G. J. 1982: Collins; guide to the seafishes of New Zealand. Collins, Auckland. 343p.

Bakus, G. J. 1969: Energetics and feeding in shallowmarine waters. International revue of general andexperimental zoology 4: 275-369.

Battershill, C. N. 1987: Factors affecting the structure anddynamics of subtidal communities characterisedby sponges. Unpublished PhD thesis, Universityof Auckland. 227p.

Bernstein, B. B.; Jung, N. 1979: Selective pressures andcoevolution in a kelp canopjr community insouthern California. Ecological monographs 49:335-355.

Bortone, S. A.; Hastings, R. W.; Oglesby, J. L. 1986:Quantification of reef fish assemblages: acomparison of several in situ methods. NortheastGulf science 8: 1-22.

Brawley, S. H.; Adey, W. H. 1981: Micrograzers mayaffect macoalgal density. Nature 292: 177.

Bray, R. N. 1981: Influence of water currents andzooplankton densities on daily foragingmovements of blacksmith, Chromispunctipinnis,a planktivorous reef fish. Fishery bulletin 78:829-841.

Bray, R. N.; Ebeling, A. W. 1975: Food, activity, andhabitat of three "picker-type" rnicrocarnivorousfishes in the kelp forest off Santa Barbara,California. Fishery bulletin 73: 815-829.

Bray, R. N.; Miller, A. C ; Geesey, G. G. 1981: The fishconnection: a trophic link between planktonic androcky reef communities? Science 214: 204-205.

Bray, R. N.; Purcell, L. J.; Miller, A. C. 1986: Ammoniumexcretion in a temperate-reef community by aplanktivorous fish, Chromis punctipinnis(Pomacentridae), and potential uptake by younggiant kelp, Macrocystis pyrifera (Laminariales).Marine biology 90: 327-334.

Dow

nloa

ded

by [

Uni

vers

ity o

f Sy

dney

] at

12:

59 1

9 M

arch

201

3

460 New Zealand Journal of Marine and Freshwater Research 1988, Vol. 22

Choat, J. H. 1982: Fish feeding and the structure of benthiccommunities in temperate waters. Annual reviewof ecology and systematics 13: 423-449.

Choat, J. H.; Ayling, A. M. 1987: The relationshipbetween habitat structure and fish faunas on NewZealand reefs. Journal of experimental marinebiology and ecology 110: 257-284.

Choat, J. H.; Ayling, A. M.; Schiel, D. R. in press:Temporal and spatial variation in an island fishfauna. Journal of experimental marine biologyand ecology.

Choat, J. H.; Bellwood, D. R. 1985: Interactions amongstherbivorous fishes on a coral reef: the influence ofspatial variation. Marine biology 89: 221-234.

Choat, J. H.; Kingett, P. D. 1982: The influence of fishpredation on the abundance cycles of an algal turfinvertebrate fauna. Oecologia54: 88-95.

Choat, J. H.; Schiel, D. R. 1982: Patterns of distributionand abundance of large brown algae andinvertebrate herbivores in subtidal regions ofnorthern New Zealand. Journal of experimentalmarine biology and ecology 60: 129-162.

Clarke, T. A. 1970: Territorial behaviour and populationdynamics of a pomacentrid fish, the garibaldi,Hypsipops rubicunda. Ecological monographs40: 189-212.

Clements, K. D. 1985: Feeding in two New Zealandherbivorous fish, the butterfish Odax pullus andthe marblefish Aplodactylus arctidens.UnpublishedMSc thesis, University of Auckland.201p.

Connell, J. H. 1985: The consequences of variation ininitial settlement versus post-settlement mortalityin rocky intertidal communities. Journal ofexperimental marine biology and ecology 93:11-45.

Coull, B. C ; Wells, J. B. J. 1983: Refuges from fishpredation: experiments with phytal meiofaunafrom the New Zealand rocky intertidal. Ecology64: 1599-1609.

Cowan, R. K. 1983: The effect of sheephead(Semicossyphus pulcher) predation on red seaurchin (Strongylocentrotus franciscanus)populations: an experimental analysis. Oecologia58: 249-255.

1985: Large scale pattern of recruitment by thelabrid, Semicossyphus pulcher: causes andimplications. Journal of marine research 43:719-742.

1986: Site-specific differences in the feedingecology of the California sheephead,Semicossyphus pulcher (Labridae). Environ-mental biology offish 16: 193-203.

Doherty, P. J. 1983: Tropical territorial damselfishes: isdensity limited by aggression or recruitment?Ecology 64: 176-190.

Ebeling, A. W.; Larson, R. J.; Alevizon, W. S. 1980a:Habitat groups and island-mainland distributionof kelp-bed fishes off Santa Barbara, California.In: Power, D. M. ed. Multidisciplinary sym-posium on the California Islands, pp. 403-431.Santa Barbara museum of natural history, SantaBarbara.

Ebeling, A. W.; Larson, R. J.; Alevizon, W. S.; Bray, R. N.1980b: Annual variability of reef-fishassemblages in kelp forests off Santa Barbara,California. Fishery bulletin 78: 361-377.

Ebeling, A. W.; Laur, D. R. 1986: Foraging insurfperches:resource partitioning or individualisticresponses? Environmental biology offish 16:123-133.

Ebeling, A. W.; Laur, D. R.; Rowley, R. J. 1985: Severestorm disturbances and reversal of communitystructure in a southern California kelp forest.Marine biology 84: 287-294.

Eckert, G. J. 1987: Estimates of adult and juvenilemortality for labrid fishes at One Tree Reef, GreatBarrier Reef. Marine biology 95: 167-171.

Ehrlich, P. R. 1975: The population biology of coral reeffishes. Annual review of ecology and systematics6:211-247.

Gaines, S. D.; Roughgarden, J. 1987: Fish on offshorekelp forests affect recruitment to intertidalbarnacle populations. Science 235: 479-481.

Goldman, B.; Talbot, F. H. 1976: Aspects of the ecologyof coral reef fishes. In: Biology and geology ofcoralreefs.Vol. III. Biology, 2, pp. l25-154. NewYork, Academic Press.

Handford, C. 1979: The habitat, population dynamics andsocial organisation of two tripterygiid fishes.UnpublishedMSc thesis, University of Auckland.114p.

Hixon M. A. 1980: Competitive interactions betweenCalifornia reef fishes of the genus Embiotoca.Ecology 61: 918-931.

1983: Fish grazing and community structure ofreef corals and algae: a synthesis of recent studies.NOAA symposium series on undersea research 1:79-88.

Hobson, E. S. Jr 1982: The structure of fish communitieson warm-temperate and tropical reefs. In:Huntsman, G. R.; Nicholson, W. R.; Fox, W. W.Jr. ed.. The biological bases for reef fisherymanagement, pp. 160-166. NOAA TechnicalMemorandum NMFS-SEFC-80.

Hobson, E. S. Jr; Chess, J. R. 1976: Trophic interactionsamong fishes and zooplankters near shore at SantaCatalina Island, California. Fishery bulletin 74:567-598.

Holbrook, S. J.; Schmitt, R. J. 1984: Experimentalanalyses of patch selection by foraging blacksurfperch (Embiotoca jacksoni Agazzi). Journalof experimental marine biology and ecology 79:39-64.

Dow

nloa

ded

by [

Uni

vers

ity o

f Sy

dney

] at

12:

59 1

9 M

arch

201

3

Jones—Ecology of rocky reef fish 461

Jones, G. P. 1980: Growth and reproduction in theprotogynous hermaphrodite Pseudolabruscelidotus (Pisces: Labridae) in New Zealand.Copeia 1980: 660-675.

1981a: Interrelationships between ecology,behaviour and life history in the protogynoushermaphrodite Pseudolabrus celidotus (Pisces:Labridae). Unpublished PhD thesis, University ofAuckland. 120p.

1981b: Spawning-site choice by femalePseudolabrus celidotus (Pisces: Labridae) and itsinfluence on the mating system. Behaviouralecology and sociobiology 8: 129-142.

1983: Relationship between density andbehaviour in juvenile Pseudolabrus celidotus(Pisces: Labridae). Animal behaviour 31:729-735.

1984a: The influence of habitat and behaviouralinteractions on the local distribution of the wrasse,Pseudolabrus celidotus. Environmental biologyof fishes 10: 43-58.

1984b: Population ecology of the temperatereef fish Pseudolabrus celidotus Bloch &Schneider (Pisces: Labridae). I. Factorsinfluencing recruitment. Journal of experimentalmarine biology and ecology 75: 257-276.

1984c: Population ecology of the temperate reeffish Pseudolabrus celidotus Bloch & Schneider(Pisces: Labridae). II. Factors influencing adultdensity. Journal of experimental marine biologyand ecology 75: 277-303.

1987: Competitive interactions among adultsand juveniles in a coral reef fish. Ecology 68:1534-1547.

Jones, G. P.; Thompson, S. M. 1980: Social inhibition ofmaturation in females of the temperate wrassePseudolabrus celidotus and a comparison with theblennioid Tripterygion varium. Marine biology59: 247-256.

Keough, M. J. 1984: Dynamics of the epifauna of thebivalve Pinna bicolor: interactions amongrecruitment, predation, and competition. Ecology65: 677-688.

Kingett, P. D.; Choat, J. H. 1981: Analysis of density anddistribution patterns in Chrysophrys auralus(Pisces: Sparidae) within a reef environment: anexperimental approach. Marine and ecologyprogress series 5: 283-290.

Kingsford, M. J. 1980: Interrelationships betweenspawning and recruitment of Chromis dispilus(Pisces: Pomacentridae). Unpublished MScthesis, University of Auckland. 84p.

1988: The early life history of fish in coastalwaters of northern New Zealand: a review. NewZealand journal of marine and freshwaterresearch 22: 463-479.

Larson, R. J. 1980: Competition, habitat selection, and thebathymetric segregation of two rockfish{Sebastes) species. Ecological monographs 50:221-239.

Leum, L. L.; Choat, J. H. 1980: Density and distributionpatterns of the temperate marine fishCheilodactylus spectabilis (Cheilodactylidae) ina reef environment. Marine biology 57: 327-337.

MacDiarmid, A. B. 1981: Factors influencing thedistribution and abundance of two temperateplanktivorous reef fish, Pempheris adspersa andScorpis violaceus. Unpublished MSc thesis.University of Auckland. lOlp.

Main, K. L. 1985: The influence of prey identity and sizeon selection of prey by two marine fishes. Journalof experimental marine biology and ecology 88:145-152.

McCormick, M. I. 1986: Spatial and temporal patterns ofabundance of Cheilodactylus spectabilis (Pisces:Cheilodactylidae). Unpublished MSc thesis,University of Auckland. 113p.

McCormick, M. I.; Choat, J. H. 1987: Estimating totalabundance of a large temperate reef fish usingvisual strip transects. Marine biology 96:469-478.

Meekan, M. G. 1986: The distribution and abundance ofthe herbivorous fish Odaxpullus and its influenceon its food plant Ecklonia radiata within atemperate reef environment. Unpublished MScthesis, University of Auckland. 68p.

Milicich, M. J. 1986: Aspects of the early life history ofParika scaber (Pisces: Monacanthidae). Unpub-lished MSc thesis, University of Auckland. 76p.

Munro, J. L.; Williams, D. McB. 1985: Assessment andmanagement of coral reef fisheries: biological,environmental and socio-economic aspects.Proceedings of the fifth international coral reefcongress, Tahiti 4: 544—578.

Mutch, P. G. 1983: Factors influencing the density anddistribution of the blue cod (Parapercis colias)(Pisces: Mugiloididae). Unpublished MSc thesis,University of Auckland. 76p.

Norton, T. A.; Benson, M. R. 1983: Ecologicalinteractions between the brown seaweedSargassum muticum and its associated fauna.Marine biology 75: 169-177.

Ogden, J. C ; Lobel, P. S. 1978: The role of herbivorousfishes and urchins in coral reef communities.Environmental biology of fishes 3: 49-63.

Poynter, M. R. 1980: The distribution and abundance of atemperate reef fish Parika scaber (Mona-canthidae) with emphasis on recruitment. Unpub-lished MSc thesis, University of Auckland. 65p.

Russ, G. R. 1980: Effects of predation by fishes,competition, and structural complexity of thesubstratum on the establishment of a marineepifaunal community. Journal of experimentalmarine biology and ecology 42: 55-69.

Dow

nloa

ded

by [

Uni

vers

ity o

f Sy

dney

] at

12:

59 1

9 M

arch

201

3

462 New Zealand Journal of Marine and Freshwater Research 1988, Vol. 22

Russell, B. C. 1971: Ecological relationships of rocky reeffishes of north-eastern New Zealand.UnpublishedMSc thesis, University of Auckland.220p.

1975: The development and dynamics of asmall artificial reef community. Helgolanderwissenschaftliche Meeresuntersunchungen 27:298-312.

1977: Population and standing crop estimatesfor rocky reef fishes of north-eastern NewZealand. New Zealand journal of marine andfreshwater research 11: 23-36.

1983: The food and feeding habits of rocky reeffish of north-eastern New Zealand. New Zealandjournal of marine and freshwater research 17:121-145.

Sale, P. F. 1980: The ecology of fishes on coral reefs.Oceanography and marine biology annual review18: 367-421.

1984: The structure of communities of fish oncoral reefs and the merit of a hypothesis-testing,manipulative approach to ecology. In: Strong, D.R.; Simberloff, D.; Abele, L. G.; Thistle, A. B., ed.Ecological communities: conceptual issues andthe evidence, pp. 478-490. Princeton UniversityPress, Princeton, New Jersey.

Sale, P. F.; Jones, G. P.; Choat, J. H.; Leis, J. M.;Thresher,R. E.; Williams, D. McB. 1985: Current prioritiesin ecology of coral reef fishes. Search 16:270-274.

Sale, P. F.; Sharp, B. J. 1983: Correction for bias in visualtransect censuses of coral reef fishes. Coral reefs2: 37-42.

Schiel, D. R. 1980: A demographic and experimentalevaluation of plant and herbivore interactions insubtidal algal stands. Unpublished PhD thesis,University of Auckland Library. 166p.

Schiel, D. R. 1988: Algal interactions on shallow subtidalreefs in northern New Zealand: a review. NewZealand journal of marine and freshwaterresearch 22:481-^89.

Schmitt, R. J.; Holbrooke, S. J. 1984: Ontogeny of preyselection by black surfperch Embiotoca jacksoni(Pisces: Embiotocidae): the roles of fishmorphology, foraging behavior, and patchselection. Marine ecology progress series 18:225-239.

Stephens, J. S. Jr; Johnson, R. K.; Key, G. S.; McCosker,J. S. 1970: The comparative ecology of threesympatric species of California blennies of thegenus Hypsoblennius Gill (Teleostomi,Blenniidae). Ecological monographs 40:213-233.

Stephens, J. S. Jr; Jordan, G. A.; Morris, P. A.; Singer, M.M.; McGowen, G. E. 1986: Can we relate larvalfish abundance to recruitment or populationstability? A preliminary analysis of recruitment toa temperate rocky reef. California cooperativeoceanic fisheries investment report 27: 65-83.

Stacker, L. J. 1984: Recruitment, growth, and mortality inthe subtidal ascidian Pseudodistoma novae-zelandiae (Ascidiacea: Polyclinidae). Unpub-lished MSc thesis, University of Auckland. 90p.

Sylvester, T. 1986: Food limitation: A preliminary studyon two groups of benthic feeding carnivorous fishin a temperate reef system. Unpublished MScthesis, University of Auckland. 87p.

Thompson, S. 1979: Ecological and behavioural factorsinfluencing the distribution and abundancepatterns of tripterygiid fishes with particularreference to Tripterygion varium. UnpublishedMSc thesis, University of Auckland. 80p.

1981: Fish of the marine reserve. Auckland,University Bindery. 364p.

1983: Homing in a territorial reef fish. Copeia1983: 832-834.

1986: Male spawning success and femalechoice in the mottled triplefin, Forsterygionvarium (Pisces: Tripterygiidae). Animalbehaviour 34: 580-589.

Thompson, S. M.; Jones, G. P. 1983: Interspecifictenritoriali ty and competition for food between thereef fishes Forsterygion varium and Pseudolabruscelidotus. Marine biology 76: 95-104.

Underwood, A. J. 1981: Techniques of analysis ofvariance in experimental marine biology andecology. Oceanography and marine biologyannual review 19: 513-605.

1983: Spatial and temporal problems in thedesign of experiments with marine grazers. In:Baker, J. T.; Carter, R. M.; Sammarco, P. W.;Stark, K. P. ed., Proceedings: Inaugural GreatBarrier Reef Conference, pp. 251-256,Townsville, James Cook University Press,Townsville.

Victor, B. C. 1986: Larval settlement and juvenilemortality in a recruitment-limited coral reef fishpopulation. Ecological monographs 56: 145-160.

Weinberg, J. R.; Caswell, H.; Whitlach, R. B. 1986:Demographic importance of ecologicalinteractions: how much do statistics tell us?Marine biology 93: 305-310.

Williams, D. McB.; Hatcher, A. I. 1983: Structure offishcommunities on outer slopes of inshore, mid-shelfand outer shelf reefs of the Great Barrier Reef.Marine ecology progress series 10: 239-250.

Zimmerman, R.; Gibson, R.; Harrington, J. 1979:Herbivory and detritivory among gammarideanamphipods from a florida seagrass community.Marine biology 54: 4 1 - 4 7

Dow

nloa

ded

by [

Uni

vers

ity o

f Sy

dney

] at

12:

59 1

9 M

arch

201

3