Inconsistency and variation in the development of rocky intertidal algal assemblages

Download Inconsistency and variation in the development of rocky intertidal algal assemblages

Post on 14-Sep-2016

212 views

Category:

Documents

0 download

TRANSCRIPT

LJournal of Experimental Marine Biology and Ecology,224 (1998) 265289Inconsistency and variation in the development of rockyintertidal algal assemblages*M.G. Chapman , A.J. UnderwoodCentre for Research on the Ecological Impacts of Coastal Cities and the Institute of Marine Ecology,Marine Ecology Laboratories A11, University of Sydney, NSW 2006, AustraliaReceived 11 March 1997; received in revised form 22 July 1997; accepted 2 August 1997AbstractExperimentally-cleared patches were used to test hypotheses about the relative importance ofbroad-scale biogeographic processes and small-scale historical processes in the development oflow-shore algal assemblages on wave-exposed rocky coasts of New South Wales (Australia). Inaddition, the applicability of generalizing from patterns of recruitment and development at onetime was tested by providing similar cleared patches in the algal assemblage at three-monthlyintervals and quantifying early development of the algal assemblage in these clearings fifteentimes over a period of four years.The early stages of development of these assemblages differed significantly from shore to shoreand time to time and there was no evidence for common broad-scale patterns of recruitment, norfor any biogeographic trend. Nevertheless, these assemblages developed from a limited commonpool of species, some of which recruited fairly regularly on most shores and some of which onlyarrived sporadically on some shores. Despite different starting points, assemblages convergedtowards the surrounding assemblages on most shores, so that within less than 12 years, clearedareas resembled the surrounding assemblages. Although the specific changes leading to conver-gence differed from shore to shore, there was a general pattern of early colonizers, such asephemeral algae and sessile animals being gradually replaced by larger perennial algae.These results demonstrate no simple seasonal nor clear-cut biogeographical patterns in thedevelopment of algal assemblages on these shores and indicated the relative importance of localinfluences. Results are discussed with respect to the need to do experiments at numerous places inorder to examine responses of assemblages to, or recovery from, environmental disturbances. 1998 Elsevier Science B.V.Keywords: Algal assemblages; Low-shore species; Cleared patches; Biogeograghical patterns*Corresponding author.0022-0981/98/$19.00 1998 Elsevier Science B.V. All rights reserved.PII S0022-0981( 97 )00202-5266 M.G. Chapman, A.J. Underwood / J. Exp. Mar. Biol. Ecol. 224 (1998) 265 2891. IntroductionSpecies of algae characteristically vary in abundance from time to time and place toplace, leading to assemblages that are very variable in structure and composition(Dayton, 1971; Lubchenco, 1980; Jernakoff, 1985; Foster, 1990). In an analysis oflowshore algal assemblages on wave-exposed rocky shores in New South Wales,Underwood and Chapman (1997) found variation at scales of replicate quadrats (metresapart), sites on a shore (tens of metres apart) and among shores (hundreds of kilometresapart). Differences were mainly due to the relative importance of a few dominantspecies, rather than changes in composition of the assemblage from place to place. Therewas no evidence for broad-scale, biogeographic trends in the assemblages or in theirpatterns of variability, although there were persistent differences among assemblages ondifferent shores. This is surprizing because the study spanned 68 of latitude (from TuraHead at 368 529 south to Scotts Head at 308 409 south; see locations in Fig. 1) and hasbeen described as spaning cooltemperate to subtropical biogeographical regions (e.g.Dakin et al., 1948; Dakin, 1987). Similarly, there was no evidence for consistentlong-term (years) or seasonal patterns of abundance of the dominant species or changesin the algal assemblage (Underwood and Chapman, 1997). In each site, assemblagesvaried from one sampling period to the next, again due to sporadic variations inabundance of the dominant species.In many other studies, temporal or spatial variation in the structure of intertidalassemblages has been attributed to responses to disturbances (e.g. Sousa, 1979a, 1980;Connell and Sousa, 1983; Underwood et al., 1983; Underwood and Denley, 1984).Therefore, processes of disturbance and recruitment are out of phase from place to placeso that there are always differences in occupancy of different parts of a shore and fromshore to shore.This model suggests that any biogeographical patterns in the structure or compositionof lowshore algal assemblages on the coast of New South Wales may be masked bystochastic disturbances and recoveries. This study investigates this model by testinghypotheses about responses to disturbances along the coast. Availability of propagulesand therefore recruitment may be similar everywhere, but vary through time. Ifdisturbances are by chance in different parts of the coast, temporal and spatialdifferences will occur because different places will get different recruits at the differenttimes they are disturbed. From this, it is predicted that similar assemblages will developover a wide geographic scale if cleared areas were made available at the same time ineach place. In contrast, if the sources and varieties of propagules were local, one wouldexpect assemblages to differ from place to place because of local differences in rates ofrecruitment (Underwood and Denley, 1984).Even if the early stages of development of an assemblage following disturbance aresimilar from place to place, because of similarity of recruits, locally different late stagesof assemblages may develop from a common early stage. Local stochastic events (localhistory) may subsequently disturb or alter the developing assemblages (Sousa et al.,1981).The model that local versus broad-scale geography determines observed patterns ofspatial differences in assemblages from shore to shore was tested by clearing patches inM.G. Chapman, A.J. Underwood / J. Exp. Mar. Biol. Ecol. 224 (1998) 265 289 267Fig. 1. Map of the coast of New South Wales, Australia showing locations of study sites.the algal assemblage on each shore and quantifying development of the assemblage inthese clearings over four years. If the source of propagules were widespread, one wouldpredict similar early stages of the assemblages from shore to shore (in this caseseparated by hundreds of kilometres). If processes are local, the early stages of theseassemblages should vary from shore to shore because they will depend on a local supplyof larvae.Similarly, the role of geography and local history in the long-term development ofthese assemblages was tested. It was proposed that these assemblages would either (a)develop to resemble those in the immediate surrounding area, i.e. the development of thelowshore algal assemblage is primarily determined by the position of the shore along the268 M.G. Chapman, A.J. Underwood / J. Exp. Mar. Biol. Ecol. 224 (1998) 265 289coast irrespective of early structure, or (b) be determined primarily by early patterns ofrecruitment, i.e. history primarily determines subsequent pattern.Finally, processes that influence structure of assemblages are frequently interpretedfrom single surveys or experiments done in one place at one time (Foster, 1990;Underwood and Petraitis, 1993). The applicability of generalizing from patterns ofrecruitment and development of early stages of these assemblages found at one time wastested by clearing patches in the algal assemblage at three-monthly intervals andquantifying the early development of the algal assemblage in these clearings during threemonths. This was done fifteen times over four years.2. Materials and methods2.1. Study sitesFive locations were chosen to span the length of the coast of New South Wales,approximately 800 km. Locations were chosen in areas about equally spaced along thecoast. Criteria for choosing the particular shore in each area were simply that it bewave-exposed, accessible and have assemblages at different levels that were typical ofother shores in the area. The locations were, from south to north, Tura Head, Flat RockIsland, the Cape Banks Scientific Marine Research Area (Cape Banks), Blueys Headand Scotts Head (Fig. 1). The lowshore assemblage dominated by macroscopic, foliosealgae was sampled in each of two sites on each shore four times per year (atapproximately three-monthly intervals) from 1980 to 1983. These sites were chosen atrandom along a shore and separated by 30100 m depending on the size of the shore.The only criterion for choosing a site was that it have a dense cover of foliose algae (atleast, at the first time of sampling). Sites were, however, chosen to represent the algalassemblage on the shore. Each lowshore site discussed here was as low on the shore as2could safely be reached; sites were between 10.5 and 14 m , except at Blueys Head2where the topography of the shore only allowed sites of 4.5 and 7.5 m (details arebelow).Scotts Head consists of steeply sloping basalt rocks (slopes of 708 or more) with alowshore algal area dominated by Corallina spp., although this varied greatly from timeto time. Several foliose algae were common on the shore. The two sites wereapproximately 50 m apart and each was about 4.5 m along-shore and 3 m downshore.The rocks are generally very smooth and there was little cover of encrusting algae inmidshore areas.Blueys Head lies approximately 200 km south of Scotts Head. Like Scotts Head,these shores were very steep (nearly vertical). The surfaces were, however, pitted andcorrugated by irregular erosion. Lowshore regions were subject to occasional periods ofscour by sand. The dominant algae were Corallina spp. and a mixture of red foliosealgae, including species of Gracilaria, Ceramium and Laurencia. The two sites sampledwere approximately 80 m apart; the first was 2.5 m along-shore by 3 m downshore, thesecond was 1.8 m by 2.5 m downshore. Scotts and Blueys Heads were extremelydifficult to sample because of steepness and extreme wave-action at many times.M.G. Chapman, A.J. Underwood / J. Exp. Mar. Biol. Ecol. 224 (1998) 265 289 269Cape Banks is approximately 230 km south of Blueys Head. The sandstone shores atCape Banks have been described in detail elsewhere (Underwood, 1975; Fairweatherand Underwood, 1991). The two sites in this study were on the exposed Point Shore,approximately 50 m apart, each extending 3 m downshore. The first was 4.7 m, thesecond 4 m along-shore. The lowshore algal assemblage was on a very gently slopingsurface, with extensive cover of foliose macro-algae, including about 60% of Ulvalactuca.Flat Rock Island is approximately 230 km south of Cape Banks. The two sites at FlatRock Island were approximately 40 m apart, on a moderately-sloping (30408) part ofthe sandstone shore. Midshore regions persistently had large numbers of the barnacleTesseropora rosea and a variety of grazing molluscs. None of the grazers was abundantin the lowshore areas dominated by foliose algae, where Corallina spp. and Sargassumspp. were the dominant algae. The two sites extended 3 m downshore; one was 4 m, theother 3.5 m along-shore.Tura Head is approximately 150 km south of Flat Rock Island. The sandstone shoresat Tura Head varied in shape. The lowshore regions were about 408 from horizontal.Although midshore areas were noticeably patchy in domination by the barnacles T. roseaand Catomerus polymerus, the lowshore areas were quite homogeneous in cover offoliose algae, primarily Corallina spp. and a variety of other algae, including Ptero-cladia lucida, Laurencia sp. and Ceramium spp. The two sites sampled were about 60 mapart; one was 4.5 m along-shore by 3 m downshore, the second was 4 m along-shore by3 m downshore.2.2. Development of algal assemblages on each shoreAt the first time of sampling, in one site on each shore, five randomly-chosen14-cm 3 16-cm quadrats in the algal assemblage were cleared by scrubbing and scrapingthe substratum to remove all foliose macro-algae, all sessile animals and most encrustingalgae. At each subsequent time, half of each of these quadrats was re-cleared after it wassampled, while the other half was left undisturbed. This method of clearing reproducesthe sort of patchy clearing found in lowshore areas of exposed shores in New SouthWales (unpub. data). Development of algae in cleared areas thus mimicked the naturalresponses to disturbances. Care was taken to ensure that clearing was equally effectiveon all shores. The sizes of clearings were a compromise between simulating sizes ofpatches found in nature and the need to get photographs of sufficient resolution withminimal time spent taking them because of the dangers of wave-action in these exposed,lowshore habitats. Clearings were of a size typical of many occurring naturally in theselowshore assemblages on exposed headlands.At each sampling time, prior to reclearing half of each quadrat, the cleared quadratsand five randomly-placed quadrats in the surrounding assemblage were photographed(justification for the use of photographic quadrats is given in Underwood and Chapman,1997). The percentage covers of algae and sessile animals in each quadrat wereestimated from 100 points from each photograph. Species with a cover smaller than 1%were not recorded. It was not possible to distinguish between primary and secondary270 M.G. Chapman, A.J. Underwood / J. Exp. Mar. Biol. Ecol. 224 (1998) 265 289cover, unless primary cover was visible below the secondary cover; encrusting algaewere grouped into a single taxon.The halves of the cleared quadrats that were left undisturbed after the first samplingperiod were used to test the hypotheses that early stages of development of theassemblage would be similar (regional processes determine them) or different (localprocesses determine them) and that the algal assemblages would (a) develop to besimilar to the surrounding assemblage or (b) develop differently from surroundingassemblages. These quadrats also provide estimates of the time taken for these algalassemblages to develop.The halves of the quadrats that were repeatedly cleared were used to test thealternative predictions that early stages of the developing assemblages on each shore areconsistent from time to time, or vary stochastically from time to time and place to place.2.3. Analyses of dataThe early stages of development of the algal assemblages were compared amongshores at Time 2 (i.e. three months after the quadrats were first cleared) using ANOSIMon the BrayCurtis measures of dissimilarity calculated from 4th root transformed datafor percentage cover of all taxa. These comparisons tested the alternative hypotheses thatsimilar (or different) assemblages would develop on each shore.To examine for convergence of the developing assemblages with the surroundingareas, Bray-Curtis dissimilarity measures were calculated for two set of quadrats foreach site and each time of sampling, i.e. among the independent quadrats in the algalassemblage and among the replicates of the quadrats that were cleared the first time ofsampling. To test the hypothesis that, on each shore, the assemblages in the clearedquadrats develop to resemble the surrounding algal assemblage and to measure thetime-course of this convergence, the average BrayCurtis measure of dissimilarity wascalculated among all replicate quadrats within the algal assemblage (Dis ) and for theWAreplicate halves of the quadrats that were cleared initially and then left undisturbed(Dis ). These give a measure of variability among quadrats within each assemblage.CNext, the average BrayCurtis measure of dissimilarity between quadrats in the algalassemblage and those in the cleared areas was calculated (Dis ). The numericalBdifference between the dissimilarity between these assemblages (Dis ) and the averageBdissimilarity within these assemblages [(Dis 1 Dis ) /2] is a measure of differenceWA Cbetween the two assemblages relative to variation within them. This relative measure ofdissimilarity is similar to the R-statistic given by Clarke (1993), but uses quantitativemeasures of dissimilarity rather than ranked values.Theoretically, this test statistic can vary between 100% (when all quadrats withinassemblages are the same and there are no similarities between assemblages) to 2100%,when the opposite pattern is found. When it is near 0%, differences among quadratswithin assemblages are similar to comparisons of quadrats from one assemblage toanother and the assemblages can be considered to have converged to reflect naturalwithin-assemblage variation.In order to identify those taxa contributing to changes that developed on each shore,taxa that most contributed to average measures of similarity within each assemblage andM.G. Chapman, A.J. Underwood / J. Exp. Mar. Biol. Ecol. 224 (1998) 265 289 271dissimilarity between the two assemblages were obtained from the BrayCurtismeasures of dissimilarity for each shore at each time of sampling (using SIMPER;Clarke, 1993). Temporal trends in mean percentage cover and variance among replicatequadrats were then separately examined for the most important taxa.We examined the alternative hypotheses that when areas are cleared at different times,early stages of development of the algal assemblage would (a) be similar from time totime on each shore or (b) differ unpredictably from time to time on each shore. Thesepredictions were tested, in March and October, 19801983, in the halves of the quadratsthat were cleared each time of sampling (i.e. cleared in January and July each year).3. Results3.1. Early development of algal assemblagesThe early stages of development of the algal assemblage were compared among allshores in March, 1980 using quadrats that had been cleared three months previously.The assemblages differed significantly among shores (R 5 0.75, P , 0.001; all pair-wisecomparisons significant at P , 0.01, ANOSIM, Clarke, 1993; Fig. 2). There is littlecorrespondence between these values and the average measures of dissimilaritycalculated from BrayCurtis measures for all pairs of shores (Table 1). Large measuresof R do not correspond with particularly large measures of dissimilarity, because R usesranked values and takes into account variability within the assemblages. Nevertheless,each of these measures indicates that early stages of these assemblages developeddifferently on each shore and there was no biogeographic trend in differences amongassemblages.Differences in early stages of development were, however, not due to a totallydifferent suite of species arriving on each shore. Assemblages were characterized by alimited suite of species and, on each shore, more than 50% of the average measure ofsimilarity was due to only one or two species. Differences between shores were causedby variation in the relative importance of these species from shore to shore (Table 2).Thus, the opportunistic algae, Enteromorpha intestinalis and Chaetomorpha aerea wereonly found at Cape Banks and Tura Head, whereas Ulva lactuca was important on allshores. The barnacle Tesseropora rosea was only important on the two northern shores,Scotts and Blueys Heads.3.2. Long-term development of algal assemblagesAs predicted, on all shores except Scotts Head, the relative measure of dissimilaritybetween the developing and established assemblages in quadrats that were initiallycleared and the surrounding algal assemblages (i.e. Dis 2 [(Dis 1 Dis ) /2]) wasB WA Cinitially positive and relatively large, but quickly declined to near 0%. It then persistednear this level throughout the study (Fig. 3). The rate of decline, i.e. convergence of theassemblages, varied among shores. At Tura Head (Fig. 3a) and Cape Banks (Fig. 3c),there was greater dissimilarity at the first time of sampling (i.e. three months after the272 M.G. Chapman, A.J. Underwood / J. Exp. Mar. Biol. Ecol. 224 (1998) 265 289Fig. 2. MDS plots at Time 2 of the replicate quadrats that were cleared three months previously for (s) ScottsHead, (h) Blueys Head, (n) Cape Banks, (x) Flat Rock Island and (q) Tura Head.quadrats were first cleared) than was found at Flat Rock Island (Fig. 3b) and BlueysHead (Fig. 3d). This suggests that there was faster convergence of the assemblagesduring the first three months at Flat Rock Island and Blueys Head than at Tura Headand Cape Banks. Nevertheless, on each of these four shores, relative dissimilarityTable 1(a) R values and (b) mean measures of BrayCurtis dissimilarity for all pair-wise comparisons between shoresat the early stages of development of the algal assemblage measured in March, 1980Shores Scotts Head Blueys Head Cape Banks Flat Rock Island(a) R valuesBlueys Head 0.55Cape Banks 0.94 1.00Flat Rock Island 0.65 0.82 0.96Tura Head 0.88 0.85 0.45 0.91(b) Mean measures of dissimilarity from BrayCurtis measuresBlueys Head 55%Cape Banks 62% 67%Flat Rock Island 46% 59% 56%Tura Head 75% 71% 42% 80%M.G. Chapman, A.J. Underwood / J. Exp. Mar. Biol. Ecol. 224 (1998) 265 289 273Table 2Percentage contribution by important taxa (and cover of bare space) to average measures of similarity amongquadrats cleared three months previously on each shore (measured at Time 2, March, 1980)Taxon Scotts Blueys Cape Flat TuraHead Head Banks Rock HeadIslandUlva lactuca 12 51 28 32 10Corallina spp. 32 2 1 54 0Enteromorpha intestinalis 0 0 28 0 11Chaetomorpha aerea 0 0 22 0 53Bluegreen algae 11 0 0 0 0Unidentified crusts 3 22 0 0 22Tesseropora rosea 15 25 0 0 0Bare space 20 0 21 14 2declined rapidly towards 0%, so that, after approximately one year, dissimilarity amongquadrats between the cleared areas and algal assemblages approximated dissimilarityamong quadrats within each of these assemblages. Thus, after about a year, theestablished and developing assemblages were similar in composition.In contrast, at Scotts Head (Fig. 3e), a different pattern was identified. Theassemblage in the cleared quadrats remained persistently different from the surroundingassemblage, although these differences were not particularly large.The test-statistic, Dis 2 [(Dis 1 Dis ) /2] can vary from time to time or place toB WA Cplace because of changes in variability among replicate quadrats within either or both ofthe assemblages (i.e. Dis and/or Dis ) and/or because of changes in the twoWA Cassemblages relative to each other (i.e. Dis ). To test which components of variabilityBmost influence differences between the two assemblages on each shore, the mean andvariance of each of the components of the above test statistic (i.e. Dis , Dis andB WADis ) were calculated from all sampling times. These provide an estimate of averageCdifferences of these assemblages between each set of cleared quadrats and thesurrounding algal assemblage. The variances give estimates of variability of thesedifferences from time to time.On all shores, there was more dissimilarity, on average, from quadrat to quadrat in theareas that were initially cleared than among quadrats in the surrounding algalassemblages (compare the mean values in Table 3). This reflects greater small-scalepatchiness in assemblages developing in cleared quadrats than that found in theestablished assemblages.Variability in this inter-quadrat patchiness through time was not, however, consistentlylarger in the cleared quadrats than in the algal assemblages (shown by the variances inTable 3). Mean dissimilarity among replicate quadrats in the algal assemblage remainedrelatively consistent throughout the study, i.e. there was only small-scale temporalvariability in this measure of patchiness and no trend for increasing or decreasingvariability through time (illustrated for Cape Banks and Blueys Head in Fig. 4). Therewas more variability between the initially-cleared quadrats and the algal assemblages atthe start of the study (as predicted and indicated in Fig. 2), but, as the assemblagesconverged, differences between them decreased to match variability within each or one274 M.G. Chapman, A.J. Underwood / J. Exp. Mar. Biol. Ecol. 224 (1998) 265 289Fig. 3. Relative difference between the mean BrayCurtis measure of dissimilarity between quadrats in thealgal assemblages and quadrats that were cleared at Time 1 and the mean measure of dissimilarity within eachof the these assemblages, i.e. Dis 2 [(Dis 1 Dis ) /2] at (a) Tura Head, (b) Flat Rock Island, (c) CapeB WA CBanks, (d) Blueys Head and (e) Scotts Head; n 5 4 or 5 quadrats in each assemblage at each time ofsampling.of them (Fig. 3). Each natural assemblage at Tura Head was less variable from quadratto quadrat and, on average, from time to time, than on other shores (shown by thesmaller means and variances in Table 3).Convergence of the assemblages developing in the cleared quadrats with thesurrounding assemblages is also illustrated in MDS plots (Clarke, 1993) for Flat RockIsland (Fig. 5a), Tura Head (Fig. 5b) and Scotts Head (Fig. 5c) for three time periods;three months after the quadrats were cleared (Time 2), one year later (Time 6) and at theend of the study. Note that for Flat Rock Island and Tura Head, despite some outlyingM.G. Chapman, A.J. Underwood / J. Exp. Mar. Biol. Ecol. 224 (1998) 265 289 275Table 3Means and variances of average BrayCurtis measures of dissimilarity among quadrats in the algal assemblageand in the quadrats that were initially cleared, calculated from the subsequent 15 times of sampling (n 5 13 forTura Head because of missing data)Shore Algal assemblages Cleared areasMean Variance Mean VarianceScotts Head 30.9 159.9 37.9 230.4Blueys Head 33.5 178.9 46.7 119.1Cape Banks 22.3 75.2 40.4 187.7Flat Rock Island 29.3 129.1 35.8 113.3Tura Head 24.4 61.0 37.3 47.7replicates, cleared and control quadrats were mixed together after one year. The plots inFig. 5 use ranked values, rather than average dissimilarities, but show similar trends tothose illustrated in Fig. 3. The assemblages in the cleared quadrats at Flat Rock Islandand Tura Head have clearly converged with the undisturbed algal assemblages on theseFig. 4. Mean BrayCurtis measures of dissimilarity between quadrats that were initially cleared and those inthe algal assemblages (j Dis ), among quadrats in the algal assemblage (d Dis ) and among the clearedB WAquadrats (s Dis ) at (a) Cape Banks and (b) Blueys Head; n 5 4 or 5 quadrats in each assemblage at eachCtime of sampling.276 M.G. Chapman, A.J. Underwood / J. Exp. Mar. Biol. Ecol. 224 (1998) 265 289Fig. 5. MDS plots of the replicate quadrats in the areas that were cleared at Time 1 (empty symbols) and theundisturbed assemblages (filled symbols) (i) three months after clearing, (ii) one year later and (iii) at the endof the study for (a) Flat Rock Island, (b) Tura Head and (c) Scotts Head.shores, whereas those at Scotts Head remained different four years after the areas wereoriginally cleared.3.3. Specific changes to the assemblages on each shoreAt Tura Head, the algal assemblage consisted primarily of Corallina spp., withsmaller amounts of other foliose algae, particularly Sargassum spp. and Laurencia sp.The latter two taxa were very patchy, with extensive cover in some quadrats at sometimes, but generally averaging less than 10%. Three months after the quadrats werecleared, there was little bare space (Fig. 6a) or cover of Corallina spp. (Fig. 6c). Mostspace was occupied by the opportunistic algae, Chaetomorpha aerea (mean cover 57%)and Ulva lactuca (Fig. 6b). Cover of C. aerea quickly declined and this species was notM.G. Chapman, A.J. Underwood / J. Exp. Mar. Biol. Ecol. 224 (1998) 265 289 277Fig. 6. Mean (S.E.) percentage cover of (a) bare space, (b) U. lactuca and (c) Corallina spp. in (s) quadratsthat were cleared in January, 1980 and in (d) undisturbed quadrats in the algal assemblage at Tura Head;n 5 4 or 5 quadrats in each assemblage at each time of sampling.found subsequently. Cover of U. lactuca declined more slowly over the next six months(Fig. 6b), giving rise to an increase in bare space (Fig. 6a) and pink coralline crust.These gradually disappeared under the growth of foliose macro-algae, particularlyCorallina spp. Although the barnacle, T. rosea, contributed to differences between theassemblages during the first year due to settlement in the cleared areas, cover wasalways small (mean , 8%) and patchy. Cover of Corallina spp. spread to match that inthe surrounding assemblage approximately one year after the quadrats were first cleared.Patchy distributions of other foliose macro-algae, such as Amphiroa anceps orColpomenia sinuosa, contributed to differences between the two assemblages from timeto time, but there was no consistent pattern in these differences and, in general, the278 M.G. Chapman, A.J. Underwood / J. Exp. Mar. Biol. Ecol. 224 (1998) 265 289assemblages in the cleared areas continued to resemble the undisturbed assemblages(Fig. 3a).At Flat Rock Island, most of the assemblage consisted of Corallina spp. andSargassum spp., with few other macro-algae. At the first time of sampling, theassemblage in the clearings had already converged to be relatively similar to theundisturbed algal assemblages (Fig. 3b). There was relatively little bare space (mean, 5%). Greater cover of U. lactuca (Fig. 7a) and smaller cover of Sargassum spp. (Fig.7c) in the clearings contributed most to differences between these two assemblages. Incontrast to Tura Head, C. aerea and T. rosea were not evident as early colonizers.Thereafter, cover of U. lactuca rapidly decreased and cover of Corallina spp. andFig. 7. Mean (S.E.) percentage cover of (a) U. lactuca, (b) Corallina spp. and (c) Sargassum spp. in (s)quadrats that were cleared in January, 1980 and in (d) quadrats in the algal assemblage at Flat Rock Island;n 5 4 or 5 quadrats in each assemblage at each time of sampling.M.G. Chapman, A.J. Underwood / J. Exp. Mar. Biol. Ecol. 224 (1998) 265 289 279Sargassum spp. increased so that after a further three-six months the assemblages hadconverged. Throughout the rest of the study, cover of Corallina spp. and Sargassum spp.showed relatively large temporal oscillations (Fig. 7b, c) which were not clearlyseasonal, but were tightly linked across the two sets of quadrats. In the last 18 months ofthe study, the solitary ascidian, Pyura stolonifera, appeared in both sets of quadrats, butbecame particularly abundant in those that had been initially cleared, reaching anaverage cover of nearly 20%.The algal assemblage at Cape Banks consisted primarily of Sargassum spp, withsmaller amounts of Corallina spp. Like Tura Head, at the first time of sampling, therewas relatively little bare space in the cleared quadrats (Fig. 8a), primarily due to cover ofU. lactuca (Fig. 8b), C. aerea (mean cover 22%) and E. intestinalis (mean cover 33%).Fig. 8. Mean (S.E.) percentage cover of (a) bare space, (b) U. lactuca and (c) Sargassum spp. in (s) quadratsthat were cleared in January, 1980 and in (d) quadrats in the algal assemblage at Cape Banks; n 5 5 quadratsin each assemblage at each time of sampling.280 M.G. Chapman, A.J. Underwood / J. Exp. Mar. Biol. Ecol. 224 (1998) 265 289Thereafter, the latter two opportunistic species disappeared and cover of U. lactucainitially increased and then gradually declined, although there were seasonal increases inabundance in the cleared quadrats and algal assemblages during the rest of the study(Fig. 8b). Sargassum spp. gradually grew back until, after two years, cover in the clearedquadrats resembled that in the surrounding assemblage, although was not generally asextensive (Fig. 8c). Recovery of Sargassum spp. at Cape Banks was therefore slowerthan recovery of Corallina spp. at Tura Head and Flat Rock Island. There was littlecover of T. rosea throughout the study, although there was settlement in the first year inthe cleared areas (mean , 4%). This species rapidly disappeared from these quadrats.The algal assemblage at Blueys Head was relatively diverse compared to the othershores, e.g. U. lactuca (mean cover 33%), Corallina spp. (30%), Laurencia sp. (5%),Gracilaria sp. (4%) and unidentified red algae (6%). As on other shores, the clearedquadrats were quickly colonised by U. lactuca, but cover of this species rapidly declinedduring the next few months. In contrast to the other shores, U. lactuca was, however,also relatively abundant in the undisturbed algal assemblages at Blueys Head and thedecline in cover in the cleared quadrats was mirrored by similar declines in the algalassemblage. Sessile animals, e.g. the worm Galeolaria caespitosa, P. stolonifera and T.rosea, contributed more to measures of dissimilarity between these assemblages thanwas the case on the above shores. These species were, however, very patchily distributedand were not consistently abundant in either assemblage. Gradually, many commonmacro-algae in the surrounding assemblage grew into the cleared quadrats and, duringmost of the study, temporal changes in abundances of these were similar in the twoassemblages.The changes in the cleared quadrats at Scotts Head reflected the lack of convergenceof these assemblages with those growing in the surrounding areas (Fig. 3e). Thesurrounding algal assemblage was relatively diverse, although predominantly made up ofCorallina spp. (Fig. 9b). Unlike the other shores, there was not much bare space in thecleared quadrats, but what there was persisted into the third year of the study (Fig. 9a).The barnacle, T. rosea, arrived in relatively large numbers at the start of the study andpersisted throughout, apparently augmented by repeated settlement (Fig. 9c). Simul-taneously, cover of Corallina spp. remained smaller than in the surrounding algalassemblages (Fig. 9b). Cover of Corallina spp. was also very patchy in the clearedquadrats, with more variability (variance among quadrats) in the cleared areas than in theundisturbed areas for the first two years of the study (Fig. 9d). Slow growth of algae andpersistence of T. rosea in the cleared quadrats ensured that these two assemblages didnot converge during the study (Fig. 3e). In the quadrats that were cleared and sampledthree months later, cover of T. rosea was negatively correlated with cover of Corallinaspp. (Fig. 10; P , 0.01), although there were many quadrats in which neither of thesetwo species dominated. Such quadrats had patchy cover of bare space and diversemacro-algae.3.4. Development of the early stages of the assemblage at different timesIn March, 1980, the relative importance of U. lactuca, Corallina spp. and bare spacein determining differences between the recently-cleared patches and the surroundingM.G. Chapman, A.J. Underwood / J. Exp. Mar. Biol. Ecol. 224 (1998) 265 289 281Fig. 9. Mean (S.E.) percentage cover of (a) bare space, (b) Corallina spp. and (c) T. rosea and (d) variance inpercentage cover of Corallina spp. in (s) quadrats that were cleared in January, 1980 and in (d) quadrats inthe algal assemblage at Scotts Head; n 5 4 or 5 quadrats in each assemblage at each time of sampling.algal assemblages varied between Tura Head, Flat Rock Island and Cape Banks (Figs.68). The relative differences in cover of these taxa in the cleared areas from shore toshore contributed to these measures of dissimilarity (Table 2). These taxa include aspecies that is opportunistic and rapidly recruits into disturbed areas (U. lactuca) and agenus that is slow-growing and probably colonizes clearings from regrowth from thebasal crust in addition to possible recruitment (Corallina spp.). The cover of bare spaceis an inverse measure of recruitment and growth of all taxa in the clearings.282 M.G. Chapman, A.J. Underwood / J. Exp. Mar. Biol. Ecol. 224 (1998) 265 289Fig. 10. Negative correlation between percentage cover of Corallina spp. and T. rosea in all quadrats at ScottsHead, three months after they were initially cleared (n 5 73; two photographic quadrats were lost).It was predicted that the relative rates of development of these three variables (coverof U. lactuca, Corallina spp. and bare space) in cleared quadrats among these shoreswould be similar when cleared areas are provided at different times of year and indifferent years. This prediction was tested by examining the percentage cover of thesetaxa in recently-cleared quadrats in March and October, 19801983, thereby testing thehypothesis for the same season in different years (March, 19811983) and for anotherseason (October, 19801983; Table 4).The relative covers of these variables were inconsistent from shore to shore and timeto time (Table 4). Temporal variability was unpredictable, i.e. there were no seasonal orother temporal patterns. Similarly, there was no consistent pattern in the development ofthese variables from shore to shore. When these taxa were not important in contributingto cover, a variable mixture of other macro-algae and sessile animals developed in thecleared quadrats. There was also no biogeographic or temporal pattern in which taxaoccupied space, nor in the importance of U. lactuca and Corallina spp.Second, from patterns obtained during March, 1980, it was predicted that theassemblage in the cleared quadrats at Flat Rock Island would converge with theTable 4Mean (S.E.; n 5 5 in most cases) percentage cover of U. lactuca (Ul), Corallina spp. (Cor). and bare space(bs) in quadrats cleared three months prior to sampling at Tura Head, Flat Rock Island and Cape Banks; datafrom March and October, 19801983Time Tura Head Flat Rock Island Cape BanksUl Cor bs Ul Cor bs Ul Cor bsMar. 1980 16 (6) 0 (0) 3 (2) 44 (12) 53 (12) 3 (1) 33 (5) 1 (1) 12 (3)Mar. 1981 1 (0) 0 (0) 56 (10) 0 (0) 0 (0) 8 (4) 3 (2) 7 (7) 20 (3)Mar. 1982 0 (0) 0 (0) 6 (4) 0 (0) 0 (0) 46 (15) 2 (2) 1 (1) 20 (8)Mar. 1983 0 (0) 0 (0) 4 (2) 1 (1) 0 (0) 7 (3) 52 (22) 0 (0) 13 (13)Oct. 1980 2 (2) 7 (5) 58 (12) 1 (1) 0 (0) 0 (0) 45 (10) 4 (2) 18 (3)Oct. 1981 0 (0) 0 (0) 12 (9) 0 (0) 0 (0) 60 (15) 22 (20) 4 (4) 26 (11)Oct. 1982 0 (0) 0 (0) 23 (17) 0 (0) 4 (4) 5 (3) 0 (0) 0 (0) 5 (3)Oct. 1983 0 (0) 5 (3) 5 (3) 0 (0) 0 (0) 98 (1) 32 (17) 13 (8) 10 (4)M.G. Chapman, A.J. Underwood / J. Exp. Mar. Biol. Ecol. 224 (1998) 265 289 283Table 5Relative dissimilarity [Dis 2 (Dis 1 Dis ) /2)] between the algal assemblage and the assemblage de-B WA Cveloped in quadrats which were cleared three months previously at Tura Head, Flat Rock Island and CapeBanks; data from March and October, 19801983 (except Tura Head, where there were no data for 1983)Time Tura Head Flat Rock Island Cape BanksMar. 1980 54.7 23.9 63.6Mar. 1981 59.2 28.1 54.6Mar. 1982 75.6 65.0 63.5Mar. 1983 63.9 53.2 33.7Oct. 1980 39.7 59.6 39.9Oct. 1981 39.7 48.4 30.1Oct. 1982 64.7 38.4 49.1Oct. 1983 63.3 29.2surrounding algal assemblage at a greater rate than at Cape Banks and Tura Head (Fig.3). To test this hypothesis, the test statistic [(Dis -(Dis 1 Dis ) /2)] was calculatedB WA Cfor each of these shores in March and October, 19801983 (Table 5; no statistic forTura Head in October, 1983 because of missing data). In contrast to the prediction, therewas no evidence for the rate of convergence of the cleared quadrats with the surroundingassemblage to be consistently faster at Flat Rock Island than on the other shores (Table5). Rates varied interactively; differences from shore to shore varied in an unpredictablemanner from one time of sampling to another.The long-term study also suggested that, at Scotts Head, there were repeated bouts ofrecruitment of the barnacle, T. rosea (Fig. 9), thereby ensuring that the assemblage thatdeveloped in the cleared areas did not converge on the surrounding algal assemblage. Totest the prediction that similar results would have been found had the experiment beendone at other times, the percentage cover of T. rosea and bare space in the repeatedly-cleared quadrats was measured in March and October, 19801983. At each time, exceptin October, 1982 when there was no recruitment of T. rosea, large densities of barnaclesrecruited into the recently-cleared quadrats (mean percentage cover varied between 18%and 89%).Finally, multivariate measures of difference between the algal assemblage andquadrats that were cleared three months previously were compared between March, 1980and March, 1981 and between March, 1980 and October, 1980 using ANOSIM. Thiswas done for all shores except Scotts Head, where developing assemblages in clearingsobviously did not converge on the surrounding assemblage, so the temporal comparisonwould make no sense. In all cases, there were significant differences in the assemblagesin the clearings from time to time. On some shores, e.g. Cape Banks, the surroundingalgal assemblages also differed from time to time (Fig. 11a). At Tura Head, in contrast,the undisturbed algal assemblages were not different at each time of sampling, but,nevertheless, the cleared quadrats developed significantly different early-stage assem-blages (Fig. 11b). In general, although significantly different early-stage assemblagesdiffered at each time on each shore, these assemblages were more similar to thesurrounding algal assemblage present at that time than present at other sampling times(i.e. they plotted closer to each other in Fig. 11).284 M.G. Chapman, A.J. Underwood / J. Exp. Mar. Biol. Ecol. 224 (1998) 265 289Fig. 11. MDS plots of the replicate quadrats in the areas that were cleared three months previously (emptysymbols) and the undisturbed assemblages (filled symbols) comparing (i) March, 1980 (j; h) and October,1980 (d; s) and (ii) March, 1980 (j; h) and March, 1981 (d; s) for (a) Cape Banks and (b) Tura Head.Finally, there appeared to be greater variability among replicate quadrats in the areasthat had been cleared compared to quadrats in the undisturbed algal assemblages (Fig.11), suggesting that one effect of disturbance had been to alter variability of theseassemblages (Warwick and Clarke, 1993).4. DiscussionThe early stages of development of these lowshore algal assemblages differedsignificantly from shore to shore and time to time. There was no evidence for commonbroad-scale patterns of recruitment, perhaps governed by widespread climatic conditions,such as onshore winds (Shanks, 1983; Bertness et al., 1996). Neither was there clearevidence for any biogeographic northsouth trend in these assemblages, even thoughthey spanned nearly 800 km of coast-line. Each shore developed a unique early-stageassemblage at each time of sampling. Such variation was predictable from numerousprevious studies which have demonstrated large spatial variation in recovery fromdisturbance (e.g. Dethier, 1984; Jernakoff, 1985; Menge et al., 1993; Benedetti-Cecchiand Cinelli, 1994).Nevertheless, these early assemblages developed from a limited common pool ofspecies. These were primarily opportunistic green algae, such as U. lactuca, C. aereaand E. intestinalis, bluegreen algal films, the barnacle, T. rosea and Corallina spp.M.G. Chapman, A.J. Underwood / J. Exp. Mar. Biol. Ecol. 224 (1998) 265 289 285Each of these, except Corallina spp., would have recruited from the plankton or frompropagules released by surrounding plants. Corallina spp. is more likely to have grownback from a basal crust. Note, however, that Menge et al. (1993) found that corallineswere the most frequent colonizers of experimentally-cleared plots on wave-exposedshores. In their experiments, a similar preponderance of Corallina occurred incompletely cleared plots, so colonization was rapid even when dependent on non-vegetative recovery by spores. Differences in recolonization from shore and shore andtime to time in this study were largely governed by the relative abundance of these fewspecies, rather than differences in the types of species, although some shores at sometimes received species not found on other shores. For example, the opportunistic earlycolonizer, U. lactuca, was found in varying amounts in cleared quadrats at many timesof sampling on all shores in the study. Other early colonizers, such as Chaetomorphaaerea and Enteromorpha intestinalis, were found only sporadically on some shores.Therefore, there is probably a common pool of potential recruits along the entire coast,but rates of recruitment or post-recruitment mortality vary from shore to shore.Similarly, previous authors have documented considerable small- and large-scale spatialand temporal variability in recruitment of midshore assemblages on the coast of NewSouth Wales (Caffey, 1982; Underwood et al., 1983) and in local algal assemblages(Underwood and Jernakoff, 1981; Jernakoff, 1985).Nevertheless, from these different starting points, there was rapid convergence ofdifferent early assemblages towards the surrounding assemblages on all shores, exceptScotts Head. On this most northerly shore, assemblages that developed in the clearingsdid not converge with the surrounding algae, but remained different for four years afterthe start of the experiment. Growth of algae was limited, there was repeated settlementof T. rosea and considerable cover of bare space persisted. The rock-surface on thisshore was very smooth and hard and recently-settled barnacles and algae could be easilydislodged by hand (Underwood, unpub. data). Algae did settle, but did not survive orgrow fast enough to overgrow the barnacles (Denley and Underwood, 1979; Underwoodet al., 1983). There was also relatively smaller cover of Corallina spp. in thesurrounding assemblage to regrow into the clearings. Bare space was thereforecontinuously available for settlement of barnacles leading to persistent differencesbetween assemblages.Except for Scotts Head and in contrast to Dyes (1993) findings, clearings didrecover to be quite similar to surroundings. In Dyes experiments, small clearings werekept clear by the activities of grazers. The low-shore regions of rocky shores in NewSouth Wales tend to have few grazers (Underwood, 1981; Underwood and Jernakoff,1981). Molluscs such as limpets are notably ineffective at reducing the rate ofcolonization by algae (Underwood and Jernakoff, 1981, 1984) in contrast to their greatefficiency at higher levels on the shore (Underwood, 1980). There are large chitons, butthese have limited and very localised effects on recovery of low-shore algae on thewave-exposed coasts studied here (unpub. data). Similarly, Menge et al. (1993)demonstrated experimentally that grazing had no significant role in recolonization bycoralline algae in their study sites.Rates of convergence varied from shore to shore so that, after between six and 18months, cleared areas resembled the surrounds. This is quite fast in comparison with286 M.G. Chapman, A.J. Underwood / J. Exp. Mar. Biol. Ecol. 224 (1998) 265 289some studies (e.g. Dethier, 1984). There was no biogeographic trend in these rates andrepeated experiments on the developement of the early stages in cleared areas suggestthat rates of convergence would differ unpredictably from time to time, in addition toshore to shore. Therefore, there were no clear seasonal or long-term patterns in rates ofrecruitment or growth of the ephemeral algae (cf Benedetti-Cecchi and Cinelli, 1996) orconvergence of these assemblages.Specific changes leading to convergence also differed from shore to shore, as wouldbe expected given that the established algal assemblages differed. Nevertheless, they didfollow a general pattern. As discussed previously, the clearings were rapidly colonizedby fast-growing, ephemeral green algae, such as U. lactuca and sessile animals, such asthe barnacle, T. rosea and the tubeworm, G. caespitosa. Such early-colonizing speciesrapidly exploit recent disturbances (Sousa, 1979b, 1980; Denley and Underwood, 1979).They were gradually replaced by the development of perennial algae. Corallina spp.developed quickly on most shores, probably because they were not dependent on thesupply of propagules and probably regrew from the basal crust. Other perennial algae,such as Gracilaria sp., Laurencia sp. and Sargassum spp., were more patchy andvariable from shore to shore, both in the developing and established algal assemblages.The availability of a local spore bank (Santelices et al., 1995) on each shore wasprobably important in the development and convergence of these new assemblages withthe surrounding assemblages. The present results are, however, quite consistent withBenedetti-Cecchi and Cinellis (1994) experimental studies of coralline algae; coloniza-tion differed between sites, but tended to conform to what was immediately surroundingthe experimental plots. In the experiments described here, there was also convergence tosurroundings despite the different starting assemblages.Previous studies of colonization of algal assemblages have found quite differentoutcomes. For example, Jernakoff (1985), for several common and widespread speciesin New South Wales, found no coherent seasonal cycles and that temporal variability inabundances was large and unpredictable. Emerson and Zedler (1978), however, foundthat Californian algae showed strong seasonality but little predictability in recovery.Benedetti-Cecchi and Cinelli (1994) found correspondence with surrounding assem-blages, despite different beginnings and routes. Thus, many studies have documentedunpredictability in the natural assemblages, the initial colonists, the final outcomes orany combination of these.In complete contrast, several experimental studies have documented great predicabili-ty in recovery of disturbed algal assemblages. For example, Daytons (1975) experi-ments on large low-shore species were characterized by recovery of dominance of thecanopy by Hedophyllum in all but the most wave-exposed sites. In those sites, otherspecies always out-competed Hedophyllum, but the results were predictable. Again,Sousa (1979b, 1980) manipulated regimes of disturbance in boulder-fields and was ableto show a consistent successional sequence of algae leading to domination by particularred species, provided that the area was not disturbed again en route. The initial colonists,intermediate stages and final outcome were predictable. Whether any particular boulderwould develop the complete sequence was not. Such results are also known forhigh-shore algae, such as Postelsia, which consistently colonizes cleared patches(Blanchette, 1996). Kim and de Wreede (1996) concluded, from experimental clearing,M.G. Chapman, A.J. Underwood / J. Exp. Mar. Biol. Ecol. 224 (1998) 265 289 287that different algae responded in different ways to disturbance. Nevertheless, theyconsidered the results were predictable from knowledge of the time of disturbance andlife-history characteristics (time of reproduction, whether recovery was by vegetativeregrowth, etc.).Assemblages developing in recently-cleared quadrats were more variable from quadratto quadrat than were the algal assemblages. Increased variability among patches hasbeen considered a symptom of disturbance or stress (Warwick and Clarke, 1993). In thisstudy, because the cleared areas were not sampled at intervals to monitor rates of arrivalof different taxa and post-recruitment mortality, it is not possible to determine whetherthis pattern was simply due to patchy arrival of propagules in the different clearings, orlocal post-recruitment disturbances subsequently leading to this patchiness. Increasedpatchiness among the cleared quadrats was greater soon after the clearing and didsubsequently decline. Nevertheless, in general, cleared quadrats remained more variablethan did the quadrats in the surrounding assemblages (see Fig. 3). It is unlikely thatdisturbance associated with clearing the areas would have persisted through four years,which suggests that effects of spatially variable early stages of development may persistfor relatively long periods, even though, on average, the assemblages may be seen toconverge. Clearing patches in these assemblages therefore increased patchiness of theassemblage and this effect may last up to four years.The degree to which recovery from disturbance is consistent or predictable and theinfluence of disturbance on spatial variation needs to be determined empirically in anyregion. More experimental analysis is needed about the processes influencing recoloniza-tion by the species recorded here and about the large number of other, less abundant andwidespread species that may be found on local exposed shores. Until such work has beendone, the influence of unpredictable colonization and increased variation due todisturbances on local patterns of diversity of algae will remain unknown.The present results do demonstrate quite clearly that there is no simple seasonal norclear-cut biogeographical pattern, that the surrounding assemblage has importantinfluences and that local influences out-weigh regional ones. These results confirm thatbiogeographical comparisons within a region necessitate experimental examination ofprocesses in numerous places. Without this, the relevant scales to examine cannot bedetermined. They also demonstrate that biogeographical comparisons from one region toanother need considerable underpinning by experimentally derived data (see particularlyUnderwood and Petraitis, 1993). The discovery that there are no clear geographicaltrends does, however, reduce the spatial scale over which these complex assemblagescan be investigated. Because so much of the outcome of disturbance is related to localfactors and features, studies on shores in any part of the coast will mostly be appropriate,without the necessity of having to study everything over the entire coast-line.AcknowledgementsThis research was assisted by grants from the Australian Research Council andInstitute of Marine Ecology. S. Kennelly, J. MacLulich, S. McNeill, P. Scanes and W.Steel assisted with the fieldwork at various times; J. Grayson, J. Harris and V. Mathews288 M.G. Chapman, A.J. Underwood / J. Exp. Mar. Biol. Ecol. 224 (1998) 265 289assisted with the preparation of the paper. Anonymous referees provided usefulcomments to improve the paper.ReferencesBenedetti-Cecchi, L., Cinelli, F., 1994. Recovery of patches in an assemblage of geniculate coralline algae:variability at different successional stages. Mar. Ecol. Prog. Ser. 110, 918.Benedetti-Cecchi, L., Cinelli, F., 1996. Patterns of disturbance and recovery in littoral rock pools: non-hierarchical competition and spatial variability in secondary succession. Mar. Ecol. Prog. Ser. 135,145161.Bertness, M.D., Gaines, S.D., Wahle, R.A., 1996. Wind-driven settlement patterns in the acorn barnacleSemibalanus balanoides. Mar. Ecol. Prog. Ser. 137, 103110.Blanchette, C.A., 1996. Seasonal patterns of disturbance influence recruitment of the sea palm, Postelsiapalmaeformis. J. Exp. Mar. Biol. Ecol. 197, 114.Caffey, H.M., 1982. No effect of naturally-occurring rock types on settlement or survival in the intertidalbarnacle Tesseropora rosea (Krauss). J. Exp. Mar. Biol. Ecol. 63, 119132.Clarke, K.R., 1993. Non-parametric multivariate analyses of changes in community structure. Aust. J. Ecol. 18,117143.Connell, J.H., Sousa, W.P., 1983. On the evidence needed to judge ecological stability or persistence. Am. Nat.121, 789824.Dakin, W.J., 1987. Australian Seashores. Angus and Robertson, Sydney, 411 pp.Dakin, W.J., Bennett, I., Pope, E., 1948. A study of certain aspects of the ecology of the intertidal zone of theNew South Wales coast. Aust. J. Sci. Res. Ser. B 1, 176230.Dayton, P.K., 1971. Competition, disturbance and community organization: the provision and subsequentutilization of space in a rocky intertidal community. Ecol. Monogr. 41, 351389.Dayton, P.K., 1975. Experimental evaluation of ecological dominance in a rocky intertidal algal community.Ecol. Monogr. 45, 137159.Denley, E.J., Underwood, A.J., 1979. Experiments on factors influencing settlement, survival and growth oftwo species of barnacles in New South Wales. J. Exp. Mar. Biol. Ecol. 36, 269293.Dethier, M.N., 1984. Disturbance and recovery in intertidal pools: maintenance of mosaic patterns. Ecol.Monogr. 54, 99118.Dye, A.H., 1993. Recolonization of intertidal macroalgae in relation to gap size and molluscan herbivory on arocky shore on the east coast of Southern Africa. Mar. Ecol. Prog. Ser. 95, 263271.Emerson, S.E., Zedler, J.B., 1978. Recolonization of intertidal algae: an experimental study. Mar. Biol. 44,315324.Fairweather, P.G., Underwood, A.J., 1991. Experimental removals of a rocky intertidal predator: variationswithin two habitats in the effects on prey. J. Exp. Mar. Biol. Ecol. 154, 2975.Foster, M.S., 1990. Organization of macroalgal assemblages in the Northeast Pacific: the assumption ofhomogeneity and the illusion of generality. Hydrobiologia 192, 2134.Jernakoff, P., 1985. Temporal and small-scale spatial variability of algal abundance in an intertidal rocky shore.Bot. Mar. 28, 145154.Kim, J.H., de Wreede, R.G., 1996. Effects of size and season of disturbance on algal patch recovery in a rockyintertidal community. Mar. Ecol. Prog. Ser. 133, 217228.Lubchenco, J., 1980. Algal zonation in the New England rocky intertidal community: an experimental analysis.Ecology 61, 333344.Menge, B.A., Farrell, T.M., Olson, A.M., Van Tamelen, P., Turner, T., 1993. Algal recruitment and themaintenance of a plant mosaic in the low intertidal region on the Oregon Coast. J. Exp. Mar. Biol. Ecol.170, 91116.Santelices, B., Hoffmann, A.J., Aedo, D., Bobadilla, M., Otaza, R., 1995. A bank of microscopic forms ondisturbed boulders and stones in tide pools. Mar. Ecol. Prog. Ser. 129, 215228.M.G. Chapman, A.J. Underwood / J. Exp. Mar. Biol. Ecol. 224 (1998) 265 289 289Shanks, A.L., 1983. Surface slicks associated with tidally forced internal waves may transport pelagic larvae ofbenthic invertebrates and fish shoreward. Mar. Ecol. Prog. Ser. 24, 289295.Sousa, W., 1979a. Disturbance in marine intertidal boulder fields: the nonequilibrium maintenance of speciesdiversity. Ecology 60, 12251239.Sousa, W.P., 1979b. Experimental investigations of disturbance and ecological succession in a rocky intertidalalgal community. Ecol. Monogr. 49, 227254.Sousa, W.P., 1980. The responses of a community to disturbance: the importance of successional age andspecies life histories. Oecologia (Berlin) 45, 7281.Sousa, W.P., Schroeter, S.C., Gaines, S.D., 1981. Latitudinal variation in intertidal algal community structure:the influence of grazing and vegetative propagation. Oecologia (Berlin) 48, 297307.Underwood, A.J., 1975. Comparative studies on the biology of Nerita atramentosa Reeve, Bembicium nanum(Lamarck) and Cellana tramoserica (Sowerby) (Gastropoda: Prosobranchia) in S.E. Australia. J. Exp. Mar.Biol. Ecol. 19, 153172.Underwood, A.J., 1980. The effects of grazing by gastropods and physical factors on the upper limits ofdistribution of intertidal macroalgae. Oecologia (Berlin) 46, 201213.Underwood, A.J., 1981. Structure of a rocky intertidal community in New South Wales: patterns of verticaldistribution and seasonal changes. J. Exp. Mar. Biol. Ecol. 51, 5785.Underwood, A.J., Chapman, M.G., 1997. Variation in algal assemblages on wave-exposed rocky shores. Mar.Freshw. Res. (submitted).Underwood, A.J., Denley, E.J., 1984. Paradigms, explanations and generalizations in models for the structureof intertidal communities on rocky shores. In: Strong, D.R., Simberloff, D., Abele, L.G., Thistle, A.B.(Eds.), Ecological Communities: Conceptual Issues and the Evidence. Princeton University Press, NewJersey, pp. 151180.Underwood, A.J., Denley, E.J., Moran, M.J., 1983. Experimental analyses of the structure and dynamics ofmid-shore rocky intertidal communities in New South Wales. Oecologia (Berlin) 56, 202219.Underwood, A.J., Jernakoff, P., 1981. Interactions between algae and grazing gastropods in the structure of alow-shore algal community. Oecologia (Berlin) 48, 221233.Underwood, A.J., Jernakoff, P., 1984. The effects of tidal height, wave-exposure, seasonality and rock-pools ongrazing and the distribution of intertidal macroalgae in New South Wales. J. Exp. Mar. Biol. Ecol. 75,7196.Underwood, A.J., Petraitis, P.S., 1993. Structure of intertidal assemblages in different locations: how can localprocesses be compared? In: Ricklefs, R.E., Schluter, D. (Eds.), Species Diversity in Ecological Com-munities: Historical and Geographical Perspectives. University of Chicago Press, Chicago, pp. 3851.Warwick, R.M., Clarke, K.R., 1993. Increased variability as a symptom of stress in marine communities. J.Exp. Mar. Biol. Ecol. 172, 215226.

Recommended

View more >