separating the elements of habitat structure: independent effects of habitat complexity and...

LJournal of Experimental Marine Biology and Ecology249 (2000) 29–49

www.elsevier.nl / locate / jembe

Separating the elements of habitat structure: independenteffects of habitat complexity and structural components on

rocky intertidal gastropods

*Michael W. BeckInstitute of Marine Ecology, A11, University of Sydney, Sydney, NSW 2006, Australia

Received 6 September 1999; received in revised form 1 February 2000; accepted 9 February 2000

Abstract

It has been difficult to understand the effects of habitat structure on assemblages because thedifferent elements of habitat structure are often confounded. For example, few studies considerthat the effects of structural components of a habitat (rocks, trees, pits, pneumatophores) may beseparate from the complexity (e.g. surface area hSAj) they create. From prior observations andexperiments, I developed three hypotheses about the effects of habitat structure on gastropods onrocky intertidal shores in Botany Bay, Australia. (1) The complexity of habitats positively affectsthe density and richness of gastropods. (2) The fractal dimension (D) represents elements ofcomplexity that affect the density and richness of gastropods better than other indices ofcomplexity. (3) The effects of specific structural components on the density and richness ofgastropods are independent of their complexity. To test these hypotheses, treatments composed ofpits and pneumatophores were used to independently manipulate complexity and structuralcomponents in experiments repeated at five different times on two shores. There was support forhypotheses (1) and (3) at most times and places but not for hypothesis (2). Richness, total density,and the densities of two of the three most common gastropods were greater in treatments withgreater complexity. D was not definitively better than other indices of complexity, but D and SAwere recommended for further consideration. When complexity was held constant, species richnessand the density of most gastropods, except Austrocochlea porcata, was greater in treatments withpits than with pneumatophores. A common mechanistic explanation for the effects of habitatcomplexity on rocky intertidal gastropods relies on a specific characteristic of pits; they pool waterand reduce desiccation stress. This assumption may be appropriate for many gastropods, but it wasinappropriate for A. porcata. Habitat complexity affected its density, but this was not because of acharacteristic specific to pits. The complexity and structural components of habitats have separateeffects on assemblages, and it confuses the study of habitat structure to combine them. 2000Elsevier Science B.V. All rights reserved.

*Present address: The Nature Conservancy, A316 Earth and Marine Sciences, University of California, SantaCruz, CA 95064, USA. Tel.: 1 1-831-469-3608; fax: 1 1-831-459-4882.

E-mail address: [email protected] (M.W. Beck)

0022-0981/00/$ – see front matter 2000 Elsevier Science B.V. All rights reserved.PI I : S0022-0981( 00 )00171-4

30 M.W. Beck / J. Exp. Mar. Biol. Ecol. 249 (2000) 29 –49

Keywords: Fractal dimension; Gastropod; Habitat structure; Habitat complexity; Structural components; Rockyintertidal

1. Introduction

Habitat structure may affect the diversity and abundance of species in many systems(Grinnell, 1917; Gause, 1934; Huffaker, 1958; Connell, 1961; Kohn and Leviten, 1976;Connor and McCoy, 1979; Menge and Lubchenco, 1981; Werner et al., 1983; Leber,1985; Underwood and Chapman, 1989). Studies of the effects of habitat structure arenearly as common as those of competition and predation, but there are few reviews ofhabitat structure, and it is rarely discussed in textbooks (McCoy and Bell, 1991). Thislacuna occurs because there has been little consistency in the (1) definition or (2)measurement of habitat structure between different studies or habitats. These problemsmake it difficult to compare methods and results between studies, which constrains thedevelopment of a broader understanding of the effects of habitat structure on thediversity and abundance of species.

To address the first problem, McCoy and Bell (1991) offered a definition of habitatstructure, which can help guide research and enhance interpretability of results fromdifferent studies and habitats. Their definition of habitat structure proposes that it iscomposed of at least two major factors, complexity and heterogeneity, and their effectsand measurement are scale-dependent (McCoy and Bell, 1991; Beck, 1998; Downes etal., 1998). Complexity encompasses variation in habitat structure attributable to theabsolute abundance of individual structural components. Structural components aredistinct physical elements of the habitat, e.g. rocks, trees, pits, and pneumatophores(McCoy and Bell, 1991; Downes et al., 1998). Heterogeneity encompasses variation inhabitat structure attributable to variation in the relative abundance of different structuralcomponents (McCoy and Bell, 1991; Beck, 1998; Downes et al., 1998). The basis of aneffect of heterogeneity is that the effects of different structural components areindependent of their complexity.

McCoy and Bell (1991) helped clarify the difference between complexity andstructural components, but these two elements of habitat structure are often confounded.Some studies suggested that particular structural components affected the density anddiversity of animals but did not account for differences in complexity (e.g. Inglis, 1994;Dittel et al., 1996; Primavera, 1997). For example, Miller and Carefoot (1989) comparedthe effects of two different structural components, rocky pits and adult barnacles, on thedensity of juvenile barnacles. They concluded that adult barnacles had a greater effect onthe density of juvenile barnacles than pits but did not consider that these results might belargely explained by the greater complexity (e.g. SA) of their treatments with adultbarnacles.

Other studies suggested that complexity affected the density and diversity of animals,but their manipulations of complexity were confounded with differences in structuralcomponents (e.g. Russ, 1980; Coull and Wells, 1983; Gilinsky, 1984; Leber, 1985; Bellet al., 1987; Pennings, 1990; Diehl, 1993). In these studies, complexity was manipulated

M.W. Beck / J. Exp. Mar. Biol. Ecol. 249 (2000) 29 –49 31

in each treatment by changing either the type of structural component (e.g. one speciesof algae vs. another) or the number of different structural components (e.g. one vs. twotypes of vegetation). These latter studies confounded complexity, structural components,and heterogeneity. When studies of habitat structure are confounded, it is not possible toidentify the important elements of habitat structure or the mechanistic basis for theireffects on species.

Habitat complexity and structural components have been experimentally separated injust a few studies (McGuinness, 1984; Dean and Connell, 1987; Jenkins and Sutherland,1997). Stoner and Lewis (1985) showed that the total abundance of epifaunalcrustaceans was affected primarily by the surface area of vegetation. It mattered littlewhether this complexity was created by seagrass or algal components. A few species,however, had greater densities in plots with algae than in plots with similar complexitiesof seagrass (i.e. there were effects of structural components over and above the effects ofcomplexity).

The lack of distinction between complexity and structural components is tied to thesecond major problem that limits our understanding of habitat structure; there is littleconsistency in the measurement of complexity between different studies and habitats(McCoy and Bell, 1991; Beck, 1998). This problem arises because most indices ofcomplexity are based on specific structural components (e.g. counts of number of treesor pits), which severely limits their use and interpretation. A number of indices ofcomplexity, e.g. D and SA, can be used more broadly, because their measure ofcomplexity is independent of specific structural components. At present, there is littlejustification for the use of one index over another (but see Carleton and Sammarco,1987; Underwood and Chapman, 1989; McCormick, 1994; Beck, 1998), and experi-ments are necessary to determine which index best represents the elements ofcomplexity that affect species (e.g. Jacobi and Langevin, 1996).

In previous work, I showed that the effects of complexity on gastropods could bemeasured and compared in rocky intertidal and mangrove habitats (Beck, 1998). Thesehabitats had very different components (pits and pneumatophores, respectively) andcomplexities as measured by several indices of complexity (Fig. 1): D, vector dispersion

2(VD), chain-and-tape (Chain), and consecutive substratum height difference (odh ). Inthe rocky intertidal habitat, the density of gastropods was significantly and positivelycorrelated with most indices of complexity, and D had the highest correlations withdensity.

From these observations, three hypotheses of the effects of habitat structure on thedensity and richness of gastropods in this rocky intertidal habitat were formulated. (1)The complexity of habitats positively affects the density and richness of gastropods. (2)D represents elements of complexity that affect the density and richness of gastropodsbetter than other indices. (3) The effects of specific structural components on the densityand richness of gastropods are independent of the effects of complexity. I manipulatedcomplexity and structural components to directly test predictions from these hypotheses.These manipulations test whether complexity, structural components, or both affect thedensity and richness of gastropods, and they help identify the appropriate indice(s) ofcomplexity. If the elements of habitat structure can be clearly defined and measured, itwill make information from different studies and habitats more commensurable and help


Fig. 1. Measurement of the indices of structural complexity. (a) Transect showing profile of habitat (profile2can be obtained from a real habitat or an experimental mimic). (b) Illustration of the calculation of VD, odh

nand Chain at the 5-mm interval. VD is a measure of the variance in u ; VD 5 n 2 o a /c Y n 2 1 , n is thes h s djd s d1

number of separate triangles along the transect. (c) Illustration of the calculation of D by the dividers method.The points are the apparent length of the transect measured by ‘dividers’ of increasingly greater intervals. Theequation for the line and the calculation of D are shown.

reduce the confusion that hinders our understanding of the effects of habitat structure oncommunity structure.

2. Methods

2.1. Natural history

All observations and experiments were done in Botany Bay, Australia on two rockyintertidal shores, Bare Island and Sutherland Point (Fig. 2). The study sites on theseshores were on moderately sheltered, mid-shore sandstone rock benches. The rocksappeared to be mostly bare with some encrusting algae; there was little or no foliosemacroalgae. The obvious structural components were shallow pits (often hemispherical)in the rock. Gastropods, mainly herbivores, were the most abundant, macroscopicanimals in the study areas. Three species were common, the limpet Cellana tramosericaand the snails Bembicium nanum and Austrocochlea porcata (ex. constricta). A. porcata


Fig. 2. Map of Botany Bay showing rocky intertidal shores.

also occurs in nearby mangrove habitats, but the other species do not. There are detaileddescriptions of these shores, their habitat structure, and the gastropods (e.g. Underwood,1975, 1976; Underwood and Chapman, 1996; Beck, 1998).

2.2. Habitat mimics and experimental design

To test predictions from hypotheses 1–3, complexity and structural components were2manipulated with 0.12-m polyester resin plates (Fig. 3). These plates were designed

from computer models to have levels of complexity similar to those observed in rockyintertidal habitats (Beck, 1998) and to have components that were similar to those foundin rocky intertidal and mangrove habitats (Fig. 3). A gold pigment paste which was inertwhen dried was added to the resin so that the color of the plates approximated the colorof the natural rock surface (rust-colored sandstone) (e.g. Walters and Wethey, 1996). Theplates were fastened on to the shore with stainless steel screws that were countersunkinto the plates and screwed into wall plugs in the rock.

Complexity was measured with a variety of indices (Fig. 3). The calculation of D,VD,2odh , and Chain is explained in Fig. 1 and Beck (1998). It is necessary to choose a clear

scale for measurements of complexity (McCoy and Bell, 1991). In prior work it wasshown that the gastropods at these sites were most affected by complexity measured at5-mm intervals as compared to greater intervals (Beck, 1998). Therefore, 5 mm was thefinest scale of measurement used in the development of these treatments. Onlygastropods larger than 5 mm in size were included in analyses. This size cutoff alsohelped to focus analyses on gastropods that moved onto treatments not larvae that settledon them. There were few gastropod or barnacle settlers on the plates during theexperiments.

These plates were used to test one assumption about the suitability of the plates as


Fig. 3. Experimental design for manipulation of complexity and structural components. Values underneathtreatments indicate their complexity as computed by different indices. The low complexity treatments arecoded so that the first letter indicates diameter (4 or 6 cm) and the second indicates depth of the pits (0.75, 1.0,or 1.5 cm); WM5wide, medium; NS5narrow, shallow; ND5narrow, deep; Pneum5pneumatophores. Pits onthe High complexity treatment were 4 and 6 cm in diameter and 1.7 cm deep. Pneumatophores were 3 cm talland 1.0 cm in base diameter. These dimensions were similar to the dimensions of natural structuralcomponents in these habitats.

habitat mimics and hypotheses 1–3 from the Introduction. The response variables were2 2the density (no. of individuals /0.12 m ) and richness (no. of species /0.12 m ) of

gastropods on each treatment.

2.2.1. Assumption 1: The plates were reasonable mimics of the habitatIf this assumption is correct, it is predicted that the density and richness of gastropods

on the high and low complexity treatments would be similar to the density and richnessof gastropods in undisturbed reference quadrats ( 5 control treatment) on the surround-ing rocky intertidal platform. The high and low complexity plates were explicitlydesigned to have complexities (Beck, 1998) and components similar to those measuredpreviously on the rocky intertidal platforms at these sites. The reference quadrats had thesame dimensions as the plates (30 3 40 cm) and their locations were haphazardlydetermined.

2.2.2. Hypothesis 1: Habitat complexity positively affects the density and richness ofgastropods

If this hypothesis is correct, the density and richness of gastropods is predicted to begreater in the high complexity treatment than in the group of three low complexitytreatments with pits.


2.2.3. Hypothesis 2: D represents elements of complexity that affect the density and2richness of gastropods better than other indices, i.e. VD, odh , surface area, number

of pitsThe three low complexity treatments with pits were explicitly designed to separate

among these indices. If D best represents elements of complexity that affect gastropods,as predicted by Beck (1998), their density and richness on the treatments should beNS , ND , WM (Table 1, Fig. 3). The other indices make different predictions aboutthe rank order of density and richness on treatments (Table 1, Fig. 3). D and Chainmeasure complexity in similar ways (Beck, 1998), and it was not possible to constructmanipulations from which contrasting predictions about their effects on the density andrichness of gastropods could be derived.

2.2.4. Hypothesis 3: Specific components have effects on the density and richness ofgastropods that are independent of complexity

This hypothesis is tested by manipulating only structural components and holdingcomplexity constant. Mimics of pits and pneumatophores of the mangrove Avicenniamarina were used to manipulate structural components. Pneumatophores were chosen,because (i) they are a definitively different component from pits, (ii) gastropods occuron pneumatophores on mangrove shores around Sydney, and (iii) in a companion study,manipulations of the complexity of pneumatophores affected the density and diversity ofgastropods in mangrove habitats (Beck, unpublished data). If this hypothesis is correct,the density and richness of gastropods is predicted to be greater on the three lowcomplexity treatments with pits than on the treatment with pneumatophores. It was notpossible to hold all indices of complexity constant on these treatments (Fig. 3), and anemphasis was placed on holding SA and D constant based in part on prior results (Beck,1998).

Experiments with these treatments were repeated as many as five different timesduring 1996 and 1997 at two different sites within each of the two shores. There weretwo replicates of each treatment at each site used in experiments. Not all treatments,sites, and shores were used in each experiment (Appendix 1). Experiment 3 included alltreatments at all sites and shores, but during the course of the experiment vandalsdestroyed all of the plates at Sutherland Point site 2. These plates were not replaced, anda second site was not used in experiments 3, 4, or 5 at Sutherland Point. Before each

Table 1aPredicted responses of gastropods to low complexity treatments based on different indices of complexity

Index Predicted density and richness

D NS,ND,WMVD ND,NS,WM

2odh NS,ND5WMChain NS,ND,WMSA NS5ND,WMNo. of pits ND,WM,NS

a The specific values of the indices on these treatments are in Fig. 3.


experiment, plates were scrubbed clean and randomly allocated to positions at the sites.Each experiment lasted approximately two months with variation in duration becauseweather and tides limited accessibility.

The five experiments were concluded in August, December 1996, February, April, andJune 1997. The experiments examine the reproducibility and temporal consistency ofthese results at different random times; they were not intended to test seasonalhypotheses per se. The experiments do, however, cover periods with different prevailingclimatic conditions.

2.3. Data analysis

The total density and richness of gastropods were analyzed for each experiment byANOVA. Treatment was a fixed factor and site was a random factor. In experiment 2,site was nested within shore, which was a random factor. The factor, shore, could not beused in ANOVAs for experiments 3–5, because the plates at the second site atSutherland Point were destroyed. Densities were transformed (log x) to meet conditionse

of homoscedasticity. In experiment 4, one replicate plate was lost, and in experiment 5,four plates were lost in storms (Appendix 1). To balance the ANOVAs for these missingreplicates, the data were replaced with the value from the other replicate at that time andplace. The residual df were reduced by 1 and 4, respectively. Mean square values wereobtained from SAS (Release 6.04, SAS Institute Inc., Cary, NC, USA).

When the effect of treatment was significant, a priori contrasts were done to test thepredictions of the assumption and hypotheses 1–3 above. To test the predictions fromhypothesis 2 (Table 1), it was necessary to do all the pairwise comparisons among thethree low complexity treatments. These comparisons were non-orthogonal. To hold thetype I error constant at a 5 0.05 for this group of comparisons each comparison wastested at a9 5 0.017 following Dunn–Sidak’s procedure (Underwood, 1997). When therewere significant treatment 3 site interactions, contrasts were done within each site withStudent–Newman–Keuls (SNK) tests (Underwood, 1997).

2.4. Combining data from experiments.

These separate experiments provided independent tests of the predictions identifiedabove, and it was possible with rank-order statistics to combine the data across thedifferent experiments. Data were combined across experiments, because in some casesthe results from individual experiments did not indicate significant differences amongtreatments, but there appeared to be clear patterns at most times and places. It also waspossible to summarize the results briefly and clearly by combining data acrossexperiments. In addition, it was possible to examine the effects of the treatments on thedensities of the three most abundant species across the experiments. ANOVAs were notdone on the densities of the individual species, because these data did not meet theassumptions of the test.

Two non-parametric tests were used to combine data; one proposed originally byAnderson (Anderson, 1959; Winer et al., 1991), which will hereafter be referred to as

2Anderson’s Q and a binomial test. The basis of both these tests is that responses to


treatments are ranked at the different times and places (trials) for each contrast orcomparison, and the likelihood that the rankings are randomly distributed amongtreatments is assessed. In these tests, each site in each experiment was a separate trial(i.e. each row in Appendix 1 represents a separate trial), and the responses to thetreatments were ranked within each trial for each of the a priori contrasts identified inassumption 1 and hypotheses 1–3. There was thus a maximum of 14 possible trials atseparate places and times (Appendix 1).

2Anderson’s Q was used to examine distributions of ranks in the contrasts (one vs.three treatments) when there were no tied ranks. In cases with tied ranks and forcomparisons (e.g. high vs. control), binomial tests were used to combine results acrossexperiments. If there was no difference in density or richness between treatments, eachtreatment should be ranked first on average 50% of the time (i.e. binomial p 5 q 5 0.5).When the binomial test was used for contrasts that involved the three low complexitytreatments with pits (i.e. low vs. control, high vs. low, pits vs. pneumatophores), densityand richness were averaged among the three treatments to condense the contrast to acomparison of two groups. In trials with tied values, a tie was counted against thepredictions (i.e. in support of the null prediction) if there was just one tie. When therewere two tied values, one value was counted in support of predictions and one against.In the analyses of the a priori contrasts for the individual species, a few trials weredropped from consideration, because no individuals of that species were observed on anytreatment in the trial. The number of omitted trials was noted on every occasion whenthis procedure was done.

3. Results

The ANOVAs indicated that there were differences in density and to a lesser extentrichness among the different sites and treatments (Tables 2, 3). The three most commonspecies accounted for most of the density of gastropods on the treatments (Fig. 4). Thedifferences among sites were not surprising (Fig. 4), because these sites werehaphazardly chosen. There were a few site by treatment interactions (Tables 2, 3); theseinteractions generally occurred because of variation in the magnitude, but not direction,of differences in treatments among sites. The differences among treatments wereexamined in the a priori contrasts.

3.1. Assumption 1: control vs. high and low complexity treatments

In most cases, the results were consistent with the predictions of no differencebetween control and complexity treatments. Total density was generally similar betweencontrol and high complexity treatments in the individual experiments (Table 2, Fig. 5a).When the data were combined across the different times and places, there was nosignificant difference in the ranking of density among trials (Table 4, Appendix 1,binomial P . 0.1). There was not a significant difference between the density ofgastropods in control vs. low complexity treatments in most of the individualexperiments (Table 2, Fig. 5a), but overall the density in the control treatments ranked

38M

.W.

Beck

/J.

Exp.

Mar.

Biol.

Ecol.

249(2000)

29–49

Table 2aComparisons of total density among treatments, sites, and shores in five separate experiments

(a) ANOVAs

Experiment 1 Experiment 2 Experiment 3 Experiment 4 Experiment 5

Source df MS F Source df MS F Source df MS F df MS F df MS F

Treatment 3 0.53 0.82 Treatment (5Tr) 4 4.62 7.74* Treatment 4 4.65 24.11*** 5 3.35 3.90* 5 3.27 3.06

Residual 4 0.64 Shore 1 0.31 0.07 Site 2 2.10 13.65*** 2 1.49 7.68** 2 2.54 11.61***

Site (Shore) 2 4.59 33.01*** Tr3Site 8 0.19 1.25 10 0.86 4.44** 10 1.07 4.90**b bTr3Shore 4 0.60 2.42 Residual 15 0.15 17 0.19 14 0.22

Tr3Site(Shore) 8 0.25 1.77

Residual 20 0.14

(b) Summary of a priori contrasts of density among treatmentsc cCondition Experiment 1 Experiment 2 Experiment 3 Experiment 4 Experiment 5

Assumption 1 Low5Control High5Control, No test B112 — Control5High, Low All — Control5High, Low

Control.Low S1 — Control.High, Low

Hypothesis 1 No test High.Low High.Low All — High5Low All — High5Low

Hypothesis 2 WM5ND5NS WM5ND5NS WM.NS, All — WM5ND5NS B11S1 — WM5NS5ND

WM5ND, NS5ND B2 — NS.WM5ND

Hypothesis 3 No test No test Pits.Pneumatophores B112 — Pits5Pneumatophores B112 — Pits5Pneumatophores

S1 — Pits.Pneumatophores S1 — Pits.Pneumatophores

a In the F-ratios, the mean squares for most terms are tested over the Residual. In experiment 2, Shore is tested over Site(Shore), Treatment is tested overTreatment3Shore, which is tested over Treatment3Site(Shore). In experiments 3–5, Treatment is tested over Treatment3Site.

b Df reduced to account for replacement of missing replicates (see text).c B1, Bare Island site 1; B2, Bare Island site 2; S1, Sutherland Point site 1; All, B1, B2, and S1.

M.W

.B

eck/

J.E

xp.M

ar.B

iol.E

col.249

(2000)29

–4939

Table 3aComparisons of richness among treatments, sites, and shores in the five experiments

(a) ANOVAs

Experiment 1 Experiment 2 Experiment 3 Experiment 4 Experiment 5

Source df MS F Source df MS F Source df MS F df MS F df MS F

Treatment 3 0.0 0.0 Treatment (5Tr) 4 2.54 3.03 Treatment 4 5.08 12.45** 5 2.05 2.32 5 5.31 6.42**

Residual 4 0.5 Shore 1 0.03 0.00 Site 2 7.03 8.44** 2 0.58 1.00 2 1.69 6.10**

Site (Shore) 2 5.63 11.84*** Tr3Site 8 0.41 0.49 10 0.88 1.51 10 0.83 2.98*b bTr3Shore 4 0.84 1.03 Residual 15 0.83 17 0.58 14 0.28

Tr3Site(Shore) 8 0.81 1.71

Residual 20 0.48

(b) Summary of a priori contrasts of richness among treatmentsc cCondition Experiment 1 Experiment 2 Experiment 3 Experiment 4 Experiment 5

Assumption 1 Low5Control High5Control, No test High5Control, B1 — High.Control

Control5Low Control5Low B2, S1 — High5Control

All — Low5Control

Hypothesis 1 No test High5Low High5Low High5Low B1 — High.Low

B2, S1 — High5Low

Hypothesis 2 WM5ND5NS WM5ND5NS WM5ND5NS WM5ND5NS B1, S1 — WM5ND5NS

B2 — NS.WM5ND

Hypothesis 3 No test No test Pits.Pneumatophores Pits5Pneumatophores B1 — Pits5Pneumatophores

B2,S1 — Pits.Pneumatophores

a The mean squares in the F-ratios are the same as in Table 2.b Df reduced to account for replacement of missing replicates (see text).c B1, Bare Island site 1; B2, Bare Island site 2; S1, Sutherland Point site 1; All, B1, B2, and S1.


Fig. 4. Total density of gastropods by species, treatment and site. Sites differed in total density and in therelative abundance of individual species. These values are pooled across all five experiments. Sutherland Pointsite 2 is not included because it was only used in one experiment. The Low column is the average density onthe three low complexity treatments with pits. Pneum5treatment with pneumatophores. Error bars indicate onestandard error for total density.


Fig. 5. Density and richness of gastropods by treatment and experiment. Data were averaged first within sitesbefore calculating the average and standard error among sites. The value is the mean11 S.E. Data fromexperiment 1 are not included, because only one site was used in the experiment. Lines underneath the barsindicate results of tests when data are combined across all experiments. Treatments on the same solid line arestatistically similar. Pneum5treatment with pneumatophores.

first in most trials above densities on all three low complexity treatments (Table 4,2Appendix 1, Q 5 14.09, df 5 3, P , 0.01).

Species richness was similar between control and complexity treatments in mostexperiments (Table 3, Fig. 5b). When the data were combined across the different timesand places they were statistically similar (Table 4, high . control in five of ten trials,binomial P . 0.1; low . control in four of 11 trials, binomial P . 0.1).

Of the three most abundant species, the densities of A. porcata and B. nanum were

42M

.W.

Beck

/J.

Exp.

Mar.

Biol.

Ecol.

249(2000)

29–49

Table 4Summary of results when data were combined across experiments: non-parametric comparisons of assumption and hypotheses 1–3

Condition Variable

Density: Richness Density

All species A. porcata B. nanum C. tramoserica

Assumption 1 Control5High, Control.Low Control5High, Low Control5High, Low Control5High, Low Control.High, LowHypothesis 1 High.Low High.Low High.Low High.Low High5LowHypothesis 2 WM.ND, WM5NS, NS5ND WM5NS5ND WM5NS5ND WM5NS5ND WM5NS5NDHypothesis 3 Pits.Pneumatophores Pits.Pneumatophores Pits5Pneumatophores Pits.Pneumatophores Pits.Pneumatophores


similar between complexity and control treatments, but the density of C. tramosericawas not (Table 4, Fig. 5c–e). The density of A. porcata was similar between the controland the complexity treatments (high . control in six of ten trials, binomial P . 0.1;low . control in three of 11 trials, binomial P . 0.1) as was the density of B. nanum(high . control in two of six trials after four all-zero trials dropped, binomial P . 0.1;low . control in four of eight trials after three all-zero trials dropped, binomial P . 0.1).The density of C. tramoserica was, however, significantly greater in the controltreatment than on the complexity treatments when the responses were combined acrossexperiments (high . control in zero of ten trials, P , 0.001; low . control in one of 11trials, P , 0.01). Inferences about the effects of treatments on C. tramoserica must beregarded with caution, because they may not accurately reflect the effects of habitatstructure on this species in unmanipulated habitats.

3.2. Hypothesis 1: high vs. low complexity treatments

Total density of gastropods was greater on high than on low complexity treatments atall times and places. When the data were examined for each experiment, the ANOVAsindicated significant differences between high and low complexity treatments only inexperiments 2 and 3 (Table 2, Fig. 5a). When the data were combined acrossexperiments, the density of gastropods on the high complexity treatment was ranked firstabove the densities on the three low complexity treatments in all 13 trials; a result that is

2highly unlikely to occur by chance (Table 4, Appendix 1, Q 5 39, df 5 3, P , 0.001).Species richness was greater on the high than low complexity treatments at most times

and places (Table 3, Fig. 5b). There was, however, only one statistically significantdifference among these treatments when the data were analyzed for each experiment(Table 3). When the responses to treatments were combined across experiments, richnesswas greater on the high complexity treatment than on the low complexity treatments inten of 13 trials (Table 4, binomial P , 0.05).

Of the three most abundant species, the densities of A. porcata and B. nanum wereaffected by complexity, but the density of C. tramoserica was not (Table 4, Fig. 5c–e).When the data were combined across the experiments, the densities of A. porcata and B.nanum were significantly greater on high than on low complexity treatments after fourall-zero trials were dropped (A. porcata — high . low in nine of nine trials, binomialP , 0.01; B. nanum — high . low in eight of nine trials, binomial P , 0.05). Thedensity of C. tramoserica was not affected by complexity (high . low in five of 13trials, P . 0.1).

3.3. Hypothesis 2: comparison of indices among low complexity treatments

There were few significant differences in density among the three low complexitytreatments (Tables 2, 4, Fig. 5a), and thus D was not significantly better than otherindices at predicting responses of gastropods to complexity. The ANOVAs revealed onlythree significant differences in total density among these treatments (Table 2b). Whenthe data were combined across experiments, the density of gastropods on the WMtreatment was greater than on the ND treatment in most trials (Appendix 1, binomial


P , 0.01). There were no significant differences in rankings of density among trialsbetween WM and NS treatments or NS and ND treatments (Table 4, binomial P . 0.05for both) (Appendix 1).

The ANOVAs revealed only two significant differences in richness among the threelow complexity treatments (Table 3, Fig. 5b). When the data were combined across theexperiments, there were no significant differences in richness among the three lowcomplexity treatments (Table 4).

There were also no significant differences in the density of any of the three mostabundant species on the low complexity treatments when the results were combinedacross experiments (Table 4, Fig. 5c–e).

3.4. Hypothesis 3: pits vs. pneumatophores

Total density was greater on treatments with pits than on treatments withpneumatophores at most times and places (Table 2, Fig. 5a). The ANOVAs for theindividual experiments indicated that there were significant differences in total densitybetween these treatments at most times and places (Table 2b). When the data werecombined across experiments, the density of gastropods was greater on treatments withpits in seven of eight trials (Table 4, Appendix 1, binomial P , 0.05).

Richness was greater on treatments with pits than on treatments with pneumatophoresat all times and places. The ANOVAs indicated that richness was significantly greater ontreatments with pits at most times and places (Table 3, Fig. 5b). When the responses totreatments were combined across experiments, richness was greater on treatments withpits in eight of eight trials (Table 4, binomial P , 0.01).

Of the three most abundant species, the densities of B. nanum and C. tramosericawere significantly greater on treatments with pits than on treatments withpneumatophores (Table 4, Fig. 5d,e; B. nanum — pits . pneumatophores in six of sixtrials after two all-zero trials dropped, binomial P , 0.05; C. tramoserica — pits .

pneumatophores in seven of eight trials, P , 0.05). The density of A. porcata was notsignificantly different between treatments with pits vs. pneumatophores (Table 4, Fig.5c; pits . pneumatophores in four of six trials after two all-zero trials dropped, P . 0.1).

4. Discussion

Complexity and structural components affected the richness, total density, and thedensities of two of the three most common gastropods in this mid-shore rocky intertidalhabitat. These experiments, which were repeated at many different times and places,indicated that the effects of habitat structure were generally consistent in space and time.The consistency in the effects of complexity in these experiments, and in the priorcorrelations between density and complexity on these shores (Beck, 1998) suggests thatcomplexity can have strong effects on the local distributions of some of thesegastropods. Specific structural components also affected the density and richness ofgastropods at most times and places, and these effects were independent of the effects ofcomplexity. Although they had similar complexities, plates with pits had a greater total


density and richness of gastropods than did plates with pneumatophores. This resultsuggests that the effects of habitat heterogeneity are likely to have important effects onsome of these gastropods, because the basis of an effect of heterogeneity is that theeffects of different structural components are independent of their complexity (McCoyand Bell, 1991).

The effects of complexity and structural components were species specific. Complexi-ty affected the densities of A. porcata and B. nanum, but not C. tramoserica. The lack ofa response by C. tramoserica could be an experimental artifact, because its density onplates was less than on controls. The lack of a response was, however, consistent withprior results that showed no correlation between the density of C. tramoserica andcomplexity on these shores (Beck, 1998). Specific structural components affected thedensities of B. nanum and C. tramoserica but not the density of A. porcata.

Few studies define, measure, or manipulate complexity and structural componentsindependently of one another, and their effects are often confounded. The complexitycreated by pits, pools, and crevices is thought to have strong effects on many rockyintertidal species including some of the gastropods examined in this study (e.g. Emsonand Faller-Fritsch, 1976; Underwood, 1976; Raffaelli and Hughes, 1978; Menge et al.,1983; Moran, 1985; Fairweather, 1988; Underwood and Chapman, 1992). Explanationsfor these effects of complexity often hinge on features that are specific to pits per se, e.g.they pool water and reduce desiccation stress. If just the complexity of pits had beenmanipulated in this study, it would appear that pits had strong effects on the density ofA. porcata. A. porcata, however, responded in similar abundances to pits andpneumatophores; it was complexity not pits per se that affected the density of thisspecies. Explanations for the behavior of A. porcata that rely on pooled water would beinappropriate. On the other hand, C. tramoserica did not respond to changes incomplexity, and it might have been suggested that habitat structure did not have strongeffects on this species. Specific structural components did, however, affect the density ofC. tramoserica; pits were important but complexity per se was not.

Each of the three main species responded differently to changes in complexity andstructural components. The ability to separate the effects of these elements will helpimprove studies that examine the mechanistic basis for the effects of habitat structure onspecies (e.g. Coull and Wells, 1983; Fletcher and Underwood, 1987; Miller andCarefoot, 1989; Beck, 1995, 1997; Persson and Eklov, 1995). In this study, B. nanumwas affected by complexity and specific structural components, pits. This species mayuse pits in particular to avoid the direct effects of desiccation on themselves (e.g.McGuinness and Underwood, 1986) or the indirect effects of desiccation on theirmicroalgal food. C. tramoserica was not strongly affected by changes in complexity, butit was affected by specific structural components when complexity was held constant.The complexity of pits may be less important to this limpet than to the snails, because itcan clamp tightly to flat surfaces to reduce the direct effects of desiccation stress orwave forces. C. tramoserica may, however, have avoided treatments withpneumatophores, because they hindered its movements or provided fewer places forattachment. Thus, the complexity of pits may have neutral or negligible effects on C.tramoserica, but pneumatophores may have a negative effect on its density. In the rockyintertidal habitat, A. porcata was affected by complexity but did not differentiate


between pits and pneumatophores; a similar result was observed in a comparable studyin nearby mangrove habitats (Beck, unpublished data). The same mechanism mayunderlie the response of A. porcata to complexity in both habitats. A. porcata mayrespond to any component that creates complexity to avoid predators in both habitats. Itis also possible that there may be separate explanations for the similarity in responses ofA. porcata between habitats, as Crowe (1996) has shown for another gastropod.

In addition to confounding of the elements of habitat structure, lack of realisticmanipulations or controls further constrain the development of our understanding of theeffects of habitat structure on species. In some experiments complexity has beenmanipulated with structural components that bear little or no relationship to naturalcomponents (e.g. Bourget et al., 1994; Jacobi and Langevin, 1996; Lemire and Bourget,1996). Artificial components often are assumed to resemble natural components, but thisassumption is rarely tested (e.g. Hart, 1978; Russ, 1980; Gilinsky, 1984; Bell et al.,1987; Gunnarsson, 1992; Caley and St. John, 1996).

D was not definitively better than the other indices of complexity examined in thisstudy. When differences in complexity were great enough to affect these gastropods (i.e.high vs. low complexity treatments), all indices predicted their responses. Whendifferences in complexity between treatments were small enough to separate among theindices (i.e. low complexity treatments), these differences were generally not greatenough to affect the gastropods. D did predict the result that the density of gastropods

2would be greater on WM than on ND treatments, but all indices except odh predictedthis response. The lack of a definitive result in comparisons of indices is common(Carleton and Sammarco, 1987; Underwood and Chapman, 1989; McCormick, 1994)

In the absence of a more definitive experimental result, D and SA are recommendedfor further examination. The prior observation that D was more highly correlated withthe density of gastropods than other indices still suggests that D could be better thanother indices at describing features of the habitat that affected these gastropods (Beck,1998) and that of other rocky shore gastropods (Kostylev et al. 1997). D is also useful,because it is calculated explicitly across different intervals of measurement, which forcesconsideration of the concordance in scale between measures of complexity and the sizeof the species in question (Gee and Warwick, 1994; Beck, 1998). The use of D does notimply that surfaces are ‘fractal’ or self-similar; it only implies that D can usefullydescribe some features of the habitat (e.g. Avnir et al., 1998). SA is intuitivelyappealing, because researchers are familiar with it. Caution is urged, because this appealdoes not necessarily make SA relevant to other species, and SA can be difficult tomeasure in the field.

It seems unlikely that any one index will be best in most habitats. Measures of severalindices should be provided in studies of habitat complexity. It will be counterproductiveto become mired in measures of complexity, but consideration of multiple indices willreiterate that the observation or manipulation of habitat structure involves manycorrelated features not just those that seem obvious to the observer. This informationwill also increase the possibility of comparisons of complexity between different studiesand habitats (Beck, 1998).

Lack of clear definitions and measures has hindered the development of ourunderstanding of the effects of habitat structure on community structure. The use ofgeneral indices of complexity made it possible to differentiate between the effects of the


complexity and structural components of habitats. Complexity and structural componentshave separate effects on assemblages, and it confuses the study of habitat structure tocombine them.

Acknowledgements

This work was supported by a Fulbright Fellowship and an ARC PostdoctoralResearch Fellowship. I thank T. Glasby, M. Haddon, B. Kelaher, I. Montgomery forfield assistance and G. Housefield for technical advice on the manufacture of fractalsurfboards. This paper benefited from discussion with and reviews by L. Airoldi, M. J.Anderson, B. Gillanders, and A.J. Underwood. [RW]

2Appendix 1. Average density of gastropods (no. /0.12 m ) on treatments (n 5 2/cell).

Not all treatments were used at each time and place. Bare Is. 5 Bare Island, Suth.Pt. 5 Sutherland Point.

Experiment Shore Site Treatment

Control High WM NS ND Pneum.

1 Bare Is. 1 13.8 – 47.0 18.0 30.5 –2 Bare Is. 1 10.3 10.0 3.5 0.5 0.5 –

Bare Is. 2 37.8 22.5 8.5 6.0 3.5 –Suth. Pt. 1 37.0 9.5 5.5 4.5 3.0 –Suth. Pt. 2 23.8 8.0 3.5 3.5 1.5 –

3 Bare Is. 1 – 7.0 6.0 1.0 4.5 0.0Bare Is. 2 – 21.5 11.0 6.5 9.0 3.0Suth. Pt. 1 – 15.5 12.5 3.5 5.0 0.0

4 Bare Is. 1 9.5 12.0 6.0 5.5 8.5 4.5aBare Is. 2 18.5 24.0 23.5 9.0 13.5 5.0

Suth. Pt. 1 72.5 26.5 12.5 13.0 8.0 1.0a5 Bare Is. 1 9.0 23.0 6.5 4.0 6.5 7.0

a bBare Is. 2 27.0 35.0 10.0 33.0 9.0 –Suth. Pt. 1 78.0 30.5 21.5 28.0 21.0 2.0

a One missing replicate.b Two missing replicates.

References

Anderson, R.L., 1959. Use of contingency tables in the analysis of consumer preference studies. Biometrics 15,582–590.

Avnir, D., Biham, O., Lidar, D., Malcai, O., 1998. Is the geometry of nature fractal. Science 279, 39–40.Beck, M.W., 1995. Size-specific shelter limitation in stone crabs: a test of the demographic bottleneck

hypothesis. Ecology 76, 968–980.Beck, M.W., 1997. A test of the generality of the effects of shelter bottlenecks in four stone crab populations.

Ecology 78, 2487–2503.


Beck, M.W., 1998. Comparison of the measurement and effects of habitat structure on gastropods in rockyintertidal and mangrove habitats. Mar. Ecol. Prog. Ser. 169, 165–178.

Bell, J.D., Westoby, M., Steffe, A.S., 1987. Fish larvae settling in seagrass: do they discriminate between bedsof different leaf density. J. Exp. Mar. Biol. Ecol. 111, 133–144.

Bourget, E., Deguise, J., Daigle, G., 1994. Scales of substratum heterogeneity, structural complexity, and theearly establishment of a marine epibenthic community. J. Exp. Mar. Biol. Ecol. 181, 31–51.

Caley, M.J., St. John, J., 1996. Refuge availability structures assemblages of tropical reef fishes. J. Anim. Ecol.65, 414–428.

Carleton, J.H., Sammarco, P.W., 1987. Effects of substratum irregularity on success of coral settlement:quantification by comparative geomorphological techniques. Bull. Mar. Sci. 40, 85–98.

Connell, J.H., 1961. Effects of competition, predation by Thais lapillus, and other factors on the distribution ofthe barnacle Balanus balanoides. Ecol. Monogr. 31, 61–104.

Connor, E.F., McCoy, E.D., 1979. The statistics and biology of the species–area relationship. Am. Nat. 113,791–833.

Coull, B.C., Wells, J.B.J., 1983. Refuges from fish predation: experiments with phytal meiofauna from the NewZealand rocky intertidal. Ecology 64, 1599–1609.

Crowe, T., 1996. Different effects of microhabitat fragmentation on patterns of dispersal of an intertidalgastropod in two habitats. J. Exp. Mar. Biol. Ecol. 206, 83–107.

Dean, R.L., Connell, J.H., 1987. Marine invertebrates in an algal succession. III. mechanisms linking habitatcomplexity with diversity. J. Exp. Mar. Biol. Ecol. 109, 249–273.

Diehl, S., 1993. Effects of habitat structure on resource availability, diet and growth of benthivorous perch,Perca fluviatilis. Oikos 67, 403–414.

Dittel, A., Epifanio, C.E., Natunewicz, C., 1996. Predation on mud crab megalopae, Panopeus herbstii H.Milne Edwards: effect of habitat complexity, predator species and postlarval densities. J. Exp. Mar. Biol.Ecol. 198, 191–202.

Downes, B.J., Lake, P.S., Schreiber, E.S.G., Glaister, A., 1998. Habitat structure and regulation of local speciesdiversity in a stony, upland stream. Ecol. Monogr. 68, 237–258.

Emson, R.H., Faller-Fritsch, R.J., 1976. An experimental investigation into the effect of crevice availability onabundance and size-structure in a population of Littorina rudis (Maton): Gastropoda: Prosobranchia. J. Exp.Mar. Biol. Ecol. 23, 285–297.

Fairweather, P.G., 1988. Movements of intertidal whelks (Morula marginalba and Thais orbita) in relation toavailability of prey and shelter. Mar. Biol. 100, 63–68.

Fletcher, W.J., Underwood, A.J., 1987. Interspecific competition among subtidal limpets: effects of substratumheterogeneity. Ecology 68, 387–400.

Gause, G.F., 1934. The Struggle for Existence, Hafner, New York.Gee, J.M., Warwick, R.M., 1994. Metazoan community structure in relation to the fractal dimensions of marine

macroalgae. Mar. Ecol. Prog. Ser. 103, 141–150.Gilinsky, E., 1984. The role of fish predation and habitat heterogeneity in determining benthic community

structure. Ecology 65, 455–468.Grinnell, J., 1917. The niche relationships of the California Thrasher. The Auk 34, 427–433.Gunnarsson, B., 1992. Fractal dimension of plants and body size distribution in spiders. Func. Ecol. 6,

636–641.Hart, D.D., 1978. Diversity in stream insects: regulation by rock size and microspatial complexity. Proc. Int.

Assoc. Theor. Appl. Limnol. (Verhandlungen) 20, 1376–1381.Huffaker, C.B., 1958. Experimental studies on predation: dispersion factors and predator–prey oscillations.

Hilgardia 27, 343–383.Inglis, G.J., 1994. Contrasting effects of habitat structure on the recruitment and mortality of an epibiotic

macroalga. Oecologia (Berl.) 99, 352–365.Jacobi, C.M., Langevin, R., 1996. Habitat geometry of benthic substrata: effects on arrival and settlement of

mobile epifauna. J. Exp. Mar. Biol. Ecol. 206, 39–54.Jenkins, G.P., Sutherland, C.R., 1997. The influence of habitat structure on nearshore fish assemblages in a

southern Australian embayment: colonisation and turnover rate of fishes associated with artificialmacrophyte beds of varying physical structure. J. Exp. Mar. Biol. Ecol. 218, 103–125.

Kohn, A.J., Leviten, P.J., 1976. Effect of habitat complexity on population density and species richness intropical intertidal predatory gastropod assemblages. Oecologia (Berl.) 25, 199–210.


Kostylev, V., Erlandsson, J., Johannesson, K., 1997. Microdistribution of the polymorphic snail Littorinasaxatalis (Olivi) in a patchy rocky shore habitat. Ophelia 47, 1–12.

Leber, K.M., 1985. The influence of predatory decapods, refuge, and microhabitat selection on seagrasscommunities. Ecology 66, 1951–1964.

Lemire, M., Bourget, E., 1996. Substratum heterogeneity and complexity influence micro-habitat selection ofBalanus sp. and Tubularia crocea larvae. Mar. Ecol. Prog. Ser. 135, 77–87.

McCormick, M.I., 1994. Comparison of field methods for measuring surface topography and their associationswith a tropical reef fish assemblage. Mar. Ecol. Prog. Ser. 112, 87–96.

McCoy, E.D., Bell, S.S., 1991. Habitat structure: the evolution and diversification of a complex topic. In: Bell,S.S., McCoy, E.D., Mushinsky, H.R. (Eds.), Habitat Structure: The Physical Arrangement of Objects inSpace, Chapman and Hall, New York, pp. 3–27.

McGuinness, K.A., 1984. Communities of organisms on intertidal boulders: the effects of disturbance andother factors. PhD thesis, University of Sydney, Australia.

McGuinness, K.A., Underwood, A.J., 1986. Habitat structure and the nature of communities on intertidalboulders. J. Exp. Mar. Biol. Ecol. 104, 97–123.

Menge, B.A., Ashkenas, L.R., Matson, A., 1983. Use of artificial holes in studying community development incryptic marine habitats in a tropical rocky intertidal region. Mar. Biol. 77, 129–142.

Menge, B.A., Lubchenco, J., 1981. Community organization in temperate and tropical rocky intertidal habitats:prey refuges in relation to consumer pressure gradients. Ecol. Monogr. 51, 429–450.

Miller, K.M., Carefoot, T.H., 1989. The role of spatial and size refuges in the interaction between juvenilebarnacles and grazing limpets. J. Exp. Mar. Biol. Ecol. 137, 157–174.

Moran, M.J., 1985. The timing and significance of sheltering and foraging behaviour of the predatory intertidalgastropod Morula marginalba Blainville (Muricidae). J. Exp. Mar. Biol. Ecol. 93, 103–114.

Pennings, S.C., 1990. Predator–prey interactions in opisthobranch gastropods: effects of prey body size andhabitat complexity. Mar. Ecol. Prog. Ser. 62, 95–101.

Persson, L., Eklov, P., 1995. Prey refuges affecting interactions between piscivorous perch and juvenile perchand roach. Ecology 76, 70–81.

Primavera, J.H., 1997. Fish predation on mangrove associated penaeids: the role of structures and substrate. J.Exp. Mar. Biol. Ecol. 215, 205–216.

Raffaelli, D.G., Hughes, R.N., 1978. The effects of crevice size and availability on populations of Littorinarudis and Littorina neritoides. J. Anim. Ecol. 47, 71–83.

Russ, G.R., 1980. Effects of predation by fishes, competition, and structural complexity of the substratum onthe establishment of a marine epifaunal community. J. Exp. Mar. Biol. Ecol. 42, 55–69.

Stoner, A.W., Lewis, III F.G., 1985. The influence of quantitative and qualitative aspects of habitat complexityin tropical sea-grass meadows. J. Exp. Mar. Biol. Ecol. 94, 19–40.

Underwood, A.J., 1975. Intertidal zonation of prosobranch gastropods: analysis of densities of four coexistingspecies. J. Exp. Mar. Biol. Ecol. 19, 197–216.

Underwood, A.J., 1976. Analysis of patterns of dispersion of intertidal prosobranch gastropods in relation tomacroalgae and rockpools. Oecologia (Berl.) 25, 145–154.

Underwood, A.J., 1997. Experiments in Ecology: Their Logical Design and Interpretation Using Analysis ofVariance, Cambridge University Press, Cambridge.

Underwood, A.J., Chapman, M.G., 1989. Experimental analyses of the influences of topography of thesubstratum on movements and density of an intertidal snail, Littorina unifasciata. J. Exp. Mar. Biol. Ecol.134, 175–196.

Underwood, A.J., Chapman, M.G., 1992. Experiments on the topographic influences on density and dispersionof Littorina unifasciata in New South Wales. In: Mill, P.J., Reid, D.G. (Eds.), Proc. 3rd Int. Symp. onLittorinid Biology, The Malacological Society of London, London, pp. 181–195.

Underwood, A.J., Chapman, M.G., 1996. Scales of spatial patterns of distribution of intertidal invertebrates.Oecologia (Berl.) 107, 212–224.

Walters, L.J., Wethey, D.S., 1996. Settlement and early post-settlement survival of sessile marine invertebrateson topographically complex surfaces: the importance of refuge dimensions and adult morphology. Mar.Ecol. Prog. Ser. 137, 161–171.

Werner, E.E., Gilliam, J.F., Hall, D.J., Mittelbach, G.G., 1983. An experimental test of the effects of predationrisk on habitat use in fish. Ecology 64, 1540–1548.

Winer, B.J., Brown, D.R., Michels, K.M., 1991. Statistical Principles in Experimental Design, 3rd Edition,McGraw-Hill, New York.

separating the elements of habitat structure: independent effects of habitat complexity and...

Documents