ENVIRONMENTAL ASSESSMENT Assessing Anchor Damage on ...

ENVIRONMENTAL ASSESSMENTAssessing Anchor Damage on Coral Reefs: A CaseStudy in Selection of Environmental IndicatorsELIZABETH A. DINSDALE*School of Tropical Environment Studies and GeographyJames Cook UniversityTownsville, Queensland 4811, Australia

VICKI J. HARRIOTTSchool of Environmental Science and ManagementSouthern Cross UniversityLismore, New South Wales 2480, Australia

ABSTRACT / Because environmental conservation can re-move scarce natural resources from competing uses, it is im-portant to gain support for conservation programs by demon-strating that management actions have been effective inachieving their goals. One way to do this is to show that se-lected significant environmental variables (indicators) vary be-tween managed and unmanaged areas or change over timefollowing implementation of a management regime. However,identifying indicators that reflect environmental conditions rele-vant to management practices has proven difficult. This paper

focuses on developing a framework for choosing indicators ina coral reef habitat. The framework consisted of three phases:(1) information gathering to identify candidate variables; (2)field-testing candidate variables at sites that differ in intensityof human activity, thus identifying potential indicators; and (3)evaluating potential indicators against a set of feasibility criteriato identify the most useful indicators. To identify indicatorssuitable to measure the success of a management strategy toreduce anchor damage to a coral reef, 24 candidate variableswere identified and evaluated at sites with different intensitiesof anchoring. In this study, measures that reflected injuries tocoral colonies were generally more efficient than traditionalmeasures of coral cover in describing the effects of anchoring.The number of overturned colonies was identified as the singlemost useful indicator of coral reef condition associated withanchoring intensities. The indicator selection framework devel-oped here has the advantages of being transparent, cost effi-cient, and readily transferable to other types of human activi-ties and management strategies.

Historically, selection and management of protectedareas has been based on a belief that managementstrategies benefit the target species or habitats. At timeswhen resource management for conservation is expen-sive, and there are many competing demands for lim-ited resources, there has been increased emphasis ondemonstrating the success of management strategies.Implementation of protected areas can impinge uponpeople’s recreation and livelihood. To maintain sup-port for the management strategies, it is imperative thatindicators measure and communicate positive achieve-ments for conservation. In addition, evaluating the suc-cess or otherwise of environmental management im-proves conservation of natural resources by allowing foradaptive planning, raising awareness of success andidentifying areas of concern (Dudley and others1999a).

Evaluations describe and record changes, if any, tothe resource that are linked to management strategies.Until very recently, evaluations of protected areas havereviewed the implementation and management pro-cesses but have not measured their effectiveness inmaintaining or improving environmental condition(Alder 1996, Attwood and others 1997). A lack of avail-able data and tools to measure the environment hasrestricted these evaluations to superficial levels (Dudleyand others 1999b, Hershman and others 1999, Hock-ings 2000). This also applies to protected areas on coralreefs, which lack long-term data that would allow anal-ysis of the effectiveness of management strategies (Mc-Clanahan 1999, Wells 1999).

Natural environments include a vast array of biota,biological interactions, and physical processes, thuseliminating the possibility of measuring every variableduring an evaluation. A few essential variables or indi-cators that describe environmental condition need tobe identified. An indicator is used to measure whetherthe objectives of a management strategy have beenachieved, i.e., has the indicator “improved” in the areawhere management has been implemented. Since man-agement plans have ecological, social, and economicobjectives, indicators could communicate trends in any

KEY WORDS: Indicator selection; Coral reefs; Anchor damage; Man-agement evaluation

Published online November 21, 2003.

*Author to whom correspondence should be addressed, email:[email protected]

DOI: 10.1007/s00267-003-3056-9

Environmental Management Vol. 33, No. 1, pp. 126–139 © 2004 Springer-Verlag New York Inc.

of these processes. Thus, an indicator is defined as anyvariable that measures changes to the environmentassociated with human activities, both protective andextractive, and can provide information about the sys-tem beyond its face value (see Waltner-Toews 1996,Griffith 1998, Margoluis and Salafsky 1998).

While lists of potential indicators have been pro-duced (Dahl 2000), identifying the right indicator foreach management strategy has proved difficult. Manyindicators measure broad-scale processes that are notappropriate for specific human activities or manage-ment strategies. For example, measuring the extent ofcoral reefs as an indicator of the amount of coral reefhabitat within Australian waters (Ward 2000) wouldonly detect change after catastrophic events and wouldnot be suitable to identify the localized effects of an-choring.

Often, multiple variables are suggested to measureenvironmental change without considering correla-tions between variables, leading to redundancy in datacollection. Yu and others (1998) compared the amountof information gained from 14 different indicators.Four indicators explained 62% of the variation, suggest-ing that little information is gained for the additionaleffort of measuring the extra 10 variables. Allegationsof bias are another problem encountered during theselection of indicators. To avoid these allegations, indi-cators need to be selected against a set of criteria that isopen to public scrutiny (Crabtree and Bayfield 1998,Dale and Beyeler 2001).

Some indicators have proven to be of little value inseparating areas affected by human activities from areasunaffected. Four variables, including measurements ofepiphytes, nominated as important to perceptions ofseagrass health were found not to differ significantlybetween healthy and perceived degraded sites (Woodand Lavery 2000). In this study, a combination of thevariables shoot density, canopy cover, shoot height,aboveground biomass, productivity, and leaf area indexwere found to be useful indicators of seagrass health(Wood and Lavery 2000). Similar problems have beenencountered during the selection of indicators to assessthe affects of human activities on coral reefs. Edingerand others (2000) expected a reduction in coral growthrates with increased input of nutrients and sedimentsrelated to human activity. However, corals were able tocompensate for lack of light in more turbid conditionswith increased consumption of particulate matter, socoral growth rates were maintained. Therefore, coralgrowth was not a useful indicator of water qualitychanges.

Variables that measure social changes may provideuseful indicators of environmental condition. Pollnac

and others (2000) expected a correlation between highhuman pressure and poor coral reef condition. How-ever, the opposite occurred because people movedfrom low condition reefs and set up business on reefsthat could provide a high-quality product. Therefore, aone-time measure of the number of people using aresource is not a useful indicator of its condition.

To address the problems in indicator selection, aframework is required to filter through the large num-ber of potential variables and identify those that areappropriate to measure changes to the environmentassociated with a specific human activity. Several frame-works have been suggested to increase transparencyand effectiveness of indicator selection. Belnap (1998)measured all possible variables at terrestrial sites tra-versed by people on their way to view scenic areas. Twosites were selected to measure of a range of variables,one site with high use and one site with low use. How-ever, the lack of replication means that indicators iden-tified may have differed because of natural variability,rather than as a result of human activities. Lorenz(1999) used available data as a starting point for indi-cator selection, but noted that such a restriction onvariables may cause other simple and cost effectivevariables to be overlooked. Similarly, Hockings (1998)identified indicators for evaluating management pro-grams by reviewing the literature and holding discus-sions with managers and experts, but did not conductfield tests. The lack of field-testing will make it difficultto identify the best indicators.

To select indicators, the framework developed hereintegrates three major phases: (1) information gather-ing, (2) on-site measurement, and (3) evaluation ofvariable usefulness (Figure 1). The information-gather-ing phase identifies how the human activity affects theresource, the objectives of management strategies, andthe availability of baseline data. Information from man-agers and stakeholders identifies their vision for thestate of the resource after implementation of the man-agement strategy. Information gathering generates alist of candidate variables that may make useful indica-tors. Candidate variables are then measured at sites thatvary in the intensity of the human activity of interest.Replicate sites selected must be similar in ambient con-dition, thereby reducing confounding results. The po-tential indicators are evaluated against a set of selectioncriteria for their usefulness to managers and clientgroups.

Here, we focus on identification of indicators ofcoral condition that are useful in assessing the benefitsof a program to protect coral reefs from anchor dam-age in the Whitsunday Islands, Great Barrier Reef

Indicators of Anchor Damage 127

(GBR). The framework developed could be modifiedfor application in other management contexts.

Management of Anchoring—A Case Study fromthe Great Barrier Reef

Many factors, both natural and anthropogenic,cause damage to coral reefs. Natural changes includestorm damage, extreme temperature events, predation,competition, and disease. Anthropogenic factors in-clude pollution, sedimentation, fishing, mining, tram-pling, anchoring, and diver damage (review by Brown

and Howard 1985). The intensity of natural distur-bance in marine environments often masks the effectsof anthropogenic disturbance (Keough and Quinn1998, Short and Wyllie-Echeverria 1996), so selectingindicators that detect changes associated with anthro-pogenic activities is difficult.

Anchors cause damage to coral reefs during setting,retrieval, and while at anchor. Corals are broken, frag-mented, or overturned as the anchor drops to thesubstratum. Once set, further damage occurs by thechain dragging across the substratum or wrappingaround reef structures. If the anchor lodges under a

Figure 1. Framework for selecting indica-tors. Steps for choosing indicators are de-scribed in plain text and examples particularto the case study are in italics.

128 E. A. Dinsdale and V. J. Harriott

coral colony, overturning occurs during the retrievalprocess, particularly if an electronic winch is used.Coral reefs that experience high intensities of boatingactivities have higher levels of broken corals. Highernumbers of fragmented coral were found at CarysfortReef (Florida Keys), which has high intensities of boat-ing, compared to nearby reefs with less boating activity(Dustan and Halas 1987). These fragmented areas alsocontained large amounts of broken propellers, lines,personal effects, and other debris, further reducing thecondition of the coral reef. Four high-use coral reefs inthe Egyptian Red Sea had higher levels of broken coraland rubble compared with rates of natural damagerecorded in the literature (Jameson and others 1999).

Anchor damage has been identified as a manage-ment problem on the GBR at sites that receive highlevels of boating activity. The Whitsunday Islands is oneregion with high levels of both recreational small boatusage and commercial charter of yachts. The manage-ment response of the Great Barrier Reef ManagementPark Authority (GBRMPA) and community organiza-tions to perceived anchor damage has been to developa reef protection program. The ecological objectives ofthe program are “to protect fringing reefs in popularbays and anchorages from the cumulative impacts ofanchoring so that natural coral communities in thearea are maintained” (GBRMPA 1999). The programhas two components; reef markers and moorings. Reefprotection markers are placed on the surface at strate-gic positions around the bay, denoting where the coverof coral reef organisms is low and the substratum iscomprised of sand. Boat operators line up two markersand drop their anchor seaward (only) of the imaginaryline between markers. Therefore, reefs shoreward ofthe markers are placed within a no-anchor area. Moor-ings were installed to ensure that people have access tothe bays while protecting the corals.

The process described below is intended to identifya suite of indicators that can be effectively used tomeasure the success of the reef protection program.The selected indicators can be used to determine thesuccess of the management strategy to reduce anchordamage. After a period of time, if the strategy is suc-cessful, the indicators in the managed area should re-semble that for an area with low human use. As a firststep to identify outcome indicators, appropriate vari-ables must be identified, field tested, and then evalu-ated using a set of transparent selection criteria. Theselected indicators will be used to examine the effec-tiveness of the specific management program in thefuture.

The indicators identified in this study focus on theoutputs of management, answering the question “has

the management strategy improved the condition ofthe resource?” Other indicators that measure the suc-cess of management inputs and processes are also re-quired during the evaluation (Hockings 1998). How-ever, these indicators are outside the scope of thispaper.

Methods

Phase 1: Generation of Candidate Variables

A review of the literature and discussion with coralreef managers allowed the identification of a suite ofcandidate variables, which might be used to measurechanges to coral reef condition associated with anchor-ing. These candidate variables were broadly groupedinto five categories: injury types; coral cover; substra-tum cover; community processes; and coral colony size.

Injury types. Corals injured by anchors show manydifferent symptoms of damage, including abrasion ofsurface tissue and skeletons, death to portions of thecoral colony, fragmentation, and removal of the coralcolony from the substratum. Furthermore, coral colo-nies with different morphologies are likely to vary ininjury type. Branching species are more likely to frag-ment, while massive species suffer injuries to surfacetissue and skeleton (Marshall 2000). Measuring injurytypes seems likely to provide useful indicators of anchordamage on a coral reef.

Coral cover. Percent cover of benthic biota is themost widely measured and reported variable for moni-toring reefs across the world (Sweatman and others2000, Wilkinson 2000). A direct symptom of coral reefdegradation is decline in coral cover and increase inalgae (Hughes 1994). However, coral cover has highnatural variability, so detecting changes in coral covermay only be possible after catastrophic events. Hawkinsand Roberts (1993) compared relative abundance ofcorals on trampled and non-trampled reefs and foundno significant differences. No difference to coral coverwas found on coral reefs in Kenya that experienceddifferent numbers of visitors (Muthiga and McClana-han 1997). While total coral cover may not vary, thecover of some species or family groups may show adifference with anchoring intensity and requiresinvestigation.

Substratum cover. In extreme conditions, on a physi-cally damaged reef, coral reef biota may be replaced bydead rubble or limestone pavement from which coralshave been removed. A change from high cover of reefbiota to other substratum type would indicate severereef damage. Furthermore, if physical damage contin-ues, the pavement or reef structure itself could alsodecline to produce more rubble or sand.


Community processes. Coral species differ in their abil-ity to resist damage (Marshall 2000), so it is possiblethat susceptible species may be lost from coral reefswith high intensities of boating activity. In a study of theeffects of trampling on reef flat coral colonies, massivespecies were more resistant to physical damage thanrobust branching species, whilst delicate branching spe-cies were highly susceptible (Liddle and Kay 1987).Therefore, changes in numbers or types of coral speciespresent may be a useful indicator. Changes in speciescomposition could lead to a change in diversity andevenness measures and these should be consideredduring indicator selection.

Colony size. The sizes of coral colonies may changewith different frequency of disturbance (Bak andMeesters 1998). Reduction in coral size has occurredwith intensive boating activity, because corals are bro-ken into smaller pieces and energy is spent on repairrather than growth (Hawkins and Roberts 1993). Tomaintain coral communities and provide quality visitorexperiences, large corals need to be conserved (Done1995). Therefore, measuring the size structure of thecoral community is potentially a useful indicator ofchanges associated with boating activity.

Phase 2: Field Testing of Candidate Variables

The Whitsunday Islands (149°E, 20°S, Figure 2) con-sist of approximately 74 individual islands surroundedby fringing coral reefs. The Whitsunday Islands were

chosen for the study for two reasons: first there aremultiple reefs that are exposed to similar environmen-tal conditions, such as current and wave regimes. Sec-ond, the islands are ideal for boating activities becausethey are close to the mainland, protected from prevail-ing winds, and provide spectacular scenery. Therefore,the region contains replicate coral reefs that were ex-posed to similar environmental conditions and influ-enced by different anchoring intensities. All sites usedin the study were westerly facing and protected fromprevailing southeasterly winds by prominent headlands.The sites have high turbidity associated with a large (3to 4 m) tidal range and are located approximately 24km from the mainland (Figure 2).

Field testing of candidate variables took place inJuly/August 2000 and February 2001. Coral reefs thatreceived high and low intensities of boating activitywere selected using information from GBRMPA, thecommunity, and on-site observations. Anchoring levelsranged from 0.09 to 4.0 boats anchoring per day, theequivalent of 36–1460 anchors per year. These coralreefs were typically visited by yachts, which use largesand-wedge anchors and electronic winches. The boatswere either skippered by commercial operators orhired by people with minimal experience. There ap-peared to be no preference for any coral reef site bydifferent types of operators. The six sites were dividedinto two treatments, high anchoring activity (site withgreater than two anchors dropped per day) and low

Figure 2. Location of survey siteson the Whitsunday Islands. Lowlevels of anchoring occurred atsites 1–3 and high levels of an-choring occurred at sites 4–6.Note the similarities of the sixsites with respect to distance fromshore and protection from south-east prevailing winds. Coral reefsare the darkly stippled areas andmangrove areas are lightly stip-pled.


anchoring activity (sites with less than one boat anchor-ing per day). At each site, two depths were surveyed: thelower slope (8–11 m) and the crest (1–3 m). Largeanchors and chain primarily affect the lower slope. Thecrest is affected by small reef anchors and associatedchain or rope. At each of the six sites, 24 candidatevariables in the five broad categories identified in phase1 were measured as described below.

Injury types. Six types of coral injuries were sampledwithin ten, 10 � 1 m belt transects laid haphazardly atthe two depths within the six sites. Within each transect,the number of fragments, overturned colonies, breaks,gouges, pieces of rubble, and diseased individuals werecounted. Fragments are defined as the “live portions ofa coral colony that has become physically separated,due to breakages of the skeleton, from the rest of thecolony” (Highsmith 1982) and remained unattachedfrom the substratum. Overturned colonies are coralsthat are dislodged from the substratum and move whentouched. They may have living tissue or be dead.

Breaks are the loss of individual branches on branch-ing colonies and appear as bright white round circleson the tops of branches. More than one break percolony is possible. Pieces of rubble are dead fragmentslarger than 10 cm lying on the substratum. Gouges arelarge pieces of missing tissue and skeleton, often withcrushing damage evident (Hawkins and Roberts 1993).Colonies displaying signs of white band or spot disease(Green and Bruckner 2000), black band disease (Dins-dale 2002) or pink spot (Aeby 2000) were counted.

Coral cover. While the belt transect was in position,the percent cover of benthic components was measuredusing the line intercept method (English and others1997). The percent cover of benthos was measured inseven categories. The percent cover of four scleractin-ian coral families, including Acroporidae, Poritidae,Faviidae, Pocilloporidae, and a combined group con-sisting of rarer hard corals (Mussidae, Dendrophylli-idae, Fungiidae, Caryophylliidae) was recorded. Thepercent cover of two more benthic categories weremeasured: (1) the hydrozoan Millepora species and (2)soft corals. Included in the soft coral category weresponges and zooanthids; however, these componentscontributed minimal cover.

Substratum cover. Substratum cover was measured us-ing the line intercept method. The substratum type wasclassified as reef biota, rubble, sand, or pavement. Rub-ble was selected as an important category because it isgenerated by damage to corals. Sand, for the purposeof this study, included all fine-grain material. Consoli-dated reef matrix with no benthic organisms (exceptturfing algae) was classified as pavement.

Community processes. Shannon-Weaver index andevenness measures were used to calculate diversity us-ing the percent cover data divided into 16 categoriesadapted from Hughes and others (2000). Two diveswere devoted to identifying the scleractinian coral spe-cies present at each site. Field classification is inher-ently difficult; therefore, underestimation of the num-bers of species in the genus Montipora and familyPoritidae was possible.

Coral size. To compare the average size of coral col-onies, the largest diameter (length) and perpendiculardiameter (width) of 50 colonies in four taxa were mea-sured to the nearest centimeter. Surface area was cal-culated for each colony using the formula for an el-lipse. Species measured were Acropora loripes, Seriatoporahystrix, Pocillopora damicornis, and branching Milleporasp. These species were selected because they were rel-atively abundant and had different growth morpholo-gies, which suggests they may vary in susceptibility tophysical damage. Colonies were measured between 6and 10 m depth and the first 50 colonies of each speciesencountered during a 50 meter swim were measured.

Statistical analysis. Variables were tested using multi-ple analyses of variances (MANOVA) with anchoringtreatment (high and low) and sites as fixed factors. Siteswere nested within the anchoring treatments. A usefulvariable shows a significant difference with the intensityof the human activity. Therefore significance levelswere evaluated for differences between the anchoringtreatments. Interaction terms were not tested becausethe sampling design is not fully orthogonal. The met-rics for candidate variables differed, for example, inju-ries were count data and coral cover was measuredusing percentages. Therefore, MANOVAs were con-ducted to test the significance of candidate variablesgrouped into their broad categories, as describedabove. Depth as a factor was tested separately for tworeasons. First, corals show distinct zonation patternswith depth and, second, the intensity of human activityvaried with depth. To achieve normality of the data,numeric data were log (x � 1) transformed and per-centage data were arc-sin square root transformed (Un-derwood 1997).

Canonical discriminant analyses (CDA) illustratesmultivariate data in a reduced set of dimensions. CDAwere used to determine the contributions of different“injury” and “cover” variables in describing the differ-ences in coral condition associated with anchoring. Thefirst CDA used the five injury types and the second CDAused percent cover of the eight coral cover groups. Therelationship between sites or groups of sites was dis-played on canonical axes (Tabachnick and Fidell


2001). The influence of each candidate variable wasidentified using a bivariate combination plot (biplots).

Consecutive CDAs were used to test the ability ofdifferent combinations of candidate variables to de-scribe the changes in coral condition associated withdifferent levels of anchoring. Four CDAs were per-formed, including combinations of: (1) injury typesand coral cover; (2) overturned corals and coral cover;(3) overturned corals, cover of Acroporidae, and softcorals, and (4) a stepwise CDA. Cross validation wasused to identify the combination of candidate variablesthat correctly classified the most replicates (Tabachnickand Fidell 2001). The data are divided into 12 catego-ries (six sites by two depths) of equal numbers; there-fore, the number of replicates correctly classified bychance alone is 8.3%. The percent correct classificationhas to be substantially larger than 8.3% for the classi-fying variables to be useful.

Phase 3: Evaluation Against Selection Criteria

The candidate variables that showed a differencebetween coral reef sites with different intensity of an-choring (potential indicators) were evaluated against aset of feasibility criteria. Potential indicators wereranked 1–5, depending on the number of criteria theymet, where 1 was the most useful and 5 was the leastuseful. The criteria were selected to identify indicatorsthat measure the condition of the coral reef and aresimple enough to be effectively and efficiently moni-tored and modeled. Building upon discussions by Bel-nap (1998), Dale and Beyeler (2001), Lorenz (1999),and local stakeholders, each potential indicator wasevaluated against the following feasibility criteria:

1 Reliable repeatable measure: indicators that arereadily identifiable regardless of ambient condi-tions.

2 Relevance: indicators that can be linked causallywith a specific human activity.

3 Respond to management: indicators that have apredictable and rapid response to the implementa-tion of management strategies would be most use-ful.

4 Ease of measurement and lack of ambiguity: allow-ing for measurements to be conducted by volun-teers.

5 Robustness: indicators should be unambiguous andlow in variation.

6 Persistence of variables in the environment: vari-ables that persist in the environment allow for im-pacts to be detected for a longer time period fol-lowing the impact.

7 Time required for data collection: indicators thatare quick to measure reduce the cost of data collec-tion.

8 Availability of baseline data: knowing the history ofa coral reef gives better understanding of its presentcondition.

9 Nondestructive measuring techniques: ensuresmonitoring does not cause further damage to pro-tected sites.

Results

Multivariate analysis revealed that significant differ-ences in coral condition occurred between coral reefsassociated with different anchoring intensities. Of thefive types of injuries to corals, overturned colonies andgouges were significantly higher at the three sites withhigh intensities of anchoring for both depths (Figure3A and B). The amount of rubble was higher withgreater anchoring intensities on the crest, but not onthe lower slope (Figure 3C). Fragments generatedshowed no trend with anchoring intensity (Figure 3D).Breaks differed between anchoring intensities on thecrest; however, there were higher numbers of breaks onthe low anchoring intensity sites compared with highanchoring intensity sites, so this was not a useful indi-cator. Breaks did not vary on the lower slope (Figure3E). Disease incidence was extremely low and did notvary with anchoring intensity.

Percent cover of some coral family groups variedwith anchoring intensity. Soft coral cover was lower onboth the crest and lower slope of intensely anchoredsites (Figure 4A). Acroporidae and Milleporidae hadlower cover on the crest of intensely anchored sites, butnot the lower slope (Figure 4B and C). The cover ofPocilloporidae was not significantly different on thecrest, but was lower at the deeper site with intensiveanchoring (Figure 4D). Cover of Faviidae and Poritidaewere higher on the crest of the intensely anchored sites,but did not vary on the lower slope (Figure 4E and F).The cover of rarer hard corals did not vary with anchor-ing intensity (Figure 4G). Variation between coral fam-ily groups suggests that anchoring affect some familygroups more than others.

At both depths, cover of total reef biota was lower onintensely anchored sites (Figure 5A). Cover of rubbleand sand was higher on the crest of intensely anchoredsites, but there was no difference on the lower slope(Figure 5B and C). Cover of pavement was similar onsites with different anchoring intensities (Figure 5D).

Between 96 and 108 scleractinian coral species werefound at the six sites, and species richness showed notrend with anchoring intensity. It appears that no spe-


cies were lost as a result of anchor damage. Neitherdiversity nor evenness varied with respect to anchoringintensity (diversity: crest F � 9.467 � 10�3, df 1:4:54, P

� 0.785, lower slope F � 1.361, df 1:4:54, P � 0.248;evenness: crest F � 4.648 � 10�3, df 1:4:54, P � 0.572,lower slope F � 0.011, df 1:4:54, P � 0.327).

Figure 3. Mean number (� 1 SE) of injuries to corals on the crest and lower slope of coral reefs (y axes; numbers) influencedby high (clear bars) and low (black bars) intensities of anchoring. Results of MANOVA are presented, degrees of freedom are1:4:53 and alpha levels were set at P � 0.05. Note the different scales on the y axes.

Figure 4. Mean percent coral cover (� 1 SE) on the crest and lower slope of coral reefs (y axes; percent cover) influenced byhigh (clear bars) and low (black bars) intensities of anchoring. Results of MANOVA are presented, degrees of freedom are 1:4:54and alpha levels were set at P � 0.05. Note the different scale on the y axes.


Mean colony size of the four coral species was highlyvariable (Figure 6). One of the four coral species ex-amined, Millepora sp., displayed different colony sizebetween anchoring intensities (Figure 6A).

CDA Analysis

Complete spatial separation of sites influenced bydifferent anchoring intensities was achieved by de-scribing the condition of coral reefs using five injurytypes (Figure 7). Sites influenced by higher anchor-ing intensities were grouped to the right and sitesinfluenced by low anchoring intensities were grouped

to the left. The relative abundance of each type ofinjury determined the position of each site in multidi-mensional space. Sites influenced by higher intensity ofanchoring had relatively more overturned corals com-pared to the sites with lower anchoring intensities. Frag-ments, rubble, and breaks had less influence on theposition of the sites compared with overturned coralsand gouges (Figure 7). Describing the condition ofcoral reefs using injury type explained 87.9% of thevariation and correctly classified 49.6% of the repli-cates, substantially more than the 8.3% expected bychance (Table 1).

Figure 5. Mean percent cover of substrate type including total reef biota (� 1 SE) on the crest and lower slope of coral reefs(y axes; percent cover) influenced by high (clear bars) and low (black bars) intensities of anchoring. Results of MANOVA arepresented, degrees of freedom are 1:4:54 and alpha levels were set at P � 0.05. Note the different scales on the y axes.

Figure 6. Mean colony area (� 1SE) for four coral species on reefs(y axes; size in cm) influenced byhigh (clear bars) and low (blackbars) intensities of anchoring. Re-sults of MANOVA are presented,degrees of freedom are 1:4:280 andalpha levels were set at P � 0.05.Note the different scales on the yaxes.


In comparison, CDA performed on coral cover vari-ables did not successfully separate coral reef sites influ-enced by different intensities of anchoring (Figure 8).Describing the condition of coral reefs using coralcover explained less of the variation (77.3%) and cor-rectly classified fewer replicates (34.2%) compared withmeasuring injury types (Table 1). The variable that hadmost influence on the separation of sites was theamount of substratum (Figure 8).

The CDA conducted using both injury and coralcover variables best described changes in coral condi-tion associated with anchoring and correctly classifiedthe highest number of replicates. However, collectingdata on all these variables was time consuming (Table1). A stepwise CDA identified eight variables (over-turned corals, gouges, rubble, cover of soft corals, Acro-poridae, Faviidae, Poritidae, and Milleporidae), as themost useful variables to describe the condition of coralreefs influenced by anchoring. The stepwise CDA cor-rectly classified 61.3% of replicates, but collecting dataon each of these variables was also time consuming

(Table 1). However, a combination of the number ofoverturned corals and coral cover correctly classified52.1% of replicates and was time efficient. Further re-duction in data collection time with minimal loss ofcorrect classification was achieved by measuring over-turned corals plus the cover of soft corals and Acropo-ridae (Table 1).

Evaluation of Potential Indicators Against FeasibilityCriteria

Eleven candidate variables of the initial 24 differedsignificantly between anchoring treatment and are con-sidered potential indicators. A matrix was constructedto evaluate the 11 potential indicators against the fea-sibility criteria (Table 2). Overturned colonies metseven of nine feasibility criteria and were efficient inseparating sites with high and low anchoring intensity,ranking it the most useful indicator. There were twodrawbacks to using overturned corals to evaluate man-agement action. First, overturned colonies persist for along time, so may be slow to respond to protective

Figure 7. Results of CDA using five injuryvariables on reef sites influenced by low (darkgrey) and high (light grey) intensities of an-choring. The length of the biplot line reflectsthe relative influence of each variable on thepositioning of the sites. The number in eachcircle denotes the group centroid and isequivalent to the site number from Figure 2.Crest sites (C), lower slope (S). The diameterof the circle is equivalent to one standarderror.

Table 1. Comparison of ability and time required for combinations of variables to describe the changes in coral reefcondition associated with different intensities of anchoringa

VariablesPercentvariation explained

Cross-validation

Collection time(minutes/transect)

Injury types 87.9 49.6 14Coral cover 77.3 34.2 6Injuries and coral cover 74.5 63.0 20Stepwise 77.2 61.3 16Overturned � cover 84.2 52.1 9Overturned � soft corals � Acroporidae 96.3 45.4 Not measured, but estimated as � 9

aMeasuring overturned corals plus cover of soft coral and Acroporidae provided efficient classification. Variables selected in the stepwise analysiswhere overturned corals, gouges, rubble, cover of soft corals, Acroporidae, Faviidae, Poritidae, and Milleporidae.


strategies introduced by managers. Second, there arefew baseline data available on the number of over-turned colonies on coral reefs on a local scale.

Gouges were present in a relatively low numbers butshowed a significant difference between anchoring in-tensities. Gouges were quickly covered in sedimentmaking their identification less reliable, increasingtraining requirements. Therefore, gouges are consid-ered less suitable than some other measures. Countingpieces of rubble was difficult and unreliable. Further-

more, rubble would take a long time to respond tomanagement strategies, reducing its usefulness as anindicator.

Coral cover variables were not as efficient in distin-guishing sites with different anchoring intensities asvariables that measured injuries to corals (Figures 7 and8). However, they relate directly to management objec-tives of maintaining coral communities and are impor-tant to measure. Soft corals and corals in the familyAcroporidae were identified in both the stepwise CDA

Figure 8. Results of CDA using eight coral cover variables on reef sites influenced by low (dark grey, sites) and high (light grey)intensities of anchoring. The length of the biplot line reflects the relative influence of each variable on the positioning of thesites. The number in each circle denotes the group centroid and is equivalent to the site number from Figure 2. Crest sites (C),lower slope (S). The diameter of the circle is equals one standard error.

Table 2. The potential indicators were evaluated against nine feasibility criteriaa

Potentialindicators

Reliablerepeatablemeasures Relevance

Time toresponse tomanagement

Ease ofmeasure

Naturalvariability(SE)

Samplingwindow

Ease oftraining

Baselinedata

Time tomeasure(minutes/transect)

Rank(1–5)

DamageOverturned High Yes Slow Easy 0.77 Long Easy Regional 3.1 1Gouges Low Yes Medium Difficult 0.07 Short Medium Regional 2.2 3Rubble Low Yes No Difficult 2.3 Long Easy None 5.4 4

Coral CoverSoft corals Yes Yes Not predictable Easy 1.9% Long Medium Local 6.1 1Acroporidae Yes Yes Not predictable Easy 2.0% Long Medium Local 1Milleporidae Yes Yes Not predictable Medium 1.3% Long Medium Local 3Pocilloporidae Yes Yes Not predictable Medium 0.4% Long Medium Local 3

Substratum coverReef biota Yes No Not predictable Easy 3.7% Long Easy Local 3.2 3Rubble Yes No Slow Easy 2.3% Long Easy Local 3Sand Yes No Slow Easy 3.0% Long Easy Local 3

Colony sizeMillepora sp. Yes Yes Slow Medium 186.3cm2 Long Medium None 32.3 5

aThe potential indicators were ranked 1–5 for their usefulness as indicators, where 1 was the most useful and 5 the least.


and MANOVA as varying with anchoring intensity at bothdepths for soft corals and on the crest for Acroporidae.These coral groups are easily identified and could bemonitored by volunteers. For rapid assessment of coralreefs, measuring these two coral groups plus overturnedcolonies described changes occurring to coral reef condi-tion associated with anchoring (Tables 1 and 2).

Collecting information at coral family level takeslonger and requires more training compared with col-lecting data on substratum categories (Table 2). How-ever, the increased costs associated with collecting dataat a higher resolution may be worthwhile, as it commu-nicates the effects of anchoring at an earlier stage.Recognizing the effects of anchoring earlier allowsmanagement strategies to be implemented before deg-radation has proceeded to a stage where recovery timesare long.

The average colony size of selected species was vari-able and only Millepora sp. varied significantly with an-choring intensity. The high variability and extensivemeasurement times, suggests that measuring sizes ofcoral colonies is not a usable indicator (Table 2).

Discussion

Anchor damage is a common disturbance to coralreefs (Jameson and others 1999, Rogers and Beets 2001).Indicators that measure the effects of anchoring on coralreef condition can be used to evaluate management strat-egies implemented to protect coral reefs from such dam-age. Selecting useful indicators to measure changes in thecondition of coral reefs, as with other natural resources,has proven difficult. To meet the requirements of man-agers, indicators need to respond to the human activity inquestion, be cost effective, and relate to managementobjectives. Therefore, we have developed a frameworkthat evaluates a range of candidate variables to identify themost useful indicator(s). Following the indicator selectionframework, 24 variables of coral reef condition were eval-uated and three were identified as being of the most useas indicators.

Increases in the number of injured coral coloniesoccurred on the intensely anchored sites, similar toresults reported from coral reefs that are associatedwith high levels of human activities elsewhere (Allison1996, Davis 1977, Hawkins and Roberts 1992, Jamesonand others 1999, Muthiga and McClanahan 1997,Schleyer and Tomalin 2000). In this study, measuringthe number of overturned colonies was the most usefulindicator in separating sites that were influenced bydifferent intensities of anchoring. Coral colonies areoverturned in cyclonic or storm conditions (Done1992), but not during typical weather patterns, so over-

turned coral are usually present on coral reefs in onlylow numbers. Overturned corals are a usable indicatorbecause they are easily identified underwater; there-fore, data collection is quick and could be conductedby volunteers further reducing costs.

Hawkins and Roberts (1992) measured the numberof fragmented corals and breaks to describe changes tocoral reef condition associated with human activities.Fragments did not vary in our study with anchoringintensity. A possible explanation for the lack of varia-tion is that fragments are regularly generated on theGreat Barrier Reef (Rouphael and Inglis 1995, Smithand Hughes 1999, Wallace 1985). Therefore, the dif-ference between the numbers of fragments generatedwith and without the influence of anchoring needs tobe high to detect a difference. The numbers of breakswas not related to anchoring intensities either, butappeared to be related to the abundance of branchingcorals, such as Acroporidae and Milleporidae.Rouphael and Inglis (1997) reported an increase of thenumber of breaks caused by divers on coral reefs withhigh cover of branching corals.

Measuring injuries to corals was more efficient thanthe more traditional measures of coral cover in describ-ing the condition of coral reefs influenced by anchor-ing. Since changes may be detected earlier using otherindicators, relying on coral cover may result in a delayin identifying damage or recovery of coral communi-ties. Comparing levels of injuries to coral rather thancoral cover identified a decline in condition of coralreefs associated with high levels of human activities inKenya (Muthiga and McClanahan 1997).

Since coral cover relates directly to the managementobjectives and provides beneficial information for com-parative studies, it was explored further by comparinghow anchoring intensity influenced the abundance ofdifferent coral families. Coral family groups respondeddifferently to anchoring, primarily because they havedifferent types of growth patterns. Soft corals lack ahard skeleton, are highly susceptible to physical dam-age, and therefore were highly responsive to anchoring.The cover of Acroporidae and Milleporidae on thecrest and Pocilloporidae on the lower slope was re-duced with anchoring intensity. These corals have abranching morphology that is susceptible to physicaldamage (Hall 1998, Marshall 2000) such as that causedby anchors. Corals in the family Poritidae and Faviidaehave a massive morphology, which is more resistant tophysical damage (Marshall 2000) and the cover of thesecorals varied little with anchoring intensity.

The process followed in our study identified that mea-suring a combination of injury variables and the mostresponsive coral cover variables would be the most useful


indicators to describe changes to coral reef conditionassociated with different anchoring intensities. The indi-cators identified were overturned corals plus the cover ofsoft corals and corals in the family Acroporidae. The threeindicators could be used to evaluate the effectiveness ofthe reef protection program implemented in the Whit-sunday region at a relatively low effort for data collection.If the management program is effective, indicators forheavily used reefs where the program is implementedshould change over time to resemble values for reefs withhistorically low anchoring rates.

Selecting appropriate indicators to evaluate manage-ment is difficult (Crabtree and Bayfield 1998). Theindicator selection framework developed has addresseddifficulties in identifying indicators by evaluating arange of variables at sites with high and low intensity ofthe human activity of interest. The framework is trans-parent and could easily be adapted to other naturalresource management strategies. Where indicators areeasy to measure, volunteer organizations can collectdata, reducing cost and increasing stakeholder collab-oration. The framework developed, using anchoring oncoral reefs in the Whitsunday Islands as a case study,identifies indicators to measure and communicate theachievements of management strategies. For these in-dicators to be useful in evaluating similar managementstrategies implemented in other coral reef locations,some adaptations may be required. The type of indica-tor selected will vary, depending on the dominant taxain the local community. Nevertheless, the frameworkdeveloped here for selection of indicators can be ap-plied to other coral reefs, or even terrestrial situations.

AcknowledgmentsResearch was supported by a scholarship from the

CRC Reef Research Centre and a grant from theGBRMPA. Fieldwork was made possible with the sup-port of Queensland Parks and Wildlife Service and theHayman Island Resort. Special thanks to Mark Fentonand Peter Valentine for their support of the researchand to numerous field assistants.

References

Aeby, G. S. 2000. Spatial and temporal patterns of Trematodeinfections on the reef coral, Porites compressa, in Hawaii.Abstracts of the Ninth International Coral Reef Symposium,Bali, p. 280.

Alder, J. 1996. Have tropical marine protected areas worked?An initial analysis of their success. Coastal Management24:97–114.

Allison, W. R. 1996. Snorkeler damage to reef corals in theMaldive Islands. Coral Reefs 15:215–218.

Attwood, C. G., B. Q. Mann, J. Beaumont, and J. M. Harris.1997. Review of the state of marine protected areas in SouthAfrica. South African Journal of Marine Science 18:341–367.

Bak, R. P. M., and E. H. Meesters. 1998. Coral populationstructure: the hidden information of colony size-frequencydistributions. Marine Ecology Progress Series 162:301–306.

Belnap, J. 1998. Choosing indicators of natural resource con-dition: a case study in Arches National Park, Utah, USA.Environmental Management 22:635–642.

Brown, B. E., and L. S. Howard. 1985. Assessing the effects of“stress” on reef corals. Advances in Marine Biology 22:1–55.

Crabtree, B., and N. Bayfield. 1998. Developing sustainabilityindicators for mountain ecosystems: a study of the Cairn-gorms, Scotland. Journal of Environmental Management52:1–14.

Dahl, A. L. 2000. Using indicators to measure sustainability:recent methodological and conceptual developments. Ma-rine and Freshwater Research 51:427–433.

Dale, V. H., and S. C. Beyeler. 2001. Challenges in the devel-opment and use of ecological indicators. Ecological Indicators1:3–10.

Davis, G. E. 1977. Anchor damage to a coral reef on the coastof Florida. Biological Conservation 11:29–34.

Dinsdale, E. A. 2002. Abundance of black-band disease oncorals from one location on the Great Barrier Reef: a com-parison with abundance in the Caribbean region. Proceed-ings of the Ninth International Coral Reef Symposium, Bali2:1239–1243.

Done, T. J. 1992. Effects of tropical cyclone waves on ecolog-ical and geomorphological structure on the Great BarrierReef. Continental Shelf Research 12:859–872.

Done, T. J. 1995. Ecological criteria for evaluating coral reefsand their implication for managers and researches. CoralReefs 14:183–192.

Dudley, N., B. Gujja, B. Jackson, J. P. Jeanrenaud, G. Oviedo,A. Phillips, P. Rosabel, S. Stolton, and S. Wells. 1999a.Challenges for protected areas in the 21st century. Pages3–12 in S. Stolton, and N. Dudley. Eds, Partnerships forprotection: new strategies for planning and managementfor protected areas. Earthscan, London.

Dudley, N., M. Hockings, and S. Stolton. 1999b. Measuringthe effectiveness of protected area management. Pages249–261 in S. Stolton, and N. Dudley. Eds, Partnerships forprotection: new strategies for planning and managementfor protected areas. Earthscan, London.

Dustan, P., and J. C. Halas. 1987. Changes in the reef-coralcommunity of Carysfort Reef, Key Largo, Florida: 1974 to1982. Coral Reefs 6:91–106.

Edinger, E. N., G. V. Limmon, J. Jompa, W. Widjatmoko, J. M.Heikoop, and M. J. Risk. 2000. Normal coral growth rateson dying reefs: are coral growth rates good indicators ofreef health. Marine Pollution Bulletin 40:404–425.

English, S., C. Wilkinson, and V. Baker. 1997. Survey manualfor tropical marine resources, 2nd ed. Australian Instituteof Marine Science, Townsville 390.

GBRMPA. 1999. Whitsunday Plan of Management. Great Bar-rier Reef Marine Park Authority, Townsville.

Green, E. P., and A. W. Bruckner. 2000. The significance of


coral disease epizootiology for coral reef conservation. Bio-logical Conservation 96:347–361.

Griffith, J. A. 1998. Connecting ecological monitoring andecological indicators: a review of the literature. Journal ofEnvironmental Systems 26:325–363.

V. R. Hall (1998) �Injury and regeneration of common reef-crest corals at Lizard Island, Great Barrier Reef.� Doctoralthesis, James Cook University, Townsville, 118 pp.

Hawkins, J. P., and C. M. Roberts. 1992. Effects of recreationalSCUBA diving on fore-reef slope communities of coralreefs. Biological Conservation 62:171–178.

Hawkins, J. P., and C. M. Roberts. 1993. Effects of recreationalSCUBA diving on reefs: trampling on reef-flat communities.Journal of Applied Ecology 30:25–30.

Hershman, M. J., J. W. Good, T. Bernd-Cohen, R. F. Goodwin,V. Lee, and P. Pogue. 1999. The effectiveness of coastalzone management in the United States. Coastal Management27:113–138.

Highsmith, R. C. 1982. Reproduction by fragmentation incorals. Marine Ecology Progress Series 7:207–226.

Hockings, M. 1998. Evaluating management of protected ar-eas: integrating planning and evaluation. EnvironmentalManagement 22:337–345.

Hockings, M. 2000. Evaluating effectiveness: a framework forassessing management of protected area. IUCN, 58 pp

Hughes, T. P. 1994. Catastrophes, phase shift, and large-scaledegradation of a Caribbean reef. Science 265:1547–1551.

Hughes, T. P., A. H. Baird, E. A. Dinsdale, N. A. Moltschani-wskyj, M. S. Pratchett, J. E. Tanner, and B. L. Willis. 2000.Supply-side ecology works both ways: the link betweenbenthic adults, fecundity, and larval recruits. Ecology81:2241–2249.

Jameson, S. C., M. S. A. Ammar, E. Saadalla, H. M. Mostafa,and B. Riegl. 1999. A coral damage index and its applica-tion to diving sites in the Egyptian Red Sea. Coral Reefs18:333–339.

Keough, M. J., and G. P. Quinn. 1998. Effects of periodicdisturbances from trampling on rocky intertidal algal beds.Ecological Applications 8:141–161.

Liddle, M. J., and A. M. Kay. 1987. Resistance, survival andrecovery of trampled corals on the Great Barrier Reef.Biological Conservation 42:1–18.

Lorenz, C. M. 1999. Indicators for sustainable river manage-ment. Doctor of Philosophy thesis. Vrije Universiteit. 257 pp.

Margoluis, R., and N. Salafsky. 1998. Measures of success.Designing, managing and monitoring conservation and de-velopment projects. Island Press, Washington 362.

Marshall, P. A. 2000. Skeletal damage in reef corals: relatingresistance to colony morphology. Marine Ecology ProgressSeries 200:177–189.

McClanahan, T. R. 1999. Is there a future for coral reef parksin poor tropical countries?. Coral Reefs 18:321–325.

Muthiga, N. A., and T. R. McClanahan. 1997. The effect ofvisitor use on the hard coral communities of the KisiteMarine Park, Kenya. Proceedings of the Eighth InternationalCoral Reef Symposium, Panama 2:1879–1882.

Pollnac, R. B., J. W. McManus, A. E. del Rossario, A. A.Banzon, S. G. Vergara, and M. L. G. Gorospe. 2000. Unex-pected relationships between coral reef health and socio-economic pressures in the Philippines: reefbase/RAMP ap-plied. Marine Freshwater Research 51:529–533.

Rogers, C. S., and J. Beets. 2001. Degradation of marineecosystems and decline of fishery resources in marine pro-tected areas in the US Virgin Islands. Environmental Conser-vation 28:312–322.

Rouphael, T., and G. Inglis. 1995. The effects of qualifiedrecreational SCUBA divers on coral reefs. Rep. 4. CRC ReefResearch Centre, Townsville 39.

Rouphael, A. B., and G. J. Inglis. 1997. Impacts of recreationalSCUBA diving at sites with different reef topographies.Biological Conservation 82:329–336.

Schleyer, M. H., and B. J. Tomalin. 2000. Damage on SouthAfrican coral reefs and an assessment of their sustainablediving capacity using a fisheries approach. Bulletin of MarineScience 67:1025–1042.

Short, F. T., and S. Wyllie-Echeverria. 1996. Natural and hu-man-induced disturbances of seagrasses. Environmental Con-servation 23:17–27.

Smith, L. D., and T. P. Hughes. 1999. An experimental assess-ment of survival, re-attachment and fecundity of coral frag-ments. Journal of Experimental Marine Biology and Ecology235:147–164.

Sweatman, H., A. Cheal, G. Coleman, B. Fitzpatrick, I. Miller,R. Ninio, K. Osborne, C. Page, D. Ryan, A. Thompson, andP. Tomkins. 2000. Long-term monitoring of the Great BarrierReef. Australian Institute of Marine Science, Townsville 117.

Tabachnick, B. G., and L. S. Fidell. 2001. Using MultivariateStatistics, 4th ed. Allyn and Bacon, Boston 966.

Underwood, A. J. 1997. Experiments in ecology. Their logicaldesign and interpretation using analysis of variance. Cam-bridge University Press, Cambridge 504.

Wallace, C. C. 1985. Reproduction, recruitment and fragmen-tation in nine sympatric species of the coral genus Acro-pora. Marine Biology 88:217–233.

Waltner-Toews, D. 1996. Ecosystem health: a framework forimplementing sustainability in agriculture. BioScience46:686–689.

Ward, T. J. 2000. Indicators for assessing the sustainability ofAustralia’s marine ecosystems. Marine and Freshwater Research51:435–446.

Wells, S. 1999. Tackling the paper park problem. Reef Encoun-ters 26:24–26.

Wilkinson, C. 2000. Status of coral reefs of the world: 2000.Australian Institute of Marine Science, Townsville 363.

Wood, N., and P. Lavery. 2000. Monitoring seagrass ecosystemhealth: the role of perception in defining health and indi-cators. Ecosystem Health 6:134–148.

Yu, C. C., J. T. Quinn, C. M. Dufournaud, J. J. Harrington,P. P. Rogers, and B. N. Lohani. 1998. Effective dimension-ality of environmental indicators: a principal componentanalysis with bootstrap confidence intervals. Journal of Envi-ronmental Management 53:101–119.


ENVIRONMENTAL ASSESSMENT Assessing Anchor Damage on ...

Documents

Transcript of ENVIRONMENTAL ASSESSMENT Assessing Anchor Damage on ...