estimating three-dimensional surface areas on coral reefs
TRANSCRIPT
Journal of Experimental Marine Biology and Ecology xxx (2008) xxx–xxx
JEMBE-48680; No of Pages 7
Contents lists available at ScienceDirect
Journal of Experimental Marine Biology and Ecology
j ourna l homepage: www.e lsev ie r.com/ locate / jembe
ARTICLE IN PRESS
Estimating three-dimensional surface areas on coral reefs
Glen Holmes ⁎Centre for Marine Studies, The University of Queensland, St. Lucia Queensland 4072, Australia
⁎ Tel.: +61 7 33651475; fax: +61 7 33654755.E-mail address: [email protected].
0022-0981/$ – see front matter © 2008 Elsevier B.V. Aldoi:10.1016/j.jembe.2008.07.045
Please cite this article as: Holmes, G., Estimjembe.2008.07.045
a b s t r a c t
a r t i c l e i n f oArticle history:
One of the main obstacles Received 6 February 2008Accepted 31 July 2008Available online xxxxKeywords:Coral morphologyEcological functionLaser scanningSurface areaSurface index
for biological assessments of coral reefs over large spatial scales is the ability tolink data obtained at the laboratory scale to spatially large data sets. This is particularly the case when tryingto assess the ecological function of microbial processes following dramatic large-scale events such as masscoral bleaching. To be able to infer ecological function of field corals from laboratory measurementstandardised to surface area it is imperative to be able to measure the actual surface area of corals in-situ.There have been several approaches proposed to estimate the three-dimensional surface area of field corals.While these have been shown to be reliable for simple coral growth forms, large degrees of error areintroduced when applying them to complex growth forms. This paper refines a technique for calculating thethree-dimensional surface area based on the projected surface area, with errors associated with complexgrowth forms reduced to b5%. Once developed, the simple mathematical relationship (called the surfaceindex) can be used to estimate the three-dimensional surface area of field corals from photograph or videoimagery, allowing physiological parameters of corals determined at the sub-colony scale to whole colony andspatially large data sets of coral reefs. The effectiveness of using laser scanning techniques to derive three-dimensional images of corals is also discussed.
© 2008 Elsevier B.V. All rights reserved.
1. Introduction
To improve the understandingof coral reef ecosystems, it is essentialthat studies are conducted over a wide range of temporal and spatialscales. It is equally important that scientists are able to apply (or infer)findings from studies conducted at one scale to studies at other scales(i.e. the ability to scale up (or down)). Indeed, the ability to extrapolatesmall scale processes has been identified as the next challenge inmicrobiology (Paerl and Steppe, 2003). With the rapidly increasinginterest in microbial processes and how they influence ecosystems atdifferent scales, this need has become evenmore urgent and necessaryto resolve. Inter-study evaluations and extrapolations require thatmeasurements are normalised to a variable that allows for simplecomparisons. In the case of coral biology, several possibilities have beenused for normalising physiological parameters such as tissue biomass,zooxanthellae density, chlorophyll density and respiration. Previouslyproposed parameters include colony (Yonge et al., 1932), colony weight(Kawaguti, 1937) and coral polyp (Marshall, 1996). The most commonstandardising parameters however are surface area and tissue biomass(Edmunds and Gates, 2002). While there are many arguments for theuse of biomass as a suitable standardising parameter (Edmunds andGates, 2002), surface area is most suited for the integration of in-situmeasurements and spatially large scale data sets as biomass measure-ments are not feasible in these situations.
l rights reserved.
ating three-dimensional surf
The ability to estimate the surface area of both biotic and abioticsurfaces is an essential component of many facets of biology includingthe capacity to relate findings across spatial scales. Nowhere is thismore evident than in the field of coral reef biology and there havebeen severalmethods proposed to estimate three-dimensional surfaceareas of coral surfaces (Table 1). Traditionally, the two-dimensionalprojected areawas used as ameasure of surface area for the calculationof ecological budgets (Kanwisher and Wainwright, 1967; Odum andOdum, 1955) with “correction factors” introduced in order to bringvalues up to more realistic levels (Webb et al., 1975; Wilkinson et al.,1984). For example, Odum and Odum (1955) used the projected areafor calculation of biomass and chlorophyll levels but scaled by a factorof three for bacterial estimates in their study on Eniwetok Atoll.In 1970, Marsh (1970) described a method for estimating the actualthree-dimensional surface area. This involved covering an object withaluminium foil and estimating the surface area either from the weightdifference of the object beingmeasured (with theweight per unit areaof the foil previously determined) or by unwrapping the object andspreading the foil flat to make a direct measurement. For many yearsthiswas themost popularmethodof determining surface area (Hoegh-Guldberg, 1988). Other methods based on this idea have since beendeveloped including coating with latex (Meyer and Schultz, 1985), dye(Hoegh-Guldberg, 1988) or paraffin wax (Stimson and Kinzie, 1991),the latter being one of the common methods currently used. Whileeach of the currently utilisedmethods provide a good estimation of thethree-dimensional surface area, they are only practical at the relativelysmall scale of the coral nubbin or small colony and fewmethods can bemade on live, in-situ colonies. While this does not present a problem
ace areas on coral reefs, J. Exp. Mar. Biol. Ecol. (2008), doi:10.1016/j.
Table 1Published methods for estimating the surface area of corals
Methodology In-situ measurements Colony scale measurements Reef scale measurements Accuracy Reference
Projected area Yes Yes Yes Poor Kanwisher and Wainwright (1967),Odum and Odum (1955)
Aluminium foil No Limited No Good Marsh (1970)Calculation Yes Yes Yes Poor Chancerelle (2000), Dahl (1973),
Alcala and Vogt (1997),Courtney et al. (2007),Fisher et al. (2007)
Latex No Limited No Good Meyer and Schultz (1985)Scanning No Limited No Very Good Kaandorp and Kuebler (2001)Dye uptake No Yes No Good Hoegh-Guldberg (1988)Waxing No Limited No Good Stimson and Kinzie (1991)Photogrammetry Yes Limited No Time constrained Bythell et al. (2001)3-D reconstruction using video Yes Limited No Time constrained Cocito et al. (2003)
2 G. Holmes / Journal of Experimental Marine Biology and Ecology xxx (2008) xxx–xxx
ARTICLE IN PRESS
for many applications of three-dimensional surface area data, whenestimates over large colonies or even whole reefs are required, thesemethods become impractical due to the inability to make unobtrusivemeasurements.
It is now recognised that like many terrestrial and aquatic eco-systems, coral reefs are under significant pressure throughout theirdistribution, with many showing signs of rapid decline (Hughes et al.,2003). The Global Status of Coral Reefs 2004 Report (Wilkinson, 2004)“predicts that 24% of the world's reefs are under imminent risk ofcollapse through human pressures; and a further 26% are under alonger term threat of collapse”. These large-scale changes within coralreef ecosystems are driving the need for ecologically relevant esti-mations of physiological parameters. Field survey techniques such asquadrats and belt transects using digital photo and video techniquesare able to assess 100's m2 of benthos (English et al., 1997; Rogerset al., 1994; Wilkinson and Hill, 2004), and with the rapidly increasingcapacity for remote sensing tools (Mumby et al., 2004), data overentire reef systems can be obtained. Similarly, there many toolsavailable for researches to assess coral reef parameters at the sub-colony scale but there are as yet none that allow the integration ofmeasurements between the two data sets. This is a critical hindranceto developing knowledge on ecosystem function and habitat interac-tion at different scales.
The concept of a surface index (SI) introduced into marine biologyby Dahl (1973) represents a potentially powerful tool to be able toapply measured parameters to more meaningful spatial scales.However, the methodology adopted by Dahl (1973) in developingthese indices, reconstructing the target object from a series is simplegeometric shapes. This resulted in an easily calculated SI, representinghowever, only a relatively crude approximation. Alcala and Vogt (1997)tested Dahl's theory on a range of coral growth morphologies andconcluded that while the method of calculating the surface area fromgeometric shapes lacked reliable accuracy, the SI concept was a poten-tially useful methodology. Chancerelle (2000) took the SI concept astep further testing six species of coral, of differing morphology, forthe existence of a SI relationship. Chancerelle (2000) and later Holmeset al. (in review) showed that a species specific linear relationship(i.e. SI) existed for each of the species tested, between the projectedarea of the coral and the actual three-dimensional surface area. Theseinvestigations have paved the way for the current study, wherebySI functions have been developed for gross coral growthmorphologies.These SI functions provide a link between spatially large data setssuch as belt transects and physiological measurements on coralsurfaces.
2. Materials and methods
In selecting a methodology for measuring the actual three-dimensional surface area of corals, the issue of resolution was first
Please cite this article as: Holmes, G., Estimating three-dimensional surfjembe.2008.07.045
considered. As outlined in Dahl (1973), there are several levels ofcomplexity associated with coral reefs. At the level of the coral colonyitself, there may be assumed to be two levels, the gross morphologyand the complexity associated with corallite structure. This study hasfocussed on gross morphology as a potential link between spatiallylarge data sets and small scale in-situ field measurements. Twomethodologies were utilised (and compared): coating with paraffinwax (Stimson and Kinzie,1991); and digital technology in the form of ahandheld laser scanner at a resolution of 2.5 mm.
158 coral skeletons representing more than 25 genera up to 75 cmwide were sourced from collections at the Queensland Museum, andthe University of Queensland. Although the majority of skeletons usedwere completely intact, some colonies with minor breakages werealso included to broaden the applicability of the determined relation-ships as coral breakage is a natural phenomenon.
2.1. Surface Area Determination – Paraffin Wax
No discrimination was given to collection location or depth inselecting skeletons for analysis. The three-dimensional surface areasof colonies were calculated using a modified version of the paraffinwax method (Stimson and Kinzie, 1991). Skeletons at room tempera-ture were weighed and dipped into a paraffin wax bath (Paraplast®Tissue Embedding Medium, Tyco Healthcare Group) maintained at65 °C for 2 seconds. When removed from the bath they were rotatedto optimise evenness of the coating. The initial coating seals theskeleton, reducing the influence of corallite rugosity and filling anycavities resulting from infaunal burrowers. The skeletons were re-weighed before and after being dipped for a second time in theparaffin wax.
Calibration objects of known surface area but of varying levelsof complexity and surface texture (wood, plastic, coral, plastacene,n=14), were measured using the paraffin wax procedure. Usingcalibration objects of varying texture provides a test of whether sur-face roughness influences the conversion function (as would beidentified by a poor correlation). The relationship between the weightof the second wax layer and the actual surface area of the objects isthen used to calculate the three-dimensional surface area of the coralskeletons.
Regression analysis of the weight of the second wax coating tothe three-dimensional surface area showed a strong correlation be-tween the calibration objects and the weight of the second layerof paraffin wax deposited (R2N0.99, F1,12=1744.96, Pb0.0001). Theexistence of a very strong correlation coefficient indicates that theinitial surface properties of the objects being coated do not play anysignificant role in calculating the three-dimensional surface area of adipped object from the weight of the second coating of paraffin. Theaverage thickness of the initial layer of waxwas 2±0.5mm (mean±SE)while the second layer had an average thickness of 0.3±0.01 mm.
ace areas on coral reefs, J. Exp. Mar. Biol. Ecol. (2008), doi:10.1016/j.
Fig. 1. Comparison of waxing and scanning technique on calibration objects fordetermining surface areas (n=13). Final resolution of scanned image≈2.5 mm.(Relationship: Scanned area=0.93×waxed area, R2N0.99). 95% confidence (dashed)and prediction (solid) lines are shown.
3G. Holmes / Journal of Experimental Marine Biology and Ecology xxx (2008) xxx–xxx
ARTICLE IN PRESS
The regression analysis was used to obtain the following calibrationrelationship:
SurfaceArea ðcm2Þ ¼ 34:32ðcm2=gÞ �WeightðgÞ
2.2. Surface Area Determination – Laser Scanning
The second methodology adopted was to use a handheld laserscanner (Polhemus Fastscan™). This scanner uses a broad beam of laserlight to illuminate an object while two CCTV cameras record a cross-sectional profile of the object. A reference signal transmitter is placedclose to the subject enabling the supplied software to reconstruct thefull three-dimensional surface of the object. Scanning can be used togenerate digital three-dimensional images of non-metallic objects up toapproximately 1 m across (i.e. longest dimension).
The raw scan data file was then be exported as a stereolithographyformat file, describing a solid object as a list of triangular surfaces.This data file was then imported into a three-dimensional modellingsoftware package for post-processing (Maya 7.0 PLE, Autodesk Inc.).Post-processing involved cleaning the image (removal of duplicatesurfaces andmerging vertices), filling any holes that the scanner couldnot see and removing noise or superfluous data from the image.The resolution of the images after post-processing of the data is ap-proximately 2.5 mm. This is comparable with CT scanning resolutionsused in other studies of coral morphology (Kaandorp et al., 2005). Thethree-dimensional surface area of the scanned corals is then readdirectly from the modelling software.
2.3. Determination of Projected Area
The projected area of coral skeletons was estimated using digitalphotography. The skeletonswere positioned in their natural orientationand photographed from above against a grid of 1 cm squares.Perspective error was minimised through maximising the distancebetween the camera and the subjectwhile utilizing the zoom lens of thecamera to maintain a high resolution image. A polygon was digitizedaround the skeletons and the area of the polygon calculated by thedigitizing software (Didger 2.0, Golden Software Inc). Digitizing ofprojected areas was all performed at the same resolution regardless ofskeleton size and only sections of skeleton that would have containedlive tissue at the time of collection were included. This allows thedeveloped SI relationships to be applied to live tissue areas excludingerrors introduced by the influence of the abiotic basal zones of colonies.Only the perimeter-based polygon was used in the analysis and nocompensation was made for areas within the polygon that may nothave contained coralmaterial (i.e. gaps between coral branches internalto the perimeter of the colony).
The projected area of scanned images was determined by con-structing a polygon around the image in the plan view. Using the plan,rather than model view, eliminates any perspective error. The area ofthe traced polygon is then read directly from the software.
2.4. Data Analysis
All data were analyzed using PopTools v 2.6.7 (Hood, 2005),Statistica v6.0 (StatSoft Inc.) and SigmaPlot 2001 v7.101 (SPSS Inc.).Regression analysis was used to determine relationships betweenthe projected and three-dimensional area of corals. The analysis wasperformed at two levels, one very broad level incorporating all coralmorphologies, and a more refined analysis where the corals wereseparated into gross growthmorphologies based on Veron (2000). Themorphologies included massive, sub-massive (these included colum-nar), open branching, complex branching (i.e. dense branches), folioseand tabular/plate growth forms. Solitary and encrusting corals wereexcluded from the analysis. In all cases with the exception of the
Please cite this article as: Holmes, G., Estimating three-dimensional surfjembe.2008.07.045
tabular growth form, the regression analysis was forced through theorigin as it has been assumed that the combined effect of the multiplespecies examined will result in an essentially linear pattern from theorigin. This is in contrast to species specific relationships where theassumption of linearity at very small projected areas must be tested(Holmes et al., in press). Tabular growth forms were not forced as it isassumed that in all cases there is a dramatic difference in morphologybetween very young corals were growth is predominantly vertical, toestablished colonies where growth in the horizontal plane dominates.
Regression analysis was also used to compare the laser scanningtechnique and calibration of the paraffin wax technique for surfacearea estimation. In all cases, residuals were analyzed to assess for anydeparture from linearity or the impact of any outliers on the finalanalysis.
3. Assessment
Regression analysis shows a very strong correlation between thescanning and waxing techniques used to assess three-dimensionalsurface area of coral colonies (F1,12=1759.9, R2N0.99, Pb0.0001, Fig. 1).Residuals analysis indicated no significant departure from linearity orundue influence of outliers on the regression results (Wald-WolfowitzRuns Test, P=0.55). The results show that the paraffin wax coatingtechnique gives slightly larger estimates of surface area than thescanning technique, having a slight smoothing effect on the corallitestructures. As the scanning procedure resulted in a final imageresolution of 2.5 mm, thewaxing procedure used can therefore be saidto result in a resolution less than 2.5 mm as the slope of the regressionis slightly less than 1.
Table 2 presents the results of regression analysis for each of thevarious coral growth morphologies. In all cases with the exceptionof tabular growth forms, a significant linear relationship is shown.As would be expected, the complex branching structures have thehighest ratio of actual to projected area at almost 6.5 times. The foliosegrowth form shows a low average ratio at just over three times butalso has a lower correlation coefficient due to the large amountof variation within this category from almost plate-like to complexformations as exhibited by many Pavona spp. The smallest ratio isexhibited by the tabular corals. This growth form is best describedusing a linear functionwith an intercept as, as mentioned above, thereis a distinct shift in the growth pattern from young corals where thereis more growth in the vertical plane, to older corals where horizontalspread dominates the growth pattern. The resulting function shows a
ace areas on coral reefs, J. Exp. Mar. Biol. Ecol. (2008), doi:10.1016/j.
Table 3Typical coral reef surface rugosity values
Reef Location Rugosity values Reference
Davies Reef, GBR 1.48 – 2.04 Klumpp and Mckinnon (1989)Glover's Reef, Belize. Large patch reef: 1.183(±0.06) Acosta and Robertson (2002)
Small patch reef: 1.134(±0.026)
Carrie Bow Bay, Belize 1.2 – 4 Dahl (1973)Hawaiian archipelago 1.12 – 2.46 Friedlander et al. (2003)Lau group, Fiji 1.4 – 1.7 Dulvy et al. (2002)Kenya 1.12 – 1.3 Mcclanahan (1994)USVI 1.18 – 2.28 Grober-Dunsmore et al.
(2007)Mafia Is, Tanzania 1.18 – 1.5 Garpe and Öhman (2003)
Table 2Relationship between projected and three-dimensional surface area for differentcoral growth forms as determined using linear regression analysis (SI=ratio of three-dimensional to projected surface area)
Growth Morphology SI±SD R2 F p-level
Massive 3.20±0.07 0.99 2042.7 (1, 28) b0.0001Sub-massive 5.90±0.21 0.97 815.3 (1, 22) b0.0001Foliose 3.04±0.21 0.91 209.3 (1, 20) b0.0001Open branching 6.16±0.27 0.97 538.1 (1, 15) b0.0001Complex branching 6.43±0.20 0.96 1066.8 (1, 40) b0.0001Tabular See Note A
Notes:A. Tabular colonieswere notwell describedwith a simple linear functiondue to thedifferent morphology exhibited by younger corals. For this growth form the regressionanalysiswas not forced throughzero. The resulting function isArea3D=2.47×Area2D+0.018all units m2 (F(1,21)=137.9; pb0.001).
4 G. Holmes / Journal of Experimental Marine Biology and Ecology xxx (2008) xxx–xxx
ARTICLE IN PRESS
slope of 2.47 with an intercept of 0.018 m2. This relationship ishowever only valid for colonies with a projected area N50 cm2. Forvalues below this level the slope of the curve must be variable astheoretically the slope must pass through the origin.
4. Discussion
4.1. Laser scanning as a tool for coral biologists
As a tool for assessing the gross morphology of large, simplestructured coral skeletons, the hand-held laser scanner is very useful,being both portable and cost effective. Scanning at a final resolution of2.5 mm is comparable with the paraffin waxing technique for surfacearea estimates while allowing for a wider range of applications thantraditional coating techniques due to its less intrusive application. Aswith other scanning technologies, it enables the user to make anaccurate digital replication of coral skeletons from which morpholo-gical measurements from the branch to colony scale can be madewithgreater repeatability and replication than can practicably be madeusing manual techniques. Laser scanning provides comparable imageresolution to CT scanning in coral morphology assessment (Kaandorpet al., 2005) at a significantly lower cost. It also has the advantage ofequipment portability. Themain disadvantage of this technique is thatto do an effective scan the surface of the object must be visible to theCCTV camera at the point the laser intercepts the object being scanned.For this reason laser scanning is impractical for complexmorphologies.Another potential problem that may be encountered if developed forin-situ scanning of corals is the surface reflectivity. Attenuation due towater or coral pigmentsmay reduce the signal from the laser light suchthat the cameras may not effectively locate the corresponding surfacein three-dimensional space. Different laser coloursmay be required fordifferent pigmented corals to overcome this problem.
4.2. Surface area of corals
As discussed in Dahl (1973), the levels of complexity on coral reefs(pertaining to the live coral zone)may be divided into three categories:benthic topography; gross coral morphology; and coral polypmorphology. Within each category it is the degree of vertical variationthat drives the scaling function for scaling up two-dimensional mea-surements. Depending on the resolution of the surface area estimaterequired, it is the product of these levels of complexity that allowsscaling of two-dimensional areas to actual three-dimensional surfaceareas.
The first scale of complexity, that of the benthic topography(excluding any corals), may be highly variable, even within reef zonesand estimations should be made on a site specific basis using fieldsurvey techniques (Klumpp and Polunin, 1989). Typically, surfacecomplexity values range from ~1 – 2 (Table 3), although values as highas 4 have been quoted.
Please cite this article as: Holmes, G., Estimating three-dimensional surfjembe.2008.07.045
Complexity due to coral polypmorphology is relevant to scaling upprocesses that occur on very small scales, such as biofilm develop-ment. Potentially it is possible to develop scaling relationships basedon polyp size and structure. However, estimations of surface area atthis level of detail are likely to be overshadowed by errors associatedwith the other two levels of complexity.
The ability to scale the second complexity level, that of gross coralmorphology, will greatly expand the abilities to extrapolate sub-colony physiological parameters to larger areas to give insights intoecosystem scale parameters of coral reefs. Analysis of the relationshipbetween projected and three-dimensional surface areas for each ofthe specific growth forms has shown that reliable estimates of three-dimensional surface area can be made from planar images. Averagingeach of the growth morphology SI's, allows for a broad assessmentcoral reef areas where coverage of live coral can be determinedwithout being classified into the various growth forms. However,caution must be exercised when using this “all corals” relationship asthe analysis for this group is based on an equal proportion of thevarious growth forms. In reality this is not likely to be the case and thismay lead to further inaccuracy in the estimation of the three-dimensional surface area.
In areas where a greater level of detail is available on thecomposition of the coral reef community, a more accurate assessmentof the three-dimensional surface area can be made. As outlined inTable 2, the coefficients of regression are much higher for specificgrowth forms, all above 0.95 with the exception of the foliosecategory. The larger degree of variation observed in this category canbe attributed to the variable complexity and vertical spread betweenspecies. Notably, the coefficients of regression for both branchingcategories, as well as the massive and sub-massive categories arecomparable with species specific relationships developed by Chancer-elle (2000) and Holmes et al. (in press).
In comparison with other techniques proposed to estimate thethree-dimensional surface area of field corals, the current studyprovides a more reliable estimate than that proposed by Fisher et al.(2007) using the surface area of a bounding box, or Courtney et al.(2007) who proposed using a log-linear model to estimate the surfacearea from three linear dimensions. Both these methodologies providea relatively simple technique to estimate the three-dimensionalsurface area of field corals. However, significantly greater errors areincurred when applying these techniques to complex coral morphol-ogies in comparison to simple ones such as massive corals. In addition,these techniques require individual corals to be manually measured inthe field, whereas the SI technique can be applied to photograph orvideo images, reducing the time required in the field. The study mostcomparable the current one, that by Alcala and Vogt (1997) derivedsimilar SI's although with a higher degree of error. (For example the SIfor branching colonies was calculated to be 6.88±5.36 c.f. 6.43±0.2 inthe current study).
As outlined above, the tabular growth form is best described by alinear function with an intercept. This is due to the difference in the
ace areas on coral reefs, J. Exp. Mar. Biol. Ecol. (2008), doi:10.1016/j.
5G. Holmes / Journal of Experimental Marine Biology and Ecology xxx (2008) xxx–xxx
ARTICLE IN PRESS
ratio of vertical to horizontal projection between small and largecolonies. The absence of a simple linear relationship complicatesthe process of estimating the three-dimensional surface area beyondthat of a single colony. To effectively estimate the three-dimensionalsurface area of a section of reef dominated by tabular corals wouldrequire an estimate of the average colony size. If this is known thenthe calculation is relatively simple. This is however, unlikely to be thecase. An alternative approach is to use the single colony relationshipand conduct a Monte Carlo Analysis with a range of colony sizes over arange of areas. This approach results in a SI of 2.58 for tabular corals(1000 replicates by area for areas between 10 and 10000 m2, colonyprojected areas 50 – 10000 cm2).
An assessment of the applicability of the relationships developed inthis study can be made by examining the “fit” of the curves to someindependent data. In a study modelling growth of the two growthmorphologies of the branchingCaribbean coralMadracismirabilisMerkset al. (2004) produced “virtual corals” showing the two distinct complexbranching morphologies (available through the supplementary mate-rial). Both of these “virtual coral's” two- to three- dimensional ratio fallwithin the 95% confidence limits of the relationship shown for complexbranching morphologies.
The analyses in this study found that the complex branchingmorphology had the highest scaling function (SI function) betweenthe planar and three-dimensional surface area, with the foliose andmassive growth form displaying the smallest. This is in contrast to thefindings of Chancerelle (2000) where (of the comparable growthforms) sub-massive and then tabular showed the highest ratios. Therelationships developed by Chancerelle (2000) are also much largerthan those developed in the current study. The methodology used todetermine the three-dimensional surface area may explain the differ-ences. Chancerelle (2000) first varnished the skeletons followed byapplying a single wax coat. Sealing skeletons with wax as per thecurrent study, rather than varnish results in a removable coating, butreduces the rugosity of corallite structures to a greater degree. Therelationship between two- and three-dimensional areas is heavilydependent on the methodology used to measure three-dimensionalsurface area, that is, the resolution that can be achieved using anygiven approach. The double wax coating method used in this studyprovides estimates comparable with a 2.5 mm resolution as demon-strated by the laser scanning comparison. It is unclear what thecomparable resolution would be for the Chancerelle (2000) study butat finer resolutions corallitemorphologywould complicate the results.
4.3. Application in ecosystem scale assessments
The SI method developed here provides a technique for bridgingthe scaling gap between traditional methods of estimating surfacearea (and the smaller scale elements of coral surface communities)and coral colony contribution to reef processes. Studies have demon-strated that estimates of reef scale metabolism can be made using
Table 4Average coral physiological parameters per m2 (projected) based on sub colony scale measu
GrowthMorphology
Biomass(g m-2)
Chl(g
All corals 3820 0.5Massive 2690 0.3Sub-massive 4960 0.6Tabular 2710 0.3Foliose 2550 0.3Open branching 5170 0.6Complex branching 5400 0.6Reference basis Odum and Odum (1955) An
FittSor
The “All corals” category is based on an unweighted average of each growth morphology.
Please cite this article as: Holmes, G., Estimating three-dimensional surfjembe.2008.07.045
remote sensing imagery combinedwith some simple assumptions andin-situ measurements taken at the scale of the reef zone (Andrefouetand Payri, 2001; Brock et al., 2006). Such reef scale analyses currentlyrequire data from measurements taken over large areas of reefproducing a measurement that is the net result of the “black box” ofinterest (for example reef crest, lagoon, rubble zone). While improvedmethodologies in this area continue to be investigated (Yates andHalley, 2003), it will remain difficult to tease apart the contributionsof the various components within the “box” and thus confidentlyapply these measurements to areas with differing compositions. Inaddition, while reef zone measurements are able to provide someinsights into whole reef processes, the ability to scale-up sub-colonyscale parameters such as tissue biomass has thus far remained elusive.
In their seminal paper on trophic structure, Odum and Odum(1955) calculated, among many other things, the average biomass ofcorals to be 0.084 g cm-2 (including both polyp and sub-polyp zones).When applying their biomass values to quadrat data, they acknowl-edged that they used projected area and that actual area may besignificantly larger. This average value can now be used to calculate amore meaningful average coral biomass per unit (projected) area(Table 4) that can then be applied to larger scale datasets.
Other physiological parameters can now be scaled up fromorganism levelmeasurements andwhilemany parameterswill exhibitsome level of variability both within and between individuals, averagevaluesmay be used to produce reasonable estimates over spatial scaleswhere it is not feasible tomakemore directmeasurements. Although abroad range of area normalised chlorophyll a levels have been quotedin the coral literature (Anthony and Hoegh-Guldberg, 2003; Fitt et al.,2000; Muscatine et al., 1989; Sorokin, 1995; Thieberger, 1995), a valueof around 10 µg.cm-2 may be used as a general estimate of coralchlorophyll content. In a study examining corals within different lightmicrohabitats, Anthony and Hoegh-Guldberg (2003) found that whenstandardised to surface area, chlorophyll a content showed nosignificant difference across a range of light habitats, suggesting thatusing area normalised chlorophyll contentmaymeaningfully be scaledto assess ecosystemwide properties.
Calcification rates have also been reported to have a broad rangedepending on both the species of interest and the prevailing environ-mental conditions (Houlbreque et al., 2003; Johannes et al., 1972;Lough and Barnes, 2000; Rinkevich and Loya,1984). An approximationof reef scale rates can however be made by assuming an average rateof 1 g.cm-2.yr-1. This produces a rate of 125 g.m-2.day-1 (or 46 kgCaCO3m-2yr-1) when scaled using the general “all corals” relationship(Table 4). This compares well to reef zone measurements presentedin Kinsey (1985) and Sorokin (1995) bearing in mind that these ratesreflect that of the coral only rather than the benthos as a whole.
Perhaps one of the most useful applications of these scaling rela-tionships is the ability to extrapolate data to estimate changes inecosystem function following large scale mortality events. Davey et al.(2008) demonstrated that in the immediate period following coral
rements
orophyll am-2)
Calcification Rate(g m2 day-1)
125881627183169176
thony and Hoegh-Guldberg (2003),et al. (2000), Muscatine et al. (1989),okin (1995), Thieberger (1995)
Houlbreque et al. (2003),Johannes et al. (1972),Lough and Barnes (2000),Rinkevich and Loya (1984)
ace areas on coral reefs, J. Exp. Mar. Biol. Ecol. (2008), doi:10.1016/j.
6 G. Holmes / Journal of Experimental Marine Biology and Ecology xxx (2008) xxx–xxx
ARTICLE IN PRESS
mortality from thermal bleaching, there is a rapid increase in the rate ofnitrogen fixation by biofilms forming on the newly available substrate.In areas of high coral coverage, large scale mortalities will equate tolarge amounts of nitrogen being fixed into the ecosystem. For example,during the 1998 bleaching event in the Maldives, live coral coveragein the central atolls was reduced from 42% to just 2% (Edwards et al.,2001). This equates to an average daily increase in ecosystem nitrogenof 13 (kg of N).(km2 of reef)-1 and a maximum of 48 (kg of N).(km2 ofreef)-1 during the first three weeks following mortality. These figuresassume that an equal proportion of each coral growth formdied duringthe bleaching event. This is however an unlikely scenario. Typically,the branching species are more susceptible to bleaching mortality(Loya et al., 2001). For the above scenario, assuming the mortalitieswere comprised wholly of branching (open and complex) species thedaily rates increase to 18 and66 (kg ofN).(km2 of reef)-1. This same areamayalso be said to have lost 2115 (103 kg).(km2 of reef)-1 of coral tissuebiomass and the daily calcification rate was reduced by 69 (103 kg).(km2 of reef)-1day-1 as a result of the bleaching induced mortality.
5. Conclusions
As noted by Edmunds and Gates (2002), care must be taken whenusing size normalised parameters in scaling and that it should also beremembered that the errors associated with the initial measurementare also scaled accordingly. The relationships presented in Table 2 arenot intended to present a highly accurate approach to extrapolatingmicro processes. They do however allow for reasonable estimates andprovide a basis for applying physiological measurements of coralstaken at the sub colony scale to larger areas of reef. The issue ofresolution however must not be dismissed. Processes that occur onvery small scales where knowledge of polyp/corallite structure is re-quired may be underestimated by more than an order of magnitude ifapplied directly to the relationships presented in Table 2.
There currently exists a wide range of remote sensing tools avail-able for the assessment of coral reefs (Mumby et al., 2004; Andrefouetand Riegl, 2004) and with ongoing development and reducing costs,these are opening the door ever wider for large-scale assessments ofcoral reef environments. While there are clearly still many hinders inapplying the SI methodology to remote sensing data sets, as the abilityof remote sensing technology to classify coral reef benthos increases,the relationships presented in Table 2 may potentially be used tocalculate ecological parameters over entire reef systems. For example,integrating a range of remote sensing tools, such as LIDAR to deter-mine benthic rugosity (Brock et al., 2004) and hyperspectral imageryto classify the benthos (Karpouzli et al., 2004), researchers may soonbe able to make estimations of coral reef parameters on a scale notpreviously possible.
Acknowledgments
I would like to thank Ron Johnstone, Olga Pantos, Lazaros Kastanisand Ove Hoegh-Guldberg from The University of Queensland, andMerrick Ekins from the Queensland Museum for their assistance inproviding support and materials for the completion of this study. [SS]
References
Acosta, A.C., Robertson, D.N., 2002. Diversity in coral reef fish communities: the effectsof habitat patchiness revisited. Mar. Ecol. Prog. Ser. 227, 87–96.
Alcala, M.L.R., Vogt, H., 1997. Approximation of coral reef surfaces using standardizedgrowth forms and video counts. Proc. 8th Int. Coral Reef Symp., vol. 2, pp.1453–1458.
Andrefouet, S., Payri, C., 2001. Scaling-up carbon and carbonate metabolism of coralreefs using in-situ data and remote sensing. Coral Reefs 19, 259–269.
Andrefouet, S., Riegl, B., 2004. Remote sensing: a key tool for interdisciplinaryassessment of coral reef processes. Coral Reefs 23, 1–4.
Anthony, K.R.N., Hoegh-Guldberg, O., 2003. Variation in coral photosynthesis, respira-tion and growth characteristics in contrasting light microhabitats: an analogue toplants in forest gaps and understoreys? Funct. Ecol. 17, 246–259.
Please cite this article as: Holmes, G., Estimating three-dimensional surfjembe.2008.07.045
Brock, J.C., Wright, C.W., Clayton, T.D., Nayegandhi, A., 2004. LIDAR optical rugosity ofcoral reefs in Biscayne National Park, Florida. Coral Reefs 23, 48–59.
Brock, J.C., Yates, K.K., Halley, R.B., Kuffner, I.B., Wayne Wright, C., Hatcher, B.G., 2006.Northern Florida reef tract benthic metabolism scaled by remote sensing. Mar. Ecol.Prog. Ser. 312, 123–139.
Bythell, J.C., Pan, P., Lee, J., 2001. Three-dimensional morphometric measurements ofreef corals using underwater photogrammetry techniques. Coral Reefs 20, 193–199.
Chancerelle, Y., 2000. Méthodes d' estimation des surfaces développées de corauxscléractiniaires à l'échelle d'une colonie ou d'un peuplement. Oceanol. Acta 23,211–219.
Cocito, S., Sgorbini, S., Peirano, A., Valle, M., 2003. 3-D reconstruction of biologicalobjects using underwater video technique and image processing. J. Exp. Mar. Biol.Ecol. 297, 57–70.
Courtney, L.A., Fisher, W.S., Raimondo, S., Oliver, L.M., Davis, W.P., 2007. Estimating 3-dimensional colony surface area of field corals. J. Exp. Mar. Biol. Ecol. 351, 234–242.
Dahl, A.L., 1973. Surface-Area in Ecological Analysis - Quantification of Benthic Coral-Reef Algae. Mar. Biol 23, 239–249.
Davey, M., Holmes, G., Johnstone, R., 2008. High rates of nitrogen fixation (acetylenereduction) on coral skeletons following bleaching mortality. Coral Reefs 27,227–236.
Dulvy, N.K., Mitchell, R.E., Watson, D., Sweeting, C.J., Polunin, N.V.C., 2002. Scale-dependant control of motile epifaunal community structure along a coral reeffishing gradient. J. Exp. Mar. Biol. Ecol. 278, 1–29.
Edmunds, P.J., Gates, R.D., 2002. Normalizing physiological data for scleractinian corals.Coral Reefs 21, 193–197.
Edwards, A.J., Clark, S., Zahir, H., Rajasuriya, A., Naseer, A., Rubens, J., 2001. CoralBleaching and Mortality on Artificial and Natural Reefs in Maldives in 1998, SeaSurface Temperature Anomalies and Initial Recovery. Mar. Pollut. Bull. 42, 7–15.
English, S., Wilkinson, C.R., Baker, V., 1997. Survey Manual for Tropical MarineResources, 2nd ed. Australian Institute of Marine Science.
Fisher, W., Davis, W., Quarles, R., Patrick, J., Campbell, J., Harris, P., Hemmer, B., Parsons,M., 2007. Characterizing Coral Condition Using Estimates of Three-dimensionalColony Surface Area. Environ. Monit. Assess 125, 347–360.
Fitt, W.K., Mcfarland, F.K., Warner, M.E., Chilcoat, G.C., 2000. Seasonal patterns of tissuebiomass and densities of symbiotic dinoflagellates in reef corals and relation tocoral bleaching. Limnol. Oceanogr. 45, 677–685.
Friedlander, A.M., Brown, E.K., Jokiel, P.L., Smith, W.R., Rodgers, K.S., 2003. Effectsof habitat, wave exposure, and marine protected area status on coral reef fishassemblages in the Hawaiian archipelago. Coral Reefs 22, 291–305.
Garpe, K.C., Öhman, M.C., 2003. Coral and fish distribution patterns in Mafia IslandMarine Park, Tanzania: fish–habitat interactions. Hydrobiologia 498, 191–211.
Grober-Dunsmore, R., Frazer, T.K., Lindberg, W.J., Beets, J., 2007. Reef fish and habitatrelationships in a Caribbean seascape: the importance of reef context. Coral Reefs26, 201–216.
Hoegh-Guldberg, O., 1988. A Method for Determining the Surface-Area of Corals. CoralReefs 7, 113–116.
Holmes, G., Ortiz, J., Kaniewska, P., Johnstone, R., in press. Using three-dimensionalsurface area to compare the growth of two Pocilloporid coral species. MarineBiology. doi:10.1007/s00227-008-1040-x.
Hood, G.M., 2005. PopTools version 2.6.7. CSIRO.Houlbreque, F., Tambutte, E., Ferrier-Pages, C., 2003. Effect of zooplankton availability on
the rates of photosynthesis, and tissue and skeletal growth in the scleractinian coralStylophora pistillata. J. Exp. Mar. Biol. Ecol. 296, 145–166.
Hughes, T.P., et al., 2003. Climate change, human impacts, and the resilience of coralreefs. Science 301, 929–933.
Johannes, R.E., et al., 1972. Metabolism of Some Coral Reef Communities - Team Studyof Nutrient and Energy Flux at Eniwetok. Bioscience 22, 541–543.
Kaandorp, J.A., Kuebler, J.E., 2001. The algorithmic beauty of seaweeds, sponges andcorals. Springer-Verlag.
Kaandorp, J.A., Sloot, P.M.A., Merks, R.M.H., Bak, R.P.M., Vermeij, M.J.A., Maier, C., 2005.Morphogenesis of the branching reef coral Madracis mirabilis. Proc. R. Soc. Lond.Ser. B. 272, 127–133.
Kanwisher, J.W., Wainwright, S.A., 1967. Oxygen Balance in Some Reef Corals. Biol. Bull.133, 378–390.
Karpouzli, E., Malthus, T.J., Place, C.J., 2004. Hyperspectral discrimination of coral reefbenthic communities in the western Caribbean. Coral Reefs 23, 141–151.
Kawaguti, S., 1937. On the physiology of reef corals. 1. On the oxygen exchanges ofreef corals. Palao Tropical Biology Station Study, vol. 1, pp. 187–198.
Kinsey, D.W., 1985. Metabolism, calcification and carbon production I: System levelstudies. Proc. 5th Int. Coral Reef Congress, pp. 505–526.
Klumpp, D.W.,Mckinnon, A.D.,1989. Temporal and Spatial Patterns in Primary Productionof a Coral-Reef Epilithic Algal Community. J. Exp. Mar. Biol. Ecol. 131, 1–22.
Klumpp, D.W., Polunin, N.V.C., 1989. Partitioning among grazers of food resourceswithin damselfish territories on a coral reef. J. Exp. Mar. Biol. Ecol. 125, 145–169.
Lough, J.M., Barnes, D.J., 2000. Environmental controls on growth of the massive coralPorites. J. Exp. Mar. Biol. Ecol. 245, 225–243.
Loya, Y., Sakai, K., Yamazato, K., Nakano, Y., Sambali, H., Van Woesik, R., 2001. CoralBleaching: the winners and the losers. Ecol. Lett. 4, 122–131.
Marsh, J.A., 1970. Primary Productivity of Reef-Building Calcareous Red Algae. Ecology51, 255–263.
Marshall, A.T., 1996. Calcification in hermatypic and ahermatypic corals. Science 271,637–639.
Mcclanahan, T.R., 1994. Kenyan coral reef lagoon fish: effects of fishing, substratecomplexity, and sea urchins. Coral Reefs 13, 231–241.
Merks, R.M.H., Hoekstra, A.G., Kaandorp, J.A., Sloot, P.M.A., 2004. Polyp orientedmodelling of coral growth. J. Theor. Biol. 228, 559–576.
ace areas on coral reefs, J. Exp. Mar. Biol. Ecol. (2008), doi:10.1016/j.
7G. Holmes / Journal of Experimental Marine Biology and Ecology xxx (2008) xxx–xxx
ARTICLE IN PRESS
Meyer, J.L., Schultz, E.T., 1985. Tissue Condition and Growth-Rate of Corals Associatedwith Schooling Fish. Limnol. Oceanogr. 30, 157–166.
Mumby, P.J., et al., 2004. Remote sensing of coral reefs and their physical environment.Mar. Pollut. Bull. 48, 219–228.
Muscatine, L., Falkowski, P.G., Dubinsky, Z., Cook, P.A., Mccloskey, L.R., 1989. The Effectof External Nutrient Resources on the Population-Dynamics of Zooxanthellae in aReef Coral. Proc. R. Soc. Lond. Ser. B. 236, 311–324.
Odum, H.T., Odum, E.P., 1955. Trophic Structure and Productivity of a Windward CoralReef Community on Eniwetok Atoll. Ecol. Monogr. 25, 291–320.
Paerl, H.W., Steppe, T.F., 2003. Scaling up: the next challenge in environmentalmicrobiology. Environ. Microbiol. 5, 1025–1038.
Rinkevich, B., Loya, Y., 1984. Does light enhance calcification in hermatypic corals? Mar.Biol. V80, 1–6.
Rogers, C.S., Garrison, G., Grober, R., Hillis, Z., Franke, M.A., 1994. Coral Reef MonitoringManual for the Caribbean and Western Atlantic. National Parks Service - VirginIslands National Park.
Sorokin, Y.I., 1995. Coral Reef Ecology. Springer-Verlag.Stimson, J., Kinzie, R.A., 1991. The temporal pattern and rate of release of zooxanthellae
from the reef coral Pocillopora damicornis (Linnaeus) under nitrogen-enrichmentand control conditions. J. Exp. Mar. Biol. Ecol. 153, 63–74.
Please cite this article as: Holmes, G., Estimating three-dimensional surfjembe.2008.07.045
Thieberger, Y., 1995. A Novel, Nondestructive Bioassay for Assessing Areal Chlorophyll-ain Hermatypic Cnidarians. Limnol. Oceanogr. 40, 1166–1173.
Veron, J.E.N., 2000. Corals of the world. Australian Institute of Marine Science.Webb, K.L., Dupaul, W.D., Wiebe, W., Sottile, W., Johannes, R.E., 1975. Enewetak
(Eniwetok) Atoll: Aspects of the Nitrogen Cycle on a Coral Reef. Limnol. Oceanogr.20, 198–210.
Wilkinson, C.R., 2004. Status of the coral reefs of the world: 2004. Australian Instituteof Marine Science.
Wilkinson, C.R., Hill, J., 2004. Methods for Ecological Monitoring of Coral Reefs.Australian Institute of Marine Science.
Wilkinson, C.R., Williams, D.M., Sammarco, P.W., Hogg, R.W., Trott, L.A., 1984. Ratesof Nitrogen-Fixation on Coral Reefs across the Continental-Shelf of the CentralGreat Barrier-Reef. Mar. Biol. 80, 255–262.
Yates, K.K., Halley, R.B., 2003. Measuring coral reef community metabolism using newbenthic chamber technology. Coral Reefs 22, 247–255.
Yonge, C.M., Yonge, M.J., Nicholls, A.G., 1932. Studies on the physiology of corals. VI. Therelationship between respiration in corals and the production of oxygen by theirzooxanthellae. Scientific Report - Great Barrier Reef Expedition 1928-29, pp. 213–251.
ace areas on coral reefs, J. Exp. Mar. Biol. Ecol. (2008), doi:10.1016/j.