A Rapid Ecological Assessment (REA) Quantitative …...A Rapid Ecological Assessment (REA)...
Transcript of A Rapid Ecological Assessment (REA) Quantitative …...A Rapid Ecological Assessment (REA)...
A Rapid Ecological Assessment (REA) Quantitative Survey Methodfor Benthic Algae Using Photoquadrats with Scuba1
Linda B. Preskitt, 2 Peter S. Vroom,3 and Celia M. Smith 2
Abstract: The challenge of assessing seldom-visited, benthic substrates hascreated the need for a method to describe benthic communities quickly and efficiently. Macroscale rapid ecological assessments (REAs) of algal assemblagesprovide managers of coral reefs and other benthic ecosystems with the fundamental descriptive data necessary for continued yearly monitoring studies. Thehigh cost of monitoring marine communities, especially remote sites, coupledwith the time limitations imposed by scuba, require that statistically valid data becollected as quickly as possible. A photoquadrat method using a digital camera,computer software for photographic analysis, and minimal data collection inthe field was compared with the conventional method of point-intersect (grid)quadrats in estimating percentage cover in subtidal benthic communities. Intimed studies, photoquadrats yielded twice the number of quadrats (and an almost infinite number of data points) as conventional methods, provided permanent historical records of each site, and minimized observer bias by having onlyone observer identifying algae in the field. However, photoquadrats requiredmore post-collection computer analyses of digital photographs than conventionalmethods. In the manual method, observer bias in algal identification can occurdepending on the degree of experience of individual divers. On the other hand,photoquadrats rely on one observer in the field and one observer in the laboratory, standardizing algal identification. Overall, photoquadrats do not yield thefiner resolution in diversity that was found using point-intersect quadrats but doprovide a more precise estimate of percentage cover of the abundant species, aswell as establishing a permanent visual record in the time allowed by work withother teams.
EXTENDING OVER 1930 km northwest ofKaua'i in the main Hawaiian Islands is theNorthwestern Hawaiian Islands (NWHI)Coral Reef Ecosystem Reserve: 257,700 km2
of marine reserve that was created by presidential executive order in 2000. Since the
1 This project was supported by the Botany Department, University of Hawai'i at Manoa; Hawai'i CoralReef Initiative (HCRI); and the National Oceanic andAtmospheric Administration. Manuscript accepted 27May 2003.
2 Department of Botany, University of Hawai'j atManoa, 3190 Maile Way, Honolulu, Hawai'i 96822.
3 Coral Reef Ecosystem Investigation, HonoluluLaboratory, National Marine Fisheries Service, 1125BAla Moana Boulevard, Honolulu, Hawai'j 96814.
Pacific Science (2004), vol. 58, no. 2:201-209© 2004 by University of Hawai'i Press
reserve's instatement, the Coral Reef Ecosystem Investigation (U.S. National MarineFisheries Service) and the Division of AquaticResources (Department of Land and NaturalResources, State of Hawai'i) have conductedyearly and biyearly, respectively, assessmentand monitoring research expeditions to thenumerous islands and reefs situated in thisremote part of the Pacific. These trips haveincluded oceanographers and algal, coral, fish,and invertebrate biologists in a cooperativeeffort to quantitatively assess this isolated andrelatively untouched ecosystem. Teams of biologists have conducted surveys to record theflora and fauna using rapid ecological assessment (REA) protocols with scuba. As thename implies time was limited, thus requiring carefully designed methods for quantitative data collection. The goal of the algalsurveys was to describe the algal community
201
202
structure and prepare a comprehensive species list for each site. Working at depths of 3to 16 m with teams using other rapid assessment methods, we recognized the need for amethod of algal assessment that would minimize time in the water yet yield the moststatistically sound data.
Conventional methods using point grid,point-intersect methods, or visual estimationhave been compared with the use of photographic and video quadrats (Bohnsack 1979,Foster et al. 1991, Meese and Tomich 1992,Roberts et al. 1994) with mixed results. Conventional methods yielded more taxa (Bohnsack 1979, Foster et al. 1991) and could bemanipulated to provide data for estimatingpercentage cover at over 100% to account forcanopy effects. However, these methods werefound to be less precise, less accurate (Fosteret al. 1991), and more time consuming inthe field (Bohnsack 1979, Meese and Tomich1992, Dethier et al. 1993, Leonard and Clark1993, Aronson et al. 1994). Photographic andvideo survey methods were found to be fasterand more objective (Bohnsack 1979, Meeseand Tomich 1992, Leonard and Clark 1993,Aronson et al. 1994), to decrease variationamong observers (Meese and Tomich 1992)but be more time consuming in the laboratory (Meese and Tomich 1992, Leonard andClark 1993). Some difficulty in identifyingsmaller individuals was experienced, resultingin a lower taxa resolution (Foster et al. 1991,Roberts et al. 1994). In comparing video withphotography, video was less time consumingand provided a better permanent record, butphotography provided higher resolution foridentifying macroalgae to species (Aronsonet al. 1994). Based upon these studies we initially considered photographic quadrats fortheir timesaving capabilities, better precisionand accuracy, less observer variation, and permanent historical records. However, the twodimensionality of the photoquadrat methodhas been criticized because canopy effects arenot considered (Foster et al. 1991) and substrate rugosity in coral reef regions makeslaboratory algal identification difficult (Bohnsack 1979, Leonard and Clark 1993, Robertset al. 1994).
With these considerations, a REA survey
PACIFIC SCIENCE· April 2004
method was designed to quantitatively assessalgal cover using photoquadrats, collect herbarium vouchers, and map cryptic species.The objective of this study was to analyze thecost/benefit ratio of the photoquadrat methodversus the conventional point grid quadratmethod often used in field assessments. Thisstudy did not attempt to test statistical applications that may be applied to the data gathered with the REA method, but ANOSIMstatistical analyses with PRIMER softwarewere done to compare the results of the twomethods. The goal of the REA surveys forwhich this method was devised is to gatherquantitative data from which significantchanges in overall community structure andlevels of abundance can be detected. Therefore, ANOSIM, or analysis of similarities, wasused to test for significant differences in thealgal community structure where fluctuationsin abundance of species or species composition might indicate environmental change.This method also provides flexibility for otherstatistical applications that are covered in otherpapers (see Dethier et al. 1993, BenedettiCecchi et al. 1996, Miller and Ambrose 2000;C. Bird, unpubl. data).
MATERIALS AND METHODS
This comparative study was conducted on 29November 2002 at Honaunau Bay, Hawai'iIsland, Hawai'i. The southwest-facing site iswell protected from all but southern swellsand is well known for large coral formationsstarting at 3 m deep and extending well past30 m. The substrate comprises mainly filamentous algal turfs, the corals Porites spp.and Pavona varians, and macroalgae locatedin crevices between coral heads. Parallel toshore, two 25-m transects were set 50 m apartat 10 m deep. To compare efficiency and dataconsistency between the two methods, twotwo-person scuba teams, one team using thephotoquadrat method described in this paperand the other using a conventional grid quadrat method, surveyed both transects. The primary purpose was to compare the number ofquadrats completed in situ in 30 min andthe consistency of the data between the twomethods.
REA Survey Method for Benthic Algae . Preskitt et at. 203
FIGURE 1. PVC photoquadrat framer with dimensions for an Olympus 3030CZ camera in Olympus PT007 housing.
Photoquadrat Method
A digital camera (Olympus 3030CZ) with ahousing (Olympus PT007) was mounted ona O.l8-m2 polyvinyl chloride (PVC) photo-
quadrat framer (Figure 1). The camera wasset at SHQ resolution and forced flash with adigital slave strobe (Ikelite Substrobe DS50)mounted on one leg of the framer. The lowerperimeter of the framer was marked with
204 PACIFIC SCIENCE· April 2004
Sne: TtanseeU Depth: Obs: Date: fndes: f1u...,o", 1II0'_)~Q~1'--------r============lFnQ~6----~"------r========;1~:Ana19aglaneraa
AS: Panttolflcra s~dfera
ID-, [OdunIF-' CaJerl'il11:_: U0l"'l'lmiaIJC UolOa,_: Galaw'aIL-, i1jilreda
~..--------'-1=======~m------------\;=========!;j 1fI-, ~lJ1eaI IA Jaiaa:rerens~ ~ ~-
rJ': Lrbcpluavaieg:la"_: McrodaymPH: A:nieria t-DmemariPI: PeysscnrtJialnCJ'l'\(fl1aPRPeySSO'lrd'larulra1_: la-ga;...,1li: Taj~ocIadagIOTlera.a
h"'----------j=======~DR--------~=======~ U_:UYa~ ~ ~M
(A: G:rclline
:imIlIcs;CR: [or~ IltbleLP: Urregorepa.emEr1so: I""fl(:ll:J:k
hu---~~~~h;._____---~==========~ _:Q4 Q9 E[; Edirahrixalancris
ED: EdlitlJlh"ixdiairnaEM: EdllrorretramatheitM H3erocentrotus
manillaL6TRIi: Tripl'f1JSte5g-<1l11a
fmj~
h"'----------\========#n:u;------------i~======~ K: r1nt~oPt·a05 Q10 K: Mlrtipcrafl<tdlaareP: MlrtlJXIrapa:ulaP£: Paies ccmpressaPD:POOllq:oradcrnlcomisff: PaRes evemmriPI..: PcrkesldJaaPMftrlllq:orarrearrl"lnaP5:Ftrhesrus1'1': l'aJ(fBvcrin
FTGURE 2. Photoquadrat data sheet with space for observational notes for 10 quadrats. Boxes for drawing maps indicating the location of cryptic species are positioned within each larger quadrat box. A list of codes for commongenera and species is on the right-hand side of the data sheet.
quadrat number and centimeter scales toprovide location and scale information ineach photograph for later computer analysis.The quadrat data were recorded on a waterproof data sheet that included space for recording the species found in the quadrat, a"map" area for identifying cryptic species indepressions that might prove difficult in computer analysis, and codes for the most common genera and species of macroalgae, corals,invertebrates, and substrate types to standardize and shorten note taking (Figure 2).
Two trained observers moved along eachtransect, with one placing the framer and operating the camera and the other taking notes
(Figure 3). The photoquadrat framer's shortedge was placed along the transect with thelower left corner of the quadrat positioned atthe meter mark and a picture taken. Beforemoving the photo quadrat, the second diveridentified algae within the photoquadrat, recorded the relative abundance of the fivemost abundant algae on a scale of 1 to 5 (with1 being most abundant), drew a quadrat maplocating species that may be hard to identifyin a photograph, and collected in numberedself-sealing plastic bags representative samples of the algal species in the quadrats fromoutside the framer for later identification inthe laboratory. Once data were recorded, the
REA Survey Method for Benthic Algae . Preskitt et at. 205
FIGURE 3. Kim Peyton and Brooke Stuerke using photoquadrat to measure algal cover, Puako, Hawai'i.
photoquadrat was moved to the next metermark and the procedure repeated for a totalof 30 min. To prevent redundancy, only samples of new algal species found in subsequentquadrats were collected.
The photographs were downloaded to apersonal computer, renamed with a uniquelocation code, then cropped and color autocorrected in image-editing software. Eachphoto was analyzed for percentage coverusing PhotoGrid (c. Bird, Department ofBotany, University of Hawai'i), a softwareprogram capable of random and stratifiedrandom point analysis on digital photography. To be comparable with the conventionalmanual method, 30 random points were generated for the photographic analysis. Once allphotos were analyzed, data were imported
into a standard spreadsheet for further statistical application.
Manual Quadrat Method
To be consistent with the photoquadratmethod, a 0.18-m2 PVC quadrat similar tothe photoquadrat frame was double-strungwith nylon line to form 49 intersection pointsin grids on both sides of the quadrat. Thirtypoints were randomly selected and markedwith colored string on the grid. To ensureobjectivity and compensate for parallax, anobserver lined up the marked intersect onboth sides of the grid and recorded theorganism or substrate under that particulardouble intersect. Data sheets included a tablefor recording the data on each point in each
206
Category
Melanamamia glomerataDictyota ceylanicaLobophora variegataRhipidosiphon javemisTolypiocladia glomerulataCrustose corallineTurfMontipora capitataPavona variamPocillopora meandrinaPorites spp.Unknown coralSandSea urchins
Total no. of categories
PACIFIC SCIENCE· April 2004
TABLE 1
List of Species/Substrate Type and Percentage Cover by Method
% Cover
All 10 First 10 AllManual Quadrats Photoquadrats Photoquadrats
0.7 0.0 0.0070.0 0.0 0.0020.3 0.0 0.0000.0 0.0 0.0021.0 0.035 0.0280.5 0.002 0.003
30.8 0.367 0.3670.2 0.008 0.008
14.5 0.067 0.0710.2 0.002 0.001
51.3 0.515 0.4690.0 0.002 0.0010.0 0.002 0.0400.5 0.002 0.001
10 10 13
Note: ANOSIM testing for similiarity between the two methods resulted in an R value of 0.078% (P < 1.8%).
quadrat and the codes for the most commongenera and species of macroalgae, corals, invertebrates, and substrate types to standardizethe note taking and speed up recordingtime.
Each diver was assigned 15 random intersect points. The two observers swam togetheralong the transect, one placing the lower leftcorner of the quadrat on each meter mark inthe same location that the photoquadrat hadbeen placed. Each observer then recorded thedata of their assigned intersect points andproceeded to the next quadrat, working downthe transect for 30 min.
Data were entered into a standard spreadsheet and transformed into percentage cover.An ANOSIM test using PRIMER software(Clarke and Gorley 2001) was conducted onthe data to determine if there was a significantdifference in similarity between the manualversus photoquadrat methods in the algalcommunity structure.
RESULTS
Before the camera batteries were depleted,the two-person team using digital photoquadrats completed 22 quadrats on transect 1
in 30 min and 16 quadrats on transect 2 in 21min, an average of 1.36 and 1.31 min perquadrat, respectively. The conventional manual method was completed on 10 quadrats oneach transect in the allotted 30 min, averaging3 min per quadrat. Laboratory analysis of thealgal samples and the 38 photographs averaged 8 min per photo to rename, crop, colorcorrect, analyze with random points, identifyaccompanying algal samples, and export datainto a worksheet.
The number of categories (algal species,coral, invertebrate, and substrate type) foundin the two methods varied slighdy (Table 1).Ten categories were identified in the 20quadrats of the grid quadrat method and thefirst 20 photoquadrats. The category list increased to 13 when all 38 photoquadrats wereconsidered. Macroalgal species identified inthe grid quadrat method included Melanamansia glomerata, Lobophora variegata, Tolypiocladia glomerulata, turf and red crustosecoralline algal forms, and the corals Montiporacapitata, Pavona varians, Pocillipora meandrina,and Porites spp. Utilization of all 38 photoquadrats added the algal species Dictyota ceylanica and Rhipidosiphon javensis and a newsubstrate, sand, to the category list.
REA Survey Method for Benthic Algae . Preskitt et at. 207
The percentage cover data from both transects, using all 10 quadrats per transect fromthe manual method and the same first 10 quadrats from the photoquadrat method, wereanalyzed for similarities using PRIMER software. The data were fourth root-transformedand a one-way ANOSIM was conducted. Theanalysis resulted in an R value of 0.078(P < 1.8%). No differences were found inthe community composition or levels of abundance of species between the two methods,or similarities between and within the sites.The near zero R value shows a very highlevel of similarity between the quadrats fromthe two methods.
DISCUSSION
The digital photoquadrat method clearly allowed data to be collected from more quadrats than the conventional point grid quadratmethod. Photoquadrats are not new; theyhave been compared with a myriad of pointintersect, visual estimation, and grid quadratmethods and have often been found wanting in scale and in diversity measurements.This study was an attempt to refine thestandard photoquadrat method by adding atwo-observer team, note taking, collectionof samples, and quadrat mapping to addressknown concerns with cryptic species. Thoughnot used in this comparison of quantitativemethods, the REA protocol includes finishingthe dive with a general algal collection of thesite in which rare, cryptic, and less-abundantspecies not present in the photoquadrats werecollected for the comprehensive species list.The photoquadrat method described heresatisfies the REA protocol requirements andallows algal teams to complete robust sampling measures in the same time periods asthe methods for other dominant benthiccover in the NWHI and the main HawaiianIslands.
Practical advantages of the photoquadratmethod over manual quadrats were numerous. The method produced twice as manyquadrats (and data) that were as statisticallyrigorous as the conventional grid quadratmethod; quadrats located on vertical wallsand pinnacles were more accessible; and little
field training and algal identification experience were required for half of the team. Theframer is easily modified for closer resolutionand smaller organisms, many of which wouldbe too small to sample manually with a smallgrid quadrat underwater; samples collectedprovided more accurate identification in thelaboratory and doubled as voucher samplesfor previously unvisited sites; and the photographs provided permanent records availableto other researchers for a variety of additionalin-depth analyses. Finally, observational biasis lessened with the photoquadrat method.The biyearly trips required six phycologists(three teams of two) in the field simultaneously, necessitating taking less-trained observers for some of the positions. With thisphotoquadrat method, variation among fieldobservers was decreased because only oneteam member needed to be trained in macroalgal identification. In addition, observational bias and error were greatly lessenedwith only one person analyzing digital photographs in the laboratory.
It was clear that limitations exist with thephotoquadrat method. Underwater digitalequipment is costly and susceptible to flooding or breakage, requiring more monetaryinvestment in equipment and backups thanconventional methods. However, such costsare already standard with other organismalassessments. The best photoquadrats weretaken on relatively flat surfaces; with greaterrugosity comes greater difficulty in the identification of species in crevices and holes onthe reef. Adding a digital slave strobe increased the focal point of the camera in theselow light conditions and made these darkareas more interpretable in the photographs.In addition, recording which species weremost abundant in the field helped whenidentifying species on the computer screen:matching the most abundant species fromfield data sheets to the species observed indigital images minimized misidentification.The mapping of quadrats with greater rugosity also increased the level of interpretation ofhard-to-see areas in the photographs. However, even with these added steps, rare, verysmall, and cryptic species were often overlooked in photoquadrats.
208
The two-dimensionality of photoquadratswill always prove costly in areas of algal canopy. Excessive canopy completely concealsthe understory in two-dimensional photography. Removing biomass and taking a series ofpictures (Bohnsack 1979, Foster et al. 1991)is not possible within the REA time limits.However, field notes by divers in areas withdense algal canopies will mark these regionsfor future research under less-stringent timeconstraints. Finally, laboratory analyses ofsamples and photographs can take as much as8 more minutes per quadrat.
The most noticeable advantage of photoquadrats over grid quadrats was flexibility instatistical analysis. Grid quadrats are static:the number of random points are fixed, usually with too few points due to time limitations, and sometimes the observer mustquickly identify algae underwater in less thanideal conditions. In comparison, the photoquadrat method is more flexible in laboratoryanalysis. The method of data collection onthe computer can vary to accommodate investigators' questions. New developments inphotographic analysis software (such as PhotoGrid) allow the observer to either determine the style (stratified or not) and numberof random points for the calculation of percentage cover or measure actual area forstatistical analysis. The added flexibility innumber of quadrats, number of points, andnumber of photographic analysis methodsprovides more choices in statistical applications of the photoquadrat data. To maximizestatistical analyses but minimize effort, poweranalyses should be conducted to estimate thesampling effort needed to ensure that thestudy is powerful enough to detect the desired sensitivity to change.
Ecological assessment and monitoringmethods are powered by the purpose of theproject. Iffine taxa resolution and an in-depthdescription of an algal community including canopy effect are desired, then twodimensional photoquadrats will not suffice.However, the objectives of REAs are broader.The goal of an REA is to quantitatively describe an ecosystem at a macro-communitylevel. The photoquadrat method providesadequate quantitative data, analysis flexibility,
PACIFIC SCIENCE· April 2004
and permanent algal specimens that enablethe investigators to determine the patternsin distribution and abundance of macroalgaeand turfs in these remote, inaccessible regions. A standardized REA protocol not onlyprovides the quantitative data needed to establish baselines for these communities butalso ensures that comparable data are collected during the ongoing monitoring neededfor management decisions.
ACKNOWLEDGMENTS
We thank Ryan Okano, Kim Page, Kim Peyton, Cheryl Squair, and Koa Shultz for fieldtesting the photoquadrat; Jen Smith and theMarine Macrophytes seminar group for theircritical comments and encouragement; andChris Bird for his statistical insight and use ofhis PhotoGrid program.
Literature Cited
Aronson, R B., P.]. Edmunds, W. F. Precht,D. W. Swanson, and D. R Levitan. 1994.Large-scale, long-term monitoring of Caribbean coral reefs: Simple, quick, inexpensive techniques. Atoll Res. Bull. 421:1-19.
Benedetti-Cecchi, L., L. Airoldi, M. Abbiati,and F. Cinelli. 1996. Estimating the abundance of benthic invertebrates: A comparison of procedures and variability betweenobservers. Mar. Ecol. Prog. Ser. 138:93101.
Bohnsack, ]. A. 1979. Photographic quantitative sampling of hard-bottom benthiccommunities. Bull. Mar. Sci. 29 (2): 242252.
Clarke, K. R, and R N. Gorley. 2001.PRIMER-E Ltd. Plymouth Marine Laboratory, Plymouth, u.K.
Dethier, M. N., E. S. Graham, S. Cohen, andL. M. Tear. 1993. Visual versus randompoint percent cover estimations: 'Objective' is not always better. Mar. Ecol. Prog.Ser.96:93-100.
Foster, M. S., C. Harrold, and D. D. Hardin.1991. Point vs. quadrat estimates of thecover of sessile marine organisms. ]. Exp.Mar. BioI. Ecol. 146:193-203.
REA Survey Method for Benthic Algae . Preskitt et at. 209
Leonard, G. H., and R. P. Clark. 1993. Pointquadrat versus video transect estimates ofthe cover of benthic red algae. Mar. Ecoi.Prog. Ser. 101:203-208.
Meese, R. J., and P. A. Tomich. 1992. Dotson the rocks: A comparison of percentcover estimation methods. J. Exp. Mar.BioI. Ecoi. 165:59-73.
Miller, A. W., and R. F. Ambrose. 2000.
Sampling patchy distributions: Comparison of sampling designs in rocky intertidal habitats. Mar. Ecoi. Prog. Ser. 196:114.
Roberts, D. E., S. R. Fitzhenry, and S. J.Kennelly. 1994. Quantifying subtidal macrobenthic assemblages on hard substratausing a jump camera method. J. Exp. Mar.BioI. Ecoi. 177:157-170.