A Dynamic Model for Wave-Induced Light Fluctuations in a Kelp Forest Author(s): Stephen R. Wing, James J. Leichter, Mark W. Denny Reviewed work(s): Source: Limnology and Oceanography, Vol. 38, No. 2 (Mar., 1993), pp. 396-407 Published by: American Society of Limnology and Oceanography Stable URL: http://www.jstor.org/stable/2837818 . Accessed: 29/03/2012 21:03 Your use of the JSTOR archive indicates your acceptance of the Terms & Conditions of Use, available at . http://www.jstor.org/page/info/about/policies/terms.jsp JSTOR is a not-for-profit service that helps scholars, researchers, and students discover, use, and build upon a wide range of content in a trusted digital archive. We use information technology and tools to increase productivity and facilitate new forms of scholarship. For more information about JSTOR, please contact [email protected]. American Society of Limnology and Oceanography is collaborating with JSTOR to digitize, preserve and extend access to Limnology and Oceanography. http://www.jstor.org Limnol. Oceanogr., 38(2), 1993, 396-407 ? 1993, by the American Society of Limnologyand Oceanography,Inc. in a lightfluctuations A dynamicmodelforwave-induced kelpforest StephenR. Wing Davis 95616 ofCalifornia, Studies,University DivisionofEnvironmental JamesJ. Leichter 01908 EastPoint,Nahant,Massachusetts University, MarineScienceCenter,Northeastern Mark W. Denny 93950 HopkinsMarineStation,PacificGrove,California University, ofBiologicalSciences,Stanford Department Abstract wave ofwaveheight, to predicttheeffects We formulated a dynamicmodelwithlinearwavetheory canopy.This pyrifera coverin a Macrocystis ofthelocalsurface and waterdepthon modulation length, (PAR) activeradiation ofphotosynthetically modelforattenuation intoa general modelwasincorporated algae.The modelpredicts understory emphasison lightreaching withparticular in a M. pyrifera forest waveperiodandthat ofPAR on thebottommatchesthatofthedominant thattheperiodoffluctuations ofa arealstainduringthetrough ofirradiance peaksis increasedbypositivelocalsurface theamplitude video and surfaceelevationand simultaneous irradiance of instantaneous wave. Field measurements analysison Cross-correlation ofthesepredictions. canopyallowedinvestigation ofthesurface recording betweensurface correlation negative showeda significant elevationmeasurements irradiance andsurface tothedecreaseinlightcannotbe attributed Thiscorrelation theunderstory. elevationandPAR reaching surface cover.Measuredchangesin alone,butcan be explainedbythechangeinM. pyrifera pathlength valuesbya factorofas muchas predicted fractional canopycoverexceededtheoretically instantaneous as of mechanisms, A combination oflightflecksexceededpredictions. theintensities 3. Consequently, and ofpredicted ofthemodel,mayexplainthedeviationbetweenthemagnitudes wellas assumptions in the periodofoceanswellsis typically in canopycoverand light.The dominant measuredfluctuations gainsin light significant rangeof 5-20 s, and lightflasheswiththeseperiodshavebeenshownto effect bycertainalgae. efficiency utilization fluxdensityand frequencycan changethephotosynthetic capacity of understory plants (Pearcy 1988, 1990; Chazdon 1988). For instance,physiologicalresponsesto sun flecksby rain forestunderstoryplantssuggestthatthese use ofhighintensityphoplantsmake efficient active radiation(PAR) of short tosynthetically duration (5-20 s) (Chazdon and Pearcy 1986a,b). Similar resultshave been foundfor some marine algae (Dromgoole 1987, 1988; Greene and Gerard 1990; Wing unpubl. data). Acknowledgments and marinesystems,forests In both terrestrial We acknowledgethe advice and assistance of the folPAR may supporthigher variable highly with Patterson, Mark Quinn, Jim Pearcy, Bob lowingpersons; productivitythanforestswithsimLou Botsford,JimDykes, Laurie Sanderson,JimWatan- understory abe, Rodger Phillips,Emily Bell, JohnLee, and PeterEd- ilar average irradiancebut lower variance. munds. Data on wave parametersfrom MontereyBay, In the marine environmentmuch of the and lightmeasurementsat Pt. Joe were provided by the in PAR is caused by surfacewaves variance MontereyBay Aquarium Research Department. This workwas supportedin partby a Sigma Xi Grant- that can focus PAR (Dera and Gordon 1968; in-Aid of Research,Universityof CaliforniaReserve Sys- Snyderand Dera 1970; Fraser et al. 1980) or tem grant54354-8, and a Universityof CaliforniaJastro- increase its penetrationinto algal stands by Shield Grant-in-AidofResearchto S.R.W., and NSF OCE mechanical disruptionof self-shading(Koehl 90-16721, and NSF OCE 87-16688 to M.W.D. J.J.L.was supportedby a Friend's of Hopkins Marine Station fel- and Alberte1988; Leigh et al. 1987). Consider, forexample, the "forest"formedby the giant lowship. 396 In the understoryof multilayeredforests, transientburstsof intensedirectillumination (knownas sun flecks)occur forseconds to several minutesas gaps in thecanopy line up with the sun (Pearcy 1988). These sun flecksmay compose a largeportionofthedaily irradiance reachingunderstoryspecies in both terrestrial hardwood forests and marine kelp forests (Pearcy 1990; Gerard 1984), and theirphoton Lightflashesin kelpforests kelp Macrocystispyrifera.A large portion of PAR attenuationoccursin thefloatingcanopy, where light penetration decreases exponentiallywith canopy cover (Gerard 1984). ConsequentlyPAR beneath undisturbedM. pyrifera canopies may be reduced to only 1-5% of surface irradiance (Reed and Foster 1984). However,undernormalconditionsthecanopy is constantlydisturbedby surfacewaves and PAR may be highlyvariable in theunderstory. We have noticed that photon fluxdensitybeneath the canopy varies on a scale of minutes as largepatches of canopy driftback and forth with the arrival of sets of waves (surfbeat). Lightintensityalso varies on a scale ofseconds as the distributionof gaps between floating algal blades varies due to the action of individual surfacewaves. These latterfluctuations occur over a frequencyrange(1-0.05 Hz) that may affectgains in the efficiency of lightfleck utilization of understorymacroalgae (Dromgoole 1987; Greene and Gerard 1990). We modeled the effectsof surfacewaves on light levels under floatingalgal canopies and compared the predictionsof the model with data collected beneath a M. pyriferacanopy and froma nearbyarea withoutkelp in Monterey Bay, California. A dynamic model of PAR in an understory kelp environment Instantaneousphotonfluxdensityat a point below a floatingmacroalgal canopy is a function of fourfactors:ambientirradianceabove the air-waterinterface,penetrationof the water's surface,attenuationof lightby algal tissues at thesurface,and attenuationwithdepth. Lightpenetrationof a flatwatersurfacedepends on ambient levels of directdownward irradiance Ed(sun), diffusedownward irradiance Ed(sky),and Ts,the transmittance forthe totaldownwardirradiancethroughthesurface: Ed(O) = [Ed(sun) + Ed(sky)]Ts. (1) Here Ed(O) is downwardirradiancejust below m-2 s-'). the water's surface(,MEinst Transmittanceof lightthroughthe surface (Ts) is a functionofalbedo (definedas theratio of upward radiation fluxto the amount of radiation incident on a surface.)In most cases reflectedradiation dominates the upward radiationflux(i.e. albedo 1 p) (Jerlov1976). The 397 albedo of a smooth ocean surfacevaries with sun zenith angle from0.04 at zenith angle of 00 (solar noon) to 0.28 at zenith angle of 800 (Holmes 1957). If the water's surfaceis not flat,Eq. 1 is not strictlyaccurate.Typical surfacewaves, suchas thosewe measuredin MontereyBay, may influencealbedo at highzenith angles and albedo may increase as much as 10% on a choppyday withwhitecaps (Lobban et al. 1981). However, whitecapsare not common in dense kelp beds as shortperiod choppy waves are damped by floatingalgal canopies. Transmittance by light-absorbingobjects floatingat the water's surfacedepends on the percentcover of the objects and theirabsorption properties: (2) TC= (1 -bA) where Tc is transmittancethroughthe canopy, b the fractionalcanopy cover, and A absorptance by the canopy. Absorptanceof singleM. pyriferablades has been measured as 66% of incidentlight(Gerard 1984). In addition, attenuance may be enhanced by local reflection and backscatteringby plant tissue in the canopy (Gerard 1984). Following penetrationof the water surface and algal canopy, PAR decreases exponentially in water with uniformparticulateconcentration(Jerlov 1976): Ed(Z) = Ed(O)exp(-zkd) (3) where Ed(z) is irradiance at depth z (AEinst m-2 s-') and kdthe verticalirradianceattenuation coefficientfor zenith sun (m-l). This equation assumes thatattenuationacrossdepth stratais dominated by absorptionand thereis only a small percentageof backscatterin the total radiation flux (Jerlov 1976). Over the range of PAR, extinctionis usually highestin the top fewmetersof waterand fallsprogressively lower with depth as high-energy wavelengthsare attenuated.Transmittancethrough the water column (Tm) is represented by exp(-zkd). Seasonal changes in PAR attenuation throughthe water column in M. pyriferaforests are correlatedwith changes in sediment flux(Dean 1985). However, attenuationmay also be influencedbyverticalM. pyrifera fronds in the water column, particularlyclose to the uprightportionof the fronds(Gerard 1984). MultiplyingEq. 1, 2, and 3, we can approx- 398 Wing et al. imate steadystateirradiancefora pointbelow the surfaceas a functionof depth: Ed(z) = [Ed(sun) + Ed(sky)]TsTcTw. (4) The parametersEd(sun), Ed(sky), Ts (as a functionofsolar elevation),and kd(in Tw)vary over long periods of time relativeto the wave period. For example Ed(sun), Ed(sky),and T, have componentsthatchangein a predictable fashion over a diel cycle, and kd may vary betweendays. However, these factorsas well as waterdepth (z), alignmentof surfacefronds with the sun, and fractionalcanopy cover (b) may change over shorterperiods as well, for example as waves pass throughthe canopy. These short period fluctuationsconspire to produce a highlycomplex and variable light environmentin the understoryof kelp forests. Major components of this variance may be correlated,however, with wave period and amplitude. Two wave-induced mechanisms for irradiance fluctuationsare due solely to the optical propertiesof seawater. First, path lengthof directlightchangesover the period of a wave (the path lengthis long under a crest, short under a trough),producing fluctuationsthat are 1800 out of phase with the water's surface elevation when the sun is directlyoverhead. The magnitudeof thesefluctuationsis directly relatedto attenuanceof thewatercolumn (Nikolayev et al. 1971). Second, waves may act as lenses that focus light at some depth dependenton wave height,wave length,and the zenithangleofthesun (Snyderand Dera 1970). The combination of these two mechanisms produces a spectrumof lightfluctuationswith two conspicuous peaks at shallowdepths(<21 m) (Nikolayev et al. 1971). The firstpeak is correlatedwiththe dominant wave frequency and is caused by the fluctuatinglight path length.The second peak, at higherfrequency, is apparentlydue to lens effectsfromsurface waves (Nikolayev et al. 1971; Nikolayev and Khulapov 1976). Over therangeofdepthsconsidered here (5-1 0 m), these two frequency peaks are distinguishable. Two othermechanismsdrivingshortperiod irradiancefluctuationsare producedby movement of kelp fronds.First,on cloudless days clumpsofblades mayblockdirectlightingfrom the sun. When waves move the canopy as a whole, irradiance on the bottom varies accordingto the spatial distributionoffrondson the surfaceand alignmentof clumps of blades withthesun.Under dense canopies,whenthere are only small gaps betweenblades, we expect that wave-induced translationof the canopy will produce lightfluctuationsof a frequency much higher than the dominant wave frequency. Second, as a wave passes throughthe canopy,we expectb to varyas the water'ssurface is stretchedand compressed, producing lightfluctuationsat the same frequencyas the waves. In this study,we concentrateon the second mechanism of variation in canopy cover as a source of fluctuationin subcanopy PAR. We model the variationin kelp canopy cover with a hypotheticalcanopy consistingof a uniform distributionoflight-absorbing fronds.Initially frondsfloaton a still water surface,to which we then impose a sinusoidal wave form. A wave crest is an area where water is pushed togetherand thereforeup; converselya trough is an area wherewateris pulled aside and the surfaceis lowered locally. As a resultof these wave-induced motions of the surfacewater, floatingfrondsare pushed togethernear the crestsand pulled apart in the troughs.In our calculationswave-inducedmovementoffronds is approximatedby assuming that frondsare freeto move withthewaterin whichtheyfloat, thus the horizontaland verticaldisplacement of frondsis modeled by the horizontal and verticaldisplacementof the surfacewater. The equations of linearwave theorycan be used to predictvertical(z) and horizontal(x-y plane) particledisplacementduringpassage of a wave,givena setofcrucialassumptionsabout the nature of the wave form and the fluid throughwhich it moves (Denny 1988). We assume thatwave heightH is small relativeto the wave lengthL, that L is large enough so thatcapillaryeffects are negligible(L > 0.1 m), thatall variationin surfaceelevation is in the plane of wave propagation(x-z) (i.e. variance in surfaceelevation along the y-axis, perpendicular to propagation,is negligible),and that watermotionis irrotational.As definedby linear wave theory,instantaneoushorizontal(x), and vertical(z) displacementsfora particleat thewatersurfacemovingalong its orbitalpath are Lightflashesin kelpforests horizontaldisplacement 2 tanH() 01 sin(kx - wt), - (5) and verticaldisplacement H cos(kx 2 = - - 399 0~~~~~~~~0 .e wt) (6) D:L =0.01 0 05 0.25 0. 0.75P 0 -0 051 - - - __ < surface elevation 05 - -- whered is the depth of waterbelow stillwater level (in m), k = 2w,L, w = 2,,l/T,and T is the wave period. Consider a small rectangulararea of thewatersurfacewithlength(1) along thex-axis and width (w) along the y-axis. As a wave crest passes, the lengthand thereforethe area of the rectangledecreases. As a wave troughpasses, the length(and area) expand. Because an ideal wave has no components of velocityparallel to the wave crest,the width of the rectangle remainsconstant.The magnitudeofexpansion and contractionof the lengthdepends on the strain(ex)imposed on local watersurfacearea duringa wave cycle. Now envisiontwopointson theundisturbed watersurface,one at x = 0 and the otherat a distanceD along the x-axis. In the presenceof surfacewaves thedistanceT betweenthepoints along the x-axis varies as points are moved horizontallyby waves. Equation 5 gives the instantaneoushorizontaldisplacementsforthe two points x = 0 and D. Thus at any time t D = m ~~~~~~~~~~~ - ~~~~~-0.5 - - - -- - - -1 0 n/2 n Phase (ot) 3n/2 2n Fig. 1. Relationship betweensurfaceareal strain(e) and surface elevationas predicted (dimensionless) byEq. 9. w(radianwavefrequency) multiplied bytime(t) gives thephaseofa waveforminradians.Amplitude andphase ofmaximumstrainis dependent on theratioofgap distance(D) and local wave length(L). For thesimulation, H = 1.0 m, L = 100 m, and d = 10 m. Surfaceelevation is represented bythedashedline. When D is small, engineer'sstrainplotted against time varies sinusoidally throughthe period of the wave, withthe maximum in the troughwhen particlesare below theirstillwater position and the minimum strain at the wave crest as the particles are above it (Fig. 1). Maximum strainincreaseswithH and decreases with an increase in L. Note thatthewavelengthis a monotonically increasingfunctionof wave period: L = gT2 -y tanh(kd) (10) so that straindecreases withwave period. The magnitudeof strain(e,) is sensitiveprimarilyto changes in H and water depth (Fig. 2). Note also the dependence of strainon the [2 tanh(kd) sin(-wt)]. scale of D/L, evident fromFig. 1. The larger The change in distance between the points is the D one chooses, the smaller the average theirinstantaneousdistanceT minus the orig- strainover that lengthas surfaceswithdifferinal distance D. Thus, notingthat sin(-wt) = ent phases are added together.Note also that -sin(wt) and rearrangingterms, thephase at whichthe maximumstrainoccurs depends on the scale of D/L. If D/L is small, =-H but not zero, there is a slightphase shiftin =2tanh(kd) [sin(kD - cot)+ sin(cot)].(8) maximum strainaway from 1800. To take an Engineer'sstrain(Ex),the fractionalstretching extremeexample,ifD = L thetwopointsmove of the water'ssurfacein thex-direction,is de- in phase and Ec = 0. For kelp canopies, D is a measure of the distance between clumps of finedas At/D, so functionallyunconnected blades along the -H -Fsin(cot)].(9) x-axis. This distance is usually small relative ,Ex= 2D_n(d [sin(kD - wot) to wave length(D/L < 0.01). Thereforein our [2 tanh(kd) sin(kD-t) = + (7) Wing et al. 400 model the strainis estimatedas D approaches 0. Noting by algebraicidentitythat Effect of varying wave height, H 0.15- H = 2.0 0.1 sin(kD 0.05 0. and thatas D approaches 0, sin(kD) approaches kD and cos(kD) approaches 1, forsmall D, sin(kD - wt) ; kDcos(wt) - sin(wt). -0.05 ' -0.1 -0.15 0 x/2 Phase (ot) Therefore 3i/2' 2xK 0.1 0.05 d = 5.0 10.0 20.0 40.0 EXDfx,- -irH an(d = -0.05 7 =-cos(-wt). 2 x/2 X Phase (ot) 3m/2 2x 0.1 0.05 - (12) L = 50 100 200 400 L tanh(kd) -0.1 -0.15 - . . . . . . . . . . . . 2x 3n/2 X 0 x/2 Phase (Ot) arealstrain(,E)(dioflocal surface Fig.2. Sensitivity fromEq. 9 to changesin wave height(H) mensionless) (m), waterdepth(d) (m), and wave length(L) (m). , (radian bytime(t) givesthephaseof multiplied wavefrequency) tochanges a waveforminradians.Strainis mostsensitive in H, followedbyd, and L. Normalvalues(exceptwhen changed)areH = 1.0 m,d = 10.0 m,and L = 100.0m. (14) In otherwords,the strainis 1800out of phase with the surfaceelevation, as we expect, and the extremelocal strainis 0-0.05- (13) in Thus, solvingforcos(-wt) and substituting Eq. 11, Effect of varying wave length, L 0.15 - cos(- wt). L tanh(kd) From Eq. 6, we calculatethatat x = 0 (where we calculate the strain) the surfaceelevation forthe wave is 0- -0.15 0 -Hk_ o(t. ( 2 tanh(kD) (11) (w) Notingthatk = 27r/L(definition)and cos(wt) = cos(-wt), we arriveat an expressionforstrain as D approaches zero: X,DO- Effect of varying water depth, d 0.15- wt) = sin(kD)cos(wt) - cos(kD)sin(wt), - irH L tanh(kd) Ex,max (15) Note also thatthetotalchangein strainthrough the wave period is 27rH L tanh(kd) x,max (6 (16) In deep water,where d > L/2, tanh(kd) z 1, so 2irH Ex, max L (17) In shallowwater,whered < L/20, tanh(kd) 2ird/L,so A'Exa H -(18) Lightflashesin kelpforests If we assume thatblades of the algal canopy are uniformlight-blockingobjects that never overlap,the area of surfacecanopy is constant while the area of the water's surfacechanges with time, dependent on local wave-induced strain.As a result,local fractionalcanopy cover (b) varies withtime. Instantaneousb is thus a functionof surfacestrain: b(t) (1 Eo) (19) where b is a functionof time t, bois b at still water,and (1 + Ex)the extensionratio. It now becomes clear thatthe ratio of direct [Ed(sun)]to diffuse[Ed(sky)]radiationincident on the water surfaceis importantforthe appropriatechoice of an equation for strain in Eq. 19. On a clear day where directradiation is the dominant componentof PAR fromthe surfacemost of the PAR reachingan understorykelp comes fromone particularangle(i.e. effectiveD is small), and Eq. 12 is an appropriate estimate forthe strainused in Eq. 19. However, if the PAR reachingan understory kelp is derivedfroma largearea of the surface, such as on a cloudyday whendiffuseradiation is the dominantcomponentof PAR, and D is effectively increased,the effectof local canopy strain on irradiance at the bottom will be increased. In this smaller, as D is effectively case, D is a measure of the total gap distance along the x-axis fromwhich lightcomes, and Eq. 9 allows fora more appropriateestimate of average surfaceareal strain over this distance. A fundamentalsimplificationin our model is that blades of kelp do not overlap and are freeto move withthe surfacewater.In reality, blades can overlap when the surfaceis compressed,and theirmotion can be constrained by theirstipes,which are tetheredto the bottom. However, our video recordstakenin the fieldshow thatstipesare typicallylong enough to allow surfacefrondsto "go withthe flow," and thatthesimplifications made in our model are appropriateforpresentpurposes.Note that in very shallow water horizontal displacementsin waves are magnified(Eq. 5) and waves may break. Under these conditions, the assumptionofindependentmotionbythefronds is clearlyviolated and our model is inappropriate. Field measurementsof instantaneouspho- 401 ton fluxdensity,surfaceelevation,and simultaneous video recordingof the surfacecanopy allowed us to compare the generalpredictions ofour model againstactual wave-inducedlight variation beneath a M. pyriferacanopy. Methods We measured PAR under M. pyriferacanopies to determinetherelativeinfluenceofsurfacefronds(Eq. 2) vs. midwaterverticalfronds on PAR attenuationthroughthe water column. PAR was measured at a depth of 13 m at threeadjacent sites (A, B, and C) under a monospecificM. pyrifera canopyat Pt. Joenear Monterey(35036.5'N, 121?57.5'W). The sites wereof similaralgal density(- 0.5 individuals m-2), depth(within0.5 m), and bottomprofile and were separated by no more than 100 m. PAR measurementson the bottom were collected simultaneouslyat each studysite with threeLiCor LI-193SA 4-rlightsensors interfaced withLI- 1000 data loggersin waterproof housings.Site A was cleared of M. pyriferaby removingalgae at theholdfastin a circulararea withradius 20 m around the lightsensor.At siteB, thefloatingsurfacefrondswerecutaway leaving the uprightfrondsand holdfastintact over an area of similar size around a second lightsensor. At site C the canopy was leftuntouched over a third light sensor. PAR was integratedfor0.5-h intervalsat each site fora period of 2 weeks. These data were then used to partitionthe effectsof uprightand floating algae on attenuationof lightbeneath the canopy. Differencesin irradiance between sites were tested with a Friedman's rank sum test (Sokal and Rohlf 1981). Hours were averaged over the 2-week period and considered as blocks, and sites as the treatmenteffects.For the pairwise tests between sites a sequential Bon Ferronicorrectionprocedurewas used to minimize type 1 statisticalerror(Rice 1989). To demonstratetherelativeamount of variance in irradiancebetween the threesites we calculated the mean C.V. (SD x 100/mean) foreach daylighthour forthe 2-week period. This measure allows us to compare the variance of sites with greatlydifferent mean irradiance. To investigatethepredictionsofour dynamic model oflocal areal strainwe measuredPAR on the bottom under a surfacecanopy of M. pyriferawhile simultaneouslymeasuring the 402 Winget al. height of waves passing throughthe canopy and recordinga video image of the percent cover of the canopy. A 12-bitA/D converter and microprocessorsupporting28 kilobytesof staticRAM (Onset ComputerCorp.) sampled PAR froma LiCor 190SA 2ir irradiancesensor, and hydrostaticpressure (a measure of wave height) from an Omega PX176-50psi pressuretransducerat a frequencyof 10 Hz. The A/D converter-microprocessor, light,and pressuresensorswere housed inside an Ikelite 58 10 strobehousing and the samplingperiod was startedby closing a magneticproximity reed switchthat completes a circuitpowering theA/D converter.Calibrationofthelightmeterwas performedagainsta NBS-traceablecalibrated LiCor LI-193SA 4ir quantum sensor in a collimated beam of light.In collimated lightthese two sensors measure the same irradiance. The pressure transducerwas calibratedagainsta ScubaPro oil-filleddepthgauge. Calibration was performedfor 10 points for both lightand pressuresensorsacross a range of values exceedingthatmeasured in the field and both calibration curves were significant withr2values in excess of 0.97 (lightr2= 0.99, P < 0.0001, pressurer2= 0.98, P < 0.0001). Hydrostaticpressure on the bottom is an accuratepredictorofwave height,but because wave-induced changes in pressureare attenuated withdepth,the pressurerecordmustbe correctedwitha depth-dependentscaling factor to determineactual wave heights(Denny 1988): erage densityof 0.41 individuals m-2 and average drybiomass of 91.5 g m-2. In all cases therewas a dense monospecificsurfacecanopy with> 50% blade cover. Waterclarityafforded 10-12 m of visibility.The instrumentswere positioned during midday (1130-1330) with SCUBA and sampling lasted 20 min at each site (12,000 data points for each sensor per sample). In all cases the skywas freeof clouds duringthemeasurementperiod.To investigate the predictionsof the surfacestrainmodel, we compare measured changes in lightlevels on thebottomwithpredictionsbased on bothpredicted (Eq. 19) and measured changes in surface canopy cover. Predictedlightlevels were calculated withEq. 4 assumingthatthe terms [Ed(sun) + Ed(sky)], T, and T, remain constantover shorttime periods while T, changes with canopy cover accordingto Eq. 2. The studysitesare characterizedby a dense growthof M. pyriferawith a surfacecanopy and numerous understoryspecies including Rhodymenia arborescens,Dictyoneuropsisreticulata, Dictyoneurumcalifornicum,Laminaria dentigera, and Desmarestia ligulata (Harrold et al. 1988). The portionof the shore closest to Hopkins Marine Station is partly shelteredfromocean waves by Pt. Pinos; the area closest to Pt. Pinos is more exposed to open ocean swell (Harrold et al. 1988). We performedcross-correlationanalysis on the surfaceelevation-irradiancedata fromall sites to determinecorrelationbetweensurface elevationand PAR. These analysesenabled us to test our predictions(Fig. 1) that the two series are cyclic with a phase lag of 1800. APmn,mx = -pgH 2[ [cosh(ks)1 (20) o] APmin,max 20 Cross-correlationvalues were computed for 2 [cosh(kd)J lags (r) fromzero to twicethe dominantwave where s is vertical distance fromthe bottom period (T) as (in m) at which the sensor is located, g the gravitationalacceleration (m s-2), and p the Xt ) MAyt (21) MT RXY(r)= densityof seawater(kg m-3). 11 Lightand wave data werecollectedfrom sites underneathM. pyriferacanopies ranging where n equals the number of data points in in depth from 6 to 10 m and situated along the series,ax is the SD of the PAR values, ay the shore fromHopkins Marine Stationto Pt. the SD of the surfaceelevation values, mythe Pinos, 3 km to the north.Lightand pressure mean of the surfaceelevation values, mx the data were also collected fromtwo sites at 8-m mean of the PAR values, and t the time indepth withno M. pyriferapresent.Maximum crementof the data set (Kope and Botsford wave heightsrangedfrom0.34 to 1.28 m. The 1990). The correlationvalues were then testedfor lightand pressuresensors were positioned on the bottom 2-4 m away fromthe nearestM. significanceby computingthe variance of the pyriferaholdfastunder canopies with an av- cross-correlationwith the statistic - Lightflashesin kelpforests t - ' (1 - p!).p (i)p,(i) (22) 403 gram Image 1.29. Frames were digitizedof an 10 m2square patchofcanopy directlyabove the sensorarrayat a frequencyof 0.5 Hz. Each framewas thenimportedinto the image analysis programand convertedto grayscale. An index of percentcover was determinedfrom thegray-scaledimageswhere0% is open water at the surfaceand 100% is total frondcover. The frameswere analyzed with a column-averagingalgorithmthataverages the densityof pixels across a specifiedarea oftheframe.Values for percent cover were then temporally matched with the lightand pressurerecord. where-yis the numberof data pairs (n - r), i varies between 1 - n and n - 1, p,, is the autocorrelationof the PAR values, and p, the autocorrelationofthewave-heightvalues. The test statistic uses the function [(1 - Ii I/ 'y)p.x(i)pYY(i)]to reduce errorimposed on the cross-correlationvalues by positive autocorrelationin each time seriesx and y. Thus the intraseriescorrelationof each data set is accountedforin theteststatistic(Kope and Bots- Results ford 1988). Note thatthis functionrepresents Shading byfloating vs. midwateralgal 1 SE of a Gaussian distributionaround zero. fronds-Averaged hourly irradiancesat each Therefore,in order for the cross-correlation of the threesitesat Pt. Joe were computed for values to be considered significant at the 0.05 level theymust be greaterin magnitudethan the 2-weeksamplingperiod and pairwisecom1.96 timesthe variance computedhere.In the parisonsweremade betweensites.Resultsfrom the Friedman's rank sum tests of A vs. B, B absence of intraseriescorrelationthe expresvs. C, and C vs. A indicate that site C (the sion becomes control,all frondsintact) is significantly differentfromsite B (all surfacefrondsremoved) 1 -. (23) (df = 1, P < 0.005) and fromsite A (surface var[RXY(r)] and subsurfacefrondscompletelyremoved)(df Fast Fourier transformations(FFT) were = 1, P < 0.005). Sites A and B are not signifperformedon theautocorrelation fromeach other(df = 1, P > functionsand icantlydifferent the cross-correlationfunctionto obtain the 0.5). These analyses suggestthat the upright cross-spectral densitybetweenthetwodata sets frondsof M. pyriferahave relativelylittlein(PAR and surfaceelevation) and the spectral fluence on light penetration to the bottom densitywithineach data set (PAR and surface compared with the floatingcanopy fronds.In elevation). This analysis allows one to deter- lightofthisresult,in our model (Eq. 4) we deal mine the frequencyrange of variance that is only withfractionalcover changes in the canpositivelycorrelatedbetweenthetwo data sets opy (b) and ignore the comparatively small as indicated by the cross-correlationanalysis. influenceof uprightfrondson the extinction To furtherour analysis of the strainmodel coefficient of light(kd) in the watercolumn. we examined the correlationsbetween local Averaged irradianceat each hour of the day surfacearea strain,Ex(as indicatedby changes is more variable between days in areas with in percentcover), and irradianceand between kelp than in clear areas. With the exceptionof EXand surfaceelevation. Video records were a dip duringthe 1600-hour record forsite C, collected simultaneouslywith the light and theaveragedC.V. forthe2-weekperiodis high pressurerecordat fivesites to determinetime during all hours of the day for sites C (kelp variance of local surface canopy cover. An intact),intermediateforsite B (surfacefronds 8-mm underwatervideo camera was posi- removed), and consistentlylow in site A tionedon thebottompointingup at thecanopy (cleared of all kelp) (Fig. 3). Cross-correlation and synchronizedwith the light-pressure rec- analysis on wave and light data collected at ord. Frames fromthe video recordwere then two 8-m sites where M. pyriferawas absent digitizedon an Apple Macintosh IIcx through showedno significant correlationbetweenlight a RasterOps 24-bit color framegrabberand and pressure. analyzed forchanges in percentcover of surTestingthepredictions ofthesurfacestrain face frondswiththe NIH image analysis pro- model-Autocorrelation of wave height and 404 Wing et al. 0.9 0.4 A 0.3 0.7 * 00.2 Kelp ~. o01 0.5 CanopyCleared 0.3 Cleared Area -0.3 0.1 800 1000 1200 1400 1600 1800 TimeofDay Fig. 3. Plot of C.V. of PAR (,uEinstmr-2s- ') vs. time of day forthreesites in a Microcystispyriferaforestnear Pt. Joe, MontereyBay. Site A was cleared of all kelp and shows the lowest variationin PAR. Site B was cleared of floatingsurfacefrondsand shows intermediatevariation in PAR. At site C the canopy was leftintact and shows high variation duringmost of the day, except fora drop at 1600 hours forunknownreasons. Each site had a dense canopy of M. pyriferabeforemanipulations. -0.4 0 5 10 15 20 25 Time(s) 30 35 40 Fig. 4. Averagedcross-correlationcoefficients (line C) between surfaceelevation and PAR (,AEinstm-2 s-') on the bottomand averagedautocorrelationcoefficients (line A) of surfaceelevation are - 1800 out of phase. Values were averaged from 11 sites, withdepth rangefrom6 to 10 m, averageMicrocystispyriferadensity0.4 individuals m-2 at each site. The sample size was 12,000 measurementsforeach sensor. Dashed lines indicatethe variance in the cross-correlationstatisticas calculated by Eq. 22 (cross-correlationvalues 1.96 times the variance are significantat the 0.05 level). values of the cross-correlationof wave height and PAR [withlag (T) from0 to twicethewave period (I)] are cyclicand 1800 out of phase inverse relationship with the light record. with each other. The peak cross-correlation However, the amplitude of changes in fracvalues ranged from 0.01 to 0.35. Cross-cor- tional canopy cover (b) tended to exceed therelation and autocorrelationaveraged for all oretical predictions.For example, Fig. 5 dis11 samples weredeterminedto be significantly plays a representativesample of data and different fromzero witha triangularweighting correspondingtheoreticalestimatesofchanges function(Kope and Botsford 1990) (Fig. 4). in b. The phase of the changes in measured b FFT analysisoftheautocorrelationoflightand correspondswiththe theoreticalpredictionof pressure and cross-correlationbetween light Eq. 14-19. While the range of measured inand pressure produce spectral density and stantaneouscanopy cover around the mean of cross-spectral densityfunctionsofthedata. The 0.46 is 0.34-0.57, the predictedrangeis 0.42spectraldensityresultsshow thatthe variance 0.49. The rangeoflightlevels aroundthemean in the lightrecordthat is correlatedwith sur- of 54.7 ,uEinstm-2 s-1 is 36-111, with tranface elevation occurs at the same frequencyas sient peaks occurringon a frequencyrangeof the dominant wave period. 5-10 Hz. When these highfrequenciesare filVideo images of the surfaceenabled us to tered out with a 5-s moving average (dashed measure an index of percentcover at fivesites line in Fig. 5), the range is 41.5-72.2 ,uEinst and to synchronizethis record with the light m-2 s-1. The range of predicted light levels and pressurerecord.We hypothesizethat the calculated fromthe predictedchanges in canindex of percentcover is an inverse measure opy cover is 54.0-55.6 ,uEinstm-2 s-1, while of the engineersstrain(,E) at the water's sur- the range of predicted light levels from the face. Afterimage analysis the data forpercent measured changes in canopy cover is 52.3cover were plottedagainst the lightand pres- 57.3 ,uEinstm-2 s-. sure records to explore the relationshipbetween the percentcover of frondson the sur- Discussion face directlyover thelightand pressuresensor The model-Models of lightin M. pyrifera and the lightand pressurerecord.In all cases forests(e.g. Jackson 1987; Nisbet and Bence investigated,thepercentcover index shows an 1989) and empiricalstudies(Dean 1985) have Lightflashesin kelpforests described steady state extinction of light throughthe canopy with depth by means of simple exponential models (e.g. Eq. 3). This generalapproach has been questioned by investigatorsconcernedwith photosynthesisby plants exposed to transientburstsof sunlight (e.g. Pearcy 1990). For example,Gerard(1984) examined the variabilityof irradiance in M. pyriferaforestsand concluded that temporal and spatial variabilityof PAR is extremely high and may influencethe light utilization ofalgae in theseenvironments.Here efficiency we investigatedexplicitmechanismsby which waves moving throughthe canopy may produce fluctuationsin photonfluxdensityon the bottom. The results are most applicable to studies concerned with lightenvironmentin the understoryof kelp forestsand utilization of fluctuating lightby understorykelp species. Analysisof our fielddata supportstwo generalpredictionsofthesurfaceareal strainmodel. Time-seriesanalysis demonstratesthatthe dominantcomponentsoflightfleckperiodbeneath the algal canopy matches the period of waves passing throughthe canopy and that peaks in PAR are correlatedwithwave troughs (Figs. 1, 4, and 5). That no significantcorrelation between lightand pressureexisted for two areas withoutM. pyriferasuggestvariation in lightpath lengthand focusingby waves can account for only a small fractionof the fluctuation in PAR. However, we do not dismiss fluctuating path lengthas a factorcontributing to the total variance in light. Video records demonstrate that gaps between M. pyrifera blades widen in wave troughsand narrowas strain is relaxed over wave crests. However, themagnitudesofmeasuredchangesin canopy cover and of light flecksare greaterin most cases than our model of strain predicts. Althoughthe surfacestrainmodel is supported bythetemporalcorrelationbetweenlocal strain and PAR on the bottom, a combination of mechanisms may explain the observed patternsand the magnitudeof lightflecks. One alternative explanation for the observedtemporalchangesin local percentcover and PAR on the bottom is that wave-driven horizontaltransportofthecanopymoves dense patches of canopy in and out of the reference frameof the video camera or back and forth in frontof the sun. From the predictionsof linearwave theory,one can see thatthe period 405 06 - 05 04 03 ... ,- . . . , 03 04 C 02 12 025 -02 100 -0.4 0 ~~~~~~~~~~~ -. - 0 20 30 Time 40 -- 25 50 (s) Fig. 5. Theoreticalvalues forfractionalcanopy cover (line A, upper panel) are plottedwithfieldmeasurements. sample offractionalcanopycover (line This representative B, upper panel), surfaceelevation (C), and PAR (D, solid line) shown in real time demonstratesthe cyclicnatureof each data series. Dashed-line D shows PAR withhighest frequenciesof variation filteredout with a 5-s moving average.These data weretakenat an 8-m sitenearHopkins Marine Stationon 3 June 1990 at 1330 hours.The sensors were placed within2-3 m of the nearestMicrocystispyriferaholdfastin an area with an average densityof 0.4 individualsm-2. Notice thatthereis a phase shiftbetween the crests of waves and the correspondingpeak in fractional canopy cover (b) and drop in PAR. The theoretical values forb were computed withEq. 14 and 19 withsurface elevations fromcurve C. of horizontalparticledisplacementis matched withthe period of surfacewaves. Consequently,one mightpredictthatPAR on the bottom will fluctuatewith the same period as surface elevation. Considera kelp canopy withb = 0.5 and haphazard placementof frondsin thecanup fromthebottom,there opy.Lookingstraight is an equal probabilitythat a gap, or a clump of fronds,will line up with the sun at solar noon understillwaterconditions.Thereis thus an equal probabilitythatthesunwillbe blocked duringthe crestof a wave or duringthe trough ofa wave as the surfacecanopy is washed back and forth.Ifhorizontaldisplacementofclumps offrondsgovernslightpenetrationthroughthe canopyin thismanner,theinverserelationship between PAR and surface elevation consistentlyobserved in our data sets (Fig. 4) would not be expected. However, random phase lag caused by thismechanismis indistinguishable fromphase lag caused by changes in solar elevation or variance in the D L ratio. Fluctuationsin the path lengthof lightas a 406 Wing et al. wave passes may introduce changes in irra- blade movementundertheinfluenceofsurface diance on thebottomout ofphase withsurface waves could be an interestingfocus of future elevation. For example, in previous studiesof investigations.However, the details of this lightin the ocean (e.g. Nikolayev et al. 1971), problem are likelyto be extremelycomplex. oflightfluctuationsThe biologicaleffects investigatorsfound significantinverse crosscorrelationsbetweenlightand pressureat the Distributionand abundance ofbenthicmarine dominant wave period for the depths with algae are stronglyinfluencedby submarineirwhich we are concerned(5-10 m). Further,in radiance (Luning 1981). Within multilayered a 2-yr study of submarine irradiance at a stands of kelp, understoryspecies may be 13.2-m site adjacent to a southernCalifornia shaded out bydensegrowthsofcanopyspecies, kelp forest,Dean (1985) recordedcoefficients indicating that interspecificcompetition for of extinction(k) rangingfrom -0.2 to 0.5. lightis an importantstructuringmechanism These values will produce changes in irradi- in theseassemblages(Dean and Jacobsen1984; ance at 5-10 m rangingfrom47 to 69% fora Dayton 1985; Dean et al. 1989). Because sur1-m change in lightpath length.In our study face waves predictablyaffectthe delivery of M. pyriferaforestsare we recordedcyclicchangesin irradiancefrom lightto the understory, 50 to 250% forsimilarchangesin depth(6-10 an ideal systemforthe study of the physical m) and path length(0.5-1.25 m). Thus, fluc- dynamicsof lightflecksin an algal systemand tuatingpath lengthalone cannot explain the of the utilizationof fluctuatingirradianceby magnitudeof these observed cyclicchangesin understoryspecies. While lightflecksbeneath terrestrial plantcanopies may be stochastically irradiance. of gap sizes in the At the depth rangeconsidered here, fluctu- generatedby heterogeneity ations in irradiancefasterthan the dominant canopy or disruptionof the canopy by wind wave frequencymay be caused by lens effects (Pearcy 1990), waves passingthroughM. pyrifrom surfacewaves (Nikolayev et al. 1971). fera canopies typicallyproduce lightfleckson The focal lengthof these waves is subject to the bottomof relativelypredictable,shortduchange with the zenith angle of the sun and rationand regularperiod (4-16 s). Lightflecks the azimuthal orientationof the wave form in thisrangeofdurationare knownto enhance of lightfleckutilizationof three waves and the efficiency withthesun. Small, high-frequency the correspondinglyhigh-frequencyfocusing species of benthicmacroalgae: Carpophyllum patternsthattheyproduce may be damped in maschalocarpum,Hormosira banksii,and dense kelp forests,but focusingpatternsmay Ecklonia radiata (Dromgoole 1988). In these of lightfleckutilizationis a well contributeto fluctuationsobservedin our studies,efficiency rateundata at frequenciesin excess of the wave fre- dimensionlessratio of photosynthetic quency, and we cannot dismiss this mecha- der fluctuatingPAR and the photosynthetic nism as a contributorto some of the high- rate predicted from steady state values (e.g. Chazdon and Pearcy 1986a). Greene and Gefrequencyfluctuationswe observed. An explanation of the model's tendencyto rard (1990) found thatgrowthrates of Chonunderestimatefractionalcanopy changes may druscrispuswereelevated underhigh-frequenlie in the assumptionthatkelp blades are uni- cy (1 Hz) lightfluctuations,above saturation formlight-blocking objects thatnever overlap irradiance,but not underlightfluctuationsbeand thatthe area of the surfacecanopy is con- low saturationirradiance.The resultsofGreene stantwhile thatof the water's surfacechanges and Gerard (1990), and Dromgoole (1987, with surface strain (Eq. 19). Clearly, surface 1988) have ecological significancein the conblades can overlap, and the area of floating textofa multilayeredstandofmacroalgaewith canopy may not be entirelyindependent of environmentalfrequencyand amplitudemodsurfacestrain.If blades are partiallyoverlap- ulation of PAR by ocean waves. Differences ping at theirstill water position (zero strain), in the period and amplitude of wave-modutheiroverlap will increase at maximum strain lated light flecksalong a wave exposure grain wave crest and decrease in wave troughs, dient formedby geomorphologicalcharacters in the yieldinga greaterchange in fractionalcanopy oftheshorelinemay resultin differences cover thanthenonoverlappingmodel can pre- photosyntheticcapacity of understoryalgae dict.An understandingofthedynamicsofkelp across these environments. Lightflashesin kelpforests References CHAZDON,R. L. 1988. Sunflecksand theirimportance to forestunderstoryplants.Adv. Ecol. Res. 18: 1-63. , AND R. W. PEARcy. 1986a, b. Photosynthetic responses to lightvariation in rain forestspecies. 1. Induction under constantand fluctuatinglightconditions. 2. Carbon gain and light utilizationduring lightflecks.Oecologia 69: 517-523, 524-531. DAYTON,P. K. 1985. Ecologyofkelpcommunities.Annu. Rev. Ecol. Syst. 16: 215-245. DEAN,T. A. 1985. The temporaland spatial distribution of underwaterquantum irradiationin a southernCalifornia kelp forest.Estuarine Coastal Shelf Sci. 21: 835-844. , AND F. R. JACOBSEN. 1984. Growthofjuvenile Macrocystispyrifera(Laminariales) in relationto environmentalfactors.Mar. Biol. 83: 301-311. , K. THIES, AND S. LAGOS. 1989. Survivalofjuvenile giantkelp: The effectsof demographicfactors, competitors,and grazers.Ecology 70: 483-495. M. W. 1988. Biologyand mechanicsofthewaveDENimr, sweptenvironment.Princeton. DERA,J., AND H. R. GORDON. 1968. Light fieldfluctuations in thephoticzone. Limnol. Oceanogr. 13: 697699. DROMGOOLE,F. I. 1987, 1988. Light fluctuationsand photosynthesisin marinealgae. 1. Adjustmentof the ratein constantand fluctuating lightregimes.2. Photosyntheticresponse to frequency,phase ratio and amplitude. Funct. Ecol. 1: 377-386; 2: 211-219. FRASER, A. B., R. E. WALKER,AND F. C. JURGENS. 1980. Spatiai and temporalcorrelationof underwatersunlightfluctuationsin the sea. IEEE J. Ocean. Eng. 5: 195-198. GERARD, V. A. 1984. The lightenvironmentof a giant kelp forest:Influenceof Macrocystispyrifera on spatial and temporalvariability.Mar. Biol. 84: 189-195. GREENE,R. M., AND V. A. GERARD. 1990. Effectsof high-frequency lightfluctuationson growthand photoacclimationof the red alga Chondruscrispus.Mar. Biol. 105: 337-344. HARROLD, C., J. WATANABE,AND S. LisIm. 1988. Spatial variationin the structureof kelp forestcommunities along a wave exposure gradient.Mar. Ecol. 9: 131156. HOLMES,R. W. 1957. Solar radiation, submarinedaylight,and photosynthesis, p. 109-128. In J.W. Hedgpeth [ed.], Treatise on marine ecology and paleoecology. V. 1. Geol. Soc. Am. Mem. 67. JACKSON,G. A. 1987. Modeling the growthand harvest Mar. Biol. yieldofthegiantkelpMacrocystis pyrifera. 95: 611-624. 407 N. G. 1976. Marine optics. Elsevier Oceanogr. JERLOV, Ser. V. 14. KOEHL,M. A. R., AND R. ALBERTE. 1988. Flow, flapping, A funcand photosynthesisof Nereocystis leutkeana: tional comparison of undulate and flatblade morphologies. Mar. Biol. 99: 435-444. KOPE, R. G., AND L. W. BOTSFORD. 1988. Detection of environmentalinfluenceon recruitmentusing abundance data. Can. J. Fish. Aquat. Sci. 45: 1448-1458. . 1990. Determinationof factorsaf, AND fectingrecruitment of chinook salmon Oncorhynchus in centralCalifornia.Fish. Bull. 88: 257tshawytscha 269. LEIGH,E. G., R. T. PAINE,J.F. QuiN, AND T. H. SUCHANEK. 1987. Wave energyand intertidalproductivity.Proc. Natl. Acad. Sci. 84: 1314-1318. AND M. J.DuNcAN. 1981. LOBBAN, C. S., P. J.HARRISON, The physiologicalecology of seaweeds. Cambridge. LUNING,K. 1981. Light,p. 326-355. In C. S. Lobban, and M. J.Wynn[eds.],The biologyofseaweeds. Univ. Calif.. NIKOLAYEV,V. P., AND M. S. KHULAPOV. 1976. Use of a nonstatisticalmodel to explain the mechanism of underwater illumination fluctuations.Izv. Atmos. Ocean Phys. 12: 993-997. ,. 0 I. PROKOPOV, G. V. ROZENBERG, AND V. I. SHEVERNEV. 1971. Statisticalpropertiesof the underwaterillumination. Izv. Atmos. Ocean Phys. 8: 936-944. NISBET,R. M., AND J. R. BENCE. 1989. Alternativedynamic regimesforcanopy-forming kelp: A varienton density-vaguepopulation regulation.Am. Nat. 134: 377-408. PEARCY, R. W. 1988. Photosynthetic utilizationof lightflecksby understoryplants.Aust. J.Plant Physiol.15: 223-238. . 1990. Sunflecksand photosynthesis in plantcanopies. Annu. Rev. Plant Physiol. Plant Mol. Biol. 41: 421-453. REED, D. C., AND M. S. FOSTER. 1984. The effectsof canopy shadingon algal recruitment and growthin a giantkelp forest.Ecology 65: 937-948. RICE, W. R. 1989. Analyzingtables of statisticaltests. Evolution 43: 223-225. SNYDER, R. L., AND J. DERA. 1970. Wave-induced lightfieldfluctuations in thesea. J.Opt. Soc. Am. 60: 10721079. SoKAL, R. R., AND F. J. ROHLF. 1981. Biometry,2nd ed. Freeman. Submitted: 6 February1991 Accepted:30 June1992 Revised:23 October1992
© Copyright 2026 Paperzz