Health & Medicine

Modeling the timing of spring phytoplankton bloom and biological production of the Gulf of St. Lawrence (Canada): Effects of colored dissolved organic matter and temperature

Description
Modeling the timing of spring phytoplankton bloom and biological production of the Gulf of St. Lawrence (Canada): Effects of colored dissolved organic matter and temperature
Published
of 16
All materials on our website are shared by users. If you have any questions about copyright issues, please report us to resolve them. We are always happy to assist you.
Related Documents
Share
Transcript
  Modeling the timing of spring phytoplankton bloom andbiological production of the Gulf of St. Lawrence (Canada):Effects of colored dissolved organic matter and temperature Zhi-Ping Mei a,  ,1 , Franc - ois J. Saucier a,2 , Vincent Le Fouest b , Bruno Zakardjian c ,Simon Sennville a , Huixiang Xie a , Michel Starr d a Institut des Science de la Mer de Rimouski (ISMER), Universite´ du Que´bec    a Rimouski, Rimouski, Que´bec, Canada G5L 3A1 b Laboratoire d’Oceanographie de Villefranche, BP 8, CNRS & Univ. Pierre et Marie Curie (Paris VI), 06238 Villefranche-sur-Mer Cedex, France c Universite´ du Sud Toulon-Var, LSEET-LEPI, Bˆatiment F, BP 20132, 83957 La Garde Cedex, France d Institut Maurice-Lamontagne, Direction des Sciences Oce´aniques Minist   ere des P ˆeches et Oce´ans CP 1000, Mont-Joli, Que´bec, Canada G5H 3Z4 a r t i c l e i n f o  Article history: Received 14 January 2010Received in revised form23 September 2010Accepted 5 October 2010Available online 14 October 2010 Keywords: Gulf of St. LawrenceMarine ecosystem modelTemperatureCDOMPhytoplankton bloomBiological production a b s t r a c t The effects of colored dissolved organic matter (CDOM) from freshwater runoff and seasonal cycle of temperatureonthedynamicofphytoplanktonandzooplanktonbiomassandproductionintheGulfofSt.Lawrence (GSL) are studied using a 3-D coupled physical-plankton ecosystem model. Three simulationsareconducted:(1)thereferencesimulationbasedonLeFouestetal.(2005),inwhichlightattenuationbyCDOM is not considered and maximum growth rate ( m max ) of phytoplankton and zooplankton are nottemperature-dependent (REF simulation); (2) light attenuation by CDOM is added to REF simulation(CDOM simulation); and (3) in addition to CDOM, the  m max  of phytoplankton and zooplankton areregulated by temperature (CDOM+TEMP simulation). CDOM simulation shows that CDOM substantiallyreduces phytoplankton biomass and production in the Lower St. Lawrence Estuary (LSLE), but slightlyreduces overall primary production in the GSL. In the LSLE, the spring phytoplankton bloom is delayedfrom mid-March to mid-April, resulted from light attenuation by CDOM. The CDOM+TEMP simulationshowsthatthespringphytoplanktonbloomintheLSLEisfurtherdelayedtoJuly,whichismoreconsistentwith observations. Annual primary production is reduced by 33% in CDOM+TEMP simulation from REFandCDOMsimulations.Zooplanktonproductionisthesameinallthreesimulations,andexportoforganicmattertodepthisreducedinCDOM+TEMPsimulation,suggestingthattemperaturecontrolledgrowthof phytoplankton and zooplankton enhances the coupling between primary production and zooplanktonproduction under the seasonal temperature cycle of the GSL. &  2010 Elsevier Ltd. All rights reserved. 1. Introduction TheGulfofSt.Lawrence(GSL)islocatedineasternCanadaatthelower limit of Subarctic region. It is an important fishing ground of Maritime provinces of Canada, producing about 25% of totalcommercial fish catch by weight of Canada (Dickie and Trites,1983; Chadwick and Sinclair, 1991), and the transport corridorbetween central and eastern Canada and North Atlantic.Biogeochemically, it connects the Great Lakes of North Americato the Atlantic Ocean. The GSL receives terrestrial organic matteralong with freshwater from Great Lakes upstream, and numerousriverssurroundingtheGSL(KoutitonskyandBugden,1991)(Fig.1). The impact of nutrient flux from the GSL reaches the Scotian Shelf (Petrie and Yeats, 2000), and Georges Bank (Houghton and Fairbanks, 2001). Freshwater pulses from the rivers are the mainforcing of buoyancy-driven circulation of the GSL and govern thedistribution of phytoplankton in the estuary (Savenkoff et al.,1997). Freshwater runoff induces cross-frontal current andupwelling in the Gaspe´ currents, which contribute to enhancedbiologicalproductivity(Bugdenetal.,1982).Levasseuretal.(1992) estimated about 85–90% of seaward nutrient flux was fromSt. Lawrence estuary. That may explain why the crustacean andfinfish landing in the Gulf are correlated with freshwater runoff.The CDOM and other suspended materials carried into the GSL withriverinefreshwaterdischargehaveastrongimpactontheopticalproperties of the surface water of the GSL (Nieke et al., 1997; Yeats,1988), reducing the irradiance available to phytoplankton Contents lists available at ScienceDirectjournal homepage: www.elsevier.com/locate/csr Continental Shelf Research 0278-4343/$-see front matter  &  2010 Elsevier Ltd. All rights reserved.doi:10.1016/j.csr.2010.10.003  Corresponding author. Tel.: +1 410 221 8288. E-mail addresses:  zmei@mta.ca, zmei@umces.edu (Z.-P. Mei), lefouest@obs-vlfr.fr (V. Le Fouest), Bruno.Zakardjian@lseet.univ-tln.fr(B. Zakardjian), simon_senneville@uqar.qc.ca (S. Sennville),huixiang_xie@uqar.qc.ca (H. Xie), Michel.Starr@dfo-mpo.gc.ca (M. Starr). 1 Current address: University of Maryland Center for Environmental Science,Horn Point Laboratory, PO Box 775, Cambridge, MD 21613, USA. 2 Deceased.Continental Shelf Research 30 (2010) 2027–2042  photosynthesis and interfering with the retrieval of chlorophyllconcentration (Chl) from satellite remote sensing (Branco andKremer, 2005; Blough and Del Vecchio, 2002; Coble et al., 2003;Nelson and Siegel, 2002; Smyth et al., 2005). CDOM may also beformed through photochemical transformation of dissolved organicmatter of terrestrial srcin and phytoplankton photosynthesis, andconsumed through photobleaching (Kieber et al., 1997; Whiteheadet al., 2000).The timing of spring phytoplankton blooms in the GSL may varywith interannual variability of ocean physics, but usually the springphytoplanktonbloominthe lowerSt. Lawrence Estuary (LSLE,Fig. 1)islater(June–July)thaninthenortheastgulf(NEG)andsoutherngulf (Magdalen shallow, MS) in April–May (de Lafontaine et al., 1991;Therriault and Levasseur, 1985; Levasseur et al., 1984; Roy et al.,2008). The timing of spring phytoplankton bloom has importantconsequences to the survival of zooplankton and thus fishes larvaeandtheirrecruitmentincoastalwaters.InthesoutheastoftheScotiaShelf, the timing of the spring phytoplankton bloom explains 89% of the interannual variability in the survival of larval haddock( Melanogrammus aeglefinus ) (Platt et al., 2003). This is expected to apply to GSL as well, based on Cushing’s match and mismatchhypothesis (Cushing, 1990).There are two hypothesis regarding the later spring phyto-plankton bloom in the LSLE than in gulf in the GSL. Firstly, theturbidity mainly due to CDOM from freshwater runoff delayes thespringphytoplanktonbloomintheLSLE(deLafontaineetal.,1991;Therriault and Levasseur, 1985). Observations in April and May,however,suggest that during these monthsmean irradiance inthemixed layer in the LSLE is sufficient for phytoplankton bloom tostart based on Sverdrup’s model (Levasseur et al., 1984). Analternative hypothesis is also related to freshwater runoff. Strongcurrent resulted from freshwater runoff during spring peak of iceand snow melt in the catchment area prevents phytoplanktonbiomass from building up (Zakardjian et al., 2000).The GSL has a strong seasonal and interannual variability inphysics.Itiscompletelyicecoveredduringwinter,andopenwaterbegins in spring-summer. The water column is strongly mixed inwinter, with maximum mixed layer depth of 100 m, and becomeshighlystratified in summer. Water temperature at thesurface is atfreezingpointinwinter,andreachesashighas  4 25  3 Catsurfaceinsummer (Saucier et al., 2003). Interannual variability of climateforcing changes due to freshwater input from rivers, ice cover,watercolumnmixing,and temperature.The largevariationsof thephysicsoftheGSLwillstronglyaffectthegrowthofphytoplanktonand zooplankton at spatial, seasonal and interannual scales.Levasseur et al. (1984) showed that the phytoplankton biomassin the LSLE is strongly correlated with water temperature. It hasbeen reported that zooplankton biomass is reduced in colderyears, and their production increases with boreal-temperatespecies developed in warmer years in the GSL (de Lafontaineet al., 1991, and references therein). These indicate planktoncommunities respond sensitively to temperature change inthe GSL.In order to understand the physical mechanisms controlling thedynamicofplanktonicecosystemintheGSL,a3-Dcoupledphysical-plankton ecosystem model of the GSL was developed by Le Fouestet al. (2005). Seasonal development of primary and secondaryproduction of the GSL was simulated, and heterogenous planktonecosystem processes were revealed to associate with mesoscalevariability of hydrodynamics across the GSL (Le Fouest et al., 2005).The model, however, did not consider the effects of CDOM andtemperature dependence of the maximum growth rates ( m max  d  1 )ofphytoplanktonandzooplankton.Inasubsequentstudy,LeFouestet al. (2006) implicitly simulated the effect of riverine CDOM,which is diagnosed from salinity based on the linear relationshipbetween light attenuation coefficient and salinity, on primaryproductivity in the estuarine region of the GSL. However, thevariability of salinity and CDOM are driven by different processes.Freshwater discharge, precipitation and melting of sea-ice decreasesalinity proportional to their amount as their salinity are close to 0,but CDOM concentrations vary depending on the sources of freshwater and their respective amount, and the season of runoff. Fig.1.  The map of Gulf of St. Lawrence. Four subregions, lower St. Lawrence estuary (LSLE), Magdalen shallow (MS)and northeast gulf(NEG), are identified for detailed timeseries analysis of phyto- and zooplankton biomass and production. The time series station of Rimouski Station is located as a star in the LSLE. Squares are locations of thestationssampledforCDOM.NoteonestationforSt.LawrenceRiverandonestationinSaguenayfjordareoutsidethemodeldomain,butservetoprovideCDOMinfreshwaterflowing to GSL.  Z.-P. Mei et al. / Continental Shelf Research 30 (2010) 2027–2042 2028  For example, CDOM concentration in the Saguenay River on thenorthshoreoftheLSLEismuchhigherthanotherrivers,eventhoughtheir salinity is the same (Xie, unpublished). Therefore, it isnecessary to simulate CDOM explicitly as a passive tracer in thecoupled model.Temperature– m max  relationship of phytoplankton has been estab-lished since Eppley (1972). Recently, Rose and Caron (2007) confirm thetemperature– m max  relationshipofphytoplankton,byincludingnewdata after Eppley (1972), and further establish temperature– m max relationships for bacteria and zooplankton at different trophic levels.They find the slopes for the temperature– m max  relationships aresignificantly different between phytoplankton and zooplankton.The  m max  of autotrophic plankton is higher than that of zooplanktonatlowertemperatureranges,butclosertoorlowerthanzooplanktonathigher temperature ranges. That affects the timing and extent of phytoplankton bloom in spring–summer in the oceans of differentlatitudes. Such temperature-based parameterization of   m max  of plankton is considered robust and mechanistically based (Caron andRose, 2008; Lo´pez-Urrutia, 2008), and when used, may reduce thenumber of free parameters of plankton ecosystem model. However,differential temperature dependence of phytoplankton and zoo-plankton growths has rarely been taken into account in marineecosystem models (Tian, 2006). Inter-model comparisons show thatdivergence of model performance is strongly linked to how theeffects of temperature on phytoplankton production is formulated(Carr et al., 2006).The objectives of the present study are to understand theimportance of CDOM and temperature in affecting the timing of spring phytoplankton bloom in different regions of the GSL, andbiological productivity of the entire GSL. In order to test the abovehypothesis regarding the timing of spring phytoplankton bloom,we simulate the CDOM dynamically in the coupled physical-plankton ecosystem model of  Le Fouest et al. (2005) and toinclude light attenuation by CDOM. Furthermore, the  m max  of phytoplankton and zooplankton is defined as temperature-dependent parameter based on Rose and Caron (2007). Theinterannual variability of CDOM and temperature will haveconsequences to that of biological productivity of the GSL. 2. Model ThegoalofthisstudyistounderstandhowCDOMandtemperatureaffect the timing of spring phytoplankton bloom and biologicalproduction. We compare the seasonal cycles of phytoplankton andzooplankton biomass and production simulated with the 3-D coupledphysical–biologicalmodeldevelopedfortheGSL(LeFouestetal.,2005),after the model is modified to account for CDOM in regulating watercolumn irradiance and temperature in regulating  m max  of phyto-plankton and zooplankton. Particularly, we want to understand whythe spring phytoplankton bloom is later in the LSLE than in the Gulf.Therefore, we compare the seasonal cycles of phytoplankton andzooplanktonbiomassandproductionofdifferentregions,includingtheLSLE, NEG and MS in the three simulations as detailed below. Thosesubregionsaredefinedbasedontopographyandbiologicaldynamicsof the GSL (Steven, 1971; de Lafontaine et al., 1991). The LSLE ischaracterized with low salinity, cold, and high nutrients from riversandupwellingattheheadofLaurentianChannel.TheNEGisnorthernhalf of the gulf, north of the Laurentian Channel. It is relatively deep( 4 200 m), and receives Labrador water through Strait of Belle-Isle,with salinity  4 30. The MS is the southern half of the gulf, and isshallow (average depth of 50m) and warmer than the LSLE and NEG.Three numerical experiments are conducted as shown inTable 1. Basic model parameters for the reference simulation(REF simulation) are listed in Table 2, and are the same as Le Fouest et al. (2005). In the REF simulation, light attenuation iscontributed from Chl, ( k Chl ), water ( k w ) and non-Chl material ( k  p )(Table2). k  p wastunedtoobtaintheeuphoticzonedepthsofabout40–50 mobservedintheoceanicGulfwaterby Doyonetal. (2000)(Le Fouest et al., 2005), and thus represents attenuation of light byCDOM and other detritus in the oceanic Gulf water, but much toolow for the LSLE, where light attenuation by CDOM is muchstronger and the euphotic zone depths range from 10 to 20 m(Nieke et al., 1997). The CDOM simulation is a modification of theREF simulation in which  k  p  of the REF simulation is replaced withlight absorption by CDOM, the latter is explicitly simulated as apassive tracer. It is assumed here that the attenuation of non-Chl amaterial are dominantly due to CDOM.  k  p  in the REF simulation isquite close to the attenuation by CDOM simulated for the oceanicNortheast Gulf (Fig. 4), suggesting CDOM dominates theattenuation of non-Chl a material in oceanic water of the GSL. Inthe LSLE, the highest attenuation due to CDOM reaches 0.30 m  1 ,which agrees with the maximum attenuation coefficient in thedataset of  Le Fouest et al. (2006). The intercept (0.98) for the relationshipbetweensalinity( 4 27)andlightabsorptionbyCDOM(  A CDOM  )(Eq.(3))isclosetothatfortherelationshipbetweensalinity( 4 24) and  k  p  of  Le Fouest et al. (2006). Therefore, it is justified toreplace  k  p  of REF simulation with light absorption by CDOM torepresent non-Chl light attenuation for the GSL. With CDOM beingsimulatedasapassivetracer,CDOMsimulationcanresolvespatial,andtemporaldynamicofnon-Chllightattenuationassociatedwithfreshwaterrunoff.Inthefuturelightattenuationbyotherparticles,such as suspended sediments need to be quantified. However,suspended sediment might have settled in the upstream of theestuary,sinceNiekeetal.(1997)findthattheeuphoticzonedepthsin the LSLE are mainly determined by phytoplankton biomass andCDOM.TheCDOM+TEMPsimulationisamodificationoftheCDOMsimulation, where  m max  of phytoplankton and zooplankton aretemperature dependent (see details in Section 2.3).  2.1. Coupled physics-biological model A simple biological model composed of nutrients, phytoplankton,zooplankton and detritus (NPZD) is coupled to the 3-D regional ice-oceancirculationmodeldevelopedfortheGSL(Saucieretal.,2003)asin Le Fouest et al. (2005). The ice-ocean circulation model of  Saucier etal.(2003)isforcedwithairtemperature,windintensity,dewpoint,cloud cover, precipitation minus evaporation, daily river runoff andarea-averaged accumulated and compacted snowdepthoverthe ice.Those atmospheric and hydrologic forcing, plus radiation that forcesecosystem model, are the output of regional and global multiscaleclimate models (Saucier et al., 2003).ThereareafewmodificationstotheNPZDmodelof LeFouestetal.(2005). In Le Fouest et al. (2005), surface irradiance of each grid was discounted by the fraction of a model grid covered with ice, and theirradiancecorrectedforicecoverwasusedtocalculatelightlimitationto phytoplankton growth of each grid. In the present study,phytoplankton growth for ice covered and ice free parts of a gridarecalculatedseparately,andthenweightedmeangrowthrateofice-coveredandice-freeareasofeachgridiscalculated.Thisistotakeinto  Table 1 The design of numerical experiments.Simulations  m max  (d  1 ) Light attenuation componentsREF Constant Water+Chl+ k  p a CDOM Constant Water+Chl+CDOMCDOM+TEMP  m max ¼  f  ð T  Þ b Water+Chl+CDOM a Attenuation due to detrital materials, set as a constant (Table 2). b See Eqs. (5)–(7) (Rose and Caron, 2007).  Z.-P. Mei et al. / Continental Shelf Research 30 (2010) 2027–2042  2029  account the fact that phytoplankton photosynthesis responds toirradiance level nonlinearly. In the present study, monthlyclimatology of nitrate (NO 3 ) profiles in the Strait of Belle-Isle andCabotStraitisusedasboundaryconditionsoftheNPZDmodelasinLeFouest et al. (2006).The detailed equations for the plankton ecosystem model can befound in Le Fouest et al. (2005). Briefly, there are two size-classes of phytoplankton,smallnon-diatoms,andlargediatoms,twosize-classesof zooplankton, microzooplankton and mesozooplankton, twonutrient sources, NO 3  and ammonium (NH 4 ), and two detritalnitrogen, dissolved detritus (DON) and particulate detritus (PON).Phytoplankton production and zooplankton grazing, and decompo-sition of PON produce DON. PON is from non-grazing death of phyto-plankton and zooplankton defecation. The planktonic ecosystemmodel is a nitrogen-based NPZD model. The specific growth rate of phytoplankton ( m ,  d  1 ) is taken as the minimum of light limitedgrowthornutrient-limitedgrowth,followingLiebig’slawofminimum: m ¼ m max min  PARPAR þ ke , N N  þ k N     ð 1 Þ where  ke  is the half-saturation constant for photosynthesis,  k N  , thehalf saturation constant of NO 3  or NH 4  uptake of large or smallphytoplankton, and N is either NO 3  or NH 4 . Primary productionreported here is total primary production, including NO 3  based newproductionandNH 4 basedregeneratedproduction,andiscalculatedasthe product of phytoplankton biomass ( N  P  ) and  m , or  PP  ¼ m N  P  ,assuming Redfield C:N ratio and constant carbon to Chl ratio (C:Chl)(Table 2). Zooplankton production, including microzooplankton andmesozooplankton,iscalculatedastheproductoftheirgrazingratesandassimilation efficiencies.  2.2. Simulation of colored dissolved organic matter (CDOM) Le Fouest et al. (2006) diagnoses the light attenuation due tonon-chlorophyll detritus (mainly CDOM) from salinity as: k  p ¼ 0.0364  S  +1.1942, and  k  p  is bounded between 0.26 m  1 atsalinity of 26 and 0.03 m  1 at salinity of 32. Previous study (Niekeet al., 1997) suggests that light absorption due to CDOM furtherincreases with decreasing salinity at salinity range  o 26. Inaddition, variability of salinity is not only caused by freshwaterrunoff from rivers, but also by melting of sea-ice in spring,precipitation and evaporation, which may not change salinityand CDOM in the same proportion. Therefore, in this study, weinitialize CDOM based on the linear relationships betweensalinity and CDOM for two salinity ranges ( o 27 and  4 27), andprescribe CDOM to the freshwater at the river mouth. As thefreshwater and CDOM enter GSL, CDOM is advected and mixed asother tracers.The concentration of CDOM is represented by its mean lightabsorption over the PAR wavelength range (  A CDOM  , m  1 ), since theecosystem model uses bulk PAR without resolving spectral lightabsorption. The  A CDOM   is initialized from salinity, based on thelinear relationship between salinity and CDOM as shown in thedata collected in winter 2005 (Xie, unpublished data). CDOM andsalinityweresampledacrosstheestuaryandextendedtotheGaspe´coast (see Fig. 1), covering salinity range between 0 and 32. Therelationship between  A CDOM   and salinity ( S  ) can be described asEqs. (2) and (3), for salinity  o 27, and  Z 27, respectively:  A CDOM  ¼ 0 : 01392 S  þ 0 : 5325  ð r  2 ¼ 0 : 9162 , n ¼ 6 , S  o 27 Þ ð 2 Þ or,  A CDOM  ¼ 0 : 02995 S  þ 0 : 9823  ð r  2 ¼ 0 : 9857 , n ¼ 6 , S  4 27 Þ ð 3 Þ The obtained relationships between  A CDOM   and salinity of differentsalinity ranges are consistent with earlier results of  Nieke et al.(1997).Thestrongcorrelationsbetweensalinityand  A CDOM  suggestthatsalinitycanbereliablyusedtopredictthe  A CDOM  infreshwaterandtoinitialize  A CDOM  acrosstheGSL.The  A CDOM  ofriverwaterissettobe0.53 m  1 ,derivedfromtheinterceptofEq.(2),correspondingto the  A CDOM   at  S   of 0.  A CDOM   from the Saguenay fjord (1.97 m  1 )  Table 2 List of symbols and model parameters for reference simulation.Symbol Variable name Parameter values Reference Light k w  Attenuation of water 0.04 m  1 Morel (1988) k  p  Attenuation by non-Chl a materials 0.04 m  1 Fitted Phytoplankton k3 LP  Half-saturation constant for NO 3  uptake of LP 1 mmol N m  3 Parsons et al. (1984)k4 LP  Half-saturation constant for NH 4  uptake of LP 0.5 mmol N m  3 k3 SP  Half-saturation constant for NO 3  uptake of SP 1 mmol N m  3 k4 SP  Half-saturation constant for NH 4  uptake of SP 0.1 mmol N m  3 ke  Half-saturation constant for photosynthesis 10 Einst m  2 d  1 Kiefer and Mitchell (1983)d t  min  Minimum doubling time of LP and SP 0.5 day Zakardjian et al. (2000) m LP, SP  Senescence of LP and SP 0.02 d  1 Fittedsed LP  Sinking speed of LP 1 m d  1 Smayda (1970)C:Chl Carbon to Chl ratio 55 gC g Chl  1  Zooplankton g  max MEZ  MEZ maximum grazing rate 0.2 d  1 Fitted  g  max MIZ  MIZ maximum grazing rate 2 d  1 Strom et al. (2001)iv MEZ  Ivlev parameter of MEZ grazing formulation 0.8 (mmol N m  3 )  1 Frost (1972) k MIZ  Half-saturation constant for MIZ grazing 0.8 mmol N m  3 Fittedass MEZ  Assimilation efficiency of MEZ 70% Kiorbøe et al. (1985)ass MIZ  MIZ growth efficiency 30% Riegman et al. (1993) m MEZ  MEZ mortality 0.05 (mmol N m  3 d  1 )  1 Fitted m MIZ  MIZ senescence 0.02 d  1 Fittedeg DON egestion by MIZ 30% Lehrter et al. (1999)ex NH 4  excretion by MEZ 0.05 d  1 Saı´z and Alcaraz (1992) Detritus sed POM  PON sinking speed 100 m d  1 Turner (2002)fg PON fragmentation rate 0.05 d  1 Fasham et al. (1990)rem DON remineralization rate 0.4 d  1 Pickard et al. (2000)  Z.-P. Mei et al. / Continental Shelf Research 30 (2010) 2027–2042 2030  is exceptionally high, and remarkably deviated from the linearrelationship described in Eqs. (2) and (3), thus the value of 1.97 m  1 was assigned to this particular river. The high  A CDOM   inSaguenay river is caused by high DOM concentrations and higherpercentage of humic substances in DOM than in LSLE (Tremblayand Gagne´, 2009).CDOM enters the GSL via river runoff, and is then transportedwithcurrents,asshowninthegeneralequationoftraceradvectionand diffusion: d C  d t   þ u d C  d  x  þ v d C  d  y  þ w d C  d  z  ¼  dd  x K   x d C  d  x   þ  dd  y K   y d C  d  y   þ  dd  z  K   z  d C  d  Z    þ source  sink  ð 4 Þ where  C   represents passive tracers, such as  A CDOM  .The strong linear relationships between  A CDOM   and  S   (Eqs. (2)and(3))inbothwinterandsummersuggestthat  A CDOM  intheGSLismainly determined by mixing of one endmember with low  S   buthigh CDOM, with another with high  S   but low CDOM. Differentslopes of the relationship for low salinity (estuarine), and highsalinity (Gulf) waters suggest different hydrodynamic processesin the mixing of two endmembers in two regions (Nieke et al.,1997). Therefore,  source  (e.g. biological production) and  sink  (e.g.photochemicaldegradationand microbialremineralization) termsin Eq. (4) for  A CDOM   are set to be 0 in the simulations.  2.3. Temperature regulation of maximum growth rate ( m max ) of  plankton In REF and CDOM simulations, the  m max  of phytoplankton andzooplanktonaresettobeconstantparameters(Table2)basedondataobtainedduringphytoplanktonbloom(Tamigneauxetal.,1997).Thatmay overestimate the  m max  before phytoplankton bloom, andunderestimate the  m max  when temperature increases. Consideringthe large spatial and seasonal temperature range in the GSL, thetemperature-dependent m max  of phytoplankton and zooplankton aretaken into account in the CDOM+TEMP simulation. In theCDOM+TEMP simulation, the relationship between temperature ( T  )and the maximum growth rate of small and large phytoplankton( m LP , SPmax  ,  d  1 ), microzooplankton ( m MIZmax ), and mesozooplankton ( m MEZmax )are defined according to Rose and Caron (2007), and are listed asEqs. (5)–(7), respectively:ln m LP  , SP  max  ¼ 0 : 06 T   0 : 5  ð 5 Þ ln m MIZ  max ¼ 0 : 10 T   1 : 0  ð 6 Þ ln m MEZ  max ¼ 0 : 13 T   3 : 0  ð 7 Þ The slope for phytoplankton growth (Eq. (5)) is similar to Eppley(1972), but the slopes for zooplankton are higher than that of phytoplankton. Consequently, under low temperature ( o 15  3 C),phytoplankton have growth advantage over zooplankton (Rose andCaron, 2007). The maximum grazing rate of micro- andmesozooplankton (g max MIZ  and g max MEZ  , respectively) are calculated from m MIZ  max  and m MEZ  max  (Eqs.(6)and(7))assuminggrowthyieldof33%formostof the zooplankton groups at their maximum growth rate (Hansenet al., 1997). This value represents an average growth yield forzooplankton across a variety of taxa groups. Even though mostvalues center around 33%, extreme values range from 10% to 45%(Hansenetal.,1997,Table7).Thatmayleadtothemaximumgrazing rate to change accordingly.  2.4. Model-data comparisons The modeled Chl concentrations are compared with Chlsampled in spring-summer (June) and fall (November). The loca-tions of sample stations can be found in Le Fouest et al. (2005,Fig. 5a). We onlycompare data within the upper 50 m of thewatercolumn.Therearetotalof375and430discretedatapointsforJuneand November, respectively. For Chl determination, discretesamples are filtered into glass fibre filter (GF/F). Chl retained onGF/F filters is extracted with 90% acetone overnight in cooltemperature, before fluorometer reading, following the protocolof  Mitchell et al. (2002).Both model output and field data are log transformed, so thatthey conform with normal distribution. Correlation coefficient ( R ),standard deviation, root mean square difference ( RMSD ), andnormalized bias ( B * ) are calculated to validate the model output. R ,  RMSD  and  B * (Friedrichs et al., 2009) are calculated as R ¼ P ni ¼ 1 ð log  C  o , i  log  C  o Þð log  C  m , i  log  C  m Þ  ffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiP ni ¼ 1 ð log  C  o , i  log  C  o Þ 2 P ni ¼ 1 ð log  C  o , i  log  C  m Þ 2 q   ð 8 Þ RMSD ¼  ffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi 1 N  X N i ¼ 1 ð log  C  m , i  log  C  o , i Þ 2 v uut  ð 9 Þ and B  ¼  log  C  m  log  C  o s o ð 10 Þ C  m , i and C  o , i aremodeledandobservedChlvalues,respectively. s o  isthestandarddeviationofobserveddata,andoverbarismeanvalue. 3. Results  3.1. Model data comparison The model output of Chl in June and November are comparedwithChldatacollectedinJune(Fig.2a–c)andNovember(Fig.2d–f) of1997,respectively.Juneand Novemberaretheseasonsofspringand fall blooms, respectively. In REF and CDOM simulations, thedatapointsofobservedvssimulatedChlfallabovethe1:1diagonalline in the range between 0.1 and 1 mg m  3 , indicating that theREF and CDOM simulations overestimate phytoplankton biomassin both June and November. The data points of CDOM+TEMPsimulations fall on the 1:1 diagonal line better than REF andCDOM simulations, particularly for Chl between 0.1 and 1 mg Chlm  3 in June, indicating that simulation of the timing of springphytoplankton bloom is improved in CDOM+TEMP simulation(Fig. 7).The statistics, including  R ,  s  and  RMSD  between the modeledand observed Chl are summarized on Taylor diagrams (Taylor,2001) (Fig. 3). Corresponding to the scatterplot in Fig. 2,  R  is higher for theCDOM+TEMP than REF and CDOM simulations in June, the springbloom season. However,  RMSD  and s for CDOM+TEMP simulationare higher than REF and CDOM simulations in June. In November,the statistics of the three simulations are quite close to each other.We further calculated the standardized bias following Friedrichset al. (2009) (Table 3). The standardized bias is the lowest in CDOM+TEMP simulation in both June and November.  3.2. Spatial distributions in water temperature and CDOM  Watertemperature isclose to freezing pointin mostof the areaof the GSL in winter (January–March), and slightly increases to  Z.-P. Mei et al. / Continental Shelf Research 30 (2010) 2027–2042  2031

P F.R H

Jan 14, 2019
Search
Similar documents
View more...
Related Search
We Need Your Support
Thank you for visiting our website and your interest in our free products and services. We are nonprofit website to share and download documents. To the running of this website, we need your help to support us.

Thanks to everyone for your continued support.

No, Thanks