CONTENTS
p re-fa ca 2 1. introduction 1
1.1. Phytoplankton composition . . . 2 1.2. Phytoplankton dynamics 3 2. Phytoplankton data analysis 5 3. problem de-Hnition 8 4. Material and Methods. I: Dascription of GREALG 9 4.1. Model ecosystem structure 9 4.2. Abiotic features 11 4.3. Microphytobenthos (benthic dxatoms) 13 5. MataHal and Mathods. II: Phytoplankton kinatic» . . . 15 5.1. ïntroduction 15 5.2. Production .' 15 5.3. Loss proces ses 24 6. Rtsults and discus»ion 33 6.1. Comparison with data 33 6.2. Comparison with CABAMOD and NUTGRE 36 6.3. Model sensitivity 37 6.4. Turbulence 38 6.5. Microphytobenthos 39 6.6. Discussion 39 7. conclusions 42 8. SUtrniwry 43 Litaratum citad 44 Add. 1. Hicrophytobtnthoa light elimata 49 Add. 2. Light «xtinction data 50 Tablts and figuras 51
PREFACE
The research descrlbed in this report is carried out within the framework of WAter BASIn Model (WABASIM). WABASIM is organized as a muljtidisciplinary coproject of the Tidal Water Division of the DeltaJ)ej£ajMynejiLJ^^ Division Water Resources and Envi-"rinïïficnT^f theDelft Hydraulics Laboratory (DHL), and is financed
by the Delta Department.
The piujuuL atffinSTthe development of aquatic ecologi'cal and water quality models, which can serve as tools in providing adequate guidelines for environmental management in the "(future) water basins in the Delta area.
This report deals with the development of a model of nutriënt dynamics and phytoplankton production and losses in Lake Grevelin-gen. Formulations concerning the dynamics of the two algal groups, evaluated in this zero-dimensional nutriënt/plankton model, will be used in a future 2-dimensional ecological model of Lake Grevel-ingen (GREWAQ). Special attent ion has been paid to the formulation of loss processes like sedimentation and grazing and their inter-action with environmental factors like nutriënt supply, light and turbulence. Besides, results of an analysis of available phyto-plankton data (provided by the Delta Institute for Hydrobiological Research) are presented,
The research and report ing are carried out by drs. J.M. Baveco (in-itially as a student project), E. Luppes and drs. I. de Vries,
The research activities are executed in cooperation with the WABA-SIH-salt project group, in which next to members of TWD and DHL, also members of the Delta Institute for Hydrobiological Research (DIHO) participate. The WABASIM-salt project group consists of the following members:
Ir, S. van de Kamer (TWD), chairman Drs. I, de Vries (DHL), secretary Drs. C. Bakker (DIHO) Drs. J.M. Baveco (DHL) Dr. B. van Eek (TWD) Drs. J.R. Heringa (DHL) Drs. P. Hofman (DIHO) Drs, C.F. Hopstaken (DHL) Dr. P. Nienhuis. (DIHO) Drs. A. Smaal (TWD) Ir. J.H.G. Verhagen (DHL) Drs. L.P.M.J, Wetsteyn (TWD)
1. INTRODUCTION
In this memo the final results are reported of the development of GREALG, a zero-dlmensional phytoplankton model. GREALG attempts to describe the dynamics of two algal groups in Lake Grevelingen, as a function of lïght, temperature and nutriënt availability, and in tcrms of production and loss processes. Besides, the results of an analysis of available phytoplankton data, of relevance to the mod-el 1 ing , are presented.
The development of this phytoplankton (sub)model forms part of modelling activities carried out within the frame work of WABASIM. Ultimately, these activities have to culminate in a more-dimen-sional ecological simulation model (GREWAQ), incorporating the most important ecological processes - especially those involved in the cycling of essential nutrients like nitrogen, phosphorus and silicon. To this purpose, various submodels will be coupled and integrated in an overall model structure, registrating all exchanges between all pools and segments.
Up till now, several models and submodels related to the aquatic ecosystem of Lake Grevelingen have already been constructed.
• An annual carbon budget model (CABAMOD) has been developed to identify the importance of the various components of the Gre-velingen foodweb (De Vries, 1984). CABAMOD (see fig, 15.1) provides a picture of the distribution and fluxes of organic carbon from and to the various ecosystem components; it is used as a franje work for the development of submodels (eelgrass, phytoplankton, microphytobenthos and macro zoobenthos) on the component level.
• A nutriënt balance model (NUTGRE) has been developed to describe and evaluate the relation between carbon cycling and nutriënt dynamics (De Vries and Hopstaken, 1984). Calculations by NUTGRE indicated a _nutrient limitation of planktonic and benthic micro-algae by reversïble storage of more than 75£ of available nitrogen and silicon in bottom detritus (mainly originating from biodeposition by benthic suspension feeders), Phytoplankton primary production was concluded to depend on the balance between storage of nutrients in bottom detritus, mineralization and uptake by benthic micro-algae. In NUTGRE measured phytoplankton biomass and product ion values were used to simulate nutriënt dynamics.
In the future ecological model for Lake Grevelingen, a submodel simulating phytoplankton kinetics will be included, Thus the development of GREALG may serve several purposes,
1. To provide a ready-made phytoplankton submoóel, which can be easily incorporated in the overall structure of GREWAQ.
2. To evaluate some possible formulations for phytoplankton pro-duction and loss processes.
3. Additionally, GREALG offers some possibilities for a further evaluation of hypotheses concerning the influence of suspen-sion feeding bottotnfauna on nutriënt regeneration (and phyto-plankton dynamics), the possible role of diatom sedimentation in the relative dominance of phytoplankton species, and the competitive interaction between benthic and planktonic algae.
Simple, robust functions are preferred and speculative differences between algal groups are avoided as much as possible, Coefficients and initial values are as far as possible derived from existing datasets of Lake Grevelingen. Otherwise literature values are used! In chapter 2. results are presented of an analysis of phyto-plankton data for 1979 and 1980, performed in cooperation with E. Luppes; these results mainly substantiate the identification of tiominant species and provide more reliable estimates of phytocar-bon.
GREALG adopted the overall structure of SEAWAQ, an existing phyto-plankton-nutrient model for the North Sea (Verhagen*, 1984). When GREALG is incorporated in GREWAQ, the nutriënt and detritus
kinet-ics formulations will be removed from the algae submodel.
1.1. PhvtQPlankton composition
Due to drastie hydrodynamical changes occurring between 1977 and 1980 (sec Bannink et al, 1984), the phytoplankton species composi-tion differed from year to year, The 1980 situacomposi-tion, with flushing only in the winter months, probably most resembles the actual situ-ation. However, a definite picture of species succession in the evolved saline lake yet does not exist.
The majority of phytoplankton species observed between 1978 and 1980 consists of diatoms, although non-diatoms (autotrophic and heterotrophic flagellates) may prevail in biomass in spring (Bakk-er and De Vries, 1984).
Dominant diatom species are Chaetoceros, SPE>. (late summer) and
(early summer). Also pennate species, mainly originating from the microphytobenthos, contribute greatly to the observed biomass; often they are not photosynthetically active, however (pers.comm, P.R.M, de Visscher).
Less dominant but frequent ly found in summer 1980 are Eucampia zoo-diacus. Licroop^ora. gragiljy. As^^r^Qnella iaponica and Rhizosole-A striking feature of the diatom dynamics is the almost complete absence of high densities in spring, when silicon and nitrogen con-centrations are still high and a diatom bloom (as in the Eastern Scheldt and North Sea) could be expected. The only species observed before or in May are Thalassios^ra flord^nskioldii f ^9^9") r Rh.izosoleniq, se^j^era (19801 t Leptocvlindricu^ danicus (1980') and
small concentrations of Skeleton^mfl cqsfiatuin and Rh,oi,pQs^-C[m.a i?Pi (fig. VIII and IX). These species are often abundant in well mixed coastal tidal waters; especially S^eJe.tiPn,?F'a costatum is
frequent-ly reported to reach peak densities in spring.
Dominant nondiatons species include Gryptophycean flagellates -Crvptomonas S P P . - with high spring densities, and Euglenophycean flagellates - Entreptiel^a S P . - with peaks in late spring, early summer. Together with unidentified cocpoid \wcells.
and largc Mesodinium S P . they account for most of the observed variation in non-diatom phytocarbon (fig. III).
A relative large part of the total phytoplankton biomass is made up by heterotrophic species, including Gvrod^n^in? spira^e. smq,],! Per-id^nitim s p p ^ Catypomopas spf and Helicostomella subulata (fig.
1.2. Phvtoplankton dvnamics
The absence of a diatom spring bloom, especially in 1978 and 1980, is an intriguing phenomenon. The observed winter silicon maxima of 0.8 to 1.6 mg Si/l could theoretically (neglecting losses) support a dLitom biomass of 50 to 100 mg Chlorophyll-a.m-' (Si/C=0.5 and C/Chlorophyll-a»30), Observed chlorophyll-a level seldomly exceeds 10 mg Chlorophyll-a.m-*.
Af ter the closure of the lake, the shift of dominance from diatoms to Cryptomonad flagellates during spring, was most evident in the years without flushing. During the flushed situation (1979) dia-tom species were more or less dominant again in the lake - as in the former tidal basin (Bakker and Vegter, 1978), This may suggest a factor like decreased turbulence, due to absence of tidal move-ments, playing an important role, by affecting underwater light climate (decreasing turbidity) and, possibly, sedimentation of diatoms.
The shift of'dominance resul ting from a decreased turbulence can be explained by several hypothetical mechanisms, possibly operating sitnultaneously.
^ f ó r si^ïeon frqtuften Planktonic and benthic diatoms. In early spring, when mineralization occurs at a slow rate and the bottom silicon pool is small, competition for dissolved s il icon in the water column is not unlikely. When the micro-phytobenthos, under the prevailing light and temperature con-ditions, is able to start exponential growth before the phytoplankton, the first raay lower dissolved silicon concen-tration to a great extent, and thus prevent a bloom of plank-tonic diatoms.
In a well-mixed waterbody phytoplankton may receive less radiant energy during the dayïight period than microphyto-benthos living at the surf ace of shallow sediments (depth Zb) and receiving a moderate, but more constant, supply of radi-ant energy. It can be deduced (see Add.1) that, for
Zb < 1/k * ln(k»Zmix) (m) (1) -k#Zmix
1- e
k * light attenuation coëfficiënt (m-1)
Zmix * depth of the mixed layer (m)
benthie diatoms have a competitive advantage over planktonic diatoms, concerning light. For k*0,2 m-1 (normal spring
val-ue) and a mixing depth of 10 in, benthie diatoms on bottoms up to a depth of 3.5 m are in advantage. Higher extinctions dur-ing summer (ks0.7 m -1) , among others caused by planktonic
chlorophyll, restrict competitive advantage of benthie algae to bottoms shallower than 2.8 m. Reduced vertical mixing by stratif ication also benefits planktonic algae relatively. Thus, the typical high transparancy in a lake with large shal-low areas, like Lake Grevelingen, especially in combination with non-stratified conditions, could have intensified compe-tition in spring between planktonic and benthie algae, for silicon and maybe nitrogen as well.
Evidence for the uptake of silicon from the water by other organisms than planktonic diatoms does exist. During spring 1978 a strong decrease in silicon content was not accompanied
by a marked diatom biomass increase (De Vries and Hopstaken, 1984). Also in the other years of investigation an increase of planktonic diatoms did not always occur synchronically with a silicon decrease,
sedjmentation loss pate in
Sedimentation of diatoms has been recognized as an important loss process, especially during periods of nutriënt limita-tion (Smayda, 1970; Bienfang and Harrison, 1984; Reynolds, 1984). The diatom species composition of Lake Grevelingen in winter is more or less characteristic for a turbulent situ-ation, and resembles North Sea or Eastern Scheldt species composition. In less turbulent water, notably in spring, these species may experience high sedimentation loss rates, increasing with increasing silicon depletion (fast removal of silicon from the upper waterlayers by sedimentation of dia-toms and diatom detritus and, possibly, by uptake by benthic diatoms), In this way diatom blooms could be suppressed untill summer species take over fDitvlum and
competition
In non-turbulent lake water, and under conditions of high transparency, increasing daylength and still low water tem-perature, motile flagellate populations can grow rapidly (Bakker and De Vries, 1984). The absence of tidal movements may thus have improved the competltive abilities of the
fla-gellates, compared to planktonic diatoms.
Salt stratification as occurred in Lake Grevelingen in 1979, on the other hand, may cause a temporary reduction in mixing depth and thus an improvement in average light cliroate in the mixed layer for the non-motile diatoms (Levasseur et al.,
2. PHYTOPLANKTON DATA ANALYSIS
To obtain more reliable simulation results from the application of' the phytoplankton model, we feit the need to dispose of coëfficiënt values on primary production, related to the specific species com-position of the lake (Baveco and De Vries, 1984). An analysis of the available phytoplankton data was performed, with two main goals:
• Identification of the dominant diatom and non-diatom species in the lake during 1979 and 1980
• oalculation of specific production and respiration coeffi-cients from samples dominated by one species.
This second goal had to be dismissed as unrealistic, since the phy-toplankton community was too diverse in most samples, and the vari-ation in measured specific production too high. Also analyses of the relation betweèn specific primary production and respiration, and temperature and light were unsuccessfull,
Phytoplankton species identification and cell counts for 1978 to 1980 were provided on sheets by DIHO (see Bakker and De Vries, 1984). In table I. the naroes of all species observed in 1979 and 1980 are shown. CelX counts and volume measurements were converted to carbon units using the regression equation of Eppley (Smayda,
1978):
log10(C) * 0.76«log10(V) - 0.352 (diatoms) (2) log10(C) • 0,94*log10(V) - 0.600 (non-diatoms)
in which c is carbon content (ug) and V is volume (urn3), As the
phytoplankton model only is concerned with autotrophic, photosyn-thetic active planktonic species, a distinction was made between autotrophic and heterotrophic species, and benthic and strictly planktonic diatoms (see table VI). In figures I to V the contrib-utions of each group to the total phytocarbon are depicted, and in figures VI to IX the biomass of all individual species during these two years is shown.
For 1979 and 1980 separately, average biomass levels of all species were calculated (table II and III), new average annual phytoplank-ton biomass was computed (tabie IV) and a new calibration phyto-plankton data file was created (table V ) . In figure 20.1 the simulation results of the nominal run are compared to this new cal-ibration data set (which is more accurate than the data file used in this report).
In an attempt to simplify the complicated pattern of alternating biomass peaks, taxonomically and/or ecologically related species were lumped in 8 diatom groups and 7 non-diatom species groups. Again, average annual biomass levels were computed f dr each group
(table VII).
Results
Heterotrophic planktonic algae appear to contribute considerably to the total observed biomass: 129 mgC.m-8 (24/S) in 1979 and 88
mgC.m* (21%) in 1980 (table IV). Peak values oecur in spring and late summer (fig. IV), In 1979 the main species are Gvrodlflium S P J -rale. smal^ Pfr^tHnium S P P . and, less dominant, Favella ehrenber-ffjj, and Helicostomella subulat^a. In 1980 absolute biomass values are lower and main heterotrophic species are laree Perid|nium S P P . and Calvcomonas S P . . and, to some extent, Strombjdium spt and Xiin-:
tinnonsis berpi^a (fig. VI and VII),
Suspended benthic diatoms contribute on average 8% to the total observed phytocarbon, 42 mgC.m-5 in 1979 and 34 mgC.m-* in 1980.
Low peaks occur throughout the year, but most frequent ly in f all and winter, probably as a result of high resuspension of bottorn material.
For the community of autotrophic and strictly planktonic species the situation is more complex (fig. III).
In both 1979 and 1980 the diatom community mainly consists of Qhae-tocftros spp. and pitvlum brjph^wQ^Ü^ (77% of diatom biomass in 1979 and 60% in 1980), DicvJ.iim, dominates in early summmer, Chae-in late summer, Besides these two species, Chae-in 1979 Thalas-siosira. spp, (spring bloom!) and Jfo^psolgnj^ S P P . are important contributors (9% and 8#), In 1980 Eugampj^ zoodiacus.
Rh^izosole-and planktonic pennates f^stgri-QFT^Vla iaponica.
serjatal are relat ively common {9%, 3% and 8/0.
The non-diatom species community consists mainly, in 1979 as well as 1980, of qocffojd y-ceHs (31% and 24% of the total autotrophic non-diatom species biomass), Crvptomonas SPP> (24% and 20%) and
Mesodinium S P P . (11% and 17%). In 1979 also Eutreptiella is an important species (16%), whereas in 1980 unidentif j,e<j| P Te e n rods
occur more frequent ly (20%),
In 1980 Crvpf.o^pnfts dominates during the first half of the year and coeooid u-celis during the second half. Greep rods are common in late summer and large Mesodinium specjes dominate in fall and1
win-ter,
conclusïon»
When yearly averaged contribution to total phytocarbon is used as a criterium, a hierarchy from important to insignificant species can be defined, The coëfficiënt values used in the model will have to represent the dominating species(groups) and literature research, if necessary, can be restricted to these species,
For the diatom community these are: Chaetocerps spp,
Ditvlum
SPP.
F.nrrampia ^ppdiaens Rhlzosolenia SPP.,
For the non-diatom species community: u-celT,s
a S P .
Unident jf ied ff^pgn rods Mosodinium spp.
Comparison uith nomina! run
In figure 20,1 simulation results (nominal run) are compared with the calculated phytocarbon amounts in this new calibratlon data
sec. (N.B. the other calculations presented in this report, are compared with the older data set.)
Apparently, simulated diatoro dynamics well approximate the data; the choosen coëfficiënt values raay thus be representative for the slimmer diatom community consisting of Chaetocerpg and
The resemblance between simulated non-diatom species dynamics and the data set is far less satisfying: only the spring bloom fCrvpto-MUSW.i Rufreptiella^ is reproduced, Observed biomass peaks in fall and winter fcoccoid ^i-cells. green rods. large Mesodinium'i are completely absent in simulation results.
3 . PffOBLEM DSMHITIOH
The main purpose of the model1ing efforts is the development of a simple planktonic algae module for the ecological model of Lake Grevelingen, GREWAQ. The following considerations directed our activities:
• Special attention has to be paid to processes by which phyto-plankton biomass is lost from the pelagic zone. *Thus, formu-lations are incorporated for phytoplankton excretion (or nutriënt release), sedimentation of diatom cells, and, more conventionally, for grazing by zoöplankton and benthic suspen-sion feeders,
• After a provisional calibration, the model ought to describe phytoplankton biomass level, production and biomass dynamics of two species groups (diatoms and other phytoplankton spe-cies) with a reasonable resemblance to measured values. The sensitivity of model results to variation in at least some of the coëfficiënt values has to be checked, in order to obtain some insight in the reliability of simulation results. In this way, the role of for example incorporated diatom sedimentation formulations can be elucidated, A thorough calibration proce-dure and sensitivity analysis will not be performed betore
incorporation of all algae modules in GREWAQ.
• Additionally, the model may be used as a tooi in approaching some questions concerning the phytoplankton of Lake Grevelin-gen, like the shift of dominance from diatoms to other phyto-plankton species during the years of investigation and the absence of a diatom spring bloom in most years.
For example, in 1.2. it is hypothesized that the decreased tur-bulence of the water, after the closure, offers an explanation for these phenomena - through an increase in diatom sinking loss rates, and improved competitive abilities of benthic dia-toms and planktonic flagellates.
The effects of increased diatom sinking rates and an improved underwater light climate can be tested in the model. Benthic diatoms are - for the time being - introduced as a f ore ing function, based on biomass and production measurements. Thus, only the impact of quantitatively different forcing functions on phytoplankton dynamics can be assessed.
Dy varying a season dependent factor (RELSD), different turbu-lence regimes can be simulated tentatively.
. MATERIAL ANP HETHOPS. I: DESCRIPTION OF GREALG
.1. Model ecosvstem structure
In figure 1 the ecosystem structure assumed in GREALG is depicted. In its essentials it is still identical to SEAWAQ's (Verhagen, 1984) conception of the North Sea ecosystem structure.
Compared to the nutrient-kinetics model NUTGRE (De Vries and Hops-taken, 1984), in GREALG both biomass kinetics of the phytoplankton as we11 as pools and fluxes of some relevant nutrients, are
simu-lated numerically.
Compared to calculations made by the overall carbon balance model CABAMOD (De Vries, 1984), an eelgrass + macro-algae and a deposit feeding + sediment surface grazing bottomfauna component are absent in GREALG.
Pools
The following pools (state variables) are taken into account; Phytoplankton groups, in mgC.m-*: PI AT OPHY Suspended PETC PETP DETN Dl AP diatoms
other phytoplankton species detritus pools, in mg.m-3:
carbon phosphorus nitrogen silicon
Bottom detritus pools, in mg.m-4:
ORSC ORSP ORSN PIAS Dissolved P N SI carbon phosphorus nitrogen silicon nutriënt pools, in mg,m-3i phosphorus nitrogen silicon
A pool of dissolved organic compounds (DOC) is not included in the model,
Basic equations
The dynamics of the mentioned state variables are calculated by means of the following, simplified, basic equations. Microphyto-benthos is not included.
For the phytoplankton species groups X(i) (i»1, diatoms, or 2, oth-er species), expressed in carbon ( m g C m -5) :
For suspended detritus, carbon fraction (mgC.m-3):
ADETC - { ECKd,.» + a . K e , . , + (l-EXZP).RGZP) .X ' } ..s
•jfi. U J U ; ( 1 ) ( 4 ) + RTSUS.ORSC/DEP
- {MINw + SED + EXZP.RGZP.FDETZP + RGZB.FDETZB}.DETC
For bottom detritus, carbon fraction (mgC.m-2):
&ORSC - {SED.DETC + l K s , . . . X . . . + (1-EXZB). { 2 X , . J.RGZB +
lij tij ti) (5)
RGZB.FDETZB,DETC}.DEP - (RTSUS + MINb),ORSC
For nutrients, f.ex. nitrogen (mgN.in-*): AN - MINw.DETN + MINb.ORSN/DEP
It (6) ,.CN( i )) + EXZP.RGZP.FDETZP.DETN
- 2 P n( i ). X( i ). C N( i )
In which:
a fraction algal excretion transformed into particulate form C N N/C ratio algae
DEP mean depth, in nu
EXZB excreted fraction zoobenthos EXZP excreted fraction zoöplankton
FPBTZB fraction detritus grazed by zoobenthos FDETZP fraction detritus grazed by zoöplankton Kd death rate algae (d-x)
Ke excretion rate algae (d-1)
Kg total grazing loss rate algae Cd-1)
Ks sedimentation loss rate algae (d-1)
HINb mineralization rate in bottom (d-1)
MINw mineralization rate in water (d-1)
Pn net specific production (d-1)
RGZB grazing rate suspension feeders (d-1)
ROZP grazing rate zoöplankton (d-1)
RTSUS resuspension rate detritus (d-1)
.g. Abiotïe features
In GREALG, Lake Grevelingen is supposed to be a homogeneously mixed waterbas in with a mean depth of 5.3 meter. From the report R1310-11 can be concluded that this assumption holds for the horizontal. In vertical direction however a nutriënt and temperature gradiënt does exist.
In figure 2 the depth to surface-area relation is depicted (after Van der Meulen, 1980). This relation was used to correct phyto-plankton product ion values for the actual bas in morphology.
In the present version of GREALG no loadings of nutrients or parti-culate organic compounds are taken into account, nor removal proc-esses like denitrification and formation of refractory silicon. The overall nutriënt budgets are completely closed. Nutriënt bal-ance calculations are made every time step,
ïrradiance and temparatime
Incident irradiance data from Oostvoorne, measured during the years 1975 to 1980 are used as input for irradiance values (see for example fig. 3.2).
For the water temperature, measurements made on sample station G17 in Lake Grevelingen, for the upper 5 nieter, are used (fig, 3.1). Besides, the possibility exists to use, instead of datasets, forc-ing functions determinforc-ing temperature (T, in °C) and irradiance (RAD, in J.cm-*.d-1):
RAD = 500 - 400*cos(2*3.1416* (TIME+10/365) (7)
T » 10 - 8*cos(2*3.1416*(TIME-35)/365) (8) The relative daylenght forcing function (fig, 4.2) is:
DL/24 - 1/2 - 1/6*cos(2*3.1416*(Time+1 D/365) (9)
Nutrientcycles
Only the nutrients N, P and Si are taken into account, without mak-ing &ny distinction between different chemical forms. Dissolved nutriënt concentrations decrease as a result of uptake by phyto-plankton and benthic diatoms and increase as a result of minerali-zation in bottom and water and excretion by zoöplankton, zoobenthos and phytoplankton. Uptake depends on growth rate and actual stoichiometry of the primary producers.
The mineralization rate of (particulate) detritus depends on tem-perature and is formulated for the waterphase as:
For all nutrients Si, P and N, and for Cf coefficients are equal
(De Vries et al,, 1984). The Q10 amounts to 2.37. Mineralization rate on and in the sediment is supposed to occur at a lower rate:
MINb(T) « RATIO • MINw(T) (d-1) (11)
For the simulation results presented here, RATIO was assumed to be 0.5. Figure 5,1 depicts the resulting values as a function of tem-perature.
Mineralization rate in water is in accordance with literature val-ues (De Vries, 1984). Compared to the previous version of GREALG (Baveco and De Vries, 1984) initially high mineralization rates could be set lower, as a result of incorporation of suspension feeding zoobenthos (and its accelerating effects on nutriënt regeneration).' Mineralization rate in bottont still has to remain higher than most literature values indicate: presumably a conse-quence of absence of separate formulations for excretion by depos-it feeders and grazing bottomfauna in the model,
Silicon mineralization or dissolution occurs at the same rate as for other nutrients (Yamada and d'Elia, 1984). Kamatani (1982) estimated the dissolution rates of about 80 7. of the diatom silica to range between 0.024 and 0.099 d-1, approximately equal to the
rate used in GREALG. We might assume some fraction of the Si-de-tritus to be refractory. However, without any loading in the pres-ent vers ion of the model, this will lead to a steady decrease of the available amount of Si. Therefore formation of refractory silicon is omitted until the incorporation of GREALG in GREWAQ. The same is true for denitrification,
sedïmentation and resuspension of detritus
Sedimentation and resuspension rates depend on turbulence of the water, mainly due to wind action and currents. A reflection of overall variation in turbulence may be found in the season depend-ent factor RELSD (Verhagen, 1984), see figure 4.1:
RELSD = 1 + (RAD-100)/800 (-) (12) where RAD represents the incident irradiance forcing function (7) A verification of the factor RELSD may prove difficult, as for example the behaviour of suspended solids concentrations in winter (a resultant of turbulence-determined processes: resuspension and sedimentation) is erratic and hardly correlated to wind velocities
(Veul and De Vries, 1982).
Sedimentation rate of suspended detritus, assuming an intensively mixed waterbody, is then related to (minimal) settling velocity VSETM (m.d-1) in the following way:
SED » VSETM • RELSD / DEPTH (d-1) (13)
Resuspension rate of bottom detritus, related to (maximal) resus-pension rate RTSUSM (.d-1), amounts to:
The actual coëfficiënt values VSETM and RTSUSM are calibrated on the total annual fluxes calculated in CABAMOD (De Vries, 1984).
4.3. Microphvtobenthos (benthic dïatomsï
Biomass and product ion of benthic diatoms are not simulated in the model. Instead measured values are used (fig. 3.4 and 3.5). These data (Nienhuis and De Bree, 1984) concern the years'1976 - 1980, and are treated in the same way as was done in NUTGRE (De Vries et al., 1984).
Besides, some simulations were done using a uniform microphyto-benthos biomass forcing function for each simulated year (calcu-lated from Nienhuis and De Bree, 1984) in mgC.m-3 (fig. 3.6):
XMFBC « 471.7 + 94.3*sin(2*3.1416*TIME/365) (15) or, with a higher amplitude and highest value in late winter (fig. 3.7):
XMFBC « 471.7 + 283.*sin(2*3.14i6*(TIME+6i)/365) (16) Mortality is calculated from the difference between expected biom-ass (based on production measurements) and encountered biombiom-ass:
AX/At » X(t+&t) - X(t) (mgC.m-'.At-1) (17)
M(t) « p(t) - AX/At (mgC.m-'.At-1)
In which M(t) is mortality, P(t) is net primary production and X(t) is biomass on time t.
At the moment, benthic diatoms are supposed to meet their nutriënt requirements by üptake from the waterphase. It is certainly more realistic to assume them to, profit also from bottom fluxes of min-eralized nutrients.
Dead benthic diatoms are allocated to the bottom detritus pool. Grazing bottomfauna and deposit feeders are lacking in GREALG. If they should have an accelerating effect on nutriënt regeneration in the real lake ecasystem, then their absence in GREALG may be an explanation for the need of still relatively high mineralization rates for bottom detritus.
Benthic diatoms can be assigned a variable stoichiometry (see phy-toplankton), However, the reality of this phenomenon can be ques-tioned, since benthic diatoms are bound to be limited by light or C02-diffusion (Admiraal, 1984) rather than by nutrients. Thus, in the nominal run a fixed stoichiometry is used.
The forced-on production curve may provide some difficulties: especially when dissolved nutriënt concentrations are very low, the calculated uptake by benthic diatoms of nutrients from the waterphase may exceed available amounts, thus inducing negative concentrations, Besides, in spring, competition between microphy-tobenthos and planktonic diatoms may influence phytoplankton
dynamics (see 1,2.)} this interaction can not be simulated when microphytobenthos production and biomass are fixed. Therefore it is strongly recommended to introducé microphytobenthos kinetics in future versions of GREALG (or GREWAQ).
5. MATERIAL AND METHODS. II! PHYTOPLANKTOM KINETICS
5.1. Introduction
Distinction is made between two groups of planktonic algae: dia-toms and other phytoplankton species (DIAT and OPHY), Main differ-ence is the use of silicon for the cellwalls of the former.
Another often applied generalisation concerns adaptations of the species groups to different light optima. Diatoms species prevail in most coastal waters during winter and early spring (Gieskes and Kraay, 1975; Smayda, 1980). They may develop under conditions of low average light intensities, for example due to intensive mix-ing, indicating a low light saturation value (Is1) for diatoms. Flagellates, on the other hand, may develop in clear water and tol-erate higher light intensities (Parsons et al., 1978), indicating a higher light saturation (Is1) and inhibition (Is2) values for flagellates,
Other distinctions made between the species groups include a high-er respiration rate for non-diatoms, a diffhigh-ering stoichiometry and sedimentation affecting only diatoms, The coefficients of the pho-tosynthetic efficiency curve (Is1 and Is2) served - within the assumed constraints - as main calibration coefficients concerning species dynamics,
As analyses of Grevelingen phytoplankton data did not provide rea-sonable estimates of production coefficients, a more elaborate survey of literature data is needed to substantiate some of the coëfficiënt choices (Pgmax, Is1 and Is2). The results presented in chapter 2 show however that this survey needs to include only a restricted number of taxonomically or ecologically related spe-cies.
5.2. Production
5.2.1. Light clïmatt
Incident irradiance
Incident irradiance values are supplied as daily totals (W.m-8
PAR) measured from 1975 to 1980 at Oostvoorne. The photosynthetic efficiency however will be determined by actual light intensities during the daylight period (DL, in hours) and thus by (24/DL) * I (I « mean irradiance in J.cm-8.hr~») when light intensities during
the daylight period are assumed to be constant (Di Toro et al,, 1970).
underwater light elfmate
As primary production to a great extend depends on the available amount of light in the water column, the attenuation of light and thus the extinction coëfficiënt (ke) in (23) will be an important factor in determining simulation results (Peterson and Festa,
1984).
Attenuation of light is generally due to absorption by suspended sediment particles, detritus, the chlorophyll of living
phyto-plankton (self-shading), and to absorption by the water itself and its dissolved substances (background extinction).
Thus the extinction coëfficiënt ke (m-1) can be calculated as
ke = B+C1*[SUSSED3+C2*CDETC]+C3*CPHYTOC] (18) in which
B is the background extinction
SUSSED are suspended sediment particles ' PHYTOC is phytoplankton(carbon)
DETC is suspended detritus, and
et, c2 and c3 are specific extinction coefficients in m*.mg-1 (Di
Toro, 1978).
In the Dutch coastal area, the concentration of suspended sediment particles and detritus is a major factor determining underwater light climate (Verhagen, 1984), In Lake Grevelingen, with its rel-atively transparent water, this seems not to be the case. Secchi-depth measurements (fig. 10,3) show that overall fluctu-ations in disc-visibility in summer are mainly correlated with phytoplankton density. In winter however, high fluctuations in Secchi-depth are apparently not related to chlorophyll concen-trations. Then, concentrations of suspended solids and detritus are main determinants of light attenuation, and variations in Sec-chi-depth may be due to variable wind conditions influencing resuspension and sedimentation rates,
Suspended solids concentrations are not simulated dynamically in GREALG. Light attenuation by suspended sediment particles and suspended detritus are, together with the background extinction, combined in a nonchlorophyll-related component, ko (Kremer and Nixon, 1975; Verhagen, 1984),
Secchi depth (Zs) measurements, made in winter when phytoplankton concentrations are low, provide information needed to estimate this ko, using one of the two following empirically derived relationships:
k « 1.45/Zs (19) (Walker, 1980)
k • 1.29/Zs + 0.109 (20) (Vegter and De Visscher, 1984a)
A mean Zs of 5 m in winter relates to an extinction coëfficiënt amounting to 0.3 or 0.37 (m-1).
To allow for some variability due to turbulence, the season depend-ent factor RELSD (12) is introduced, varying from 1 in winter to 2
in summer (Verhagen, 1984).
When a Unear relation between attenuation and chlorophyll-af
total extinction is given by
where EPSD and EPSO are specific extinction coefficients of resp. dxatoms and other phytoplankton species (n^.mgC-1).
Specific extinction coefficients for phytoplankton(chlorophyll) vary with species, its size, concentrations of accessory pigments, e t c , but a value of abouc 0.02 m2.mgchl-a-1 is commonly used as an
average for natural populations (Peterson and Festa, 1984). Assuming a carbon to chlorophyll ratio of 40 to 1, specific extinction then amounts to 0.0005 m * . m g Ox. Values for diatoms are
assumed to be higher than for other species.
In addendum 2 the use of this extinction function (21) is more closely investigated and an alternative function is ptfoposed.
5.2.2. Photosynthetic efficiency
Maximal specific «rross production
When light and nutrients are abundantly available, photosyntb.esis solely depends on physiological properties of the algal cell. since physiological processes generally depend on temperature, it is not surprising to find a commonly agreed-upon temperature dependence of specific photosynthetic rates and subsequent growth. In literature, a Q10 ranging from 1,9 to 2.3 is frequently found. In GREALG, Eppley's (1972) Q10 of 1.89 is adopted. Maximal gross specific production rate (d-1) is formulated as
Pgmax(T) » Fgmax(20) * exp(0.0639*(T-20)) (22) Pgmax is defined here as the maximal possible production rate dur-ing a light period of 16 hours, with saturatdur-ing light intensities. In some laboratory or dialysis cultures Pnmax(20) (the maximal possible growth rate, i.e. Pgmax minus respiration, at 20°C) val-ues are measured of up to 2 div.,d-x, equivalent to 1.4 d-1
(Epp-ley, 1972; Sakshaug, 1977), but also higher values of up to 2.8 d-1
(Jorgensen, 1979), or even diatom growth rates of 3,6 d-1 are
found. Phytoplankton growth rates for natural waters may amount to 2.3 d-1 (Goldman et al., 1979). Pgmax(20), including respiration,
may then amount to circa 4 d-1. In the calibration runs we used
these rather high Pgmax values.
Maximum specific growth and production rates of phytoplankton tend to decrease with increasing unicellular algae size. Maximum rates normalized to cell size show diatoms generally grow about twice as fast as most dinoflagellates, and other algae fall between these values or below dinoflagellate values (Banse, 1982).
Although the actual distinction between diatoms and non-diatoms made in GREALG, does not coincide with any consequent trend in cell sizes, in the nominal run a somewhat higher Pgmax value for diatoms is assumed. In figure 5.2 the Pgmax curve for Pgmax(20) = 4 d-1 is
depicted.
Photosynthetic efficiency curves
When all other requirements are met, photosynthesis will be-light-limited until a certain irradiance intensity (Is1) is reached, Experimentally derived P/I-curves generally show a line-ar relation between P and I for low light intensities. The dP/dl value is supposed to be relatively constant among all species, when expressed per unit chlorophyll-a, However, conversion to per mg G, the dimension used in GREALG, results in considerable greater
var-iations, due to differences in chlorophyll-a content per cell (Reynolds, 1984).
In GREALG, diatoms experience light saturation at lower light intensities than other phytoplankton species. In Colijn (1982), table V, some irradiance values of light saturated growth are assembled from various sources. Most planktonic diatoms species experience light saturated growth at about 1.8 to 5,4 J.cm-^.hr-* (mean daily irradiance: actual irradiance values during the day-light period being 3/2 to 3x as high). Non~diatoms, comprising mainly flagellate species, are supposed to have a higKer light sat-uration value Is1 and to tolerate higher light intensities (Is2 higher) than diatoms (Parsons et al., 1978),
The values of light saturation constants to a great extent deter-mine the population its 'competitive ability' in the model. As they seem to be species-specific it might.be doubted whether simu-lation with only two species groups may yield results comparable to observed population dynamics, If the assumed general picture, reflected in the choice of Isl and Is2 values for diatoms and non-diatoms, is more or less realistic, some resemblance may nonethe-less be expected, as biomass trends are set by only a limited amount of species (chapter 2 , ) ,
Photo-inhibition, presumably a relevant phenomenon in the of ten highly transparant water of Lake Grevelingen, occurs when irradi-ance intensities exceed a critical value (Is2). A setback of the assumption of a constant light level during the daylight period (Di Toro et al., 1970) is the loss of midday brightness in a daily average; when mean light intensity is lower than the inhibition
light intens ity, product ion may be overestimated.
On the other hand, under conditions of high surface light intensity (on the average much greater than Is2) and low light attenuation, mean production may be severely underestimated, due to the absence of dawn and dusk periods with optimal light conditions (Kremer and Nixon, 1975). For Lake Grevelingen this may be a factor of impor-tance.
The photosynthetic efficiency curve used in GREALG to calculate average efficiencies over the whole water column, is linearized for all light intensities (see fig. 7.1). An easy integratable curve, well-suited for the present purposes, results,
If it is assumed that temperature determines maximal gros"s specif-ic production (Pgmax), while it does not affect the dP/dl-ratio, it will be necessary to adjust the photosynthetic efficiency curve for temperatures (T') deviating from a fixed Standard temperature (here 15°C). This may be done either by recalculating Is1 and Is2 or by simply rescaling the actual light intensity lo, using Pgmax(15)/Pgmax(T') as conversion factor.
As is done in BLOOM II (see Los, 1982, for a more elaborate description) we chose the latter possibility. In figure 3.3 the rescaled incident irradiance values are shown, for one year. The choosen saturation and inhibition coefficients have to be compared with these irradiance values, to obtain information about the rel-ative importance of saturation and inhibition. It appears fhat inhibition will be of importance mainly in spring and early summer. Average photosynttofie efficiency
The average efficiency per day can be calculated by integrating the efficiency values (E) over depth and during the whole day.
Light intensity I(Z,t) at all depths as related to surface incident irradiance, follows from the Lambert-Beer law:
lo(t) * exp(-ke*z) (J.cm-*.hr-i) (23) Using this I(Z,t), the average efficiency over depth (EDEP) at each
time t may be calculated as
/
Zmax
E(Io*exp(-ke*Z)) dZ (-) (24) u
yielding (for a more elaborate description see Los, 1982) when Io a Is2:
E » (1 - (Io/Is1)* exp(-ke*Zmax) (25) + In(ls2/Is1) - (EI*(Io-Is2))/(Im-Is2)
+ (1+(EI*Is2)/(Im-Is2))*ln(Io/Is2))/(ke*Zmax) when Isl £ Io < Is2:
E * (1 - (Io/Is1)*exp(-ke*Zmax) (-) (26) + ln(Io/Is1))/(ke*Zmax)
when Io < Is1:
E * (Io/Is1 - (Io/Is1)*exp(~ke*Zntax))/(ke*Zmax) (27) In which:
ïo is resscaled incident irradiance (J.cm-a.hr-1),
isl and Is2 are saturation and inhibition light intensities
(J.cm-*.hr-x)t
BI is relat ive inhibition at high light intens ity lm (-), ke is the total extinction coëfficiënt (m-1), and
Zmax is the maximal water depth or mixing depth (m).
comparable formulations are included to calculate average effi-ciency in case of I(Zmax) surpassing saturation or even inhibition values.
Averaging over time, assuming a constant daylight intensity (Di Toro et al., 1970) and production efficiency being a function of DL/16, yields
EAVG » DL/16 * E (-) (28) or, as an option in GREALG
where DLSAT represents the time (hr) after which maximal growth rate on a daily base is attained. If it appears to be possible to assign different values of growth rate saturation to different species groups, this feature tnay be incorporated more elaborately in GREALG, in a way comparable to the procedure followed in recent BLOOM II versions (Los and Postma, 1984).
5.2.3. Correction for bathymetry of the lake
When the morphology of the lake is not taken into account, and a mean depth, Zm, is used in calculations of average specific pro-duction over the water column, total productivity tends to be over-estimated.
Mean depth is calculated from surface/depth relationships, Deep water columns (Z > Zm) however comprise a much greater volume of water than shallow water columns (Z < Zm), and when an average pro-duction efficiency is calculated for the whole lake the column efficiency of the deep water columns is decisive. Due to the expo-nential attenuation of light, deeper water columns yield a much lower column-averaged production, when the algae are assumed to be homogeneoüsly distributed, Thus, the resulting overall average production of the whole lake will be lower than calculated using Zm.
The depth-surface area relation of Lake Grevelingen is depicted in figure 2 (Van der Meulen, 1980). In GREALG, average productivity in 6 imaginairy lake segments (columns) with about equal surface areas is calculated separately. The mean of these six values, weighted by volume, represents overall specific production.
If the relation between surface area (A in ha) and depth (z in m) is assumed to be according to the exponent ial function:
A(z) * Ao • exp(-q*z) (30) an analytical solution of (24) becomes available. The q represents the surface reduction coëfficiënt (m-1), This approach is
recom-mended for GREWAQ-2D.
However, for zero-dimensional GREALG, results are expected to dif-fer only slightly from those already obtained with the column approach, showing a corrected production amounting to about 2/3 of the .originally calculated yearly production. A similar value (3/4) has been suggested for the Eastern Scheldt (Klepper, 1984). Simple linear regression plots (fig. 13) show a linear relation-ship between the adjusted and original values. A reduction factor of 0,65 is found for both the diatoms and the non-diatom species group. Neither the P/I-curves (different groups), nor the light climate (different years) seem to influence this correction factor significantly. A 2/3 correction on annual primary production cal-culated for a rectangular waterbasin with mean depth Zm, may thus yield results similar to those obtained by our simple integration-al method.
5.2.4. Respiration
Respiration can be divided in dark respiration and photo respira-tion (Raven, 1976), Photo respirarespira-tion is not taken into account in GREALG as it seems to be rather small and scarcely documented. In a growing cell overall dark respiration includes production of ATP for basie cell metabolism (maintenance respiration) and of ATP, NADPH and carbon skeletons for the formation of new cell mate-rial (growth respiration) (Raven and Bear dal 1, 1981). Maintenance respiration is supposed to depend on biomass, growth respiration on the other hand will depend on specific growth rate. In litera-ture respiration is often reported to be linearly related to spe-cific growth rate (Wetsteyn, 1984). The maintenance respiration is relatively small (to 5 %) compared to growth respiration (Raven,
1976).
In GREALG both growth and maintenance respiration are included (as in Kiefer and Hitchell, 1983):
Ro + b*Pn (d-1) (31)
in which R * specific respiration rate, Ro « specific maintenance respiration rate (d-*)t Pn » specific growth rate (d-exp1.) and b = ratio C lost as C02 in growth respiration and C fixed as cell material. b may amount to 0,25 for diatoms and 0.35 for other phy-toplankton species (Wetsteyn, 1984).
Net product ion is then given by:
Pn « (Pg - Ro)/(1+b) (d-1) (32)
Maintenance respiration rate depends on temperature:
Ro(T) * R20 • exp(O.O69(T-2O)) (d-*) (33) In the present vers ion this rate is supposed to be somewhat higher for non-diatioms tlian for diatoms, namely 0.036 and 0.045 at 20°C. The Q10 amounts to 1,99 for both (values from Wetsteyn, 1984). In figure 5.3 the curves are depicted, resulting when coefficients of the nominal run are used,
The total specific respiration depends on both temperature and the realized productivity and will thus be different for each single simulation run. An indication of the magnitude is given in figure 14.2.
5.2.5. Nutriënt requiremants
In GREALG, as in most phytoplankton models, only the influence of the availability of the macro nutrients N, P and.Si on algal growth is taken into account. In Lake Grevelingen, nitrogen and silicon dynamics are supposed to be closely coupled to biological uptake and release (De Vries and Hopstaken, 1984; Bakker and De Vries, 1984). Data show N and Si concentrations frequently decreasing
below known growth limiting concentrations, Phosphorus dynamics, on the other hand, are hardly influenced by biological processes occurring in the lake, and phosphorus fluxes calculated in GREALG can by no means be correlated to any data on fluctuating
concen-trations.
One of the oldest and simplest formulations quantifying the relation between nutriënt availability and uptake by algae - and subsequently growth - has been the application of the Monod equation describing Michaelis-Menten enzyme kinetics (Dugdale,
1967):
P = Pmax • s/(Ks+S) (d-x) (34)
Where P and Pmax represent actual and maximal specif ie growth rate, S the limiting nutriënt concentration and Ks the half-saturation constant.
The Monod model has been further extended by Droop (1973), includ-ing both external and internal nutriënt concentrations in the for-mulation, thus providing for accumulation in excess of imroediate demands while nutrients are still freely available.
Evidence for luxury uptake and storage of phosphorus, and to a les-ser extent, of nitrogen, does exist. However, silicon uptake by diatoms seems to be little more than is necessary for the next cell division; only insignificant amounts of silicon can be stored in cell parts other than the cell wall (Paasche, 1980).
In GREALG the single Monod equation is maintained, partly because the buffering effect exerted by internal storage is not expected to result in different simulation results. Besides, the (extra) coefficients would be difficult to estimate for muiti-species groups.
An additional assumption made in GREALG, and in many other phyto-plankton models as well (f.ex. Kremer and Nixon, 1975) is the eKistence of one single nutriënt being limiting at a time (Liebig's law), and the absence of simultaneous, additive or multiplicative effects.
Bear ing in mind the f act that phosphorus concentrations in the mod-el stay amply above growth limiting values, the actual nutriënt restriction on growth is given by:
NUTLI « min( N , Si ) (diatoms) (35) KsN+N KsSi+Si
and
NUTLI *> min( N ) (non-diatoms) KSN+N
In general, when using this type of growth limitation formu-lations, simulation results with respect to species-groups abun-dances (relative, not absolute) appear to depend heavily on Ks-values. From literature, a considerable variation in half-sa-turation constants appears (Jorgensen, 1979). Most of them con-cern uptake of nutrients rather than growth, which is not necessarily the same.
Half-saturation constants for ammonium uptake range from approxi-matcly 1 to 1400 ugN.1-1 (with the greater part below 14 ugN.1-1),
If ammonium is available, ammonium uptake is preferred by most algae. Only when concentrations are low, nitrate and ammonium are
used simultaneously as nitrogen sources (Syrett, 1981). In the table, some of the (older) data on half-saturation constants for nitrogen uptake of diatom species present in Lake Grevelingen, are listed.
Half-saturation constants for species Asterionella japonica Asterionella japonica Chaetoceros debilis Chaetoceros gracilis Chaetoceros gracilis Ditylum brightwellii Ditylum brightwellii Ditylum brightwellii Leptocylindrus danicus Skeletonema costatum N NH4 N03 NH4 NH4 N03 NH4 N03 N03 NH4 NH4
nitrogen uptake (ugN
Ks 14.0 14.0 7.0 5.6 2.8 15,4 8.4 8.4 9.8 7.0 reference Eppley Eppley Conway Eppley Eppley Eppley Eppley Parsons Eppley Conway et al. et al. • l-1) 1969 •1969 and Harrison 1977 et al. et al. et al. et al. 1969 1969 1969 1969 j and Takahashi'73 et al. 1969 and Harrison 1977
Phosphate uptake constants measured in natural phytoplankton popu-lations range from 0.6 to 24 ug PO4-P.I-1 (Nalewajko and Lean,
1980). These values are lower than measured in laboratory cul-tures. Under non-steady-state phases, half-saturation constants for growth may be much lower than for uptake.
Paasche (1980, table 7.3) lists data on half-saturation constants for silicon limited growth of some marine diatoms. Values range from 0.02 to 0.98 umol Si(OH)4.1-x, or 0.56 to 27.4 ug Si.1-1.
Ks-values are lower for silicon limited growth than for silicon uptake.
Based on literature data it is difficult to make a justified dis-tinction in half-saturation constants for diatoms and non-diatoms. Therefore, Ks-values for N and P are supposed to be practically equal for both species groups; only in the calibration run the Ks for nitrogen is assumed to be higher for non-diatoms.
Most values used in GREALG (in mg.m-3) lay well inside the reported
ranges for marine phytoplankton and are comparable with values used by Kremer and Nixon (1975) in the Narragansett Bay model.
Half-saturation constants used (ugN.1-1)
group Diatoms other phytoplankton P 0,62 0.62 N 8.0 8.-14. Si 35.-50 * 5.2.6. variabla stoichiomotry
As in NUTGRE (De Vries et al,, 1984) the stoichiometric values of the algae are made dependent on concentrations of dissolved nutri-ents. Summer and spring values for diatoms and mean values for other phytoplankton species were taken from literature. Ratios applying to non-diatom species were supposed to vary within a com-parable range as diatom ratios.
For example the N/C ratio is calculated according to:
CN = CNSU + (N/Nmax)*(CNSP - CNSU) (-) (36) N represents the average concentration of dissolved nitrogen dur-ing the previous (maximal 4) days. Nmax is the winter maximum. CNSU and CNSP are N/C ratio's at minimal and maximal ambient dissolved nitrogen levels, respectively. In the calculations presented in this memo, benthic diatoms are given the stoichiometric ratios of spring diatoms, constant during the year, presuming no nutriënt limitation at the bottom. In figures 6.3 to 6.5 the relations between stoichiometry and ambient nutriënt concentrations are depicted.
The new ratios are calculated at the beginning of each time step. As the new values apply to both the produced, new, biontass as well as the existing biomass, the nutriënt contents of the latter have to be adjusted in the model. The resulting 'uptake' or 'excretion' of nutrients is provided by or goes to the dissolved nutriënt pools.
5.3. Loss proeesges
5.3.1. Xntroduction
During the entire course of a phytoplankton population a certain amount of losses will be sustained, keeping population density at a considerably lower level than should be expected from specific product ion rates.
Loss processes comprise all processes which actively remove biom-ass from the part of a water body under consideration: removal by herbivores, by sinking, vertical and horizontal transport and by cell mortality due to parasitic attacks or due to exposure to phys-iological extremes of light, temperature, nutriënt concentrations and toxic substances.
In GREALG allowance has been made for the rate of removal by graz-ing and sinkgraz-ing, An additional factor is included (namely excre-tion or nutriënt-release) to account for losses due to physiological stress (e.g. nutriënt limitation). All other mor-tality factors are combined in one single, temperature dependent, death rate.
As all removal processes included are instantaneously density dependent, they can be expressed in the same rate terms as growth. Total removal rate Ltot is the sum of all loss rates:
Ktot m Kg + Ks + Ke + Kd (d-1) (37)
in which Kg, Ks, Ke and Kd are loss rates due to grazing, sinking, excretion and 'death'.
5.3.2. Grazïng by Zoobanthos and Zoöplankton
Phytoplankton dynamics and biomass levels in Lake Grevelingen must be directly and to a great extent affected by grazing by zooplank-ton and suspension feeding bottomfauna. In De Vries (1984) it was calculated that, on average, the whole lake volume was filtered by suspension feeders about every 4-5 days.
Biomass
As it lays beyond the scope of the present phytoplankton modelling activities to try to incorporate suspension feeders kinetics in GREALG, the zoöplankton and bottomfauna biomass - and thus the grazing pressure - are included in the model by means of forcing functions,
For zoöplankton, this function ZP (in m g C m -5) , compiled from
Bakker et al. (1978, fig.10), amounts to
ZP • 5 + 105*sqrt(1 - (tanh((TIME-210)/40)**2) (38) with a mean biomass of 41 mgC.m-3 (=.22 gC.m-*), in accordance with
the estimate used in CABAMOD (De Vries, 1984), see also figure 4.4. For filter-feeding zoobenthos, the biomass function is rather uncertain. Some information on fluctuations during the year and estimated mean biomass in the lake are obtained from Verhagen (1983), concerning the mussel, and Wolff et al. (1975). In GREALG the CABAMOD estimation of the maximal suspension feeding bottom-fauna biomass, ZBmax, of about 1500 mgC.m-' (*8 gC.m-*) is main-tained as a guideline, yielding (see also fig. 4.5)
ZB = 1000 - 500*sin(2*3.1416*(TIME+61)/365) (39)
Filtration rates
Next to biomass, the specific filtration rate (F) of the herbivores plays an important role in determining the fraction of phytoplank-ton (and detritus) removed by grazing. In Lake Grevelingen, where phytoplankton concentrations are relatively low, suspension feed-ing bottomfauna may be assumed to be food-limited and to maintain maximal filtration rates, Filtration rates then are independent of food concentrations and functional responses (for example decreas-ing filtration rates with increasdecreas-ing food availability) are negli-gible. For motile and more selectively grazing zooplankters, this assumption is presumably less tenable.
In GREALG filtration rates of both zoöplankton and zoobenthos are assumed to be constant, with a temperature constraint (Verhagen, 1984):
TR » .4 if T i 2 TR • ,4+.1*(T-2) if 2 < T < 8 TR = 1. If T È 8
To prevent reduction of phytoplankton populations to extremely low values, a Monod term is added: comparable to a decreased filtration efficiency at very low food concentrat ions (< ± 200 mgC.m-3), In
at the bottom of suspended organic matter, during periods with reduced turbulence, will be included (Heringa et al., 1985).
The actual filtration rates (F) are taken from CABAMOD (De Vries, 1985) and amount to about 0.0015 for zoöplankton and 0.00015 (m3.mgC-l.d-1) for suspension feeding zoobenthos.
Food
Food is assumed to consist of phytoplankton (grazed without pref-erence for species groups) and a ingested fraction (FDET) of the suspended detritus (inmgC.m-*):
ZBF • DIAT + 0PHY + FDETZB*DETC (zoobenthos) (40)
2PF • DIAT + 0PHY + FDETZP*DETC (zoöplankton) (41) As zoöplankton may be able to filter much more selectively
(Dona-ghay and Small, 1979) than suspension-feeding bottomfauna: FDETZB amounts to ± 1.0, and FDETZP to < 0.1.
The resulting loss rate caused by zoobenthos grazing is;
RGZB • ZB*Fzb*TR * ZBF/CZBF+100) (d-1) (42)
by zoöplankton grazing:
RGZP e» ZP*Fzp*TR * ZPF/(ZPF+100) (d-1) (43)
The total grazing loss rate is:
Kg • RGZB + RGZP ' (d-1) (44)
Allocattan of nutMents
Up till now, no allocation of nutrients to zoobenthos and zooplank-ton biomass pools is assumed in GREALG. However, as minimal and maximal biomasses for suspension feeding zoobenthos amount to about 500 and 1500 mgC.m-1, according to a stoichiometric ratio N/C
» 0,2, the zoobenthos will be responsible for the withdrawal from possible miner&lization of 200 rogN.m-3 in thé period early spring
to late summer. Bearing in mind the low and growth limiting nitro-gen concentration in early summer, this clearly will affect phyto-plankton productivity,
To a lesser extent, the same might be true for the zoöplankton; although its biomass level is much lower, it still may cause a withdrawal of 20 mgN.m-3.
Concerning the suspension feeding zoobenthos, this shortcoming of GREALG is removed by J,Heringa (Memo R1310-77) in a further elabo-ration of the role of zoobenthos in the regeneelabo-ration of nutrients. Of the phytoplankton amount grazed by zoöplankton, a fraction
EXZP • RGZP (d-1) (45)
is excreted and allocated to the dissolved nutriënt pools, except for silicon. The remaining part
(t - EXZP) • RGZP (d-1) (46)
and all silicon, is allocated to the suspended detritus pool (faec-es).
Of the phytoplankton araount grazed by filter feeding bottomfauna, the fraction
EXZB • RGZB (d-1) (47)
of the nitrogen and phosphorus content is allocated to the dis-solved nutriënt pools» while the fraction
(1-EXZB) * RGZB (d-1) (48)
is allocated to the bottom detritus pool.
Forming part of a student project at the Delta Department, efforts are made to quantify excretion coefficients of suspension feeding bottomfauna (EXZB), for carbon, nitrogen, phosphorus and silicon fractions (Heringa et al., 1985).
All suspended detritus grazed by zoobenthos is allocated as pseu-dofaeces to the bottom detritus pool. In this way only the phyto-plankton plays a role in the diet of the filter feeding bott'omfauna
(Williams, 1981).
5.3.3. Sadimentation
Sinking of non-motile algae, especially diatoms, can be a major source of population decline, Incorporation of sedimentation of algae in phytoplankton models, however, provides difficulties, due to high variability, both interspecific and intraspecifie, in observed intrinsic settling velocities, and a frequently observed high sensitivity of model results to sinking rates.
Naturally, size is a main factor in determining sinking veloci-ties, Smayda (1970) a.o. demonstrated a higher settling velocity for large centric marine dlatoms than for morphologically similar smaller ones. The relation is however not linear and adaptive mech-anisms for depressing sinking velocities, including changes in
density and shape of larger species, compensate for increased size.
Some marine species, like Ditvlum britrhtwelHi. one of the domi-nant summer diatom species in Lake Grevelingen, may be able to reg-ulate their density by selectively accumulating univalent Na- and K-ions instead of divalent Ca- and Mg-ions (Andersson and Sweeney, 1978). Others may decrease cell density by formation of mucilagi-nous sheaths. Increasing form resistance, either by changing shape (in general all morphological changes that increase surface area)) or by formation of colonies is another available strategy
(Walsby and Reynolds, 1980).
Viewing all these possible adaptations and bearing in mind the amount of different diatom species observed in Lake Grevelingen (and the accompanying ever-changing species composition), one sin-gle constant representing settling velocity during the year seems not appropriate,
Two observations determined modelHng efforts:
• In marine environments, including the Dutch coastal area, dia-tom blooms almost invariably result in massive sedimentation of diatom cells to the bottom. Nutriënt depletion may be res-ponsable for ending the bloom, probably by increasing settling velocities,
• In Lake Grevelingen, some diatom species (mainly Chaetoceros and Dit;vlvffli) are abundant in summer, when silicon concen-trations and turbulence are low. In winter and early spring, with high silicon concentration and turbulence, the diatom coiranunity consists of other species, a.o. Tha1assiosira spp,t
Rhizosole.n1.flL PPp^ and Skele^pnema CQ.gtatWTl (Van Iwaarden,
1979; see also chapter 2 ) .
Front these observations, two hypotheses with concomitant model formulations are deduced.
• For the first, nutriënt dsplatïon> especially of süicon, increases sinking velocitias. Experimentally, sinking veloci-ties have been shown to be correlated with physiological con-ditions. Carbon depletion and photosynthetic inhibition cause short-term changes, while in the longer run nutriënt deficien-cies have been shown to increase sinking rates (Reynolds,
1984).
For example, experiments of Bienfang and Harrison (1984) with large centric marine diatoms showed a clear increase of about 70 % in sinking velocity af ter 4 days of silicate or phosphate depleted conditions.
In GREALG, therefore, nutriënt concentrations during a fixed number of (maximal 4) days are stored in an array and a moving average concentration is calculated, Using these moving aver-ages, a nutriënt stress index in relation to sedimentation is defined as (see fig. 6,6):
SEDstress = (1 - Si ) « KsS (-) (49) KsS+Si KsS+Si
where KsS is the half saturation constant for silicon uptake. Secondly» 'summer-species' «re bet ter adaptcd to conditions of
•spMng-species' (the latter will profit more from light and temperature adaptations). Under the same experimental condi-tions, in non-turbulent media, 'summer-species' then ought to display lower sinking velocities. tt is difficult to verify this statement, using literature values (given a.o. in Linge-man-Kosmerchock and Los, 1978), Growing populations of Chae.-toceros species and p^ylum. brightwq^lii show sinking velocities of ,5 m.d-1. Senescent or dead cells may sink at a
speed of 5 m.d-1, and resting spores even at 9.6 m.d-1. Data
on sinking velocities of 'spring-species1 of the lake are
unfortunately too scanty to be compared with these" values. On theoretical grounds a distinction between life-strategies of spring-bloom diatoms and 'summer-species' is not unreason-able. smetacek (1985) argues that, for spring-bloom diatoms, an increase in sinking rates at the end of a bloom (under nutriënt depleted conditions) may be ari' adaptive strategy. A fraction of all settled diatoms (possibly resting spores) may survive at the bottom during periods of unfavourable condi-tions in the pelagial, and be resuspended later.
The dlffering properties of 'summer-species' can be simulated by introduction of a new species group to the model. A prefera-ble alternative however seems to be the introduction of a time dependent coëfficiënt, the 'relative sedimentation suscepti-bility', SEDrel, varying from 1. (winter) to a low summer val-ue; for example (see fig, 4.3):
SEDrel * 0.6 + 0.4*cos(2*3.1416*(TIME+10/365) (50)
Summarized, the spring diatom community depends on high turbulence and high silicon concentrations. A gradual reduction in turbulence during the first months of the year will increase average sinking velocities, but need not stop exponential increase in spring as long as nutriënt conditions are favourable. Reduced silicon availability will result in an increasing fraction of spring-dia-tom biomass being removed from the upper waterlayer by sedimenta-tion. Ultimately, net population growth will stagnate as a result of the coupled influence of nutriënt limitation and sedimentation, and better adapted summer diatoms will take over.
Maximum sinking velocities
Maximum sinking velocities observed in vitro for nutriënt depleted diatoms amount to 10 m.d-1 (Smayda, 1970). Intact diatoms in
flakes-like or gelatinous aggregations may even settle at 50 to 100 m.d-1 (Smetacek, 1985). ïn GREALG, a maximum value DVSETM is
assumed of 2.5 to 10 m.d-1. A rough correction for the influence
of turbulence is made by applying the factor RELSD (12) varying from 1. (winter) to 2. (summer). Thus, actual maximal sinking velocity ( 20 m.d-1) amounts to:
SEDmax * DVSETM • RELSD (m.d-1) (51)
The resulting diatom loss rate due to sedimentation is given by: Ks * SEDrel*SEDstress*SEDmax/DEP (d-1) (52)
The sedimentation rate In the model thus depends on • ambient silicon concentration
• diatom community susceptibility, varying during the year
• a maximal sedimentation rate (under complete silicon depletion)
• turbulence, represented by RELSD
In this wayt the actual value of Ks during the year, will be
dif-ferent from simulation to simulation, but essentially the same pattern is supposed to emerge as depicted in figure 14*. 5.
5.3.4. Excretion/nutrient release
Organic substances excreted by algae include glycollate and many carbohydrates, simple amino acids as well as larger ntolecules (Fogg, 1966). Glycollate excretion is known to occur when growth rates are limited in oligotrophic or nutriënt depleted waters, or when the cells are metabolically stressed (Sharp, 1977), Under both types of conditions, the rate of glycollate production may well exceed the rate of metabolic consumption (Harris, 1980).
Commonly, excretion is measured with the C14-method and thus defined as extra cellular release of recently assimilated (la-belled) carbon. A conventional way of presenting excretion rates is as a percentage of total assimilated carbon (PER), These PER-values vary from 0 to 80J?, but on average they do not surpass 10£ of primary production (Wetsteyn, 1984). Some of the available information however suggests increasing PER-values under unfavour-able environmental conditions (Sharp, 1977). Lancelot (1983) found high negative correlations between PER and mineral nitrogen concentrations in flagellate dominated populations. In diatom dom-inated populations in summer, such a correlation was absent,
For Lake Grevelingen measured excretion amounts to 10 gC.m-*.yr-1
and equals about 1% of the total primary production in 1978 (Vegter and De Visscher, 1984). Occasionally, high PER-values were found at the end of spring blooms (f.ex. April 1978).
Excretion, thus defined, can be modelled easily (SEAWAQ: Verhagen, 1984)
Pn * Pg*(1-EX) - Ro-^Pg-Fn = EX*Pg + Ro (53) and does not constitute a loss process in a strict sense, since it does not act as a sink of phytoplankton biomass, but reduces actual .(net) production.
It has been shown that excretion products incorporate N and P, and even Si (pers. comm. Donaghay, 1984). As only carbon excretion is taken into account when a mode11ing approach is adopted as described above, no model information becomes available on stoi-chiometric ratios of the excreted substances: necessary informa-tion for future efforts to model the microbial utilizainforma-tion of dissolved organic substances.
Besides excretion in a strict sense, the impact of algal lysing bacteria and fungal parasites on nutriënt- or light-stressed algal
