Towards end-to-end modeling of the marine food web Wolfgang Fennel Leibniz Institute of Baltic Sea Research (IOW) Warnemünde e-mail: wolfgang.fennel@io-warnemuende.de Motivation, construction of the model system, (WFM), test experiments (incl. eutrophication), comparison with independent data, skill assessment issue (truncated food webs).
The coupled system physics N, P, Z, D,... lower part of the food web Time scales-1 12 months food mortality Fish Time scales 1.20 years loads fishery
For example, what are the reasons for the interannual variations?
1.2 1.0 Eastern Gotland Basin Jan.-April phosphate mmol/m³ ( ) 8.0 Eastern Gotland Basin Jan.-April nitrate mmol/m³ ( ) 0.8 6.0 0.6 0.4 4.0 0.2 2.0 0.0 1955 1960 1965 1970 1975 1980 1985 1990 1995 2000 0.0 1960 1965 1970 1975 1980 1985 1990 1995 2000 Bottom up? Nutrient loads
Normal fish community Courtesy S.Hanson The piscivorous fish removed S.Carpenter UW Madison
copepods copepods zooplanktivores Sprat eggs larvae year classes Herring eggs larvae year classes Fish model Baltic case piscivore Cod eggs larvae year classes Cod, herring and sprat -> 80% of fish biomass
Define size- (or mass-) classes to formulate predator-prey interaction Use Bertalanffy Formula, with fitted parameters, (H and k, carries a lot empirical information) to define consumption and growth rates Map the Bertanlanffy dynamics onto piece-wise constant effective growth rates
mass Interaction of Cod, Herring and Sprat, Feeding limited by an Ivlev function CX 6 =1500 HX 5 =150 CX 5 =800 HX 4 =60 CX 4 =200 HX 3 =30 HX 2 =10 HX 1 =5 CX 3 =60 CX 2 =30 CX 1 =5 SX 4 =20 SX 3 =15 SX 2 =10 SX 1 =5
Predator- prey interaction, example Cod-Herring the predators sees all smaller prey animals
Prey-predator interaction, example herring cod, the prey sees all larger predator animals
Further dynamic ingredients: Metabolism: respirations- and excretion rates transferring part of the ingested food (or bodymass) to nutrients and detritus (rating: good, quantification of parameters can be improved) Reproduction: off-spring approach (rating: reasonable, but needs refinement) Mortality: natural deaths and starvation rates, fishing mortalities, (rating: reasonable, but difficult, partly questionable, needs further consideration) The result is: Warnemuende Food web Model (WFM)
WMF (show just a few equations) Predator (cod): Model-equations for biomass and abundance averaged individual mass m=b/n
Prey (herring): Model-equations for biomass and abundance Feeding on herring reduces prey biomass and numbers!!!
runs over 20 years Exp=9.1 initial number / km 3 Nc = 10 4 - cod NH = 10 5 - herring Ns = 10 6 - sprat ---------------------------------- Initial vectors of Cod, Herring and Sprat MC0 = [0,0, 0,0, 0,0, 0,0, 0,0,Nc*CX5,Nc, 0,0]; MH0 = [0,0, NH*HX1,NH, NH*HX2,NH, NH*HX3,NH, NH*HX4,NH, 0,0]; MS0 = [0,0, Ns*SX1,Ns, Ns*SX2,Ns,Ns*SX3,Ns,0,0]; Pair structure: [, initial mass in gram / km 3, initial number / km 3,...] ---------------------------------------------------------------------------------------------------------- Fishing mortality applied to larger mass classes F C6 = 6.8/10 4 /d; F C7 = 8/10 3 /d; F H5 = 1/10 3 /d
General increase of ind.-number, (prey development not limited by food (Z) )
Fishing pressure affects reproduction, (no reprod. in years 12-16) Interannual variations of Catches General increase in fish biomass due to missing limitation of food for prey, (bottom up)
Link to lower food web
Light, T P Z fish Truncated model N D Truncated model l ZD = 0.03 /d adjusted zooplankton mortality,
Coupling the NPZD-model to fish - three channels Respiration of fish Feeding of fish on Z 1mmolC => 12 mgc => 100 mg = 0.1 g wetmass conversion Fish mortality feeds back into D
Exp 30.15 mor_opt=1 initial distribution No fishing mortality Z-mortality rates [1/day] l ZD =0.02/d; Total Balance
Exp 30.15 mor_opt=1.1 high fishing mortality Total balance No external loads Mortality rates [1/day] l ZD =0.02/d; Fishing mortalities [1/day] FmortC 6 = 6.8 10-4 ; FmortC 7 = 8 10-3 ; FmortH 6 = 2.7 10-4 ; FmortS 5 = 2.7 10-4 ;
High fishing mortality, No external loads Exp=30.15; option_mort=1.1; import_n=0, Indication of a trophic cascade, or just a decline due to removal of mass? W.Fennel, Jour. Mar. Syst. (2007) in press
Exp=30.15; option_mort=1.1; import_n=0
Exp=30.15; option_mort=1.1; import_n=0.0063g/m³/d (0.003 mmolc/m³/d),
Exp=30.15; option_mort=1.1; import_n=0.0063g/m³/d ( 0.003 mmolc/m³/d),
Note Increase in nutrients correlates with catches until the mid 1980ties (F. Thurow 1997)
Catches provide independent data (Volume of the central Baltic ~ 13 10³km³) Use the three example runs: NO 3 ~ 2-2.5 mmol/m³, modelled total catches: 37 tons/km³ amounts to 481 10³ tons Nitrate level was observed around 1965 Catch data ~ 500 10³ tons, in the 1960ties, --------------------------------------------------------------------------------------------------------------------- NO 3 ~ 3 mmol/m³, modelled total catches: 60 tons/km³ amounts to 780 10³ tons Nitrate level was observed around 1970-75 Catch data ~ 800-850 10³ tons, in the 1970ties, --------------------------------------------------------------------------------------------------------------------- NO 3 ~ 5 mmol/m³, modelled total catches: 100 tons/km³ amounts to 1300 10³ tons Nitrate level was observed around 1975-85 Catch data ~ 950-1000 10³ tons, in the 1980ties, in this simple model, catches are controlled by nutrients Overall values of the catches are consistent!
However The nitrate level of ~ 5 mmol/m³, was also observed in1985-95 (implying modelled total catches ~ 1300 10³ tons) but Catch data dropped to ~ 500 10³ tons, in 1985-95! Clearly, the model is in the current stage too simple, to mimic recruitment failures through combined actions of fishing and oxygen depletion in the halocline, etc.
Opportunity Exp=30.15; option_mort=1.1; import_n=0.0063g/m³/d (0.003 mmolc/m³/d), Skill assessment of truncated models virtually identical results for: N-load: dn/dt ~ 3 10-3 mmolc/m³/d, l Z = 2 10-4 /d, extra Z mortality, and D-loss - flux into sediments: dd/dt ~ - 1.25 10-3 mmolc/m³/d or extra Z-mortality, l Z = 4. 1 10-4 /d, and no D-loss
Issues & challenges: consolidation of parameter choices, step by step increase of complexity of the NPZD component oxygen dynamics phytoplankton succession state resolved copepods Higher resolution of reproduction processes (refine the off-spring approach) Higher order interaction prey feed on predator eggs cannibalism spatial explicit model migration from spawning to nursery region etc. behavior (forage, environmental preferences, etc)