Increasing temperatures change pelagic trophodynamics and the balance between pelagic and benthic secondary production in a water column model of the Kattegat

Journal of Marine Systems 85 (2011) 57–70

Contents lists available at ScienceDirect

Journal of Marine Systems j o u r n a l h o m e p a g e : w w w. e l s ev i e r. c o m / l o c a t e / j m a r s y s

Increasing temperatures change pelagic trophodynamics and the balance between pelagic and benthic secondary production in a water column model of the Kattegat Marie Maar ⁎, Jørgen L.S. Hansen National Environmental Research Institute, Aarhus University, Department of Marine Ecology, Frederiksborgvej 399, P.O. Box 358, 4000 Roskilde, Denmark

a r t i c l e

i n f o

Article history: Received 13 April 2010 Received in revised form 24 November 2010 Accepted 26 November 2010 Available online 7 December 2010 Keywords: Ecosystem model Spring bloom Warming Zooplankton Benthos

a b s t r a c t Temperature constrains the various processes in marine ecosystems differently and future climate warming of up to 6 °C will therefore change ecosystem functioning at various trophic levels. This study investigates how increased temperatures would change the present time overall trophic interactions with emphasis on the balance between pelagic primary and secondary productions and between pelagic and benthic secondary productions. A 1D coupled hydrodynamic-biogeochemical model was calibrated against data from a monitoring station in the Kattegat from 2004 to 2006 and validated in a hind-cast study for the period 1994– 1996. Climate warming scenarios (+ 3 and + 6 °C) showed that the magnitude and duration of the spring bloom were reduced due to higher grazing impact by heterotrophic plankton. Moreover, sedimentation rate decreased up to 44% due to lower sedimentation of phytodetritus after the spring bloom and higher heterotrophic respiration in the water column. The lower food supply to benthos and enhanced respiration due to higher temperatures reduced the biomass of deposit feeders by 23–66% whereas benthic filter feeders were less affected. The onset of the spring bloom was not changed since it was triggered by the light regime above a permanent halocline at 10–15 m depth. This study showed that climate warming presumably will change the trophodynamics of primary and secondary production and will alter the balance of the ecosystem towards a higher pelagic and a lower benthic secondary production. © 2010 Elsevier B.V. All rights reserved.

1. Introduction Climate warming (~ 0.74 °C) has occurred in most regions of the northern hemisphere over the past century and has been predicted to increase further by 1.1–6.4 °C over the next 100 years according to the Intergovernmental Panel on Climate Change (IPCC, 2007). The temperature of the surface of the oceans will increase correspondingly and thereby influence the marine ecosystems. Temperature constrains the various processes in the marine ecosystems differently and therefore a general warming of the water column will change trophic interactions and ecosystem functioning (Beaugrand and Reid, 2003; Alheit et al., 2005; Hays et al., 2005). In temperate seas, the seasonal succession of plankton provides one example of how the temperature influences autotrophic and heterotrophic processes differently. The winter–spring blooms of diatoms are triggered by the high nutrient availability and increasing light irradiance and not directly by temperature (Andersson et al., 1994; Iriarte and Purdie, 2004). However, thermal stratification affects bloom development indirectly by reducing the depth of the upper mixed layer and hence the vertical mixing of phytoplankton below the critical depth for photosynthesis (Sverdrup, 1953; Huisman et al., 1999). This can be seen in the western Baltic Sea, where the ⁎ Corresponding author. Tel.: + 45 4630 1200; fax: + 45 4630 1114. E-mail address: [email protected] (M. Maar). 0924-7963/$ – see front matter © 2010 Elsevier B.V. All rights reserved. doi:10.1016/j.jmarsys.2010.11.006

spring bloom typically is initiated in March–April after the convective mixing has ceased and thermal stratification starts to establish (Fennel, 1999). However, in the neighbouring Kattegat, the spring bloom is initiated earlier in February–March (Olesen, 1993; Maar et al., 2002) because there is a persistent halocline at about 15 m depth (Andersson and Rydberg, 1988). In this area, the year-to-year variability of the onset of the spring phytoplankton bloom is therefore not related to thermal stratification but depends on the weather conditions that govern the overall light regime in the upper mixed layer such as cloud cover and wind mixing (Iriarte and Purdie, 2004). In contrast, heterotrophic plankton (bacteria and zooplankton) are strongly dependent on temperature and the typical time-lag between the peak in primary production during the spring bloom and the corresponding response of the zooplankton is expected to be reduced at elevated temperatures (Oviatt, 1994; Keller et al., 1999; Scheffer et al., 2001). Zooplankton could therefore benefit from the increased temperatures due to a faster growth during phytoplankton blooms. This is the case for protozooplankton that have high growth rates and can always respond rapidly to both increasing temperatures and food conditions (Müren et al., 2005; Aberle et al., 2007). In contrast, mesozooplankton, have a lower growth response and for many species it is critical that the timing of the spring bloom matches with the timing of their reproduction otherwise there is the possibility that copepod nauplii will experience food limitation (Sommer et al., 2007; Winder and Schindler, 2004a). Mesocosm experiments with natural sea water

58

M. Maar, J.L.S. Hansen / Journal of Marine Systems 85 (2011) 57–70

have shown an increasing capability of protozooplankton and mesozooplankton to control the winter–spring phytoplankton bloom at elevated temperatures (Müren et al., 2005; Aberle et al., 2007; Sommer et al., 2007). The increased temperatures are therefore expected to increase the degree of heterotrophy in the planktonic food web although the composition of the heterotrophic plankton may change with implications for the entire pelagic food web (Müren et al., 2005). Increasing temperatures could also change the balance between pelagic and benthic secondary production. Sedimentation of phytodetritus after the spring bloom contributes with a significant part of the total annual sedimentary input to the bottom (Wassmann, 1990) and the benthic fauna depends on this input. Sedimentation rate of organic matter during the spring bloom has been shown to decrease due to a higher zooplankton grazing impact and bacterial respiration in the water column if the temperature increases (Keller et al., 1999; Müren et al., 2005). Thus, the total sedimentary input to sustain the benthos may decrease if more material is channelled through the pelagic grazing food chain (Eriksson Wiklund et al., 2009). Higher bottom water temperatures will also increase pelagic microbial remineralisation of the settling particulate organic matter and this effect will be more pronounced the deeper the water column and the longer the sinking material is exposed to pelagic respiration (Hansen and Bendtsen, 2006). Changes in the temperature could probably also change the species composition of the benthos according to their feeding ecology (Coyle et al., 2007). While benthic filter feeders have first access to the sedimenting food and therefore probably will be less affected by a lower food supply, deposit feeders are more likely to experience food limitation (Josefson and Conley, 1997). Altogether, increasing temperatures may alter the functioning of all trophic levels in a cascade from the primary producers to the higher trophic levels such as fish (Alheit et al., 2005). In order to predict the effects on the marine ecosystem of a warmer climate in the future, it is obviously essential to know how temperature affects the processes at different levels in the marine ecosystem (Winder and Schindler, 2004b; Alheit et al., 2005). In addition to changes occurring at the level of overall trophic interactions, the effect may also be seen at species level where for example species-specific temperature preferences affect the species composition and the species composition may itself shape the ecosystem function as a secondary temperature effect (Smetacek and Cloern, 2008). However, species-specific adaptations and interspecific competitive ability among organisms sharing the same overall ecological function are in general unknown, but may dampen the overall ecosystem response to climate change. However, we assumed that at the level of functional organism groups (e.g. no species specific competitive responses) the influences of increasing temperatures are more predictable and may in some instances be inferred from patterns observed across climate zone during the present day climate. The aim of the study was to investigate how increasing temperatures due to climate change would change the present time overall trophic interactions with emphasis on the balance between primary and secondary production and between pelagic and benthic secondary production. By using a coupled hydrodynamic-biogeochemical 1D water column model, we tested the following hypothesis: 1) increased temperatures will increase heterotrophic respiration relatively more than primary production during the winter–spring bloom and thereby reduce the magnitude of the phytoplankton bloom, 2) because more organic matter is respired in the water column, the sedimentary input of organic matter to the benthos will decrease if the climate gets warmer and 3) reduced sedimentary input to the bottom will affect the composition of the benthic fauna.

Baltic Sea (Fig. 1). The hydrography is characterised by a general estuarine circulation sustained by an outflow of brackish water from the Baltic Sea and a transport of high saline water from the North Sea/ Skagerrak toward the central Baltic Sea separated by a persistent halocline at 15 m depth (Andersson and Rydberg, 1988). The surface layer is well mixed and the seasonal thermocline largely coincides with the halocline. The flow through the area is mainly forced by differences in sea levels between these two areas, whereas tidal currents are weak. This study focuses on the Southern Kattegat where monitoring data for model calibration and validation were obtained from Gniben Station (56°07′N, 11° 10′E, Fig. 1) with a water depth of 48 m. The Kattegat is of estuarine character affected by man-introduced eutrophication (Richardson and Heilmann, 1995) and primary production is nitrogen limited during the summer period (Graneli et al., 1990; Richardson and Christoffersen, 1991). However, during the period from 1993 to 2006 the winter surface concentrations of DIN decreased significantly from 9 to 6 mmol m− 3 (Fig. 2b), and bottom sea water temperatures have increased significantly with about 1.5 °C at the study site (Fig. 2a).

2.2. Model set-up We applied a coupled model system consisting of the hydrodynamical model COHERENS model ver. 8.4 (Luyten et al., 1999), a pelagic biogeochemical model (Lee et al., 2002), and a one-layer benthic model (Lee et al., 2002). The model was set up in 1D for a 48 m deep water column with a vertical resolution of 1 m and a numerical time step of 300 s and calibrated for the years 2004–2006. The model was forced with hourly meteorological fields of air temperature, wind speed and direction, cloud cover and relative humidity provided by an operational atmospheric model for the considered years (Brandt et al., 2001). We selected a k–ε turbulent closure scheme without limiting conditions to achieve a more diffusive thermocline according to observations (Fig. 3). The default Total Variance Diminishing (TVD) scheme was selected for vertical movement of scalars. Tidal mixing was included using an amplitude of 0.12 m and a phase of 265° obtained from a nearby station (DMI, 2002). Salinity was included in the model by a relaxation to linear interpolated measurements (Bendtsen et al., 2006) with a timescale of 5 days. Bottom oxygen concentrations are mainly dependent on advection of bottom water from the Skagerrak–Kattegat front (Bendtsen et al., 2009) and were therefore also relaxed to linear interpolated measurements with a timescale of 10 days. This means that bottom oxygen concentration

2. Methods and material 2.1. Study area The Kattegat is a coastal sea located between northern Denmark and Sweden in the transition zone between the North Sea and the

Fig. 1. The Kattegat. The stations for bottom fauna are black and the stations for pelagic variables are white.

M. Maar, J.L.S. Hansen / Journal of Marine Systems 85 (2011) 57–70

11

59

a)

10

o

C

9 8 7 6 5 21

mmol-DIN m-3

18

b)

15 12 9 6 3 0 92

93

94

95

96

97

98

99

00

01

02

03

04

05

06

07

years Fig. 2. Data from Gniben St. a) Average (±SE) measured annual bottom temperatures. Regression analysis: df = 13, R2 = 0.49, p b 0.01. b) Measured surface concentrations of NO3 during winter (1 January–25 February). Regression analysis: df = 36, R2 0.17, p b 0.01.

Fig. 3. Measured a) salinity and b) temperature and modelled c) salinity and d) temperature.

60

M. Maar, J.L.S. Hansen / Journal of Marine Systems 85 (2011) 57–70

Fig. 4. The modified version of the 2MPPD model includes large protozooplankton (LPZ) feeding on MP1 and MP2, mesozooplankton feeding on MP1, MP2, phytodetritus and LPZ, and benthos consisting of filter feeders, surface deposit feeders and sub-surface deposit feeders. The filter feeders graze on MP1, MP2, phytodetritus and detritus.

was only partially determined by internal processes in the model, but was constraint by measured values. 2.3. Biogeochemical model The biogeochemical model was the 2-MicroPlankton-Phyto-Detritus (2MPPD) model (Lee et al., 2002; Tett and Lee, 2005) describing sea water concentrations of NO3, NH4, SiO2 and O2, the cycling of C, N and Si through microplankton and (phyto)detrital compartments and with a dynamic coupling between pelagic and benthic biogeochemical processes. The model was extended with new state variables describing growth of ‘large protozooplankton’ (LPZ), copepods and three groups of benthos (Fig. 4). The two microplankton groups relevant for the Kattegat were diatomaceous microplankton (MP1) and ‘flagellatary’ (or microbial loop) microplankton (MP2). The microplankton compartments included a constant ‘heterotrophic fraction’ consisting of bacteria and small protozooplankton that were assumed to have the same net growth rate as autotrophic organisms (Tett, 1998; Lee et al., 2002). Due to the heterotrophic fraction within each microplankton compartment, it was assumed that dissolved organic matter (DOM) and inorganic nutrients excreted by microplankton or leaked due to ‘sloppy’ feeding by protozooplankton were re-assimilated very rapidly with turn-over times of less than a day thereby neglecting the existence of a labile DOM pool. The autotrophic fraction was high (0.875) in MP1 dominated by diatoms and relatively lower (0.6) in MP2 thereby reflecting the more recycled production when flagellates dominate the plankton (Table 1). The uptake of dissolved inorganic nutrients by phytoplankton was described by Michaelis–Menten kinetics and was dependent on the internal nutrient-to-carbon ratio which allowed N and Si-excretion (negative uptake) if the internal quotas exceeded a maximum.

Microplankton growth rate was calculated on the basis of the cell quota threshold limitation theory (Droop et al., 1982) as the minimum growth rate predicted from light, nitrogen or silicate (only diatoms). The state-variables LPZ and copepods with internal varying C:Nratios (ingested Si became defecated) were included in the model to account for heterotrophic growth uncoupled from that of autotrophs. Growth was described as: dB = ðI × AE × R−β−M Þ × B dt

ð1Þ

where B = biomass, I = ingestion rate, AE = assimilation efficiency, R = active respiration, β = basal respiration and M = mortality. Ingestion rate (ILPZ) of LPZ was described as a Michaelis–Menten saturation curve: ILPZ = Imax ×

C C + Km

ð2Þ

where Imax = maximum ingestion rate, Km = half saturation constant and C = food (MP1 and MP2) concentration. Mortality of LPZ was due to the predation by copepods. Ingestion by copepods Icop was described by a modified Ivlev formulation (Fennel, 2001; Maar et al., 2009):    2 2 Icop = Imax × 1− exp −Iv C

ð3Þ

where Iv = Ivlev constant and C is the sum of MP1, MP2, LPZ and phytodetritus concentrations multiplied by the respective grazing preferences of 0.8, 0.8, 1.0, and 0.1. The parameters were based on literature values and mortality rates were found by fitting simulated biomasses of LPZ and copepods to the observed levels in the calibration period (Table 2).

Table 1 Parameter values for MP1 and MP2. The values marked with ‘*’ are fitted to observations in the present study. More details can be found in Lee et al. (2002). Functional group

Autotrophic fraction

Max. growth rate (20 °C, day− 1)

Sinking or swimming rate (m day− 1)

Stickiness (ratio)

Respiration slope (ratio)

Basal respiration rate (day− 1)

Q10

MP1 MP2

0.875 0.600*

1.5* 2.3

− 0.5–(− 5.0) 2.0

0.1* –

0.781 1.850

0.03 0.10

1.0 2.8

M. Maar, J.L.S. Hansen / Journal of Marine Systems 85 (2011) 57–70 Table 2 Maximum ingestion rate (Imax, day− 1), half saturation constant of ingestion (km, mmol C m− 3 or mmol C m− 2), assimilation efficiency (AE, fraction of ingestion), excretion or respiration (R, fraction of ingestion) and basal respiration (β, day− 1) at 20 °C, Q10 and fitted mortality rate (M, day− 1) for large protozooplankton (LPZ), copepods, benthic filter feeders, surface deposit feeders and sub-surface deposit feeders. Mortality rates were obtained by fitting the model to observations except for LPZ which was grazed by copepods. The km for copepods was estimated as 1/Iv (Eq. (3)). Functional group

Imax

km

AE

R

β

Q10

M

Large protozooplanktona, b Copepodsc Benthic filter feedersd Surface deposit feedersd Sub-surface deposit feedersd

5.0 1.23 0.12 0.15 0.15

40 15 25 165 330

0.50 0.70 0.65 0.42 0.20

0.50 0.50 0.35 0.35 0.35

0.0270 0.0270 0.0036 0.0036 0.0036

2.8 4.8 2.0 2.0 2.0

– 0.14 0.0012 0.0012 0.0018

References: a) Nielsen and Kiørboe, 1994, b) Hansen et al., 1997, c) Fennel, 2001, and d) Blackford, 1997.

The model considered slow-sinking detritus and fast-sinking phytodetritus. Phytodetritus was formed by shear-driven aggregation of diatoms, using a simple algorithm for bulk processes (Jackson, 1990). Sedimentation of (phyto-)detritus and microplankton entered a superficial fluff-layer (Fig. 4) from where the detritus got gradually buried due to bioturbation and sustained a pool of detritus in the consolidated sediment (Lee et al., 2002). The spin-up time was three years in order to obtain initial conditions for the labile organic pools in the sediment in CAL. The model did not include a large pool of old refractory carbon compounds which are normally a part of the diet of the deposit feeding macrofauna. The sediment pool of organic matter (C, N and Si) was mineralised by a temperature-dependent first-order decay rate supported by the oxidant flux of O2. If the carbon mineralisation exceeds that supported by the oxidant flux, denitrification takes place. The elementary constituents were transported back to the water column by a temperature-dependent exchange rate that took into account the enhancement of pore–water molecular diffusion by macro-benthic pumping (Lee et al., 2002). Benthos were included in this modified version of 2MPPD-model by adding new state-variables describing the dynamics of i) benthic filter feeders, ii) surface deposit feeders and iii) sub-surface deposit feeders. Growth and ingestion rates of benthos were derived from Eqs. (1) and (2), respectively, using reported values of AE, R, β, Imax and Km (Table 2). Filter feeders ingested microplankton and (phyto-) detritus in the bottom layer of the water column, while surface deposit feeders ingested the fresh detritus in the fluff layer and finally the sub-surface deposit feeders ingested buried detritus in the sediment. Bio-deposition was added to the detritus pools in the sediment, while excretion of nitrogen (Si was not assimilated) was added to the NH4 pool of the bottom layer in the water column. The mortality of benthos was found by adjusting the benthic biomass assuming semi steady-state conditions during the calibration period (Table 2) and these rates largely correspond to turnover rates described in Pedersen et al. (2008). Mortality due to hypoxia was not included in this model and was irrelevant for that particular station during CAL. All state variables can be found in Fig. 4 and Table 3. All biological pelagic and benthic processes (e.g. nutrient uptake, growth, grazing, and remineralisation) affected by temperature followed an Arrhenius temperature function given as:

FT = exp

  ðT−Tr Þ × Q10 10

ð4Þ

where T is the actual temperature, Tr is the reference temperature (20 °C), and Q10 is the temperature coefficient (Tables 1 and 2). Growth of MP1 was, however, assumed not to be temperature dependent because they are adapted to low temperatures and generally limited by light (Andersson et al., 1994). The sensitivity of

61

Table 3 Initial values for the state variables in the model (MADS, 2008). C:N-ratios were 106:16 for plankton and 106:11 for detritus. N:Si-ratios were 2:1 for MP1 and detritus. Units

Initial conditions

Pelagic variables Salinity Temperature NO3 NH4 Si MP1 MP2 Protozooplankton Copepods Detritus Phytodetritus

°C mmol mmol mmol mmol mmol mmol mmol mmol mmol

−3

Nm N m− 3 Si m− 3 C m− 3 C m− 3 C m− 3 C m− 3 C m− 3 C m− 3

Sediment variables Fluff-layer detritus Sediment detritus Benthic filter feeders Benthic surface deposit feeders Benthic sub-surface deposit feeders

0–13 m

14–48 m

17.0 2.0 4.5 2.0 13.5 0.3 3.0 0.6 0.3 0.1 0.001

29.0 8.0 6.0 0.1 9.5 0.2 1.5 0.3 0.3 0.1 0.001

Initial conditions mmol mmol mmol mmol mmol

C C C C C

−3

m m− 3 m− 2 m− 2 m− 2

14 50 480 65 65

the temperature response of selected variables was tested by increasing the Q10 values (Tables 1 and 2) by 10%. There is a net input of nutrients to the Kattegat and Belt Sea with the advection of bottom water from the Skagerrak area (Ærtebjerg et al., 2002) contributing to the maintenance of enhanced nutrient concentrations in the bottom layer. However, within the Kattegat and Belt Sea there are no strong horizontal nutrient gradients in the bottom water and thereby the conditions at the study site resemble those in the rest of the Kattegat and Belt Sea. The nutrients are gradually mixed into the productive surface layer sustaining primary productivity in the area. In order to account for advection of nutrients in the bottom water, fluxes of NH4 (0.01 mmol m− 3 day− 1 July– August) and Si (0.2 mmol m− 3 day− 1 in December) were added to the model at the bottom mixed layer obtained by fitting to observations. In addition to advective input, atmospherically depositions of nitrogen (NH4) were added to the top layer of the model based on monthly data from an air pollution model (Hertel et al., 2007). The initial values of temperature and pelagic biogeochemical state-variables corresponded to typical winter conditions observed in the area at two depth strata 0–13 m and 14–48 m (Table 3). 2.4. Model scenarios We used a coupled hydrodynamic-biogeochemical 1D water column model under the assumption that a 1D vertical model was appropriate to capture the most important biogeochemical dynamics of this particular system characterised by persistent stratification and a weak tidal forcing (Vichi et al., 2004). The model was calibrated for the period 2004–2006 (CAL) and validated in a hind-cast scenario (HIND) against observations from 1994 to 1996. HIND refers to the meteorological forcing during 2004–2006 but with 1.5 °C lower water and air temperature and 40% higher winter concentrations of nitrate and ammonium compared to CAL. Further, we wanted to disentangle the effects of increased loadings and decreasing temperatures on the food web in HIND and we therefore conducted a scenario with 1.5 °C lower air temperatures (−1.5AT) and a scenario with higher winter concentrations (+NSi) of DIN (40%) and silicate (25%) according to observations. Changes in nutrient loadings in the area were assessed by changing the winter concentrations due to the limits of a 1D set-up that ignores advective processes. The two warming scenarios were conducted in order to test how increasing temperatures may change the function and structure of the marine ecosystem according to our hypothesis. Climate models predicted a warming of air temperatures

62

M. Maar, J.L.S. Hansen / Journal of Marine Systems 85 (2011) 57–70

of 1.1–6.4 °C for this area (IPCC, 2007) and we therefore increased air temperatures and initial water temperatures + 3 and +6 °C in the scenarios 3AT and 6AT, respectively. Nutrient loadings were unchanged from CAL. The rest of the forcing variables (wind, cloud cover, etc.) were kept the same in the scenarios as in the calibration period due to the assumption that temperature and nitrogen loadings were the main drivers of the system. The results of these scenarios were compared with the model scenario of the present day conditions (CAL) as the difference between annual averages or timing in days. For benthos, there was a lag-time of about one year in the response to changed temperatures and therefore we only compared the results from the last two model years. 2.5. Error statistics The model was calibrated and validated against monitoring data that are freely available from the NERI (National Environmental Research Institute) web database (MADS, 2008). In order to evaluate the performance of the model, we compared measured and modelled physical and biological variables for CAL (2004–2006) and HIND (1994–1996). The selected variables were; i) temperature at three depths (1, 15 and 42 m), ii) ammonium, nitrate, and Chl a concentrations (surface and 15 m), iii) biomass (1–10 m) of diatoms, autotrophic flagellates, protozooplankton and copepods and iv) depth-integrated (0–48 m) net primary production. Respiration was assumed to be 14% of the measured gross primary production as evaluated in Carstensen et al. (2003). Chl a concentrations were derived from the autotrophic N-biomass in MP1 and MP2 using a conversion factor of 2.5 mg Chl a mmol-N− 1. Silicate is not limiting for primary production in this area and was not included in the analysis. The observed biomasses of protozooplankton (ciliates and heterotrophic dinoflagellates) corresponded to the modelled sum of LPZ and 50% of the heterotrophic fraction of MP2 assuming the other 50% to be bacterial biomass (Maar et al., 2002). For HIND, there were no data on phytoplankton and protozooplankton biomass.

36

The first statistical test was correlation analysis (Pearson, p b 0.05), where R2 was used as a measure of the model's capability to catch seasonal changes (Allen et al., 2007). The second statistical test was the Percentage Model Bias (PMB, the sum of model error normalised by the data) which is an indicator of the model performance: n

∑ ðO−MÞ

PMB =

i=1

ð5Þ

× 100

n

∑ O

i=1

where O are observations, M are model results and n is the number of data pair (Allen et al., 2007). PMB is a measure of whether the model systematically under- or overestimates the observations and we defined a minimum performance criterion of the model as |PMB| b 40% (Allen et al., 2007). 3. Results 3.1. Model calibration 2004–2006 The water column was stratified by a permanent halocline at 10– 15 m depth (Fig. 3a and c) except for a strong mixing event in February 2005 (Days 380–430) which was also reproduced by the model although this gave some deviations from temperature measurements (Fig. 5). Temperature stratification began in April and was mixed out in October (Fig. 3b and d). Bottom water tended to be colder during autumn in the model in comparison with measurements, probably due to southward advection of warmer bottom water from the Skagerrak during the autumn. The performance (PMB) of the calibrated model (CAL) was acceptable for the tested variables except for surface NH4 and diatom concentrations (Table 4). The seasonal variability, expressed as R2, was significant for all variables but with low values for NH4 concentrations, diatom biomass and flagellate biomass (R2 = 0.05–0.14)

a)

32 28 24

o

C

20 16 12 8 4 0 36

b)

32

measured CAL HIND 3AT 6AT

28 24

o

C

20 16 12 8 4 0 0

100

200

300

400

500

600

700

800

900

1000

1100

Days since 1/1-2004 Fig. 5. Measured and modelled temperatures at a) surface and b) bottom for CAL, HIND (− 1.5 °C), 3AT and 6AT.

M. Maar, J.L.S. Hansen / Journal of Marine Systems 85 (2011) 57–70 Table 4 Regression statistic R2 (p b 0.05) and PMB (%) between observations and CAL. All regressions were significant.

Temperature (1 m) Temperature (15 m) Temperature (42 m) NH4 (1 m) NH4 (15 m) NO3 (1 m) NO3 (15 m) Chl a (1 m) Chl a (15 m) Diatom biomass (1–10 m) Flagellate biomass (1–10 m) Protozooplankton (1–10 m) Copepod biomass (1–10 m) Primary production

N

R2

PMB

72 72 72 76 76 76 76 74 76 69 72 72 69 74

0.98 0.86 0.87 0.35 0.05 0.81 0.47 0.40 0.41 0.14 0.12 0.29 0.38 0.19

− 11 −0 1 58 − 23 − 15 − 27 − 39 0 − 58 − 11 − 37 − 16 18

(Table 4). The modelled spring diatom bloom lasted from the beginning of February until April and the autumn bloom occurred in October–November (Fig. 7a). The model overestimated diatom biomass after the spring bloom causing the too high PMB and low R2. Autotrophic flagellates and protozooplankton peaked in early summer and then varied around 20–50 mg C m− 3 in agreement with observations although the simulated spring peak not always was supported by data (Fig. 7b and c). Copepod biomass was highest from May to October and gave a good fit to observations (Fig. 7d). The pelagic heterotrophic to autotrophic biomass (H/A)-ratio was 1.04 (Table 6) estimated as the biomass of the heterotrophic parts of MP1 and MP2, LPZ and copepods divided by the biomass of the autotrophic fractions of MP1 and MP2. The seasonal pattern of Chl a-concentrations was captured by the model showing a surface bloom in mid-February of 12 mg m− 3 and again in autumn around 4–6 mg m− 3 (Fig. 8a). During summer, surface Chl a concentrations were b2 mg m− 3 while there were sub-surface blooms around the pycnocline (Fig. 8b). Primary

5

production from spring to autumn ranged from 100 to 1000 mg C m− 2 day− 1 and fitted well to observations (Fig. 9a). Sedimentation rates were 200–380 mg C m− 2 day− 1 after spring and autumn blooms and 2–4 times lower during the summer period (Fig. 9b). Average annual biomass of benthos in the model (Table 6, CAL) was within the observed range for 2003–2007 (Table 5). In conclusion, the calibrated model performed well according to the R2 and PMB reproduced the overall seasonal dynamics of physics, nutrients, plankton and primary production. 3.2. Hind-cast scenario In the hind-cast scenario (HIND), the response of the model to higher nutrient concentrations and lower temperatures was largely in accordance with previous measurements (Table 7) although R2 not was significant for primary production due to some mismatch between peak events. The annual average of primary production was on the other hand similar (PMB =−5%). In comparison with CAL, there was an increase in Chl a concentration (24%), primary production (15%), zooplankton production (12–15%) and sedimentation rate (19%) (Table 6, Figs. 10 and 11a). The duration of the spring bloom was prolonged 14 days and the peak in protozooplankton and copepod biomasses were delayed 6 to 12 days (Table 8). However, the H/A-ratio was unchanged. There was an accumulation of organic N in the sediment in the first model year but this was reduced to the same level as for CAL in the next years (Fig. 11c). The biomass of deposit feeders increased around 40% whereas the biomass of filter feeders was less affected (10%) in HIND (Table 6) in agreement with field observations during 1993–1997 (Table 5). The observed high variability in the biomass of benthic filter feeders was due to the occasional presence of the large mussel Arctica islandia. Eutrophication versus temperature effects in HIND can be separated by comparing the scenarios +NSi and −1.5AT. The effects of these scenarios reinforced each others for Chl a concentrations, sedimentation rates and benthos, whereas for the other variables there were opposing effects (Table 6).

a)

observations model

4

mmol NH4 m-3

63

3

2

1

0 10 9

b)

mmol NO3 m-3

8 7 6 5 4 3 2 1 0

Days since 1 Jan 2004 Fig. 6. Observed and modelled (CAL) surface concentrations of a) NH4 and b) NO3 at Gniben Station.

64

M. Maar, J.L.S. Hansen / Journal of Marine Systems 85 (2011) 57–70

500

a)

observations model

mg C m-3

400 300 200 100 0 250

b) mg C m-3

200 150 100 50 0 250

c)

mg C m-3

200 150 100 50 0 250

d)

mg C m-3

200 150 100 50 0 0

100

200

300

400

500

600

700

800

900

1000

1100

Days since 1 Jan 2004 Fig. 7. Observed and modelled (CAL) biomass of a) diatoms, b) autotrophic flagellates, c) protozooplankton and d) mesozooplankton at Gniben Station.

3.3. Temperature scenarios In the warming scenarios 3AT and 6AT, average Chl a concentrations decreased 16 and 31%, respectively, while average primary production increased 8 and 17%, respectively (Table 6, Figs. 10a and 11a). The spring bloom was terminated 9 to 21 days earlier than for CAL, whereas the onset and peak time of the spring bloom were unchanged (Table 8). Average pelagic secondary production increased up to 23% (Table 6, Fig. 10b and c) and the peak in biomass of protozooplankton and copepods was achieved 11–19 and 18–31 days earlier, respectively (Table 8). As a result, the average H/A-ratio increased 13 and 33% in 3AT and 6AT, respectively (Table 6). DIN accumulated in the surface layer by the end of the productive period relative to CAL (Fig. 11b). Sedimentation rates decreased 25–44% (Table 6, Fig. 10d) and the organic N-pool in the sediment was also lower than for CAL (Fig. 11c). Biomass of deposit feeders decreased 47–66% in 6AT, whereas benthic

filter feeders were less affected (− 34%) by increasing temperatures (Table 6). The most sensitive variables to a 10% increase of Q10-values were copepod production, the H/A-ratio and surface deposit feeders that all showed a response N10% (Table 9). 4. Discussion In the present study, increasing temperatures were found to alter the tropho-dynamics and benthic-pelagic coupling in the ecosystem using a dynamic water column model. Previous studies addressing the effects of climate warming on ecosystem dynamics in the Baltic Sea area were conducted as mesocosm experiments (Aberle et al., 2007; Sommer et al., 2007; Eriksson Wiklund et al., 2009), analyses of time series (Alheit et al., 2005) and ecological modelling (Neumann, 2010). Ecological modelling is constrained by the choice of functional groups, parameterisations and process descriptions. However, the advantage of using ecosystem modelling is that complex physical–biogeochemical interactions can be

M. Maar, J.L.S. Hansen / Journal of Marine Systems 85 (2011) 57–70

65

24

a)

observations model

mg Chl a m-3

20 16 12 8 4 0 14

b)

12

mg Chl a m-3

10 8 6 4 2 0 0

100

200

300

400

500

600

700

800

900

1000

1100

Days since 1 Jan 2004 Fig. 8. Observed and modelled (CAL) Chl a concentrations at a) surface and b) 15 m depth at Gniben Station.

4.1. Model performance

considered on a time-scale of years with a direct response to climate variability. This allowed us to test the response of the ecosystem to increasing temperatures of 3 and 6 °C that was within the predicted range by IPCC (2007).

The model was calibrated for the period 2004–2006 (CAL) and then validated for 1994–1996 (HIND) against measurements of

1750

a)

observations model

1500

mg-C m-2 d-1

1250 1000 750 500 250 0 500

b) mg-C m-2 d-1

400

300

200

100

0 0

100

200

300

400

500

600

700

800

900

1000

1100

Days since 1 Jan 2004 Fig. 9. Observed and modelled (CAL) a) primary production and b) sedimentation rate at Gniben Station.

66

M. Maar, J.L.S. Hansen / Journal of Marine Systems 85 (2011) 57–70

Table 5 Average (± STD) and range (min–max) of observed biomass of filter feeders, surface deposit feeders and sub-surface deposit feeders measured in May in the Kattegat (Fig. 1) for three time periods (MADS, 2008). The average biomass of sub-surface deposit feeders in 1993–1999 was significantly different (t-test, df = 8, p b 0.05) from the two other periods. For comparison with model results, see Table 6. Filter feeders Surface deposit feeders Sub-surface deposit feeders 1993–1997 6.0 ± 3.7 0.9–11.1 1998–2002 4.2 ± 4.4 0.6–11.4 2003–2007 5.8 ± 5.7 0.4–14.5

1.1 ± 0.6 0.4–1.9 0.5 ± 0.4 0.2–1.0 0.7 ± 0.4 0.3–1.1

1.1 ± 0.2 0.8–1.3 0.4 ± 0.2 0.3–0.7 0.6 ± 0.3 0.2–0.8

temperatures, nutrient concentrations (NO3, NH4), Chl a concentrations, plankton biomasses (only CAL) and primary production from Gniben Station (Tables 4 and 7). Overall, the model was able to reproduce the main seasonal patterns of the selected variables except for the underestimation of surface NH4. This suggests that remineralisation processes were to low in the model or that there was a surface advection of nutrient patches causing the observed high concentrations during autumn 2005 (Fig. 6a and b). There were no observations of secondary production and sedimentation at the study site, but modelled secondary production can be compared with observations from the nearby Lysegrund Station in 1989 (Fig. 1). This showed that the average annual production of protozooplankton and copepods (Table 6) were comparable to measured values of 57 and 12 g C m− 2 year− 1, respectively, at (Kiørboe and Nielsen, 1994; Nielsen and Kiørboe, 1994). The modelled sedimentation rate followed the same seasonal pattern as observed at the Lysegrund Station with a peak after the spring and autumn blooms and lower values during summer (Olesen and Lundsgaard, 1995). They estimated the annual sedimentation rate across the pycnocline to 63 g C m− 2 year− 1 during the productive season. Likewise, a simple descriptive model estimated the average sedimentation rate across the pycnocline to 53 g C m− 2 year− 1 during 1989–1997 at Gniben Station (Carstensen et al., 2003). Microbial remineralisation of settling matter below the pycnocline corresponded to about half of the sedimentary input from the productive surface layer (Graneli, 1992). The bottom sedimentation of 38 g C m− 2 year− 1 in HIND would then correspond to a sedimentation rate of about 76 g C m− 2 year− 1 in agreement with the previous estimates. When considering the effects of increased nutrient loadings separately, all the considered biomasses and production and sedimentation rates increased as found in previous studies of eutrophication (Cederwall and Elmgren, 1990; Josefson and Rasmussen, 2000). Decreasing temperatures counteracted this affect for pelagic primary and secondary production, whereas sedimentation rates and benthos biomass were further increased with a similar magnitude. Thus, changes in nutrients and temperatures were both important for the

Table 7 Regression statistic R2 (p b 0.05) and PMB (%) between observations and HIND. ‘ns’ = not significant.

Temperature (1 m) Temperature (15 m) Temperature (40 m) NH4 (1 m) NH4 (15 m) NO3 (1 m) NO3 (15 m) Chl a (1 m) Chl a (15 m) Copepod biomass (1–10 m) Primary production

N

R2

PMB

26 26 26 26 26 26 26 26 26 20 21

0.92 0.64 0.74 0.10 0.04 0.67 0.12 0.23 0.23 0.51 ns

−2 −4 1 42 − 25 22 − 31 − 38 8 − 15 −5

observed development of the food web from 1994–1996 to 2004– 2006. 4.2. Pelagic trophodynamics Climate warming is suggested to increase the degree of heterotrophy in marine ecosystems because the growth response to increasing temperatures is strong for heterotrophic plankton whereas the light intensity sets an upper limit of phytoplankton primary production and in particular during the winter–spring bloom (Keller et al., 1999; Scheffer et al., 2001; Sommer et al., 2007). In the warming scenarios, annual production of protozooplankton and copepods increased by up to 13% and 23%, respectively (Table 6, Fig. 10b and c). Furthermore, the first seasonal peak in biomass occurred 2–4 weeks earlier due to the faster growth rates of zooplankton (Table 8). This corresponded to a displacement of the seasonal peak of 3.4 days °C− 1 for protozooplankton and 5.5 days °C− 1 for copepods. In comparison, mesocosm experiments with natural sea water from the Baltic Sea found similar results, i.e. 2.0 and 9 days °C− 1 of ciliates and copepods, respectively, during bloom conditions (Müren et al., 2005; Aberle et al., 2007; Sommer et al., 2007). As expected, the response of the zooplankton production to warming was strongest during bloom conditions in the model when food is less limiting and this is in accordance with mesocosm experiments (Aberle et al., 2007; Isla et al., 2008) and field studies (Halsband and Hirche, 2001). Simulated copepod production was very sensitive to changes in the Q10-values and different values should probably be applied depending on the system. Thus, the balance between autotrophy and heterotrophy (the HA-ratio) changed towards a more heterotrophic pelagic ecosystem especially during the spring bloom, while on an annual basis the predicted change was less pronounced and similar to the inter-annual variability. The higher grazing pressure from zooplankton terminated the spring diatom bloom 2–3 weeks earlier in the warming scenarios (Table 8, Fig. 11a). The onset of the spring bloom was, on the other hand, unaffected by higher temperatures in the present study because

Table 6 Annual average (± STD) of Chl a concentrations, primary production, protozooplankton- and copepod production, heterotrophic versus autotrophic biomass (H/A)-ratio, sedimentation rate, biomass of benthic filter feeders, surface deposit feeders and sub-surface deposit feeders for the periods 1994–1996 (HIND) and 2004–2006 (CAL). For the different scenarios (SCE), the percentage difference in annual means relative to CAL ((SCE − CAL)/CAL × 100%) is shown.

Chl a concentration Primary production Protozooplankton production Copepod production H/A-ratio Sedimentation rate Benthic filter feeders Benthic surface deposit feeders Benthic sub-surface deposit feeders

Units

HIND average

CAL average

HIND (%)

NSi (%)

− 1.5AT (%)

3AT (%)

6AT (%)

mg m− 3 g C m-2 y− 1 g C m− 2 y− 1 g C m− 2 y− 1 – g C m− 2 y− 1 g C m− 2 g C m− 2 g C m− 2

4.0 ± 1.7 115 ± 163 63 ± 13 29 ± 85 1.04 ± 0.59 38 ± 59 5.9 ± 0.6 1.2 ± 0.2 1.1 ± 0.2

3.2 ± 1.5 100 ± 137 55 ± 12 26 ± 72 1.04 ± 0.57 32 ± 49 5.4 ± 0.5 0.9 ± 0.1 0.8 ± 0.1

24 15 15 12 0 19 10 41 42

15 21 20 19 3 3 7 10 9

6 −5 −4 −8 −1 9 1 23 11

− 16 8 7 12 13 − 25 − 14 − 42 − 23

− 31 17 13 23 33 − 44 − 34 − 66 − 47

M. Maar, J.L.S. Hansen / Journal of Marine Systems 85 (2011) 57–70

67

500

mg-C m-2 d-1

400

a)

HIND 6AT zero-line

300 200 100 0 -100 -200 100

mg-C m-2 d-1

75

b)

50 25 0 -25 -50 -75 -100 200

mg-C m-2 d-1

150

c)

100 50 0 -50 -100 -150 150

mg-C m-2 d-1

100

d)

50 0 -50 -100 -150 0

100

200

300

400

500

600

700

800

900

1000

1100

Days since 1 Jan 2004 Fig. 10. The difference between HIND and CAL and between 6AT and CAL for a) primary production, b) protozooplankton production, c) copepod production and d) sedimentation rate. No difference from CAL is indicated by the zero line.

it was determined by the light regime above the permanent halocline. However, if the system had a substantial overwintering copepod population or resting eggs, the accelerated growth rates may cause a mismatch between nauplii and food because they peak too early before the spring bloom (Sommer et al., 2007). In other areas, where the onset of the spring bloom is governed by the establishment of a seasonal thermocline, climate change could potentially influence the timing of the spring bloom as observed for temperate lakes (Winder and Schindler, 2004a; Peeters et al., 2007). The ability to respond (e.g. grazing, growth and loss rates) to an earlier spring bloom differs between zooplankton species and may lead to changes in the zooplankton community structure (Scheffer et al., 2001; Winder and Schindler, 2004a). Such changes will vary between ecosystems and may have profound implications for higher trophic levels in the food web as shown e.g. in the North Sea (Beaugrand and Reid, 2003; Alheit et al., 2005) and the Baltic Sea (Alheit et al., 2005; Hinrichsen et al., 2005). The retention efficiency of nutrients in the upper mixed layer increased due to the faster elementary turnover rate at higher temperatures in the ecosystem. As a result, primary production increased even though the Chl a concentration was lower than in the

reference scenario, CAL (Figs. 10a and 11a). The faster turnover of nutrients in the upper mixed layer caused temporary higher surface DIN concentrations by the end of the productive period and lasted until the effect was diluted during winter by vertical mixing/diffusion (Fig. 11b). In mesocosm experiments, nutrient concentrations were also found to increase when temperature was raised 5–10 °C during bloom conditions probably due to a faster remineralisation rate by heterotrophic organisms (Müren et al., 2005). In the sediment, there was a lower content of organic N (Fig. 11c) and hence a lower remineralisation relative to CAL. However, we did not include a large pool of refractory organic material in the sediment. Thus, the model could have underestimated remineralisation of nutrients due to a gradual reduction of “older” pools of refractory organic matter in the sediment at elevated temperatures. 4.3. Benthic secondary production Warming scenarios decreased the annual sedimentation rate by up to 44% (Table 6) which was similar to an estimated reduction of 45% at +5 °C in mesocosm experiments (Müren et al., 2005). The sedimentary flux out of the surface decreased at higher temperatures

68

M. Maar, J.L.S. Hansen / Journal of Marine Systems 85 (2011) 57–70

10 8

a)

mg-Chl a m-3

6 4 2 0 -2 -4 -6 5

b)

HIND 6AT zero line

mmol-DIN m-3

4 3 2 1 0 -1 -2 100

c)

75

mmol-N m-2

50 25 0 -25 -50 -75 0

100

200

300

400

500

600

700

800

900

1000

1100

Days since 1 Jan 2004 Fig. 11. The difference between HIND and CAL and between 6AT and CAL for a) Chl a concentrations, b) DIN concentrations and c) organic N in the sediment. No difference from CAL is indicated by the zero line.

because the pool of phytodetritus was maintained at a lower level by a combination of higher grazing pressure from zooplankton and a higher bacterial remineralisation in the water column (Keller et al., 1999; Hansen and Bendtsen, 2006; Eriksson Wiklund et al., 2009). This effect would become more pronounced the longer it takes for the sinking organic particles to traverse the heterotrophic bottom layer of the water column exposed to pelagic remineralisation. In temperate

Table 8 Timing (day of year) for onset of the spring bloom (Chl a N 3 mg m− 3), first seasonal peak in Chl a concentrations, end of phytoplankton bloom, and first seasonal peak in protozooplankton and copepod biomass for CAL (2004–2006). The change in timing is given as the average (± SD) difference in the number of days between CAL and the different scenarios.

Onset of spring bloom Peak in Chl a concentration End of bloom Peak in protozooplankton biomass Peak in copepod biomass

CAL (day of year)

HIND (days)

3AT (days)

6AT (days)

42 ± 15 66 ± 4 118 ± 4 130 ± 4 155 ± 1

−2±1 1±0 14 ± 6 6±2 12 ± 2

0±0 0±0 −9±1 − 11 ± 1 − 18 ± 3

2±1 −2±3 − 21 ± 2 − 19 ± 0 − 31 ± 1

seas, the spring bloom sedimentation is an important food supply to the benthos which often is food limited (Graf et al., 1982; Josefson and Rasmussen, 2000) and the model estimated reduction in sedimentation rate of 7% °C− 1 would therefore reduce the growth rate of benthos in a warmer climate. Similarly, the increase in basal respiration rate of the benthos due to higher temperatures would also have a negative effect Table 9 Sensitivity study of the response to a 10% increase of Q10 for 2MPPD and LPZ, copepods, benthos and on all Q10 values (Tables 1 and 2) for the variables in Table 4. Responses N10% is indicated by ‘*’.

Chl a concentration Primary production Protozooplankton production Copepod production H/A-ratio Sedimentation rate Benthic filter feeders Surface deposit feeders Sub-surface deposit feeders

10% 2MPPD + LPZ

10% copepods

10% benthos

10% All

2.8 − 2.7 − 1.4 − 20.4* − 9.8 8.3 3.5 8.1 9.1

1.0 0.1 − 2.2 9.0 − 7.3 0.8 − 0.2 0.1 0.6

− 2.2 − 1.1 0.3 − 19.1* 0.1 0.8 − 3.8 11.6* − 2.2

0.5 − 3.7 − 3.5 − 7.7 − 16.2* 8.9 − 1.4 19.4* 7.8

M. Maar, J.L.S. Hansen / Journal of Marine Systems 85 (2011) 57–70

on their growth rate. Benthic filter feeders were less affected by the increasing temperatures in the model because they had first access to suspended particulate organic matter in the bottom water. Later the material was deposited on the sediment surface and became available for the surface deposit feeders and, finally, it became available for the sub-surface deposit feeders once mixed into the sediment. Accordingly, the biomass of surface and sub-surface deposit feeders was severely reduced by 47–66% in 6AT compared to CAL (Table 6). Thus, our model results suggest a succession within the benthic community because organisms with different feeding ecology will experience a varying degree of food limitation if the sedimentary input becomes reduced in a warmer climate. The predicted changes of macrofauna community are difficult to compare with corresponding observations from the field. We are not aware of any field study of how a reduced food supply and starvation within the benthic community become manifested in relation to the feeding ecology of the macrofauna. This is probably because the coastal zone generally has experienced increasing eutrophication world wide (e.g. Nixon, 1990). Thus, the opposite scenario where increasing sedimentation affects the benthic community has hitherto probably occurred frequently and has been documented experimentally. For example, Quijón et al. (2008) found that when adding phytodetritus to the sea floor, suspension feeders and surface deposit feeders responded fast to the enrichment. After the pulse was processed by consumption or buried into the sediment, surface and sub-surface deposit feeders and predators characterised the community. Correspondingly, Josefson and Conley (1997) found an increase in the biomass of surface deposit feeder across pelagic production gradients in the northern Kattegat. In the Bering Sea shelf, a higher biomass of surface deposit feeders was observed during a cold period in the 1970s in comparison with a warmer period in the 1950s (Coyle et al., 2007). The suggested mechanisms were an elevated carbon flux and exclusion of predators during the cold water period. Thus, it seems very likely that major structural changes will occur in the benthic community when the overall transfer of organic matter changes radically due to climate warming. How such overall changes in the input of organic material will affect the functionality of the benthic community is a more open question. The model predictions corresponded to observations from the field, given that starvation of the community results in the opposite changes as do enrichment with organic material. However, this is not necessarily so as the total biomass and the general nutrition of the macrofauna may be important for such changes. The model does not take into account that there may be a buffering effect due to the various pools of refractory organic matter of unknown accessibility in marine sediments. Model predictions are obviously also limited by a simplified parameterisation of the different ecological groups including the temperature response. This response was indeed sensitive to the Q10-value especially for surface deposit feeders in the model. The ecological behaviour of some of the species present is largely unknown and probably does not match any of the three groups operating in the model. For example, the benthic filter feeders are represented by a parameterisation of mussels from shallow water above the halocline like the blue mussel. Another example concerns the most common macrofauna species in the Kattegat area, Amphiura filiformis, which is deposit feeding in the beginning and shifts to filter feeding later in the life cycle. Therefore, the outcome of the competition between the different ecological forms is uncertain and there is a need for more experimental studies of how starvation and temperature affect the benthic community in order to calibrate models in the future. 5. Conclusion In conclusion, the higher degree of heterotrophy in the ecosystem in response to a warmer climate was found to have implications for

69

the benthic–pelagic coupling and the benthic community structure. Changes in the benthic community will eventually affect the growth and survival of higher trophic levels in the benthic food web such as demersal fish. Thus, climate warming is likely to change the trophodynamic balance of the ecosystem towards a higher pelagic and a lower benthic secondary production. These indirect effects of climate warming on biological community composition will come on top of the more direct effects of rising temperatures which concern how the temperature affects the individual species and the interspecific competition between species.

Acknowledgements This study was supported by the EU grant: Resolving Climatic Impacts on Fish Stocks (RECLAIM, Contract no. 44133) and a grant from the Directorate for Food, Fisheries and Agri Business: Modelling the Impact of Hydrography and Lower Trophic Production on Fish Recruitment (MODREC, Contract no. 3304-FVFP-060683). We acknowledge Jørgen Brandt, Jesper Christensen and Jørgen Bendtsen for providing meteorological forcing data to the model and Jae-Young Lee for providing the source code of the 2MPPD model. Stiig Markager is thanked for detailed and fruitful comments on the MS and Anne van Acker is thanked for English improvements.

References Aberle, N., Lengfellner, K., Sommer, U., 2007. Spring bloom succession, grazing impact and herbivore selectivity of ciliate communities in response to winter warming. Oecologia 150, 668–681. Ærtebjerg, G., Andersen, J., Carstensen, J., Christiansen, T., Conley, D.J., Dahl, K., Dahllöf, I., Fossing, H., Greve, T., Hansen, J.L.S., Henriksen, P., Josefson, A.B., Krause-Jensen, D., Larsen, M.M., Markager, S.S., Nielsen, T.G., Petersen, B., Risgaard-Petersen, N., Rysgaard, S., Strand, J., Bering, N., Oversen, N., Ellermann, T., Hertel, O., Skjøth, C.A., 2002. Status of the Marine Environment 2001. Faglig rapport fra DMU nr. 419. (In Danish). Alheit, J., Mollmann, C., Dutz, J., Kornilovs, G., Loewe, P., Mohrholz, V., Wasmund, N., 2005. Synchronous ecological regime shifts in the central Baltic and the North Sea in the late 1980s. ICES J. Mar. Sci. 62, 1205–1215. Allen, J.I., Holt, J.T., Blackford, J., Proctor, R., 2007. Error quantification of a highresolution coupled hydrodynamic-ecosystem coastal-ocean model: Part 2. Chlorophyll-a, nutrients and SPM. J. Mar. Syst. 68, 381–404. Andersson, L., Rydberg, L., 1988. Trends in nutrient and oxygen conditions within the Kattegat — effects of local nutrient supply. Estuar. Coast. Shelf Sci. 26, 559–579. Andersson, A., Haecky, P., Hagström, Å., 1994. Effect of temperature and light on the growth of micro-, nano- and pico-plankton: impact on algal succession. Mar. Biol. 120, 511–520. Beaugrand, G., Reid, P.C., 2003. Long-term changes in phytoplankton, zooplankton and salmon related to climate. Glob. Change Biol 9, 801–817. Bendtsen, J., Gustafsson, K.E., Petersen, J.K., 2006. Modelling vertical mixing in the surface boundary layer using artificial age tracers. J. Mar. Syst. 60, 115–128. Bendtsen, J., Gustafsson, K.E., Söderkvist, J., Hansen, J.L.S., 2009. Ventilation of bottom water in the North Sea-Baltic Sea transition zone. J. Mar. Syst. 75, 138–149. Blackford, J.C., 1997. An analysis of benthic biological dynamics in a North Sea ecosystem model. J. Sea Res. 38, 213–230. Brandt, J., Christensen, J.H., Frohn, L.M., Palmgren, F., Berkowicz, R., Zlatev, Z., 2001. Operational air pollution forecasts from European to local scale. Atmos. Environ. 35, S91–S98. Carstensen, J., Conley, D., Muller-Karulis, B., 2003. Spatial and temporal resolution of carbon fluxes in a shallow coastal ecosystem, the Kattegat. Mar. Ecol. Prog. Ser. 252, 35–50. Cederwall, H., Elmgren, R., 1990. Biological Effects of Eutrophication in the Baltic Sea, Particularly the Coastal Zone. Ambio 19, 109–112. Coyle, K.O., Konar, B., Blanchard, A., Highsmith, R.C., Carroll, J., Carroll, M., Denisenko, S. G., Sirenko, B.I., 2007. Potential effects of temperature on the benthic infaunal community on the south-eastern Bering Sea shelf: possible impacts of climate change. Deep Sea Res. II 54, 2885–2905. DMI, 2002. Tidevand ved de danske vandstandsstationer. Technical Report 02-21. The Danish Meteorological Institute, Copenhagenhttp://www.dmi.dk/dmi/tr02-21.pdf. Droop, M.R., Mickelson, M.J., Scott, J.M., Turner, M.F., 1982. Light and nutrient status of algal cells. J. Mar. Biol. Assoc. UK 62, 402–434. Eriksson Wiklund, A.-K., Dahlgren, K., Sundelin, B., Andersson, A., 2009. Effects of warming and shifts of pelagic food web structure on benthic productivity in a coastal marine system. Mar. Ecol. Prog. Ser. 396, 13–25. Fennel, K., 1999. Convection and the timing of phytoplankton spring blooms in the western Baltic Sea. Estuar. Coast. Shelf Sci. 49, 113–128. Fennel, W., 2001. Modeling of copepods with links to circulation models. J. Plankton Res. 23, 1217–1232.

70

M. Maar, J.L.S. Hansen / Journal of Marine Systems 85 (2011) 57–70

Graf, G., Bengtson, W., Diesner, U., Schultz, R., Theede, H., 1982. Benthic response to sedimentation of a spring phytoplankton bloom. Mar. Biol. 67, 201–208. Graneli, E., 1992. Below-halocline oxygen consumption in the Kattegat. Hydrobiologia 235, 303–310. Graneli, E., Wallstrom, K., Larsson, U., Graneli, W., Elmgren, R., 1990. Nutrient limitation of primary production in the Baltic Sea area. Ambio 19, 142–151. Halsband, C., Hirche, H.J., 2001. Reproductive cycles of dominant calanoid copepods in the North Sea. Mar. Ecol. Prog. Ser. 209, 219–229. Hansen, J.L.S., Bendtsen, J., 2006. Klimabetingede effekter på marine økosystemer. Faglig rapport fra DMU no. 598, the Ministry of Environments. 50 pp. Hansen, P.J., Bjørnsen, P.K., Hansen, B.W., 1997. Zooplankton grazing and growth: scaling within the 2–2,000-μm body size range. Limnol. Oceanogr. 42, 687–704. Hays, G.C., Richardson, A.J., Robinson, C., 2005. Climate change and marine plankton. Trends Ecol. Evol. 20, 337–344. Hertel, O., Ellermann, T., Palmgren, F., Berkowicz, R., Løfstrøm, P., Frohn, L.M., Geels, C., Skjøth, C.A., Brandt, J., Christensen, J., Kemp, K., Ketzel, M., 2007. Integrated airquality monitoring — combined use of measurements and models in monitoring programmes. Environ. Chem. 4, 65–74. Hinrichsen, H.H., Schmidt, J.O., Petereit, C., Möllmann, C., 2005. Survival probability of Baltic larval cod in relation to spatial overlap patterns with its prey obtained from drift model studies. ICES J. Mar. Sci. 62, 878–885. Huisman, J., van Oostveen, P., Weissing, F.J., 1999. Critical depth and critical turbulence: two different mechanisms for the development of phytoplankton blooms. Limnol. Oceanogr. 44, 1781–1787. IPCC, 2007. Summary for policy makers. Climate Change 2007: Synthesis Report. . Iriarte, A., Purdie, D.A., 2004. Factors controlling the timing of major spring bloom events in an UK south coast estuary. Estuar. Coast. Shelf Sci. 61, 679–690. Isla, J.A., Lengfellner, K., Sommer, U., 2008. Physiological response of the copepod Pseudocalanus sp. in the Baltic Sea at different thermal scenarios. Glob. Change Biol. 14, 895–906. Jackson, G.A., 1990. A model of the formation of marine algal flocs by physical coagulation processes. Deep Sea Res. I 37, 1197–1211. Josefson, A.B., Conley, D.J., 1997. Benthic response to a pelagic front. Mar. Ecol. Prog. Ser. 147, 49–62. Josefson, A.B., Rasmussen, B., 2000. Nutrient retention by benthic macrofaunal biomass of Danish estuaries: importance of nutrient load and residence time. Estuar. Coast. Shelf Sci. 50, 205–216. Keller, A.A., Oviatt, C.A., Walker, H.A., Hawk, J.D., 1999. Predicted impacts of elevated temperature on the magnitude of the winter–spring phytoplankton bloom in temperate coastal waters: a mesocosm study. Limnol. Oceanogr. 44, 344–356. Kiørboe, T., Nielsen, T.G., 1994. Regulation of zooplankton biomass and production in a temperate, coastal ecosystem: 1. Copepods. Limnol. Oceanogr. 39, 493–507. Lee, J.Y., Tett, P., Jones, K., Luyten, P., Smith, C., Wild-Allen, K., 2002. The PROWQM physical–biological model with benthic–pelagic coupling applied to the northern North Sea. J. Sea Res. 48, 287–331. Luyten, P., Jones, J., Proctor, R., Tabor, R., Tett, P., Wild-Allen, K., 1999. A coupled hydrodynamical–ecological model for regional and shelf seas: user documentation. MUMM Report. Belgium, Management Unit of the Mathematical Models of the North Sea. Report 914 pp. Maar, M., Nielsen, T.G., Richardson, K., Christaki, U., Hansen, O.S., Zervoudaki, S., Christou, E.D., 2002. Spatial and temporal variability of food web structure during the spring bloom in the Skagerrak. Mar. Ecol. Prog. Ser. 239, 11–29. Maar, M., Bolding, K., Petersen, J.K., Hansen, J.L.S., Timmermann, K., 2009. Local effects of blue mussels around turbine foundations in an ecosystem model of Nysted offshore wind farm. J. Sea Res. 62, 159–174.

MADS, 2008. The National Database for Marine Data, The National Environmental Research Institute, Aarhus University us, DenmarkWWW page http://www.dmu. dk/International/Water/Monitoring+of+the+Marine+Environment/MADS/ 2008. Müren, U., Berglund, J., Samuelsson, K., Andersson, A., 2005. Potential effects of elevated sea-water temperature on pelagic food webs. Hydrobiology 545, 153–166. Neumann, T., 2010. Climate-change effects on the Baltic Sea ecosystem: a model study. J. Mar. Syst. 81, 213–224. Nielsen, T.G., Kiørboe, T., 1994. Regulation of zooplankton biomass and production in a temperate, coastal ecosystem: 2. Ciliates. Limnol. Oceanogr. 39, 508–519. Nixon, S., 1990. Marine eutrophication: a growing international problem. Ambio 19 (3), 101. Olesen, M., 1993. The fate of an early diatom spring bloom in the Kattegat. Ophelia 37, 51–66. Olesen, M., Lundsgaard, C., 1995. Seasonal sedimentation of autochthonous material from the euphotic zone of a coastal system. Estuar. Coast. Shelf Sci. 41, 475–490. Oviatt, C.A., 1994. Biological considerations in marine enclosure experiments: challenges and revelations. Oceanography 7, 45–51. Pedersen, T.M., Hansen, J.L.S., Josefson, A.B., Hansen, B.W., 2008. Mortality through ontogeny of soft bottom marine invertebrates with planktonic larvae. J. Mar. Syst. 73, 185–207. Peeters, F., Straile, D., Lorke, A., Livingstone, D.M., 2007. Earlier onset of the spring phytoplankton bloom in lakes of the temperate zone in a warmer climate. Glob. Change Biol 13, 1898–1909. Quijón, P.A., Kelly, M.C., Snelgrove, P.V.R., 2008. The role of sinking phytodetritus in structuring shallow-water benthic communities. J. Exp. Mar. Biol. Ecol. 366, 134–145. Richardson, K., Christoffersen, A., 1991. Seasonal distribution and production of phytoplankton in the southern Kattegat. Mar. Ecol. Prog. Ser. 78, 217–227. Richardson, K., Heilmann, J.P., 1995. Primary production in the Kattegat past and present. Ophelia 41, 317–328. Scheffer, M., Straile, D., van Nes, E.H., Hosper, H., 2001. Climatic warming causes regime shifts in lake food webs. Limnol. Oceanogr. 46, 1780–1783. Smetacek, V., Cloern, J.E., 2008. On phytoplankton Trends. Science 319, 1346–1348. Sommer, U., Aberle, N., Engel, A., Hansen, T., Lengfellner, K., Sandow, M., Wohlers, J., Zollner, E., Riebesell, U., 2007. An indoor mesocosm system to study the effect of climate change on the late winter and spring succession of Baltic Sea phyto- and zooplankton. Oecologia 150, 655–667. Sverdrup, H.U., 1953. On conditions for the vernal blooming of phytoplankton. J. Cons. Perms. Int. Explor. Mer. 18, 287–295. Tett, P., 1998. Parameterising a Microplankton Model. Report Edinburgh University. 53 pp. Tett, P., Lee, J.Y., 2005. N:Si ratios and the ‘balance of organisms’: PROWQM simulations of the northern North Sea. J. Sea Res. 54, 70–91. Vichi, M., Ruardij, P., Baretta, J.W., 2004. Link or sink: a modelling interpretation of the open Baltic biogeochemistry. Biogeosciences 1, 79–100. Wassmann, P., 1990. Relationship between primary and export production in the boreal coastal zone of the North Atlantic. Limnol. Oceanogr. 35, 464–471. Winder, M., Schindler, D.E., 2004a. Climatic effects on the phenology of lake processes. Glob. Change Biol 10, 1844–1856. Winder, M., Schindler, D.E., 2004b. Climate change uncouples trophic interactions in an aquatic system. Ecology 85 (8), 2100–2106 85(11):3178.