Impacts of natural and anthropogenic climate variations on North Pacific plankton in an Earth System Model

Impacts of natural and anthropogenic climate variations on North Pacific plankton in an Earth System Model

Ecological Modelling 244 (2012) 132–147 Contents lists available at SciVerse ScienceDirect Ecological Modelling journal homepage: www.elsevier.com/l...

3MB Sizes 4 Downloads 53 Views

Ecological Modelling 244 (2012) 132–147

Contents lists available at SciVerse ScienceDirect

Ecological Modelling journal homepage: www.elsevier.com/locate/ecolmodel

Impacts of natural and anthropogenic climate variations on North Pacific plankton in an Earth System Model Lavinia Patara a,1 , Marcello Vichi a,b,∗ , Simona Masina a,b a b

Centro Euro-Mediterraneo per i Cambiamenti Climatici (CMCC), Viale Aldo Moro 44, I-40127 Bologna, Italy Istituto Nazionale di Geofisica e Vulcanologia (INGV), Viale Aldo Moro 44, I-40127 Bologna, Italy

a r t i c l e

i n f o

Article history: Received 28 July 2011 Received in revised form 1 May 2012 Accepted 9 June 2012 Available online 31 July 2012 Keywords: Plankton model Biogeochemistry Physiological stoichiometry Natural climate variability Climate change North Pacific Earth System Model

a b s t r a c t The impacts of natural atmospheric variability and anthropogenic climate change on the spatial distribution, seasonality, structure, and productivity of North Pacific plankton groups are investigated by means of an Earth System Model (ESM) that contains a plankton model with variable stoichiometry. The ESM is forced with observed greenhouse gases for the 20th century and with the Intergovernmental Panel on Climate Change A1B Emission Scenario for the 21st century. The impacts of the two main modes of variability – connected with the Aleutian Low (AL) strength and with the North Pacific Oscillation (NPO) – are considered. When the AL is strong, primary productivity and chlorophyll concentrations are higher in the central Pacific, the seasonality of plankton is enhanced, and the classical grazing chain is stimulated, whereas in the Alaskan Gyre the model simulates a chlorophyll decrease and a shift toward smaller phytoplankton species. A stronger NPO increases productivity and chlorophyll concentration at ∼45◦ N. In the anthropogenic climate change scenario, simulated sea surface temperature is 4 ◦ C higher with respect to contemporary conditions, leading to reduced mixing and nutrient supply at middlesubpolar latitudes. The seasonal phytoplankton bloom is reduced and occurs one month earlier, the flow of carbon to the microbial loop is enhanced, and phytoplanktonic stoichiometry is nutrient-depleted. Primary productivity is enhanced at subpolar latitudes, due to increased ice-free regions and possibly to temperature-related photosynthesis stimulation. This study highlights that natural climate variability may act alternatively to strengthen or to weaken the human-induced impacts, and that in the next decades it will be difficult to distinguish between internal and external climate forcing on North Pacific plankton groups. © 2012 Elsevier B.V. All rights reserved.

1. Introduction North Pacific marine ecosystems are significantly affected by climate variability and change. Several fluctuations of North Pacific marine ecosystems have been linked to climate indices and teleconnection patterns, such as the Pacific Decadal Oscillation (Mantua et al., 1997; Miller and Schneider, 2000) and the North Pacific Gyre Oscillation (Di Lorenzo et al., 2008). At the same time, the accelerating rates of anthropogenic carbon emissions are exerting increasing pressure on North Pacific plankton communities and ecosystems. Natural climate variability and anthropogenic climate change both have the potential of affecting the dynamics, community structure, and productivity of marine ecosystems at various trophic

∗ Corresponding author at: Centro Euro-Mediterraneo per i Cambiamenti Climatici (CMCC), Viale Aldo Moro 44, I-40127 Bologna, Italy. E-mail address: [email protected] (M. Vichi). 1 Now at Helmholtz Centre for Ocean Research Kiel (GEOMAR), Düsternbrooker Weg 20, D-24105 Kiel, Germany. 0304-3800/$ – see front matter © 2012 Elsevier B.V. All rights reserved. http://dx.doi.org/10.1016/j.ecolmodel.2012.06.012

levels (Doney et al., 2012). Climate variability and change directly affect the productivity and the functioning of the planktonic structure through modifications in temperature, stratification, light, and nutrient availability. Changes in quality and quantity of the plankton community structure may have cascading effects on the higher trophic levels of the marine food webs and on the biogeochemical cycling of carbon and limiting nutrients (Finkel et al., 2010). It is thus a great challenge of today’s marine research to assess the impacts of natural climate fluctuations on North Pacific marine ecosystems, and to make projections of how ecosystem structure and composition may be impacted by scenarios of increasing fossil fuel emissions in the 21st century. There are however several difficulties in assessing these combined impacts and in making future projections. Observational data sets lack the temporal and spatial coverage to be able to distinguish between the spatial patterns associated with natural climate variability and with anthropogenic climate changes. On the other hand, climate-biogeochemistry models are often characterized by biases, and commonly include biogeochemical models which are too simplified to capture realistic changes in structure and functionality

L. Patara et al. / Ecological Modelling 244 (2012) 132–147

of the lower trophic levels of the marine ecosystem (Finkel et al., 2010). The aim of this study is to further assess the combined impacts of North Pacific natural climate variability and anthropogenic climate change on marine biogeochemistry and on the food web dynamics of the lower trophic levels of North Pacific marine ecosystems. A state-of-the-art Earth System Model (hereafter ESM), including ocean, sea ice, atmosphere, land, and marine biogeochemistry, is used to produce two simulations: (i) one is a historical 20th century simulation (forced with observed well-mixed greenhouse gases, sulphate aerosols, and ozone), and (ii) the other is a 21st century climate projection forced with the Intergovernmental Panel on Climate Change (IPCC) A1B “business-as-usual” emission scenario (Nakicenovic and Swart, 2000). The marine biogeochemistry model PELAGOS (Vichi et al., 2007) embedded in the ocean component of the ESM is well suited for capturing climate-related changes in food web dynamics and material flows, as it incorporates concepts of ecological stoichiometry, in which the balance of energy and elements affects and is affected by organisms and their ecosystem interactions. In PELAGOS, major ecological functions of producers, decomposers, and consumers – and their specific trophic interactions – are formulated in terms of material flows of basic elements (C, N, P, Si, and Fe) from the (in)organic pelagic pools through the food web as a function of organisms’ demand and trophic relationships. The originality of this study is that it incorporates ecologically oriented processes, such as the microbial-level material flows and trophic interactions, within the framework of a changing climate as obtained by multi-century simulations for the 20th and 21st century carried out with an ESM. In addition to the impacts of anthropogenic climate change, a systematic analysis of the biogeochemical and ecological changes associated with the first two modes of North Pacific climate variability is carried out. The leading mode of North Pacific atmospheric variability is associated with the modulation of the Aleutian Low (hereafter AL) pressure system. Its temporal evolution is described by the Pacific – North American pattern (PNA, Wallace and Grutzler, 1981) or by the North Pacific Index (NPI, Trenberth and Hurrell, 1994). Fluctuations in the AL strength co-vary with the Pacific Decadal Oscillation (hereafter PDO) which is the first mode of sea surface temperature (hereafter SST) variability in the North Pacific (Mantua and Hare, 2002). In the past century, the PDO index showed the tendency for decadal persistency and a few cases of abrupt sign change, as it occurred in 1976–1977 when the PDO entered in a predominantly positive phase (Hare and Mantua, 2000). In its positive phase, the PDO leads to lower-than-average SST in the central Pacific and higher-than-average SST in the Northeast Pacific. Observational and modeling studies have shown that positive PDO phases tend to be associated with higher chlorophyll concentration in the central North Pacific (Chai et al., 2003; Martinez et al., 2009) and lower nutrient and chlorophyll concentrations in the Northeast Pacific (Alexander et al., 2008; Wong et al., 2007). PDO fluctuations are also importantly related to changes in fish populations throughout the North Pacific (Hare and Mantua, 2000). The second mode of North Pacific atmospheric variability is the North Pacific Oscillation (hereafter NPO) which is associated with a seesaw of atmospheric pressure between subarctic and subtropical latitudes (Rogers, 1981). Oceanic expressions of the NPO are the “Victoria Mode” defined as the second mode of North Pacific SST variability (Bond et al., 2003) and the North Pacific Gyre Oscillation (NPGO, Di Lorenzo et al., 2008) defined as the second mode of sea surface height variability. Whereas the impacts of the PDO on North Pacific biogeochemistry and ecosystems are well documented, the impacts of the second mode of North Pacific climate variability are only recently starting to emerge. By combining observations and models, Di Lorenzo et al. (2008, 2009) found that the NPGO may even be the dominant mode of biogeochemical

133

variability in some regions of the northeastern Pacific, as positive NPGO phases enhance nutrient availability and chlorophyll concentration off the California coast due to upwelling-favorable conditions. The ecosystem impacts of the NPO–NPGO mode should therefore be considered in conjunction with those of the AL–PDO mode, even more so in the light of recent studies suggesting that the NPO–NPGO mode might gain importance in scenarios of increasing greenhouse gases (Di Lorenzo et al., 2010). Ocean warming associated with rising atmospheric CO2 concentrations may also produce large impacts on North Pacific marine ecosystems. Increased ocean temperatures enhance metabolic rates of marine organisms, and have been suggested to increase photosynthesis and primary production (Taucher and Oschlies, 2011) because of the stimulation of the nutrient cycling through the microbial loop (Doney et al., 2012). Higher ocean temperatures are also expected to enhance stratification, reduce sea ice cover, and alter patterns of ocean circulation (Meehl et al., 2007). The interplay of direct and indirect temperature effects will likely modify the nutritional status, food availability, and light conditions for planktonic communities, thus potentially leading to significant modifications in seasonal abundance, spatial distribution, biological interactions, community composition, and energy flows within the marine food web (Vichi et al., 2003; Wohlers et al., 2009; Finkel et al., 2010; Doney et al., 2012). In the North Pacific, isolating the impacts of human-induced ocean warming is complicated by the decadal climate fluctuations and teleconnection patterns associated with the AL–PDO and the NPO–NPGO, whose impacts are currently of the same order of magnitude as those suggested to be caused by anthropogenic greenhouse gas emissions (Bindoff et al., 2007). Still, long observational time series in the North Pacific suggest that anthropogenic climate change may indeed be the cause of a number of environmental and ecosystem changes. These include a decadal increase in stratification in the Northeast Pacific (Freeland et al., 1997), decreasing surface nutrient concentrations in the subarctic North Pacific (Watanabe et al., 2005; Ono et al., 2008) and in the subtropical ALOHA station (Karl et al., 2001), and a shift in phytoplankton community structure toward smaller picophytoplankton species at the ALOHA station (Karl et al., 2001). Using climate-biogeochemistry models of varying complexity, several studies have projected how marine biogeochemical cycles and plankton dynamics may be impacted by increasing atmospheric CO2 during the 21st century (Bopp et al., 2005; Schmittner et al., 2008; Henson et al., 2010; Steinacher et al., 2010; Polovina et al., 2011; Vichi et al., 2011). Henson et al. (2010) and Steinacher et al. (2010) analyzed a suite of climate model projections which overall indicated that by the end of the 21st century chlorophyll and primary productivity are likely to decrease at middle latitudes and increase at higher latitudes. Furthermore, the statistical analysis by Henson et al. (2010) suggests that the anthropogenic impact will be clearly distinguishable from natural climate fluctuations only starting from the second half of the 21st century. Schmittner et al. (2008) also detected a reduction in North Pacific middle-latitude plankton biomasses in a multi-millennial simulation with a carbon cycle model, even though at a global scale primary productivity shows an increase, owing to a temperature-related stimulation of photosynthesis. Vichi et al. (2011) projected an overall decline in North Pacific net community production under both the A1B scenario and under a substantial mitigation scenario. It is therefore a challenge to quantitatively distinguish the impacts of natural climate variability and of rising atmospheric CO2 concentrations on the structure, seasonality, spatial distribution, and production of North Pacific marine ecosystems. This work illustrates a modeling technique to make this integrated assessment. Section 2 describes the model and the experiment set up, while Section 3 compares simulated chlorophyll patterns with satellite

134

L. Patara et al. / Ecological Modelling 244 (2012) 132–147

observational estimates. The response of plankton communities to natural atmospheric variability is investigated in Section 4, whereas its response to anthropogenic greenhouse gases in the 21st century scenario is addressed in Section 5. Section 6 estimates the impacts of natural climate variability and anthropogenic impacts on the planktonic food web. Section 7 concludes by summarizing and discussing the main results of this study.

2. Model configuration and experiment set up The INGV-CMCC Earth System Model consists of an atmosphere–ocean–sea ice physical core coupled to models resolving carbon cycle on land and ocean. The technical description of the atmosphere–ocean coupling as well as of the inclusion of the carbon cycle models into the physical models is described in Fogli et al. (2009), and extensive details of the simulation results used in this work can be found in Vichi et al. (2011). The model components are: the ECHAM5 atmospheric general circulation model (Röckner et al., 2003), the SILVA land surface model (Alessandri, 2006), the OPA8.2 ocean general circulation model (Madec et al., 1998), the LIM2 sea ice model (Timmermann et al., 2005), and the PELAGOS model for the ocean biogeochemistry (Vichi et al., 2007; Vichi and Masina, 2009). The software used to couple the atmosphere (including the land-vegetation model) component and the ocean (including the biogeochemistry) is OASIS3 (Valcke, 2006). PELAGOS is the global implementation of the Biogeochemical Flux Model (http://bfm.cmcc.it) within the OPA 8.2 ocean general circulation model. The ocean model has a horizontal resolution of 2◦ (in the North Pacific) and 31 vertical levels. The model solves the dynamics of lower trophic levels and of major inorganic and organic components of the marine ecosystem from a functional biogeochemistry perspective (Vichi et al., 2007), as schematically shown in Fig. 1. The living functional groups (LFG) simulated by PELAGOS are three unicellular planktonic autotrophs (picophytoplankton, nanophytoplankton, and diatoms), three zooplankton groups (nano-, micro- and meso-), and bacterioplankton. The other chemical functional families (CFF) also depicted in Fig. 1 are nitrate, ammonium, orthophosphate, silicate, dissolved bio-available iron, oxygen, carbon dioxide, and dissolved and particulate (non-living) organic matter. Diatoms are the largest phytoplankton group and are characterized by high growth rates in cool and nutrient-rich conditions, whereas the nanophytoplankton group is adapted to more nutrient-depleted conditions. Nutrient uptake is parameterized following a Droop kinetics which allows for multi-nutrient limitation and variable internally regulated nutrient ratios. Chlorophyll synthesis is down-regulated when the rate of light absorption exceeds the utilization of photons for carbon fixation. Temperature affects all metabolic rates of simulated plankton groups through the Q10 parameter, which generally doubles the speed of the process every 10 ◦ C increase (Vichi and Masina, 2009). Net primary production in PELAGOS is parameterized as a function of light, temperature, chlorophyll, iron cell-content, and dissolved silicate concentration (Vichi et al., 2007). The cell availability of N and P does not directly control photosynthesis, but the subsequent transformation of carbohydrates into proteins and cell material. A portion of photosynthesized carbon is thereby released as Dissolved Organic Carbon (DOC) according to the internal nutrient quota (Baretta-Bekker et al., 1997; Vichi et al., 2007). Nutrient remineralization by bacteria is controlled by the quality of dissolved and particulate organic matter (i.e. the stoichiometric content of nutrient with respect to carbon), which in turn also regulates the competition of bacteria with phytoplankton for dissolved inorganic nutrients. In PELAGOS, the fluxes of carbon and limiting nutrients through the food web are not characterized by fixed values and

ratios. Finally, the ocean carbonate chemistry is solved with a simplified solution proposed by Follows et al. (2006), and sea-air CO2 fluxes are calculated using the Wanninkhof (1992) parameterization. PELAGOS has shown skill at reproducing observed climatologies and interannual variability of biogeochemistry and plankton properties (Vichi and Masina, 2009; Patara et al., 2011) as well as responses to anthropogenic emission scenarios (Vichi et al., 2011). Its flexibility in terms of material flows and elemental stoichiometry within the lower trophic levels of the ecosystem make is a flexible tool for the detection of ecological responses to climate changes. The ESM is used to perform (i) a historical 20th century simulation forced with observed well-mixed greenhouse gases (CO2 , CH4 and N2 O), sulphate aerosols, and ozone, and initialized from a 300year pre-industrial simulation (Vichi et al., 2011), and (ii) a 21st century climate projection forced with the IPCC SRES A1B emission scenario (Nakicenovic and Swart, 2000). The experimental set up follows the multi-model ENSEMBLES concerted experiment described in Johns et al. (2011), without any variation in solar and volcanic forcing. More details on the experimental design of the simulations used in this study can be found in Vichi et al. (2011). The analysis of climate mode variability was done using Principal Component Analysis (PCA, Preisendorfer, 1988), which is a standard technique in climate and ecological studies in the North Pacific (e.g. Mantua et al., 1997). The method allows extracting the dominant variability patterns and their temporal evolution, the principal components (PC) time series, from a large set of data, which in this case is the model output of sea level pressure at monthly mean resolution. Using PCA, the extracted modes of variability are orthogonal among each other and identifiable by their regional patterns and temporal behavior.

3. Comparison with observed estimates Fig. 2 (left) shows annual means of simulated present-day chlorophyll concentrations averaged over the euphotic layer (i.e. the depth where light is 1% of surface conditions) and their standard deviations. The visual comparison with satellite SeaWiFS chlorophyll estimates (McClain, 2009) for the period 1998–2006 (Fig. 2, right) indicates that the main chlorophyll patterns, i.e. higher subpolar values and lower subtropical values, are captured by PELAGOS. However, the model overestimates chlorophyll concentration and its standard deviation at subpolar latitudes (even though the patchy standard deviation in the SeaWiFS may reflect the limited size of the sample) and underestimates it along the coasts and at subtropical latitudes. The poleward displacement of the subtropical–subpolar chlorophyll transition is related to a poleward shift of the simulated westerly wind belt and of the Kuroshio Current (not shown). Also, at subpolar latitudes the winter MLD tends to be overestimated with respect to observed (De Boyer Montégut et al., 2004) whereas at subtropical latitudes it tends to be shallower-than-observed, thus accentuating the subtropical–subpolar chlorophyll gradients. Chlorophyll concentrations are underestimated in coastal regions: this may be due to insufficient upwelling (as for the California Current System and along the Washington-Oregon coast) or to overestimated sea ice in high-latitude regions. Finally the simulated net primary production is reasonably captured by PELAGOS compared with in situ data and satellite estimates of primary production (Vichi and Masina, 2009). From previous studies (Patara et al., 2011) it is known that this model exhibits a spurious drift in surface biogeochemical variables in the northern extratropical regions. This is likely caused by an unbalance between export of organic matter into deep ocean layers

L. Patara et al. / Ecological Modelling 244 (2012) 132–147

135

Fig. 1. Schematic representation of the state variables and pelagic interactions of the biogeochemistry model. Living (organic) Chemical Functional Families (CFF) are indicated with bold-line square boxes, non-living organic CFFs with thin-line square boxes and inorganic CFFs with rounded boxes. Modified after Vichi et al. (2007).

Fig. 2. Annual means (top) and their standard deviations (bottom) of present-day chlorophyll concentration (mg m−3 ). Left: Simulated chlorophyll averaged over the euphotic depth and during the years 1970–1999 of the simulation. Right: SeaWiFS satellite estimates for the years 1998–2006.

and nutrient re-supply into the euphotic layer. As it will be shown in Section 5.2, an estimate of the magnitude of the model spurious drift of 0.004 mg Chl m−3 per decade is obtained by computing the linear chlorophyll trend between 1900 and 1949, i.e. in a period when human-induced climate trends are negligible (Fig. 9a and Fig. 1a from Vichi et al., 2011).

4. Natural variability 4.1. Physical climate North Pacific atmospheric variability is investigated by means of PCA analysis on sea level pressure (hereafter SLP) anomalies

136

L. Patara et al. / Ecological Modelling 244 (2012) 132–147

PC1

2

SLP SST

0

−2 a 52% 25%

48% 34%

1900 1925 1950 1975 2000 2025 2050 2075 2100

PC2

2

SLP SST

0

−2 b

22% 18%

24% 13%

1900 1925 1950 1975 2000 2025 2050 2075 2100 Fig. 3. Bars: (a) first and (b) second standardized principal component (PC) time series of simulated winter (JFM) sea level pressure (SLP) anomalies in the North Pacific sector. Red lines: 9-year running means of the SLP PC time series. Blue lines: 9-year running means of the sea surface temperature (SST) PC time series. Numbers indicate the variance explained by the first and second mode of SLP and SST variability for the (left) 20th and (right) 21st century simulations. (For interpretation of the references to color in this figure legend, the reader is referred to the web version of the article.)

in the North Pacific sector (20–65◦ N, 120◦ E–100◦ W). We remark that this PCA-based definition is not identical to the station-based definitions by Wallace and Grutzler (1981) and Rogers (1981). However station-based and PCA-based analyses yield very similar results. The analysis of physical variables is performed for the January–March (JFM) months, which correspond to the period of maximum physical variability. Fig. 3a shows, in bars, the standardized PC time series for the first mode of atmospheric variability (hereafter AL index), describing changes in the Aleutian Low strength, whereas Fig. 3b shows the PC time series for the second mode of atmospheric variability (hereafter NPO index), describing the North Pacific Oscillation. We computed linear regression coefficients between the time series of simulated variables at each grid point and the two indices. The resulting anomalies correspond to a +1 standard deviation departure of the index, assuming a linear relationship between the two. Panels 4a and 4b show the linear regression coefficients of JFM SLP and surface wind anomalies at each model grid point onto the AL and NPO indices. In this model, AL fluctuations (Fig. 4a, colors) explain 52% of the SLP variance in the 20th century. By convention, when the AL index is in its positive polarity (hereafter AL+) the Aleutian Low is stronger than usual and the wind field exhibits an anomalous cyclonic circulation throughout the North Pacific (Fig. 4a, arrows). This involves strengthening of high-latitude easterlies and of westerly winds between 30◦ N and 45◦ N, and weakening of subtropical easterlies. In this model the NPO explains 22% of the SLP variance in the 20th century and corresponds to a seesaw of sea level pressure between subtropical and arctic latitudes (Fig. 4b, colors). Positive values of the NPO index (hereafter NPO+) indicate enhancement of the north-south SLP gradients and intensification and northward displacement of mid-latitude westerly winds between 45◦ N and 60◦ N (Fig. 4b, arrows). Even though the simulated atmospheric fluctuations are mostly characterized by interannual-scale fluctuations, they also exhibit some degree of persistency, as it can be seen from their 9-year running averages (Fig. 3, red lines). The 9-year running mean of the AL index matches quite well also the 9-year running mean of the

simulated first mode of SST variability (Fig. 3 blue lines), i.e. the PDO. In this model no correlation is evident between the NPO and the second mode of North Pacific SST variability, as also detected in several coupled climate models used for the IPCC Fourth Assessment Report (Furtado et al., 2011). Panels 4c and 4d show the linear regression coefficients of winter SST onto the AL and NPO indices (colors) and superimposed the winter SST climatology (contours). Correlation coefficients between the SST time series and the AL and NPO indices were also computed, and in Fig. 4c and d only the SST anomalies corresponding to correlations statistically significant at 90% are shown. An AL+ year is associated with lower-than-average SST (∼0.5 ◦ C) in the western middle latitudes and higher-than-average SST to the east and south, a pattern clearly identifiable as the PDO (Mantua and Hare, 2002). The winter MLD response to AL+ years (Fig. 4e) shows an east-west dipole of deeper MLD in the west and shallower MLD in the east. Simulated changes in MLD may exceed 150 m, corresponding to a 50% change with respect to the 200–300 m depth climatological value (Fig. 4e, contours). A NPO+ year leads to a north-south dipole of higher-thanaverage SST values in the western subtropical latitudes and of lower-than-average SST values at middle and subpolar latitudes (∼0.5 ◦ C changes), as shown in Fig. 4d (in agreement with Di Lorenzo et al., 2008; 2009). In this model, NPO+ winters lead to an up to 50 m deepening of the winter MLD at 50◦ N and in the eastern part of the basin, and a slight MLD shoaling in the western subtropical latitudes (Fig. 4f). 4.2. Biogeochemical and plankton community response The response of biological variables to the AL fluctuation and to the NPO is investigated using a selected set of model variables: nitrate concentration (as an example of the multiple inorganic nutrients), chlorophyll concentration (as a proxy of total phytoplankton biomass), carbon biomass of diatoms and nanophytoplankton, and net primary and community productions. Net primary production (hereafter NPP) is defined as the difference between gross photosynthetic rate and phytoplankton respiration. Net community production (hereafter NCP) is computed by subtracting all heterotrophic respiration fluxes from NPP, and is a measure of total organic carbon production which is potentially available to higher trophic levels or export. Variables are averaged (or integrated in case of rates) down to the euphotic layer depth. 4.2.1. Spatial patterns Figs. 5 and 6 show linear regression coefficients (colors) of annually averaged biological variables onto the AL index and onto the NPO index respectively. As for Fig. 4, the anomalies are shown only for grid points in which the correlation coefficients are significant at 90%, and annual climatologies are shown as contours. During AL+ years, deeper winter mixing in the central-western Pacific causes simulated nitrate (Fig. 5a) and chlorophyll concentrations (Fig. 5b) averaged in the euphotic depth to be higher-than-average (by 10–50%) between 30◦ N and 45◦ N, with chlorophyll concentrations up to 0.05 mg m−3 higher than climatological values (which range in this region between 0.1 and 0.5 mg m−3 ). The phytoplankton subpolar maximum, located on average at 50◦ N, exhibits a southward expansion toward middle latitudes, consistently with the model and satellite studies of Chai et al. (2003) and Martinez et al. (2009) who find a chlorophyll increase in the central Pacific after 1976–1977 when the PDO entered in a predominantly positive phase. The chlorophyll changes are mostly due to diatoms (Fig. 5c) which the model indicates as the dominant phytoplankton group at these latitudes. On the other hand, in the Alaskan Gyre, AL+ winters lead to a decrease of annual chlorophyll concentrations of 0.03–0.04 mg m−3 . This is caused

L. Patara et al. / Ecological Modelling 244 (2012) 132–147

AL+ anomalies SLP and wind velocity

75oN

1 m sec

75oN

4

60 N

6 −1

60oN

1 m sec

4 2

0

45 N

hPa

2 o

o

−2

o

30 N

0

45 N

−2

30oN

−4

−4 125oE

150oE

c

175oE

160oW 135oW

125oE

−6

d

SST

o 75 N

150oE

175oE

160oW 135oW

SST

o

75 N

0.48

e

o

175 E

o

−0.48

160 W 135 W

96 9 18 21 o

125 E

24o

150 E

f

o

175 E

6

°

o

30 N

MLD

o

0.24

C

o

−0.24

21

45 N

0 63 9 12 15 18 21

9 12 o

o

−0.24

21

−0.48

160 W 135 W

MLD

o

75 N

75 N

40

160

−160

200 10075 o

125 E

o

150 E

400 75

m

400 75

75

75

o

30 N

0 10 1 7500 50 o o 175 E 160 W

7550

−80

45 N

15000 20 0 75 75 0 10 50

o

135 W

0

o

0

o

150 E

80

10

o

125 E

0 10 1 7500 50 o o 175 E 160 W

5 270050 0 5 200 120007 05

o

60 N

50

200 10075

7550

o

0

45 N

10

50

o

15000 20 0 75 75 0 10

50

5 270050 0 5 200 120007 05

60oN

30 N

0

15

18

24o

150 E

0

o

18

o

125 E

9 12

18

9 18 21

0 3

18

96

6

45 N

0 63 9 12 15 18 21

15

o

0.24

C

0 3

0.48 60oN

°

60oN

30oN

−6

o

135 W

20 0

m

o

b

6 −1

NPO+ anomalies SLP and wind velocity

hPa

a

137

−20 −40

Fig. 4. 20th century JFM linear regressions onto the first SLP principal component time series (PC1, left column) and onto the second SLP principal component time series (PC2, right column) of (a, b) sea level pressure (SLP) in hPa (colors) and wind speed at 10 m height in m s−1 (arrows), (c, d) sea surface temperature (SST) anomalies in ◦ C (colors) and corresponding climatology (contours), (e, f) mixed layer depth (MLD) anomalies in m (colors) and corresponding climatology (contours). Only grid-points corresponding to 90% significant correlations with the PCs are shown; note the different color scaling between panels e and f to highlight the different response of MLD to the two modes. (For interpretation of the references to color in this figure legend, the reader is referred to the web version of the article.)

by shallower-than-average mixing (Fig. 4e) which supplies less nutrients in the euphotic layer (Fig. 5a), in agreement with the studies of Alexander et al. (2008) and Wong et al. (2007). During AL+ years, in the Alaskan Gyre the simulated diatom biomass decreases whereas the nanophytoplankton biomass slightly increases. This points to a possible change in phytoplankton community structure, even though the model is by construction rigid in simulating changes in community composition since it considers only three phytoplankton groups. A closer inspection of our model outputs suggests that the increase in nanophytoplankton is due to a topdown control (reduced micro-zooplankton abundance) rather than a bottom-up control (not shown). The dominant response of AL+ years is also a ∼30% increase in simulated NPP (Fig. 5e) and NCP (Fig. 5f) in the central Pacific, indicating that more carbon remains available in organic form for higher trophic levels transfer and/or export. This would imply higher food availability for fish, although the latter component is not explicitly simulated in this model. It must be noted that the model allows for different pathways of carbon production (as also described in Section 6) therefore the NCP increase is not just a direct consequence of the increase in NPP. In the Alaskan Gyre, the model

results show a slight reduction of NPP, whereas the NCP changes are not statistically significant at 90% level. The reduction in NPP and NCP is more evident in the region of the California Current System, where both diatoms and smaller phytoplankton decrease following the shoaling of the MLD. The simulated response of biogeochemical variables to the NPO is statistically significant on smaller areas than for the AL variability (Fig. 6). Yet at around 45◦ N it is the dominant mode explaining the simulated variations in chlorophyll (Fig. 6b), which increases by up to 0.05 mg m−3 (∼10% of the mean annual value, shown in contours). This is caused by deeper MLD (Fig. 4f) which leads to higher nutrient availability (Fig. 6a) and enhanced diatom biomass (Fig. 6c). Nanophytoplankton biomass is instead reduced at subpolar latitudes (Fig. 6d), because of lower SST which increases the sea ice coverage (even though it should be noted that the model overestimates sea ice cover over these areas). Off the coast of California, increased nanophytoplankton biomass is responsible for a small chlorophyll increase, which is consistent with the study of Di Lorenzo et al. (2009). The changes in NPP (Fig. 6e) and NPC (Fig. 6f) in response to NPO+ years are also characterized by a statistically significant increase at around 45◦ N.

L. Patara et al. / Ecological Modelling 244 (2012) 132–147

nitrate

5

10

175oE

−32

0.2

mg m−3

00.5.3 0.1

00.2 .10.02 0. 5

0.32

3

0 −0.32

mgC m−3

3 2

0.2

0

o

60 N

−0.64

0

32

0

20 o 100 20 45 N 0 2 10 520 100 50 0−20 500 0 250 0 o 30 N −2 0 −2200 −20 50 −2 0 −20 −0020 5 50 − 50 125oE 150oE 175oE 160oW 135oW

0

0 0

−1

year

−2

10

160oW 135 W

−16

gC m

10 10 50

115000

1000 50 1 o

0

00

f

0

150oE

10 1 0

0

−0.04

NCP

0 20 50

10

50

160 W 135 W

0

32 16

175 E

20

o

200 150 1000 5

10

150 E

50 000 20 1 5

e 125 E

0 10 150

10 50

o

30 N

0.1

d125 E 75oN

50

50

10

0

10

100

50

10

60 N

50

o

−0.02

16 0 −16

gC m−2 year−1

o

30 N

NPP

75oN

45 N

50 0

−2.4

160oW 135oW

o

5 0.0 0.5 0.05 0.20.5 1 0 5 51 2 . 0.2 0. 0.0 0.5 2 0.05 0.05 0.005.051.2 2 0 05o o o o o

100

1

o

60 N

0.05 0. 2

−3

−1.2

mgC m

2105 5

175oE

0

0.64

0

5

1

150oE

160oW 135oW

.25 0100.0 .5

o

0.02

0.1

0 02

175oE

0.2

c 125 E

150oE

0.04

nanophytoplankton

75oN

1

5

30 N

o 45 N

0.02

o

5

30 20 15 10 1

2 0.0

b125 E

2.4

0.7

0

0.1

0.1

−0.8

1.2

15 112500

o

30

1 10

5

45 N

0. 02

0.1

0. 1 0.1

10

1

5

1

2105 30 30 51 10 15 20 1

o

160oW 135 W

1

o

60 N

o

30 N

.5 o01

2 0.00.5 0.5 0. 3 0.2 0.1 2 0.0 0.02

0. 0.7 5 2 0.3 0 . 0 0.1

o

45 N

diatoms 11

1

15

175oE

5

o

75 N

150oE

60 N

0.02

22 0.00.

0.1

o

−0.4

0.02

0. 3

o

.2 0.1 0

0.1

a 125 E

0 1

30 N

0.5 1 0.1 0.

o

35

0.02

0.1 0.23

3

o

45 N

0.4

5 00..11

1

3 3 5 3 1 0.5

chlorophyll

75oN

0.8 −3

1 0.5

50.151 0.1

mmol m

1 3

o

60 N

3

001.5 .1 .1 0 0.51

3

50 301.5 .1 3

1

o

75 N

50

138

−32

Fig. 5. 20th century linear regressions onto the first SLP principal component time series (PC1) of annual anomalies of (a) nitrate concentration averaged over the euphotic depth (ED) in mmol m−3 , (b) chlorophyll concentration averaged over the ED in mg Chl m−3 , (c) diatom biomass averaged over the ED in mg C m−3 , (d) nanophytoplankton biomass averaged over the ED in mg C m−3 , (e) net primary production (NPP) integrated in ED in g C m−2 year−1 , (f) net community production (NCP) integrated in ED in g C m−2 year−1 . The anomalies are shown only on grid points in which the correlation coefficients with the PC1 time series are significant at 90%. Panel (b) additionally shows the area (35–50◦ N, 155◦ E–145◦ W) on which averages shown in Fig. 7a, c and e are computed.

4.2.2. Seasonal cycles The seasonal response of biological variables to AL fluctuations and to the NPO is investigated by building composites over simulation years in which AL and NPO indices are higher (or lower) than +1 (−1). Spatial averages are computed over the regions depicted in Fig. 5b (for the AL fluctuations) and in Fig. 6b (for the NPO), where each region corresponds to the areas of highest biogeochemical response to each mode. Fig. 7 shows composite seasonal cycles of MLD, nitrate, and chlorophyll concentration together with the seasonal climatologies and their standard deviations. In both selected areas, the MLD seasonal climatology (Fig. 7a and b) is close to the observed estimates from De Boyer Montégut et al. (2004), even though MLD tends to be deeper-than-observed in the central Pacific (Fig. 7a). For both modes of variability, positive index phases correspond to ∼50 m deeper winter MLD, which is in the standard deviation range. Deeper winter mixing, occurring during the positive polarity of each mode, enhances winter nutrient concentrations in the euphotic layer (Fig. 7c and d), whereas the converse occurs during negative index phases. The chlorophyll bloom peaks in May (Fig. 7e) in agreement with SeaWiFS satellite estimates averaged over 1998–2006 (black dashed line on the same plot). Going poleward, the chlorophyll peaks in June (Fig. 7f). The model tends to overestimate the amplitude of the chlorophyll seasonal cycle, maybe because of the overestimated MLD seasonality (Fig. 7a and b). During the positive

index years of both modes of variability, the chlorophyll seasonal cycle is amplified (Fig. 7e and f) because deeper winter MLD reduces light availability in late winter but enhances the nutrient supply into the euphotic layer for the eventual spring growth. During AL+ years, chlorophyll maxima are enhanced by 30–40%, even though the duration of the bloom is unchanged. On the other hand, during NPO+ years phytoplankton biomass remains higher during both May and July, indicating a later termination of the bloom. An asymmetry between positive and negative AL and NPO composites is evident, i.e. negative composites are closer to climatological values than positive index phases. This is due to the fact that in this 20th century simulation negative index phases of both modes of variability are more frequent than the positive index phases (Fig. 3). Thus, the climatology is biased by the negative index phases. 5. Anthropogenic impacts 5.1. Physical impacts The projected impacts of anthropogenic climate warming in the 21st century are analyzed by computing differences between the last 30 years of the 21st century (2070–2099) and the last 30 years of the 20th century (1970–1999). Fig. 8 shows 90%-significantchanges in annual SST, annual sea ice cover, annual sea surface

L. Patara et al. / Ecological Modelling 244 (2012) 132–147

nitrate

d125 E

−0.02

mg m−3

00.2 .10.02 0. 5

00.5.3 0.1

0.2

0

−0.04

−0.16

mgC m−3

3 2

0

3

−0.32

0

o

0

16

00

0

20 100 20 45oN 0 2 1 0 520 2500 100 50 0−20 500 o 30 N −20 0 −2200 −20 50 −2 0 −20 −0020 5 50 − 50 125oE 150oE 175oE 160oW 135oW

f

8 0 −8

gC m−2 year−1

0

50 0

−1

0.16

NCP

0

10

year

−2

−16

160 W 135 W

0

60 N

0 0

−8

gC m

115000

10 10 50

0

175 E

0

160oW 135 W

150 E

0 20 50

10

175oE

1000 50 1 o

0.2

−3

o

30 N

20

10 150oE

10 1 0

45 N

16 8

50

o

5 0.0 0.5 0.05 0.20.5 1 5 0.2 51 2 0 . 0.2 0. 0 0.5 0.05 0.05 0.005.051.2 2 0 05o o o o o

50 000 20 1 5

200 150 1000 5

10

10 50

0 10 150

160 W 135 W

0.02

0.32 o

60 N

75oN

50

50

10

0

10

100

50

10

175oE

100

−2.4

160oW 135oW

150oE

0

−1.2

1

2 0.0

0.05 0. 2

0

0.1

0.04

nanophytoplankton

75oN

mgC m

1

5 175oE

50

o

60 N

o

0.1 o

NPP

e 125 E

0.02

b125 E

0.7

50

0.1

5

150oE

75oN

o

0. 02

0.1 1

o

30 N

o

2 0.00.5 0.5 0. 3 0.2 0.1 2 0 . 0 0.02 2 o0.1 o0 0

.25 0100.0 .5

5

1

c 125 E

45 N

0.0

0.02

22 0.00.

0.2

30 20 15 10 1

30 N

o

45 N

30 N

2.4 1.2

15 112500

30

1 10

5

o

0. 0.7 5 2 0.3

o

−0.4

5

2105 30 30 51 10 15 20 1

o 45 N

−3

0.1

10

1

5

1

1

o

60 N

0.15

160oW 135oW

60 N

diatoms 11

1

15

175oE

2105 5

o

150oE

−0.2

0.02

0. 3

o

0.1

o

75 N

1

0.1

a 125 E

0 0.

o

30 N

0.2

35 0.5 1 0.1

0. 1 0.1

45 N

0.4

0.02

.2 0.1 0

3

o

5 00..11

1

3 3 5 3 1 0.5

chlorophyll

75oN

02 .1 0. 3

1 0.5

50.151 0.1

mmol m

1 3

o

60 N

3

001.5 .1 .1 0 0.51

3

50 301.5 .1 3

5

o

75 N

139

−16

Fig. 6. 20th century linear regressions onto the second SLP principal component time series (PC2) of annual anomalies of (a) nitrate concentration averaged over the euphotic depth (ED) in mmol m−3 , (b) chlorophyll concentration averaged over the ED in mg Chl m−3 , (c) diatom biomass averaged over the ED in mg C m−3 , (d) nanophytoplankton biomass averaged over the ED in mg C m−3 , (e) net primary production (NPP) integrated in ED in g C m−2 year−1 , (f) net community production (NCP) integrated in ED in g C m−2 year−1 . The anomalies are shown only on grid points in which the correlation coefficients with the PC2 time series are significant at 90%. Panel (b) additionally shows the area (38–52◦ N, 175◦ E–140◦ W) on which averages shown in Fig. 7b, d and f are computed. (For interpretation of the references to color in this figure legend, the reader is referred to the web version of the article.)

salinity (SSS), and winter (JFM) MLD. The statistical significance of the mean changes of the SST, sea ice and SSS (which are normally distributed) is assessed with a t-test, whereas the statistical significance of MLD median changes (which is not normally distributed on large parts of the North Pacific basin) is assessed with the non-parametric rank-sum Wilcoxon test, which does not assume normality of the samples. The model simulates an increase in annual SST of up to 4 ◦ C in the central North Pacific (Fig. 8a, colors), which is in the range of model projections under the A1B scenario (Meehl et al., 2007). The model also simulates a 0.2–0.4 increase in SSS in the central Pacific (Fig. 8b) because of enhanced evaporation (not shown), and a 0.4–0.8 SSS decrease in the eastern part of the basin because of increased freshwater inputs by decreased sea ice (Fig. 8a, contours). Associated with surface temperature and salinity changes, the JFM MLD at the end of the 21st century is projected to shoal in the Alaskan Gyre and in the western part of the basin by more than 100 m, whereas MLD is projected to deepen north of 55◦ N due to reduced sea ice cover (Fig. 8c). However, due to large interannual MLD fluctuations in the 20th century (Fig. 9b), the changes in the MLD medians between 21st and 20th centuries are not statistically significant at 90% on large areas of the North Pacific. Fig. 9a and b shows, respectively, annual SST and winter MLD time series averaged over the 40–55◦ N Pacific area (shown in black

box in Fig. 10b) during the 20th and 21st century simulations. In the first part of the 20th century, interannual-to-decadal variability is large compared to longer-term trends and causes fluctuations of ∼1 ◦ C amplitude in SST and of several tens of meters in MLD. Starting from the last decades of the 20th century, North Pacific SST gradually grows to a value up to 3 ◦ C higher with respect to the first part of the 20th century. At the same time, winter MLD shoals by tens of meters and its interannual variability is largely damped. We remark that these trends are driven by climate forcing and are not spurious model drifts. This can be deduced (i) from previous simulations performed under constant greenhouse gas concentrations showing stable values for upper-ocean heat content (Patara et al., 2011, 2012) and (ii) by the fact that in the period 1900–1949, when human emissions were still low, the SST and MLD time series do not show any significant long-term trend. 5.2. Impacts on marine biogeochemistry 5.2.1. Time series Time series of annual chlorophyll concentrations averaged over the 40–55◦ N Pacific area are depicted in Fig. 9c. The simulated chlorophyll concentration and its interannual variability during the last decade of the 20th century are close to SeaWiFS satellite estimates for the years 1998–2006

140

L. Patara et al. / Ecological Modelling 244 (2012) 132–147

AL composites MLD

0

0

a

50

100

m

m

50

NPO composites MLD

150

mmol m

−3

c

0.5

1.5

Jul

chlorophyll

1 0.5 Jul

Sep Nov

Sep Nov

d

1 0.5

2

e

0 Jan Mar May

1.5

Jul

nitrate

0 Jan Mar May

Sep Nov

−3

−3

2

1

2

200 Jan Mar May

Sep Nov

nitrate

0 Jan Mar May

mg Chl m

Jul

mg Chl m

−3

mmol m

1.5

100 150

200 Jan Mar May 2

b

1.5

Jul

Sep Nov

chlorophyll

f

1 0.5 0 Jan Mar May

Jul

Sep Nov

Fig. 7. 20th century seasonal cycles during years in which the (left) first SLP principal component time series (PC1) and the (right) second SLP principal component time series (PC2) are higher than +1 (red dashed lines) and lower than −1 (blue dashed-dotted lines). Climatological values (full black line) and their standard deviation (gray shading) are also shown. The seasonal cycles are averaged (left) over the area (35–50◦ N, 155◦ E–145◦ W) shown in Fig. 5b, and (right) over the area (38–52◦ N, 175◦ E–140◦ W) shown in Fig. 6b. (a, b) Mixed layer depth (MLD) in m (note reversed y-axis) and observed climatology (dashed black line) from De Boyer Montégut et al. (2004), (c, d) nitrate concentration averaged over the euphotic depth (ED) in mmol m−3 , (e, f) chlorophyll concentration averaged over the ED in mg Chl m−3 and observed SeaWiFS estimates for the period 1998–2006 (dashed black line). (For interpretation of the references to color in this figure legend, the reader is referred to the web version of the article.)

(Fig. 9c, red line), even though it has to be remarked that the almost exact matching of the interannual fluctuations is a matter of coincidence, since the climate model is not forced by observed atmospheric reanalyses. From previous studies (Patara et al., 2011) it is known that this model exhibits a negative spurious trend in near-surface biogeochemical variables. In order to separate climate-related trends from model spurious drift, the linear chlorophyll trend is computed in the period 1900–1949, when human emissions were still low (cf. Fig. 9a and Fig. 1a from Vichi et al., 2011), yielding a spurious drift of 0.004 mg Chl m−3 decade−1 (Fig. 9c, blue line). Since past studies have shown that biogeochemical drifts in this model tend to diminish in time (Patara et al., 2011), this trend provides a higher-end estimate of the chlorophyll spurious drift. Mean annual chlorophyll concentration in this area decreases from ∼0.5 mg m−3 at the beginning of the 20th century to ∼0.25 mg m−3 , indicating an average decrease of 0.13 mg m−3 per century. The estimated model

spurious drift is 0.04 mg m−3 per century, i.e. less than 30% of the total trend (Fig. 9c, blue line). Hence, this model projection shows that, between 40◦ N and 55◦ N, a chlorophyll decrease of 0.09 mg m−3 per century is unambiguously related to anthropogenic climate change. In addition to long-term human-induced chlorophyll trends, natural fluctuations at a decadal scale are also visible from 9-year running means of simulated chlorophyll concentrations (Fig. 9c, green line). During the period 1998–2006, SeaWiFS chlorophyll estimates also show a decreasing trend (Fig. 9c, red line). However, the observed chlorophyll variability is completely in the range of the natural fluctuations simulated by the ESM. The comparison with the much longer chlorophyll time series simulated by the ESM suggests that one cannot unambiguously invoke anthropogenic climate warming to explain the observed reduction in chlorophyll (Henson et al., 2010), which might instead reflect the natural variability of the system.

L. Patara et al. / Ecological Modelling 244 (2012) 132–147

45 N

−5

0

1900

o

o

175 E

o

o

160 W 135 W

m

−2

150 E

SST

C

°

6 2

30oN

−4

75

1975

2000

2025

2050

2075

2100

1925

1950

1975

2000

2025

2050

2075

2100

1925

1950

1975

2000 years

2025

2050

2075

2100

b

SSS

o

−3

75 N

mg m

0.8 o

60 N

0.4

o 45 N

175oE

160oW 135oW

MLD

75oN

150 90

60oN

m

30 o

45 N −30 o

30 N

−90

150oE

175oE

160oW 135oW

0.3

c

Fig. 9. 20th and 21st century time series averaged over the area (40–55◦ N, 155◦ E–135◦ W) shown in Fig. 10b of (a) simulated annual sea surface temperature (SST) in ◦ C, (b) simulated JFM mixed layer depth (MLD) in m, (c) simulated annual chlorophyll concentration (CHL) averaged over the euphotic layer depth in mg m−3 (black line), its 9-year running mean (green line), and SeaWiFS satellite estimates for the years 1998–2006 (red line). The simulated chlorophyll spurious drift is estimated as the linear chlorophyll trend in the period 1900–1949 (continuous blue line), and shown also during the remaining part of the time series (dashed blue line) in order to distinguish climate-related trends from model spurious drifts.

−0.8

150oE

0.4

1900

−0.4

o

0.5

0.2

0

30 N

c 125oE

1950

150

1900

b 125oE

1925

CHL average

a 125oE

8

4

MLD

−5

o

a

C

−2 −355 −35 −25 −15

60oN

10

°

o

75 N

21st minus 20th century SST and sea ice −5 −5

141

−150

Fig. 8. Differences between the simulated years 2070–2099 and 1970–1999 of (a) annual SST in ◦ C (colors) and annual sea ice cover in % (contours), (b) annual sea surface salinity (SSS) and (c) JFM mixed layer depth (MLD) in m. Means are shown for SST, sea ice cover and SSS, whereas medians are shown for MLD. Only statistically significant differences at 90% are depicted, computed with a t-test for mean changes and with the Wilcoxon non-parametric test for median changes. (For interpretation of the references to color in this figure legend, the reader is referred to the web version of the article.)

5.2.2. Spatial patterns Fig. 10 shows median changes in biogeochemical properties between 2070–2099 and 1970–1999, where only statistically significant changes at 90% are shown. Medians for the period 1970–1999 are also shown in contours. The statistical significance of the median changes is tested through the non-parametric Wilcoxon test which is suitable for non-normally distributed time series, as it is the case for biological variables. In response to the projected increase in stratification at the end of the 21st century (Figs. 8c and 9b), simulated nutrient concentrations in the euphotic layer are reduced by ∼2 mmol m−3 on large parts of the middle and subpolar latitudes, which corresponds to a ∼50% reduction with respect to the last decades of the 20th century (characterized by maxima of ∼4 mmol m−3 ). As a response, chlorophyll concentration averaged in the euphotic layer decreases by more than 0.4 mg m−3 at the end of the 21st century, which corresponds to ∼60% decreases with respect to contemporary 20th century values which are ∼0.7 mg m−3 (Fig. 10b). At higher latitudes, chlorophyll concentration is projected instead to increase

by ∼0.1 mg m−3 owing to more extended ice-free regions (Fig. 8a, contours). Chlorophyll changes at middle and subpolar latitudes are mainly due to diatoms, which decrease by more than 50% on most of the subpolar latitudes and increase in formerly ice-covered regions (Fig. 10c). Nanophytoplankton biomass exhibits a decrease in the eastern middle latitudes (Fig. 10d) and an overall increase in the subpolar area north of 50◦ N owing to an increase of the ice-free regions, to reduced competition with diatoms, and more favorable temperature conditions. The NPP (Fig. 10e) and NCP (Fig. 10f) at the end of the 21st century are simulated to decrease by >50% at middle latitudes and to increase at subpolar and subtropical latitudes. The subpolar increase in NPP is associated with the increase in ice-free regions and possibly with increased temperature which stimulates photosynthetic rates. There are large subpolar areas where NPP increases whereas chlorophyll decreases. This can be explained by considering that under nutrient-deplete conditions, phytoplankton reduce the amount of chlorophyll in their cells but increase photosynthesis and DOC exudation when exposed to more light, a metabolic adjustment that is present in the parameterizations used in PELAGOS (Baretta-Bekker et al., 1997; Vichi et al., 2007) and that was invoked as a key mechanism by Taucher and Oschlies (2011). The NPP decrease at middle latitudes is instead related to the massive reduction of nutrient supply without any other effect linked to the already favorable light availability conditions. 5.2.3. Seasonal cycles Climatological seasonal cycles in the 20th and 21st century averaged over the area [40–55◦ N, 155◦ E–135◦ W] are presented in Fig. 11, and can be usefully compared with Fig. 7. The projected winter MLD shoals by several tens of meters with respect to contemporary conditions (Fig. 11a) because of the ∼3 ◦ C increase in SST (Fig. 11b). This leads to a drastic reduction (from 1.8 to 0.5 mmol m−3 ) of euphotic-layer nitrate (Fig. 11c) and of the magnitude of the spring phytoplankton bloom (Fig. 11d). The chlorophyll

L. Patara et al. / Ecological Modelling 244 (2012) 132–147

o

150 E

d

NPP

−80

0.2

−3

mg m

0 .3 0.1

5 0.7

0

30oN o 125 E

f

o

150 E

o

175 E

o

0 −40

o

−80

160 W 135 W

−3

mgC m

year

−2

40

−1

80

gC m

0 50

0 0

−1

year

−2

gC m

10

0

0

10 50

115000

10

10

160 W 135 W

−1.2

0

o

1 o 0

−40

o

45 N

0

175 E

10

0

o

60 N

0

o

150 E

50 10 o

1050

0 510

10

30 N o 125 E

15200 1000

40

o

5200

10

o

45 N

10 100

o

160 W 135 W

0 0 −20 0 20 0 0 50 50 20 20 1 0 0 0 0 12500150 5 20 205 − 20 −20 −20

20 20 0 15000

150

0 10 15 0

50

10

0

60 N

80

1500

−0.6

5

NCP

75oN 0

0 0

o

50 10

75oN

175 E

0

0

−16

0.0o5

0.0

0

−3

30 N o 125 E

0.5 0.2

1

0.05 0.2 0.5

0.6

100 0

o

45 N

−2

160oW 135oW

5 0.00.2

0

−8

5 0.00.2 0.5

60 N

0.0.05 2

0

mgC m

8

1.2 o

o

175oE

0.

5

15 30 15 10 5 1

20

510 1

30

nanophytoplankton

0

0.1

mmol m−3

3 53 3 0.1

0.1 150oE

−0.4

0.2 55 00..0 .5 00.2

c

160oW 135oW

1

30 N 125oE

175oE

0.2

1

o

20 5

−0.2

16

15

103015

0.1 0.0 2

1 0. 0.02

150oE

b

0

0.02 0.02

205 1 o

10 5 20 1

15 1 200

15

60 N

e

−4

75oN

1 5 30 20

o

o

160oW 135oW

0.2

023.1 00..

175oE

0. 0.502

0.3 0 .1 0.2

0.7 0.20.3

30 N 125oE

diatoms

75oN

45 N

o

45 N

0.4

0.02 2 0.

0.3 0.02

60 N

o

150oE

a

−2

o

0.2 2 00.1 0.

1 0..51 0

0 0.1

o

30 N 125oE

13 00..5 1

5

3

chlorophyll

75oN

3

o

45 N

3

00..15

0.5 10.5

1

4 2

5 01.

o

60 N

3 3 3 51 0.5 .1 01

1.1 00.5 0.1

501.5 5 3 0.1

0.1 0. 0. 75

nitrate

75oN

0. 0. 02 02 0.5

142

Fig. 10. Annual median differences between the years 2070–2099 and the years 1970–1999 (colors) and annual climatologies between 1970 and 1999 (contours) of (a) nitrate concentration averaged over the euphotic depth (ED) in mmol m−3 , (b) chlorophyll concentration averaged over the ED in mg Chl m−3 , (c) diatom biomass averaged over the ED in mg C m−3 , (d) nanophytoplankton biomass averaged over the ED in mg C m−3 , (e) net primary production (NPP) integrated in ED in g C m−2 year−1 , (f) net community production (NCP) integrated in ED in g C m−2 year−1 . The anomalies are shown only on grid points in which the median differences are statistically at 90% (computed with the Wilcoxon test). Panel (b) additionally shows the area (40–55◦ N, 155◦ E–135◦ W) on which averages shown in Figs. 9 and 11 are computed. (For interpretation of the references to color in this figure legend, the reader is referred to the web version of the article.)

decrease is again mainly due to diatoms (Fig. 11e) which constitute the largest part of the phytoplankton biomass in this area. Nanophytoplankton biomass (Fig. 11f) also exhibits a drastic decrease at these latitudes and an almost complete flattening of its seasonal peaks. The model also simulates a change in the seasonality of the chlorophyll bloom in response to climate change. In late winter, shallower MLD increases the permanence of phytoplankton within the euphotic depth and allows for earlier optimal stratification conditions for the bloom initiation. On the other hand the reduction of nutrient in the euphotic layer (Fig. 11c) causes a 1-month earlier termination of the phytoplankton bloom with respect to presentday conditions. These results are in agreement with the modeling study from Hashioka et al. (2009), who find that maximum spring phytoplankton biomass at subpolar latitudes is shifted 10–20 days earlier under a doubling CO2 climate scenario. 6. Impacts on food web dynamics The impacts of climate variability and change on the material and energy flows within the simulated plankton community of the North Pacific Ocean are here investigated. The PELAGOS model has a fixed community structure (Fig. 1) but allows for a high degree

of flexibility when it comes to adjusting fluxes and stoichiometric processes to changing environmental conditions. The annually averaged fluxes of carbon (in mg C m−2 day−1 ) between different aggregated compartments of PELAGOS (phytoplankton, zooplankton, bacteria, and dissolved and particulate organic carbon) are averaged over the area [35–55◦ N, 155◦ E–135◦ W], encompassing the two areas used for Figs. 7 and 11, and integrated over the euphotic layer depth. The fluxes are averaged over the last 30 years of the 20th century and of the 21st century, and over AL+ years (AL index > 0.8) during the last 50 years of the 20th century. For the AL composite we use 50 years instead of 30 in order to have a larger sample of AL+ years; we remark however that the results are similar when taking only the last 30 years of the 20th century. To better isolate the changes in the food web dynamics, the fluxes are normalized to the gross primary production (hereafter, GPP) of each of the three scenarios (20th century, AL+ composite, and 21st century). The resulting normalized fluxes, as well as the absolute value of the GPP, are schematically shown in Fig. 12. Red (blue) boxes in Fig. 12b and c indicate an increase (decrease) of the flux with respect to 20th century conditions. Simulated NPP is 93.5 g C m−2 year−1 under contemporary conditions. This estimate compares quite well with available observations in the North Pacific: at the station KNOT (44◦ N, 155◦ E)

L. Patara et al. / Ecological Modelling 244 (2012) 132–147

MLD

a C

50 1970−1999 2070−2099

100 Jan Mar May Jul

b

10

−3

2

1 0.5

c

diatoms −3

50

Sep Nov

d

1 0.5

0.8

e

0 Jan Mar May Jul

1.5

Sep Nov

chlorophyll

0 Jan Mar May Jul

Sep Nov

mg C m

100

Jan Mar May Jul

mg Chl m

−3

Sep Nov

1.5

0 Jan Mar May Jul

mg C m−3

SST

5

nitrate

2

mmol m

15

o

m

0

143

0.6

Sep Nov

nanophytoplankton

f

0.4 0.2 0 Jan Mar May Jul

Sep Nov

Fig. 11. Simulated seasonal cycles calculated during the years 1970–1999 of simulation (blue full line) and during the years 2070–2099 of simulation (red dashed line), and averaged over the area (40–55◦ N, 155◦ E–135◦ W) shown in Fig. 9b. (a) Mixed layer depth (MLD) medians in m (note reversed y-axis), (b) sea surface temperature (SST) medians (◦ C), (c) nitrate concentration medians averaged over the euphotic depth (ED) in mmol m−3 , (d) chlorophyll concentration medians averaged in ED (mg Chl m−3 ), (e) diatom concentration medians averaged in ED (mg C m−3 ), (f) nanophytoplankton concentration medians averaged in ED (mg C m−3 ). (For interpretation of the references to color in this figure legend, the reader is referred to the web version of the article.)

Imai et al. (2002) measured a value of 90 g C m−2 year−1 in the upper part of the euphotic zone, and at the station K2 (47◦ N, 166◦ E) Elskens et al. (2008) measured a 146–183 g C m−2 year−1 during summer (i.e. during the seasonal maximum). In the 20th century (Fig. 12a) this model shows that a large part (64% of GPP) of the fixed carbon in the central North Pacific is transferred to the DOC compartment, which is almost entirely used by bacteria for stock maintenance, and which provides an important food source for zooplankton (16% of GPP). In the 20th century the ecological role of bacteria is slightly higher than the direct zooplankton grazing rate over phytoplankton (14% of GPP). The simulated system in the 20th century is therefore characterized by a multivorous food web, with equally important carbon fluxes of organic carbon from bacteria and autotrophs to zooplankton. The realized bacterial growth efficiency is about 0.4 as it can be derived by the ratio between bacterial production (carbon demand minus respiration rates) and bacterial carbon demand. Since the model has built-in parameterizations with variable

stoichiometry, it is useful to consider the concurrent changes in nutrient uptake and remineralization rates (Table 1, taking phosphate as an example). Bacteria have a positive net uptake in all cases, indicating that organic matter does not have an optimal nutrient content. The entire planktonic web is sustained by preformed nutrients, as only 4% of total nutrient uptake is from regeneration. During AL+ years, GPP increases with respect to average 20th century conditions, whereas in the future scenario it is on average reduced (as already discussed for Figs. 5e and 10e). The effects of AL+ years on the carbon flows (Fig. 12b) tend to reinforce the grazing pathway (16% of GPP) which becomes larger than the flow of carbon from bacteria to zooplankton (15% of GPP). This is consistent with the increase in nutrient regeneration by zooplankton (Table 1). During the 21st century scenario (Fig. 12c) the model simulates an increase of all microbial loop fluxes, such as production of DOC by phytoplankton, DOC consumption by bacteria, bacterial respiration, and bacterial grazing by (micro-)zooplankton.

Table 1 Mean annual uptake and remineralization rates of dissolved PO4 3− (mmol P m−2 year−1 ) integrated over the euphotic layer by phytoplankton (uptake only), zooplankton (remineralization only), bacteria (uptake minus remineralization) and total uptake from water pre-formed and regenerated P (computed as the sum of phytoplanktonic and bacterial uptake). The percentages with respect to the total uptake are shown in brackets. The rates are averaged over the area [35–55◦ N, 155◦ E–135◦ W] and over the last 30 years of the 20th century, over the last 30 years of the 21st century, and over AL+ during the last 50 years of the 20th century.

20th century AL+ years (20th century) 21st century

Uptake by phytoplankton

Remineralization by zooplankton

Net uptake by bacteria

Total uptake

21.4 (96%) 28.3 (95%) 10.4 (94%)

3.3 (15%) 4.9 (16%) 1.2 (11%)

0.9 (4%) 1.5 (6%) 0.7 (6%)

22.3 29.8 11.1

144

L. Patara et al. / Ecological Modelling 244 (2012) 132–147

organic matter, simulated here by the upper-ocean C:N ratios of diatom and nanophytoplankton groups (Fig. 13). The model is designed to adjust the cellular ratio around an optimal value (ideally corresponding to the Redfield ratio) and ranging from a minimum required for the structural existence of the cell up to about twice the optimal value, in order to simulate internal luxury storage (Baretta-Bekker et al., 1997; Vichi et al., 2007). Over the last 30 years of the 20th century, the simulated mean value of C:N averaged over the first 100 m depth is lower than the optimal Redfield ratio of 6 for both diatoms and nanophytoplankton, indicating nutrient-replete conditions. The simulated values are within the observed range (Geider and La Roche, 2002), with an expected increase in C:N ratios toward the more nutrient-depleted subtropical gyre. For the AL+ composites, all phytoplankton groups show an increase in luxury uptake with a further reduction in the mean C:N ratio (particularly for nanophytoplankton) in the southwestern part of the basin. This implies that food quality is generally high and that top-down control may become an important regulator of the standing stock. During the 21st century scenario, both nanophytoplankton and diatoms experience an increase of their intra-cellular C:N ratios over most subpolar and middle-latitude regions, although the values are only slightly higher than the optimal ones. In the 21st century, phytoplankton cells in the surface ocean are therefore more limited by nutrients than during current climate conditions, leading to an overall reduction of net community production as shown in Section 5.2.1. 7. Summary and discussion

Fig. 12. Annual carbon fluxes integrated over the euphotic layer among aggregated PELAGOS compartments (phytoplankton, zooplankton, bacteria, DOC and POC). Fluxes are averaged over the area [35–55◦ N, 155◦ E–135◦ W], integrated in the euphotic layer, and averaged (a) over the last 30 years of the 20th century, (b) for AL+ years in the 20th century (last 50 years), and (c) over the last 30 years of the 21st century. The fluxes are normalized to the gross primary production (in g C m−2 year−1 ) of each scenario, shown in brackets at the top of each graph. The red (blue) boxes indicate an increase (decrease) of the normalized flux with respect to the 20th century. (For interpretation of the references to color in this figure legend, the reader is referred to the web version of the article.)

An increase in phytoplankton DOC production under warmer conditions was also found in mesocosm experiments performed by Wohlers et al. (2009). In the 21st century, the bacterial predation becomes the main route of carbon toward zooplankton (17% of GPP) at the expenses of the direct grazing food chain which exhibits a significant decrease (9% of GPP). Consistently, a substantial reduction of both nutrient uptake and remineralization rates is simulated (Table 1). The relative changes in carbon and nutrient flows within the food web impact the stoichiometric quality of the autotrophic

In this study, the impact of natural climate variability and anthropogenic climate change on the structure, seasonality, spatial distribution, and productivity of North Pacific plankton groups was investigated by means of the INGV-CMCC Earth System Model (ESM). A novel aspect of this study was to combine the complexity of the microbial-level material flows and trophic interactions with a state-of-the-art coupled model, to obtain an assessment of the human-induced global warming impacts in conjunction with the two main modes of North Pacific atmospheric variability. The analysis was based on a historical 20th century simulation, and on a 21st century climate projection forced with the IPCC A1B scenario. The model showed significant effects associated with natural fluctuations of the Aleutian Low (AL) and of the North Pacific Oscillation (NPO). A deepening of the AL is associated with decreased SST and stratification in the central and western Pacific, and with increased SST and stratification in the Alaskan Gyre (0.5–1 ◦ C changes). A strengthening of the NPO is connected with reduced SST and stratification at subpolar latitudes. The model projection at the end of the 21st century showed an up to 4 ◦ C increase in SST, reduced sea ice cover, and significant shoaling of the mixed layer depth at mid- and subpolar latitudes. The PELAGOS biogeochemical model, embedded in the ESM model, was used to assess the impacts of these physical changes on the lower trophic levels of the ecosystem. PELAGOS is a suitable tool in this framework, since it flexibly reacts to climate changes in terms of elemental ratios and material flows within the lower trophic levels of the ecosystem. Hereafter we summarize the main impacts on North Pacific plankton found in this study: (1) Changes in spatial patterns (schematically summarized in Fig. 14): during AL+ years, chlorophyll concentrations, net primary production, and net community production are 10–50% higher-than-average in the central Pacific and lower-thanaverage in the Alaskan Gyre. NPO+ years lead to a 10% increase in chlorophyll and primary and community production at

L. Patara et al. / Ecological Modelling 244 (2012) 132–147

145

AL+ anomalies C:N ratio nanophytoplankton

AL+ anomalies C:N ratio diatoms 75° N

°

3

3

75 N

2

2

0

45° N

−1 °

30 N

60 N

mol mol−1

1

160° E

175° W

150° W

125° W

0

45° N

−1 °

30 N

−2 135° E

1

−3

mol mol−1

°

°

60 N

−2 135° E

21st century − 20th century C:N ratio diatoms

160° E

175° W

150° W

125° W

−3

21st century − 20th century C:N ratio nanophytoplankton

75° N

°

75 N

3 2

°

3 2

°

0

45 N

−1 °

30 N

mol mol−1

1 °

−2 °

135 E

°

160 E

°

175 W

°

150 W

°

125 W

−3

1 °

0

45 N

−1 °

30 N

mol mol−1

60 N

60 N

−2 °

135 E

°

160 E

°

175 W

°

150 W

°

125 W

−3

Fig. 13. Simulated annual C:N ratios (mol mol−1 ) in (a, c) diatoms and (b, d) nanophytoplankton averaged over the upper 100 m depth. Colors: (a, b) Differences between AL+ composites and 20th century average conditions; (c, d) differences between 21st century and 20th century. Contours: average values over the 20th century, where the thick line indicates the Redfield C:N ratio (equal to 6), dashed lines show C:N < 6, and thin solid lines indicate C:N > 6. The 20th and 21st century averages are over the last 30 years of each century, the AL+ composites over the last 50 years of the 20th century. (For interpretation of the references to color in this figure legend, the reader is referred to the web version of the article.)

the large increase in stratification, the phytoplankton bloom is flattened by more than 50% and it occurs one month earlier due to improved light conditions in late winter. (4) Changes in food web dynamics: During AL+ years, at middle latitudes the classical food web chain is enhanced, with larger nutrient regeneration via zooplankton, whereas the carbon channeling through the microbial loop decreases. Conversely, at the end of the 21st century the microbial loop carbon fluxes exhibit an increase at the expenses of a significant contraction of the direct zooplankton grazing chain. In the 21st century the more nutrient-depleted conditions lead to a reduction of nutrient uptake by phytoplankton and to an increase of their intracellular C:N ratios. Fig. 14. Schematic representation of the areas where net primary production is mostly impacted by anthropogenic climate change in the 21st century (solid boxes), by a deepening of the AL (AL+, dashed boxes), and by increased subtropical–subarctic pressure gradients (NPO+, dotted box). Red (blue) contours indicate net primary productivity increases (decreases). (For interpretation of the references to color in this figure legend, the reader is referred to the web version of the article.)

∼45◦ N in the eastern part of the basin. In the scenario projection, chlorophyll diminishes by 50% over the middle and subpolar latitudes and increases over formerly sea-ice covered regions. Primary production increases over subpolar latitudes, likely stimulated by the temperature increase. The impacts of natural climate variability and anthropogenic changes therefore tend to counteract each other at middle latitudes. (2) Shifts in plankton community composition: a deepening of the AL causes a decrease in diatom abundance and an increase in the smaller nanophytoplankton in the Alaskan Gyre. At the end of the 21st century, nanophytoplankton, better adapted to nutrient-depleted conditions, take advantage of the ecological niche left open by diatoms at subpolar latitudes. (3) Changes in seasonality: Both AL+ and NPO+ years lead to a ∼30% amplification of the seasonal cycle of chlorophyll concentration at middle latitudes. At the end of the 21st century, because of

There is substantial agreement between this and previous studies regarding the response of bulk biogeochemical properties, such as chlorophyll concentration and primary productivity, to variations in the AL strength (e.g. Alexander et al., 2008; Chai et al., 2003; Martinez et al., 2009), to fluctuations of the NPO (Di Lorenzo et al., 2008, 2009), and to global ocean warming projections for the 21st century (e.g. Bopp et al., 2005; Schmittner et al., 2008; Henson et al., 2010; Steinacher et al., 2010; Vichi et al., 2011). This work expands these previous studies by showing the relevance of these changes for the structure of the North Pacific planktonic food web, which in turn may affect higher trophic levels and the ocean biogeochemical cycling. Our results suggest that the increasing nutrient depletion at middle latitudes in the course of the 21st century may cause a reorganization of the carbon fluxes within the food web, with the recycling within the microbial loop becoming the dominant pathway of carbon channeling to (micro-)zooplankton and the direct grazing chain being reduced. This study also suggests significant changes in the phenology of phytoplankton blooms in a warmer world. It could be hypothesized that these shifts may cascade on higher trophic levels, with unknown consequences on abundance and migration patterns of fish and mammal populations (Moore, 2008). For instance, a shift toward the microbial loop may

146

L. Patara et al. / Ecological Modelling 244 (2012) 132–147

favor salp abundance with respect to krill, and there is evidence that this shift may already be at work in other subarctic biomes (Atkinson et al., 2004). This study also suggests that, in a warmer world, phytoplankton may produce an organic matter which is more carbon-enriched with respect to present day (Wohlers et al., 2009) with possible impacts on the efficiency of the biological pump (Schneider et al., 2004). This study however suggests that anthropogenic impacts should be considered jointly with natural climate fluctuations: indeed, the impacts of AL fluctuations on productivity and material flows are only slightly lower than those due to anthropogenic emissions at the end of the 21st century. This means that the changes in the microbial loop and classic grazing chain due to anthropogenic warming might be much more limited in years characterized by a strong AL, whereas they would possibly be reinforced in years characterized by a weak AL. In responding to natural and anthropogenic impacts, the relative role of direct and indirect temperature effects on primary productivity and plankton food web structure is a topic of current debate (Wohlers et al., 2009; Taucher and Oschlies, 2011). In simulations designed to tackle this issue, Taucher and Oschlies (2011) found that in response to human-induced global warming, the direct effect of temperature in stimulating primary productivity and the microbial loop may counter the indirect temperature effect of reduced nutrient supply. The resulting overall increase in primary production (Schmittner et al., 2008) would not be inconsistent with concomitant chlorophyll decreases (Behrenfeld et al., 2006; Boyce et al., 2010), since phytoplankton organisms in nutrient-depleted conditions reduce their chlorophyll amount whilst increasing photosynthesis. Our results show a pattern consistent with these hypotheses. In a warmer climate, we detect an overall chlorophyll decrease at subpolar latitudes whereas primary productivity is higher with respect to contemporary conditions. This can be explained by considering that in the model the parameterization of photosynthetic rates is not directly linked to nutrient availability, while temperature has a direct effect on all metabolic rates. At middle-latitudes we detect a large reduction in nutrient uptake due to increased stratification (an indirect temperature effect) which has in this case the effect of decreasing both chlorophyll and primary production. At middle latitudes, the reduced transformation of N and P into biomass is thus a dominating factor in determining the decrease in primary production (Vichi et al., 2007). At middle latitudes direct temperature effects of anthropogenic warming are possibly evident in the increase of carbon channeling to the DOC compartment and to the fast recycling within the microbial loop (Wohlers et al., 2009). The ESM used in this study is however affected by biases which, despite being in the range of IPCC climate models (Randall et al., 2007), might affect some of our results: (1) in most coastal areas the model does not reproduce the high chlorophyll values of satellite estimates (McClain, 2009), owing to unresolved upwelling and the related coastal production. This affects the model capability of fully capturing the phytoplankton response to wind fluctuations e.g. within the California Current (Chenillat et al., 2012). (2) This model underestimates SST and overestimates sea ice cover at subpolar latitudes. As a consequence, primary production in these areas is typically low in the 20th century and experiences a large boost in the warmer 21st century conditions. It is possible that both these features may be overestimated by the model. (3) The model exhibits a downward drift in chlorophyll throughout the simulation. The human-induced trend has been shown to dominate over the spurious drift, but the magnitude of the chlorophyll decrease in the 21st century may be overestimated by about 30%. An urgent question in Earth System Sciences is whether in the next decades anthropogenic trends may be distinguishable from the natural variability of the climate system (Henson et al., 2010). The comparison between the 200-year chlorophyll time series

simulated by the ESM and the 9-year time series estimated from the SeaWiFS satellite highlighted how chlorophyll temporal variations in such short observational data sets may be completely explained by the natural variability of the system. The model projections analyzed here suggest that, under the A1B scenario, in the second half of the 21st century anthropogenic impacts on marine biogeochemistry may be effectively separated from natural biogeochemical variability. In the preceding decades natural variability impacts may be as large as the human-induced ones (in agreement with Henson et al., 2010) and alternatively act to weaken or reinforce the effects of anthropogenic warming. The spatial signature of the two main modes should however be recognizable in the patterns of change over the North Pacific (Fig. 14) and the results shown here additionally provide an estimate of the amplitude of the changes related to the first two modes of climate variability. Acknowledgments We gratefully acknowledge the support of Italian Ministry of Education, University and Research and Ministry for Environment, Land and Sea through the project GEMINA. The authors thank the US National Aeronautics and Space Administration (NASA) for providing SeaWiFS chlorophyll satellite products and the Hadley Center for the sea surface temperature data available on the website www.metoffice.gov.uk/hadobs. Comments from two anonymous reviewers are gratefully acknowledged. References Alessandri, A., 2006. Effects of Land Surface and Vegetation Processes on the Climate Simulated by an Atmospheric General Circulation Model. PhD thesis 2006. Bologna University Alma Mater Studiorum, 114 pp. Alexander, M.A., Capotondi, A., Miller, F., Chai, R., Brodeur, R., Deser, C., 2008. Decadal variability in the northeast Pacific in a physical-ecosystem model: role of mixed layer depth and trophic interactions. Journal of Geophysical Research 113, C02017, http://dx.doi.org/10.1029/2007JC004359. Atkinson, A., Siegel, V., Pakhomov, E., Rothery, P., 2004. Long-term decline in krill stock and increase in salps within the Southern Ocean. Nature 432, 100–103. Baretta-Bekker, J.G., Baretta, J.W., Ebenhoh, W., 1997. Microbial dynamics in the marine ecosystem model ERSEM II with decoupled carbon assimilation and nutrient uptake. Journal of Sea Research 38 (3–4), 195–211, http://dx.doi.org/10.1016/S1385-1101(97)00052-X. Behrenfeld, M.J., O’Malley, R.T., Siegel, D.A., McClain, C.R., Sarmiento, J.L., Feldman, G.C., Milligan, A.J., Falkowski, P.G., Letelier, R.M., Boss, E.S., 2006. Climate-driven trends in contemporary ocean productivity. Nature 444 (7120), 752–755. Bindoff, N.L., Willebrand, J., Artale, V., et al., 2007. Observations: oceanic climate change and sea level. In: Solomon, S. (Ed.), Climate Change 2007: The Physical Science Basis. Contribution of Working Group I to the Fourth Assessment Report of the Intergovernmental Panel on Climate Change. Cambridge University Press, ISBN 9780521880091, pp. 385–497. Bond, N.A., Overland, J.E., Spillane, M., Stabeno, P., 2003. Recent shifts in the state of the North Pacific. Geophysical Research Letters 30 (23), 2183, http://dx.doi.org/10.1029/2003GL018597. Bopp, L., Aumont, O., Cadule, P., Alvain, S., Gehlen, M., 2005. Response of diatoms distribution to global warming and potential implications: a global model study. Geophysical Research Letters 32, L19606, http://dx.doi.org/10.1029/2005GL023653. Boyce, D.G., Lewis, M.R., Worm, B., 2010. Global phytoplankton decline over the past century. Nature 466 (7306), 591–596. Chai, F., Jiang, M., Barber, R.T., Dugdale, R.C., Chao, Y., 2003. Interdecadal variation of the transition zone chlorophyll front: a physical–biological model simulation between 1960 and 1990. Journal of Oceanography 59, 461–475. Chenillat, F., Riviere, P., Capet, X., Di Lorenzo, E., Blanke, B., 2012. North Pacific Gyre Oscillation modulates seasonal timing and ecosystem functioning in the California Current upwelling system. Geophysical Research Letters 39, http://dx.doi.org/10.1029/2011GL049966. De Boyer Montégut, C., Madec, G., Fischer, A.S., Lazar, A., Iudicone, D., 2004. Mixed layer depth over the global ocean: an examination of profile data and a profile-based climatology. Journal of Geophysical Research 109, C12003, http://dx.doi.org/10.1029/2004JC002378. Di Lorenzo, E., Schneider, N., Cobb, K.M., Franks, P.J.S., Chhak, K., Miller, A.J., McWilliams, J.C., Bograd, S.J., Arango, H., Curchister, E., Powell, T.M., Riviere, P., 2008. North Pacific gyre oscillation links ocean climate and ecosystem change. Geophysical Research Letters 35, L08607, http://dx.doi.org/10.1029/2007GL032838.

L. Patara et al. / Ecological Modelling 244 (2012) 132–147 Di Lorenzo, E., Fiechter, J., Schneider, N., Bracco, A., Miller, A.J., Franks, P.J.S., Bograd, ˜ A., Hermann, A.J., 2009. NutriS.J., Moore, A.M., Thomas, A.C., Crawford, W., Pena, ent and salinity decadal variations in the central and eastern North Pacific. Geophysical Research Letters 36, L14601. Di Lorenzo, E., Cobb, K.M., Furtado, J.C., Schneider, N., Anderson, B.T., Bracco, A., Alexander, M.A., Vimont, D.J., 2010. Central Pacific El Nino and decadal climate change in the North Pacific Ocean. Nature Geosciences 3 (11), 762–765, http://dx.doi.org/10.1038/NGEO984. Doney, S.C., et al., 2012. Climate change impacts on marine ecosystems. Annual Review of Marine Science 4, 11–37, http://dx.doi.org/10.1146/annurev-marine041911-111611. Elskens, M., Brion, N., Buesseler, K., Van Mooy, B.A.S., Boyd, P., Dehairs, F., Savoye, N., Baeyens, W., 2008. Primary, new and export production in the NW Pacific subarctic gyre during the vertigo K2 experiments. Deep Sea Research – Part II 55 (14–15), 1594–1604, http://dx.doi.org/10.1016/j.dsr2.2008.04.013. Finkel, Z.V., Beardall, J., Flynn, K.J., Quigg, A., Rees, T.A.V., Raven, J.A., 2010. Phytoplankton in a changing world: cell size and elemental stoichiometry. Journal of Plankton Research 32 (1), 119–137. Follows, M., Ito, T., Dutkiewicz, S., 2006. On the solution of the carbonate chemistry system in ocean biogeochemistry models. Ocean Modelling 12 (3–4), 290–301. Fogli, P.G., Manzini, E., Vichi, M., Alessandri, A., Patara, L., Gualdi, S., Scoccimarro, E., Masina, S., Navarra, A., 2009. INGV-CMCC Carbon (ICC): A Carbon Cycle Earth System Model. CMCC Research Paper 61. http://www.cmcc.it/ publications-meetings/publications/research-papers/rp0061-ingv-cmcccarbon-icc-a-carbon-cycle-earth-system-model. Freeland, H., Denman, K., Wong, C.S., Whitney, F., Jacques, R., 1997. Evidence of change in the winter mixed layer in the Northeast Pacific Ocean. Deep Sea Research – Part I 44, 2117–2129. Furtado, J.C., Di Lorenzo, E., Schneider, N., Bond, N.A., 2011. North Pacific decadal variability and climate change in the IPCC AR4 models. Journal of Climate 24 (12), 3049–3067, http://dx.doi.org/10.1175/2010JCLI3584.1. Geider, R.J., La Roche, J., 2002. Redfield revisited: variability of C:N:P in marine microalgae and its biochemical basis. European Journal of Phycology 37 (1), 1–17. Hare, S.R., Mantua, N.J., 2000. Empirical evidence for North Pacific regime shifts in 1977 and 1989. Progress in Oceanography 47 (2–4), 103–145, http://dx.doi.org/10.1016/S0079-6611(00)00033-1. Hashioka, T., Sakamoto, T.T., Yamanaka, Y., 2009. Potential impact of global warming on North Pacific spring blooms projected by an eddy-permitting 3-D ocean ecosystem model. Geophysical Research Letters 36, L20604, http://dx.doi.org/10.1029/2009GL038912. Henson, S.A., Sarmiento, J.L., Dunne, J.P., Bopp, L., Lima, I., Doney, S.C., John, J., Beaulieu, C., 2010. Detection of anthropogenic climate change in satellite records of ocean chlorophyll and productivity. Biogeosciences 7 (2), 621–640. Imai, K., Nojiri, Y., Tsurushima, N., Saino, T., 2002. Time series of seasonal variation of primary productivity at station KNOT (44 degrees N, 155 degrees E) in the subarctic western North Pacific. Deep Sea Research – Part II 49 (24–25), 5395–5408, http://dx.doi.org/10.1016/S0967-0645(02)00198-4. Johns, T.C., Royer, J.F., Höschel, I., et al., 2011. Climate change under aggressive mitigation: the ENSEMBLES multi-model experiment. Climate Dynamics 37 (9–10), 1975–2003, http://dx.doi.org/10.1007/s00382-011-1005-5. Karl, D.M., Bidigare, R.R., Letelier, R.M., 2001. Longterm changes in plankton community structure and productivity in the North Pacific Subtropical Gyre: the domain shift hypothesis. Deep Sea Research 48, 1449– 1470. Madec, G., Delecluse, P., Imbard, M., Levy, C., 1998. OPA 8.1 Ocean General Circulation Model Reference Manual. Note du Pole de Modélisation, vol. 11. Institut Pierre Simon Laplace, Paris. Mantua, N.J., Hare, S.R., Zhang, Y., Wallace, J.M., Francis, R.C., 1997. A Pacific interdecadal climate oscillation with impacts on salmon production. Bulletin of the American Meteorological Society 78, 1069–1079. Mantua, N.J., Hare, S.R., 2002. The Pacific decadal oscillation. Journal of Oceanography 58 (1), 35–44. Martinez, E., Antoine, D., D’Ortenzio, F., Gentili, B., 2009. Climate-driven basinscale decadal oscillations of oceanic phytoplankton. Science 326 (5957), 1253–1256. McClain, C.R., 2009. A decade of satellite ocean color observations. Annual Review of Marine Science 1, 19–42. Meehl, G.A., Stocker, T.F., Collins, P., et al., 2007. Global climate projections. In: Solomon, S. (Ed.), Climate Change 2007: The Physical Science Basis. Contribution of Working Group I to the Fourth Assessment Report of the Intergovernmental Panel on Climate Change. Cambridge University Press, Cambridge, United Kingdom and New York, NY, USA. Miller, A.J., Schneider, N., 2000. Interdecadal climate regime dynamics in the North Pacific Ocean: theories, observations and ecosystem impacts. Progress in Oceanography 47, 355–379. Moore, S.E., 2008. Marine mammals as ecosystem sentinels. Journal of Mammalogy 89 (3), 534–540, http://dx.doi.org/10.1644/07-MAMM-S-312R1.1.

147

Nakicenovic, N., Swart, R., 2000. Special Report on Emissions Scenarios. A Special Report of Working Group III of the Intergovernmental Panel on Climate Change. Cambridge University Press, Cambridge, UK, ISBN 0521804930, 612 pp. Ono, T., Shiomoto, A., Saino, T., 2008. Recent decrease of summer nutrients concentrations and future possible shrinkage of the subarctic North Pacific high-nutrient low-chlorophyll region. Global Biogeochemical Cycles 22, http://dx.doi.org/10.1029/2007GB003092, GB3027. Patara, L., Visbeck, M., Masina, S., Krahmann, G., Vichi, M., 2011. Marine biogeochemical responses to the North Atlantic Oscillation in a coupled climate model. Journal of Geophysics Research – Oceans 116, http://dx.doi.org/10.1029/2010JC006785, C07023. Patara, L., Vichi, M., Masina, S., Fogli, P.G., Manzini, E., 2012. Global response to solar radiation absorbed by phytoplankton in a coupled climate model. Climate Dynamics, http://dx.doi.org/10.1007/s00382-012-1300-9. Polovina, J.J., Dunne, J.P., Woodworth, P.A., Howell, E.A., 2011. Projected expansion of the subtropical biome and contraction of the temperate and equatorial upwelling biomes in the North Pacific under global warming. ICES Journal of Marine Science 68 (6), 986–995, http://dx.doi.org/10.1093/icesjms/fsq198. Preisendorfer, R., 1988. Principal Component Analysis in Meteorology and Oceanography. Elsevier Pub. Co., NY. Randall, D.A., Wood, R.A., Bony, S., et al., 2007. Climate models and their evaluation. In: Solomon, S., et al. (Eds.), Climate Change 2007: The Physical Science Basis. Contribution of Working Group I to the Fourth Assessment Report of the Intergovernmental Panel on Climate Change. Cambridge University Press, Cambridge, United Kingdom and New York, NY, USA, pp. 591–662. Rogers, J.C., 1981. The North Pacific Oscillation. Journal of Climatology 1, 39–57. Röckner, E., Bäuml, G., Bonaventura, L., et al., 2003. The atmospheric general circulation model ECHAM5, Part I: Model description. Max-Planck-Institute for Meteorology, Report No. 349, Hamburg, Germany. Schmittner, A., Oschlies, A., Matthews, H.D., Galbraith, E.D., 2008. Future changes in climate, ocean circulation, ecosystems, and biogeochemical cycling simulated for a business-as-usual CO2 emission scenario until year 4000 AD. Global Biogeochemical Cycles 22 (1), GB1013, http://dx.doi.org/10.1029/2007GB002953. Schneider, B., Engel, A., Schlitzer, R., 2004. Effects of depth- and CO2 -dependent C:N ratios of particulate organic matter (POM) on the marine carbon cycle. Global Biogeochemical Cycles 18 (2), http://dx.doi.org/10.1029/2003GB002184. Steinacher, M., Joos, F., Frolicher, T.L., Bopp, L., Cadule, P., Cocco, V., Doney, S.C., Gehlen, M., Lindsay, K., Moore, J.K., Schneider, B., Segschneider, J., 2010. Projected 21st century decrease in marine productivity: a multi-model analysis. Biogeosciences 7 (3), 979–1005. Taucher, J., Oschlies, A., 2011. Can we predict the direction of marine primary production change under global warming? Geophysical Research Letters 38, http://dx.doi.org/10.1029/2010GL045934. Timmermann, R., Goosse, H., Madec, G., Fichefet, T., Ethe, C., Dulière, V., 2005. On the representation of high latitude processes in the ORCA-LIM global coupled sea ice-ocean model. Ocean Modelling 8, 175–201. K.E., Hurrell, J.W., 1994. Decadal atmosphere-ocean Trenberth, variations in the Pacific. Climate Dynamics 9 (6), 303–319, http://dx.doi.org/10.1007/BF00204745. Valcke, S., 2006. OASIS3 User Guide (oasis3 prism 2-5). PRISM Support Initiative Report No 3. CERFACS, Toulouse, France, 64 pp. Vichi, M., May, W., Navarra, A., 2003. Response of a complex ecosystem model of the northern Adriatic Sea to a regional climate change scenario. Climate Research 24 (2), 141–159, http://dx.doi.org/10.3354/cr024141. Vichi, M., Masina, S., Pinardi, N., 2007. A generalized model of pelagic biogeochemistry for the global ocean ecosystem. Part I: theory. Journal of Marine Systems 64 (1–4), 89–109. Vichi, M., Masina, S., 2009. Skill assessment of the PELAGOS global ocean biogeochemistry model over the period 1980–2000. Biogeosciences 6, 2333–2353. Vichi, M., Manzini, E., Fogli, P.G., Alessandri, A., Patara, L., Scoccimarro, E., Masina, S., Navarra, A., 2011. Global and regional ocean carbon uptake and climate change: Sensitivity to a substantial mitigation scenario. Climate Dynamics 37 (9–10), 1929–1947, http://dx.doi.org/10.1007/s00382-011-1079-0. Wallace, J.M., Grutzler, D.S., 1981. Teleconnections in the geopotential height field during the Northern Hemisphere Winter. Monthly Weather Review 109, 784–812. Wanninkhof, R., 1992. Relationship between wind speed and gas exchange over the ocean. Journal of Geophysical Research 97, 7373–7382. Watanabe, Y.W., Ishida, H., Nakano, T., Nagai, N., 2005. Spatiotemporal decreases of nutrients and chlorophyll-a in the surface mixed layer of the western North Pacific from 1971 to 2000. Journal of Oceanography 61, 1011–1016. Wohlers, J., Engel, A., Zollner, E., Breithaupt, P., Jurgens, K., Hoppe, H.G., Sommer, U., Riebesell, U., 2009. Changes in biogenic carbon flow in response to sea surface warming. Proceedings of the National Academy of Sciences of the United States of America 106 (17), 7067–7072, http://dx.doi.org/10.1073/pnas.0812743106. Wong, C.S., Xie, L.S., Hsieh, W.W., 2007. Variations in nutrients, carbon and other hydrographic parameters related to the 1976/77 and 1988/89 regime shifts in the sub-arctic Northeast Pacific. Progress in Oceanography 75 (2), 326–342.