Life history and biogeography of Calanus copepods in the Arctic Ocean: An individual-based modeling study Rubao Ji1, Carin Ashjian1, Robert Campbell2, Changsheng Chen3, Guoping Gao3, Cabell Davis1, Geoffery Cowles3, Robert Beardsley1 1 Woods Hole Oceanographic Institution, MS # 33, Woods Hole, MA 02543, USA. Email: rji@whoi.edu 2 Graduate School of Oceanography, University of Rhode Island, Narragansett, RI, 02882, USA 3 University of Massachusetts Dartmouth, School for Marine Science and Technology, New Bedford, MA 02744, USA Abstract: Calanus spp. copepods play a key role in the Arctic pelagic ecosystem. Among four congeneric species of Calanus found in the Arctic Ocean and its marginal seas, two are expatriates in the Arctic (C. finmarchicus and C. marshallae) and two are endemic (C. glacialis and C. hyperboreus). The biogeography of these species likely is controlled by the interactions of their life history traits and physical environment. A mechanistic understanding of these interactions is critical to predicting their future responses to a warming environment. Using a 3-D individual-based model, we show that 1) C. finmarchicus is unable to penetrate into the Arctic Ocean under present conditions of temperature, food availability, and length of the growing season, mainly due to insufficient time to reach its diapausing stage and slow transport of the copepods into the Arctic Ocean during the growing season or even during the following winter, at the depths the copepods are believed to diapause. 2) For the two endemic species, the model suggests that their capability of diapausing at earlier copepodite stages and utilizing icealgae as a food source (thus prolonging the growth season length) contribute to the population sustainability in the Arctic Ocean. 3) The inability of C. hyperboreus to attain diapause in the central basin, as demonstrated by the model, contradicts field observations and suggests that our current estimation of either the growth parameters or the growing season length (based on empirical assessment or literature) needs to be further evaluated. 1. Introduction Calanus copepods play a key role in the Arctic pelagic ecosystem. When their biomass is high they can exert a significant impact on the primary production retaining much of the production in the pelagic food webs (refs). In contrast, most of the primary production is exported to the benthos when biomass is low (Grebmeier et al., 2006; Campbell et al., submitted). Due to their large body size and high lipid content, Calanus are an important high-quality food source for pelagic fish species such as capelin, herring, and pollack and they can also be an important part of the diet for larval and juvenile demersal fishes (e.g. cod) as well (refs). Hence, the Calanus species are critical components of the carbon cycle on Arctic shelves and basins and dictate to a large degree the extent of pelagic-benthic coupling and the composition of the pelagic ecosystem. Four congeneric species of Calanus are found in the Arctic Ocean and its marginal seas, two are expatriates in the Arctic (C. finmarchicus and C. marshallae) and two are endemic (C. glacialis and C. hyperboreus). For the endemic species, C. glacialis dominates on the shelves and slopes (refs) while C. hyperboreus is most important in the deeper basin regions (refs). Both these species can reproduce and grow in the extremely cold Arctic waters. The population centers of the expatriate species occur in more southerly waters. C. finmarchicus is advected into the Barents Sea from the Norwegian Sea and into the Arctic Basin through Fram Strait (Jaschnov, 1970; other refs )and C. marshallae, if it does enter the Arctic, passes through Bering Strait from the northern Bering Sea (Frost 1974, Springer et al. 1989, Plourde et al. 2005). Both these species are better adapted to warmer water conditions than those found in the Arctic Ocean proper. Biogeogrpahy of copepods, as of many other ectothermic animal species in the ocean, is strongly affected by temperature tolerance window they can adapt in order sustain reproduction success and other life functions. However, temperature is certainly not the only factor. Other environmental factors such as food availability can also be critical, as well as the life history traits of organisms including development and reproduction strategies under certain temperature and food conditions. For a Calanus population to complete a life cycle in the Arctic Ocean, it is necessary that individuals reach a life stage where diapause can be initiated for overwintering before the end of the growing season when the food concentration drops below certain threshold. Therefore, even if temperature and food do not affect the reproductive success, species like C. finmarchicus may still not able to colonize the Arctic Ocean, simply because development rate under low-temperature and low-food environment is too slow to reach C5 and store enough lipid for overwinter. These dynamics could be altered as the environmental conditions change in the Arctic and marginal seas. The Arctic is particularly susceptible to climate warming, seen most clearly in the recent seasonal ice retreat in the western Arctic (Serreze et al., 2003; Stroeve et al., 2005). It has been predicted that seasonal ice cover in the Arctic Basin could essentially disappear by ~2040 (Holland et al., 2006). Warming sea surface temperature and decreasing seasonal ice cover could enhance phytoplankton production and provide better growth conditions for C. finmarchicus, making it possible for C. finmarchicus to expand their range from the northern North Atlantic further into the Arctic and its marginal seas. Meanwhile, for the endemic species like C. glacialis, who may rely on ice algae to gain energy to start reproduction process in the beginning of growth season, will be negatively affected due to the loss of the sea ice and associated algae production. This change of Calanus species composition might cause a regime shift of ecosystem structure and function due to the trophic importance of Calanus populations in the Arctic Ocean. Although such a shift is not yet evident in the Arctic system (refs?), a similar shift has been observed in the North Sea, where fluctuations in the plankton, primarily a shift in the abundance of C. finmarchicus to its warmer water congener C. helgolandicus, has led, through bottom up control to long-term changes in Atlantic cod recruitment (Beaugrand 2003). It is essential to understand how the combinations of life history, physical advection, seasonality, and food environment limits the ranges of Calanus populations before we can assess how climate change might cause the shift of its biogeographic boundaries in the Arctic Ocean and marginal seas. This is a challenging question mainly due to the multiple processes involved in controlling the biogeography and limited observational data available for detailed analysis. In this paper, we present results from a biologicalphysical coupled model to explore factors that control Calanus population dynamics and biogeographic boundaries in the Arctic Ocean and marginal seas, and to provide base for further investigating the impacts of various climate warming scenarios on the biogeography. 2. Material and methods 2.1. Physical model The physical model used to drive biological model is an updated Arctic Ocean FiniteVolume Community Ocean Model (AO-FVCOM) (Chen et al., 2009). AO-FVCOM was developed based on the spherical coordinate, semi-implicit version of FVCOM with a full coupling of an Unstructured Grid version of the Los Alamos sea ice model Community Ice CodE (UG-CICE) (Gao et al., 2010; Hunke and Lipscomb, 2006). The computational domain covers the Pan-Arctic region shown in Fig 1. The non-overlapped triangular grid is used in the horizontal and a hybrid coordinate in the vertical. The horizontal resolution varies in a range of 10-50 km and the vertical resolution depends on water depth. The water column has a total of 45 layers. In regions deeper than 225 m, the s-coordinate is chosen, with ten and five uniform layers specified near the surface and bottom, respectively. The thickness of each layer is 5 m. In a shallower shelf region of 225, the -coordinate is used, which provides a vertical resolution of 5 m or less. These two coordinates merge at the 225-m isobath, where all layers have a uniform thickness of 5 m. In this study, AO-FVCOM was driven by 1) astronomic tidal forcing constructed eight tidal constituents (M2, S2, N2, K2, K1, P1, O1 and Q1), 2) the surface wind stress, 3) the net heat flux at the surface plus shortwave irradiance in the water column, 4) the air pressure gradient, 5) precipitation minus evaporation, and 6) river discharges. The meteorological forcing data represent climatologically averaged fields over 1978-1994 and the data are from the database (version 6) of the Arctic Ocean Modeling Intercomparison Project (AOMIP) derived from the ECMWF reanalysis ERA-15. The river discharge along the US and Canada coast was specified by the daily climatologic mean from USGS monitoring sites (on the website: http://www.usgs.gov and www.ec.gc.ca), while the data outside the US and Canada coast was provided by L. F. Smedstad at Navy Coastal Ocean Modeling (NCOM) Group. AO-FVCOM ran through nesting to the Global-FVCOM under the same forcing condition, which provides the surface elevation, currents, water temperature/salinity and mixing coefficients at the open boundaries. The Global-FVCOM was spin up for a 50-year run, while AO-FVCOM were initialized with the Global-FVCOM spin-up field and ran for 6-years with data assimilation of monthly climatologic temperature and salinity fields. The model-predicted fields reached the equilibrium state after 5 years, and the integration time selected in was long enough to conduct the climatologic field used for this study. The time step used to drive AO-FVCOM is 600 sec. The ice internal stress in UG-CICE was computed by 120 iterations with a time step of 5 sec. 2.2. Biological model An FVCOM-based i-state Configuration Model (FISCM) has been implemented to examine biological-physical factors affecting the biogeographic boundary of Calanus populations. The model has two modules, a Lagrangian tracking module and a generic stage-based biological module. The tracking module is driven by the current fields derived from the hourly-stored FVCOM output (so-called “offline” approach). The resulting locations of individual particles, along with the temperature field from FVCOM, provide input for the biological module. In the Lagrangian tracking module, the movement of each individual particles caused by advection (and possibly vertical migration) can be computed by solving the following equation with a classic 4th order 4-stage explicit Runge-Kutta method that has been implemented in FVCOM (Chen et al., 2006b; Ji et al., 2006; Huret et al., 2007), dx dt = v ( x (t ), t ) + vb , where x is the particle position at time t, and v is the velocity interpolated from the surrounding model grids provided by FVCOM. The biological behavior term v b can be derived from the literature/field measurements. In this study, we do not include any horizontal swimming behavior, and keep individual particles at certain depths (e.g. 0 m and 50 m below the surface for active individuals) vertically throughout model runs. In the biological module, the whole life cycle of the target zooplankton species is divided into multiple morphologically distinct stages including egg, nauplii, copepodite and adult stages. An individual copepod is represented as a vector in the model with information such as location (x,y,z), sex, age, stage, ovarian status and other population dynamic variables (referred as i-state by Metz and Diekmann, 1986). Each vector is updated at each time step according to development rate and reproductive functions derived from field measurements and lab experiments. The model starts with an initial population structure and distribution, then monitors the change of each individual by recording the i-state of individual j at any time t (Miller et al., 1998; Carlotti et al., 2000): X i , j (t ) = X i , j (t - dt ) + f ( x1, j (t - dt ),... xi , j (t - dt ),..., T , food ,...) , (1) where X i , j (t ) is the value of the i-state of individual j and f is the process modifying X i , j as a function of the values of different i-states of the organisms, and external parameters such as the temperature T and food concentration. Belehrádek’s (1935) temperature function is used to describe the development times under saturated food conditions as a function of temperature following Corkett et al. (1986). Development time (D) for any one given stage is given by D = a (T+α)β, (2) where a, α and β are constants and T is temperature. The value for α is assumed to be fixed for a given species (9.11 for Calanus), and β is taken to be -2.05 from Corkett et al. (1986), who found this to be the mean for 11 species of copepods. The value for “a” is fitted for each developmental stage. Here we take advantage of the intra-generic equiproportional rule for copepod development (Corkett 1984; Corkett et al., 1986; McLaren, 1986): namely, the proportion of time that an individual spends in a given stage relative to the entire development time is constant across genera and across different temperatures. This permits us to use stage-duration proportions determined in laboratory studies (e.g., Campbell et al., 2001) for a given species (C. finmarchicus) coupled with measured egg development times from the other Calanus species to derive stage specific development times at different temperatures for each species of interest. Parameterization for C. marshallae at Arctic temperatures was unsuccessful. The primary difficulties are that not only are the data limited, but also that the egg hatching experiments that have been reported have been carried out in temperatures greater than 9oC (Peterson 1986). Therefore, we only did a test with an assumption that C. marshallae has the same parameter as C. finmarchicus. The detailed “a” values for the Calanus species in this study are listed in Table 1. In addition to temperature, we also added food dependence to the development rate equation of C. finmarchicus and C marshallae for further simulations, using a similar approach by Speirs et al. (???), who fit the function to the observation data from Campbell et al. (2001). The development duration (D) becomes D = a (T+α)β[1-exp(-F/K)], (3) where F is food concentration (unit: g chl l-1), and K is a constant associated with the intensity of food limitation (0.8 g chl l-1 in this model). No experiments with food- dependent development were conducted for other two species (C. glacialis and C. hyperboreus), mainly because the lack of food concentration data (ice algae data is very sparse and can not be readily derived from satellite). Model runs without food limitation on development represent the best scenario for the populations. 2.3. Numerical experiments A series numerical experiments have been conducted to test our proposed hypotheses regarding the potential geographic range of Calanus populations and how temperatureand food-dependent development rates coupled with advection by the prevailing circulation might dictate its distribution. The daily temperature distribution was derived from FVCOM model output. Food concentration and the length of the growth season were derived for each location from two different sources. For locations that were only seasonally ice covered, a climatology of 8-day composites of SeaWiFS chlorophyll a data were used to estimate the dates that food was first and last available, as chlorophyll, at each grid point in the IBM modeled field (see Fig 2 top panel for the spatial distribution of the beginning of growth season). Food concentration for each day and location throughout the model run also was estimated from that climatology. Locations that are perennially ice covered, such as in the Arctic Basin, do not have SeaWiFS chlorophyll data available and hence the growing season was computed for each ice covered location using snow melt onset data (data source ???, see Fig 2 bottom panel for the spatial distribution of the beginning of growth season in ice covered area), along with AO-FVCOM model-computed sea surface shortwave radiation (to determine the end of growth season). For each location, then a specific growing season was computed, with those further to the north necessarily having shorter growth seasons. The beginning of the growing season was used to determine the onset of egg production at each of the starting model node points for Calanus finmarchicus, C. marshallae, and C. glacialis, under the assumption that food is required to fuel significant reproduction. For C. hyperboreus, the onset of the growth season coincides with the presence of nauplius stage 3, the first feeding stage, in surface waters, which were spawned at depth fueled by the lipid reserves of the adult females prior to the bloom. Once an individual is released, it is advected by the dominant circulation and develops from node to node according to the temperature at each point and, for some simulations, the food concentration. The model also permits us to simulate development and movement of individuals at different depths, each of which experiences different temperature conditions. Development is terminated when the individual reaches a location specific date where food is no longer available. Development is deemed to be successful when an individual reaches a life stage where diapause can be initiated (C5 for finmarchicus, C3 or C4 for hyperboreus, and C4 or C5 for glacialis). Most of our simulations to date have followed individuals only for one growing season to see the location of each individual who reaches diapause, with one additional simulation to track the locations of diapausing individuals. The logic behind this is that failure to reach diapause can be considered a failure to survive and reproduce. Successful colonization out of the established range would only occur when individuals can successfully reach diapause under the prevailing conditions and survive to reproduce the following year. 3. Results 3.1. Physical model results AO-FVCOM predicted climatological fields of the sea ice coverage and concentration, water temperature, salinity and currents have been validated by the comparison with the observational data from satellites, mooring, drifters/floats and climatologically averaged hydrographic database (Gao et al., 2010; Chen et al., 2010). The model captured the spatial distribution and seasonal variation of both sea ices and currents in the Arctic and its adjacent region. To focus our discussion on the FISCM results, we only include a brief description of physical model results here. Figures 3 shows examples of monthly mean water temperature and subtidal currents averaged in the upper 50 m in April (top panel) and September (bottom panel) in the Arctic region, In early spring, the surface circulation in the Arctic Basin is characterized as the anti-cyclonic Beaufort Gyre circulation and strong transpolar drifting currents. In summer, the anti-cyclonic Beaufort Gyre circulation shrinks and is much weaker and also the transpolar circulation shifts toward North American side. The coastal currents can reach 20-30 cm/s in September but is as low as 10 cm/s in April. The transports through Bering Strait for these two months are about 1.4 Sv and 0.6 Sv, respectively. It is close to the long term mooring observation 1.2Sv and 0.6 Sv in summer and winter (Woodgate et al, 2005). The net inflow into the Arctic Ocean through the Fram Strait is about 1.6 Sv, which counts for the outflow along the Greenland shelf. The model-predicted northward current is about 10-20 cm/s, which is close to the observation (Fahrbach et al., 2001). The temperature in this strait is as high as 6-8 °C in September and as low to 1-3°C in April. 3.2. Biological model results 3.2.1. Calanus finmarchicus and Calanus marshallae For C. finmarchicus, the focus is on the North Atlantic side of model domain. Under development rates that are temperature-dependent only and with the copepods at the surface, C. finmarchicus was able to successfully reach the diapausing stage of C5 only in the Barents, southern Kara Sea and southern GIN Seas (Fig 4, Atlantic side). No penetration of the species into the Arctic Basin was possible under the constraints of temperature-dependent development and the length of the growing season at each location. Note also that even the copepods that failed to reach diapause had not been advected into the Arctic Basin by the end of the growing season. If development rate is also food limited (in addition to development only occurring during the growing season), C. finmarchicus successfully reaches diapause only at locations considerably further south in the GIN and Barents Seas and in the Spitzbergen Current (Fig. 5). The C. finmarchicus individuals released and advected at 50 m, under temperaturedependent only development rate, were not able to reach diapause at locations as far north in the Barents Sea as those at the surface (not shown here). Under both food and temperature dependent development rate, successful diapause was achieved only in the southern portions of the GIN and Barents Seas and in the warm Atlantic water running north along the western side of Spitzbergen (Figure 6), Notice that some eastward advection of successfully developing copepods along the shelf-break to the north of Spitzbergen did occur. For C. marshallae, if the development is temperature-dependent only, surface individuals released in the Bering Strait and Chukchi-Beaufort shelf was able to successfully reach C5 (Fig 4, Pacific side). Under the development rate that depends on both temperature and food, very few surface individuals can reach C5 within the growth season in the Chukchi-Beaufort shelf region (Fig 5), and no individuals reached C5 if individuals stayed in 50 m below the surface (Fig 6). Notice that there is no overlap of successful individuals between the Pacific and Atlantic sides of the Arctic in all the cases. 3.2.2. Calanus glacialis and Calanus hyperboreus Neither C. glacialis and C. hyperboreus can reach even their earliest diapause stages in the central Arctic Basin (even without food limitation on development) when the growth season starts at the onset of snow melt (Fig. 7 and Fig 8). The distribution of C. glacialis C4 (youngest diapause stage) predicted by the model at the end of the growth season matches observed distributions fairly well (Fig. 7), with C. glacialis distributed along the edges of the Arctic Basin and in the marginal seas. Exceptions are the Chukchi Sea, where C. glacialis may not be present in any abundance following winter. The inability of C. hyperboreus to attain even the youngest diapause stage (C3) in the central basin (Fig. 8) is contradictory to the observation reported by ???, suggesting the growing season may be longer than used in the model, possibly due to earlier ice algal production or a heterotrophic food-web that extends the length of the growing season, or the vital rates used in the model is not correct. Alternatively, C. hyperboreus may exist as expatriates in the Central Arctic Basin as has been suggested by Olli et al. (2007). This will be further discussed in the Discussion section. 4. Discussion (Outline only for most subsection…need expansion) 4.1. Processes affecting biogeography of Calanus finmarchicus Calanus finmarchicus is believed to be an expatriate species in the Arctic Ocean and marginal seas, with the population being advected into the Barents Seas from the Norwegian Sea and the Arctic Basin through Fram Strait (Jaschnov, 1970; other refs). It can be found in the Eurasian Basin of the Arctic Ocean but cannot sustain a population there, and is not transported throughout the Arctic Ocean (refs). Compared to its two congeneric species of Calanus (C. glacialis and C. hyperboreus) that are endemic in the Arctic Ocean, C. finmarchicus is smaller, faster growing, and better adapted to warming water conditions. Low temperatures have been suggested as the major cause for the low growth and possibly the reproduction failure, and therefore the failure of C. finmarchicus to sustain itself in the Arctic Ocean and marginal seas (Jaschnov, 1970, Sameoto, 1984; Tande 1985; Hansen 1986). However, the question of whether temperature is the only factor remains to be further examined. For instance, Hirche (1990, 1997) found from lab experiments that females can continue to spawn at 0 C, and the egg production rate of female C. finmarchicus from -1.5 to 2 C is similar to that of C. glacialis. Data from field observations by Hirche and Kosobokova (2007) also showed the presence of spawning females despite low water temperature between +1 and -1 C. These lab and field work led to a hypothesis that the late availability of food in addition to low temperature in the Arctic Ocean limits reproductive success and hence the sustainability of the population (Hirche and Kosobokova, 2007). Our results seem to suggest that, even without considering the low temperatureand/or food- induced reproduction failure, the low development rate at low temperature alone can limit the expansion of C. finmarchicus to the north. If the additional fooddependent development is invoked in the model, the population is much more constrained in the southern part of the GIN Seas, Barents and Kara Seas. Based on the observation data (Kosobokova, 1998; Hirche and Kosobokova 2007; Kosobokova and Hirche; 2009), most of the individuals found along the margin of the Arctic Basin are in C5 and adult stages, especially in the north-east of the Svalbard (the Northern edge of Barents and Kara Seas, as well as Leptev Sea). The absence of young copepodids suggests that the population in those areas are either not reproducing or can not even reach copepodite stages after hatching (how about nauplii?, I didn’t see data from Kosobokova ???), and that the observed C5 and adult individual are mainly advected from Atlantic and southern part of Barents and Kara Seas (possibly as diapause individuals during the previous winter). To investigate whether advection would bring copepods that reached diapause into the Arctic Ocean during the overwintering period, we continued to advect the successfully diapausing copepods through the winter until July 1 of the following year. These individuals presumably would be able to continue development to the adult stage and reproduce. The simulation demonstrated that only a very few of the diapausing copepods were advected into the Arctic Ocean during the winter (Fig. 9). The C5 and adult individuals observed by Kosobokova and Hirche (2009) along the northern edge of Barents, Kara Sea, and Leptev seas, are probably the result of continuous advection of those individuals emerge from diapause over winter. Our results suggest that C. finmarchicus is unable to penetrate into the Arctic Ocean under present conditions of temperature, food availability, and length of the growing season because of a combination of factors. The copepods are unable to reach the diapausing stage under the conditions experienced in the northern portions of the GIN and Barents Seas. In addition, the prevailing circulation is not fast enough to advect the copepods into the Arctic Ocean during the growing season or even during the winter following, at the depths the copepods are believed to diapause. This scenario is not likely to change even if the water temperature increases by 2oC across the region (see more discussion in 4.4). 4.2. Intrusion of Calanus marshallae from Pacific to Arctic Difficulty in estimating vital rates. What is the observed distribution? If Cfin rates are applied, intrusion of cmar through Bering Sea to Beaufort shelf is possible, but not able to cross Canadian Archipelago region, and reach the Atlantic side. Argue against the Sundt & Melle (1998) on the observation of cmar in Atlantic side. 4.3. C. hyperboreous in the central basin? Endemic vs. expatriate in the central basin (discuss about observation) Case tested: start the growth season two weeks earlier (Fig. 10) Uncertainty in the vital rates estimation (->implication for future observation)? Difficulty in determining the beginning and end of growth season (-> implication for future observation). 4.4. Warming scenario Tested cfin/cmar and Chyp by increase water temperature by 2oC over climatoloty Notice that the warming scenario testing in the model is not dynamics, no changes in ice coverage and food change, only temperature change. Warming/early ice melting caused the timing mismatch: Søreide et al., (2010) talked about “Females of C. glacialis utilized the high-quality ice algal bloom to fuel early maturation and reproduction, whereas the resulting offspring had access to ample high-quality food during the phytoplankton bloom 2 months later. Reduction in sea ice thickness and coverage area will alter the current primary production regime due to earlier ice break-up and onset of the phytoplankton bloom. A potential mismatch between the two primary production peaks of high-quality food and the reproductive cycle of key Arctic grazers may have negative consequences for the entire lipiddriven Arctic marine ecosystem.” 5. Conclusion References Fahrbach, E., Meincke, J., Østerhus, S., Rohardt, G., Schauer, U.,Tverberg, V., and Verduin, J.(2001), Direct measurements of volume transports through Fram Strait, Polar Res., 20, 217–224, 2001. Woodgate, R. A., K. Aagaard, and T. J. Weingartner (2005), Monthly temperature, salinity, and transport variability of the Bering Strait through flow, Geophys. Res. Lett., 32, L04601, doi:10.1029/2004GL021880. Proshutinsky, A. Y., and M. A. Johnson (1997), Two circulation regimes of the winddriven Arctic Ocean, J. Geophys. Res., 102, 12,493–12,514, doi:10.1029/97JC00738. Table 1. Development-temperature function parameters.