ICES CM 2010/L:17

Modeling populations in the Gulf of Maine: building prediction capability through a process-oriented approach

Rubao Ji, Cabell S. Davis, Christoph Stegert

Abstract

Zooplankton are sensitive to climate change and may amplify subtle environmental signals due to their non-linear responses to environment forcing. It is critically important to understand the underlying biological-physical mechanisms that control the variability of populations, such that the predictions on how the climate change might affect plankton and higher trophic levels become possible. In this study, we use a coupled hydrodynamics/food-web/population-dynamics model to identify the key processes controlling the observed seasonality and distributional patterns of key copepod populations in the Gulf of Maine region including Pseudocalanus spp., Centropages typicus, and finmarchicus. The importance of life-history traits (e.g. development, reproduction, mortality, diel and seasonal migration) vs. physical processes (e.g. advection and diffusion) are examined. The implication of this modeling study on data-need, observing system design, and climate change scenario testing will also be discussed.

Keywords: plankton, , modeling, biological-physical interaction, life-history traits

Contact author: Rubao Ji, Biology Department, Woods Hole Oceanographic Institution, MS #33, Redfield 2-14, Woods Hole, MA 02543, USA. Email: [email protected] Introduction

Copepods have the highest abundance among the meso-zooplankton in the northern and play an important role in marine food webs (e.g. Davis, 1987; GLOBEC, 1992). Copepods are also sensitive to environmental changes and therefore can be considered as important climate indicators (e.g. Richardson, 2008). Increasing efforts have been made to build the prediction capability of coupled biological-physical models, so the influence of climate change on plankton phenology and biogeographic boundaries can be quantitatively assessed (Ji et al., 2010).

Of the several copepod species found in the Gulf of Maine (GoM), the relatively large- sized Calanus finmarchicus and the smaller Centropages typicus and Pseudocalanus sp. (the congener P. newmani and P. moultoni) are among the most dominant species (Kane, 2007). All three copepods show abundance peaks at different times of the year, though the temperature and food (considering the bulk property as -a) conditions are the same to all species (Fig. 1). Further, the spatial distributions of these species differ from each other (see Ji et al., 2009), suggesting that their spatio-temporal distributional patterns are probably related to their unique life-history trait. It is therefore expected that the response of different copepod species to climate change is likely to be different.

Figure 1: Climatological seasonal cycle of temperature, chlorophyll (satellite data, top panel) and three copepod populations (MARMAP data, lower panel) in the Gulf of Maine. With the predicted warming of the oceans the adaptation of species is challenged. Pseudocalanus is a good example for adaptation to a colder environment; its development time does not increase as much as warm water species such as Centropages typicus when they are exposed to cold temperature. Meanwhile, at a warmer temperature the development of C. typicus becomes faster than Pseudocalanus. The exponential decrease in the Temperature-Development relation also indicates that a change in temperature has a stronger impact in colder regions. For the GoM an increase of ca. 2- 4°C is predicted from IPCC simulations within this century (IPCC, 2007). This temperature change is likely to cause a shift in the species distribution as conditions become more favorable for C. typicus (Stegert et al., 2010). The fidelity of this type of empirical predication depends on how well we understand the importance of different aspects of life-history traits on population dynamics, which can be examined in a biological-physical coupled population model used in this study.

In this paper, we are interested in understanding the spatio-temporal distribution of the three target species Calanus finmarchicus, Pseudocalanus sp. and Centropages typicus. We hypothesize that the observed different seasonality is due to different life-history strategies manifested as the interplay of physiological processes and interactions (Fig. 2), including their development (from eggs to adults), reproduction and their loss by mortality. These processes are affected by environmental conditions, which can usually be expressed as empirical regressions obtained from in situ or lab measurements (e.g. the Belehradek function for temperature dependent development rate). The key questions are:

• Can we reproduce the observed spatial and temporal patterns if we use life-history parameter values from experiments for these processes (found in literatures) and use environmental conditions from a validated physical-biological model?

• What aspects of life-history strategies are critical for the formation of the observed spatio-temporal patterns?

Figure 2: Conceptual diagram of processes forming the distribution of copepods based on their specific life-history strategies and the environment.

In the following we examine the controls of population growth by different life strategies for the above-mentioned three target species, covering the following topics:

T1) Temperature-dependent development for cold- vs. warm- water species;

T2) The role of predation and cannibalism;

T3) The importance of bottom-up effect through food-dependent development and reproduction;

T4) The impact of diel and seasonal migration on seasonality and distribution; and

T5) The population persistence in an advective system. Methods

We used a process-oriented approach for this biological-physical coupled modeling study. It involves a simple population dynamics model that simulates key life-history processes including: development (stage duration D) as a function of temperature (T), reproduction R as a function of temperature (T) and food (F), and mortality M as a function of temperature. R and M determine the population’s increase and decrease, respectively. A 0-D (no spatial dimension) version of the population model that consists of four development stages, eggs (E), nauplii (N), copepodites (C) and adults (A), with mean-age calculation to minimize numerical diffusion of stages, was developed by Hu et al. (2008). The details for the 3-D coupled model implementation of the mean-age population model can be found in Ji et al. (2009).

The population model was embedded into the three-dimensional FVCOM model (Chen et al., 2003) and a --Zooplankton-Detritus (NPZD) module (Ji et al., 2008). We investigate the seasonal cycle and test the effect of different life-history strategies and their importance in shaping the species spatio-temporal distribution. The tests included several simulation runs (‘RunX-Y’) to address the mentioned topics [T1-5], (where the X-value indicates the species, 1: Pseudocalanus, 2: C. typicus and 3: C. finmarchicus and the Y-value indicates runs for each species):

• Run*-0: baseline case [T1]: * means for all three species.

• Run2-1: mortality test 2 [T2]: including a new Q10 value for temperature dependent mortality, for C. typicus.

• Run2-2: mortality test 3 [T2]: include cannibalism, for C. typicus.

• Run2-3: feeding test [T3,4]: adding microzooplankton as food source instead of phytoplankton only, for C. typicus.

• Run2-4: migration test 1 [T5]: individuals migrate to the layer with maximum chlorophyll concentration, for C. typicus.

• Run3-1: migration test 2 [T5]: include diapause behavior, individuals migrate to depth at certain temperature and food conditions, for C. finmarchicus. • Run3-2: advection test [T6]: adding an upstream influx from the Scotian Shelf, for C. finmarchicus.

Pseudocalanus

Centropages typicus Centropages

Calanus finmarchicus Calanus

Figure 3: Bi-monthly averaged distribution for May/June and Sept/Oct from observations (left) and baseline simulation Run*-0 (center) and simulated time-series at two positions (GB crest and central Jordan Basin) compared to observations (right). Results

At the baseline case (Run*-0) for all three species, the population dynamics of Pseudocalanus are similar to the observed seasonal cycle (Fig. 4): Following the spring bloom the population increased until summer. The decrease was due to increased mortality (simulations with constant mortality did not show this decrease). In contrast, the simulated annual cycles of C. typicus and C. finmarchicus at baseline runs differ significantly from observations: modeled abundances of C. typicus peak in June similar to Pseudocalanus instead of fall, while the population of C. finmarchicus goes extinct. The question is which life-history traits of these two species are missing in the model in order to get the observed distributions?

The decline of Pseudocalanus in summer is probably due to the combined factors of lower food concentration and increased mortality (correlated to temperature in the model). These factors seem not to work in C. typicus. For instance, lowering the temperature dependence for mortality (Run2-1, Q10 = 1.9) resulted in a shift of abundance peak from spring to fall, but the population on Georges Bank increases much more than observed, while the population in the deeper basins became significantly lower, resulting in a unrealistically strong spatial gradient from GB to the central GoM.

Figure 4: Simulations for C. typicus: Effect of change in temperature dependent mortality (blue line), cannibalism (green line) and omnivorous feeding (red line) on the abundance. We tested some additional life-history traits for C. typicus that are not configured in the baseline run. First, it is known that the food concentration mainly influences the reproduction rate but also has influence on development (Davis and Alatalo, 1992). Second, like some other copepods, C. typicus are reported to feed on their own or other species’ eggs (Calbet et al., 2007). Including these two aspects of life-history traits in the model reduce the peak abundance that matches the observed level on the well mixed portion of Georges Bank (Run2-2, Fig. 4, left panel), where the population size is large and the egg cannibalism could increase egg mortality for species that freely release their eggs (broadcasting strategy as in Centropages, Calanus or Temora) compared to those who carry their eggs (e.g. Pseudocalanus, Oithona) (Ohman et al., 2002). At a low-food environment (such as in deep basins), the population size is small enough that cannibalism is probably not an issue, instead the reproduction is limited by food availability. Choosing from several sources, including e.g. microzooplankton (Run2-3), allows the increase of population size in low-food deep-water basins (e.g. the Jordan Basin case, Fig. 4, right panel).

Figure 5: Simulated seasonal cycle of C, typicus adults at Wilkinson Basin: daily values (left) and May-August mean abundance (right) indicated by the red box from simulation without (upper panel) and with (lower panel) vertical migration. Most copepod species are capable of migrating into more favorite layers to search for food (e.g. Alcaraz et al., 2007). Fig. 5 (from Run2-4) shows, that in summer C. typicus population grows better in the subsurface chlorophyll maximum layer. When actively “searching” for food (i.e. migrating towards the chlorophyll maximum), the mean depth integrated abundance in April-August increased from 347 ind. m-2 to 410 ind. m-2 (18%), while the peak at 20 m depth of 10 ind. m-3 is much higher than the surface maximum of 6 ind. m-3 without vertical migration. This strategy adds to the survival in the deeper regions, though the tests showed, that migration is still less important than the extended food source. Including the food dependence of life-history traits appears to improve the model simulation of copepod distribution patterns, especially in the deep basins (Fig. 6), while the mortality remains to be the main factor in controlling both seasonality and spatial gradient.

Figure 6: Bi-monthly depth averaged distribution of Centropages typicus from simulation

Run2-4. The color scale is log10 (ind. m-3).

Figure 7: Bi-monthly distribution of C. finmarchicus adults in the Gulf of Maine. Left: simulation without upstream influx (Run3-1), right: simulation with upstream influx

(Run3-2). The color scale is log10 (ind. m-3).

Mortality also seems to be an important factor causing Calanus finmarchicus to extinct at the baseline run. Its life cycle includes the specific strategy of migrating to deeper layers at late copepodite stage (C5) and remaining at depth with reduced metabolism (diapause) during summer and fall, shaping its seasonal cycle. Several studies have been dealing with the diapause dynamics (e.g. Hirche, 1996; Tittensor et al., 2003; Saumweber and Durbin, 2006). Our modeling tests with different formulations for the onset of diapause (using different food and temperature thresholds) showed a distinct sensitivity for the population dynamics and affect the population sustainability in the GoM.

It is important to notice that the diapause may not be the only important factor for the population sustainability: even after including diapause strategy, the abundance in some region, especially on the Scotian Shelf remained much lower than observed if a zero-flux boundary condition is applied at the upstream boundary, suggesting the importance of the population contribution from the upstream region through advection input. This is shown by two simulations that have the same population dynamics for Calanus finmarchicus, with one having no influx at the boundary (Fig. 7, left) and the other including an upstream influx using Dirichlet condition (specified with observed abundance at the upstream boundary on the Scotian Shelf (Fig. 7, right). Not only the abundances on the Scotian Shelf are increased but also the summer and fall abundance in the Gulf itself. This suggests that the Calanus population is probably not self-sustainable in the advective Scotian Shelf and GoM system.

Discussion

Numerical population models are subject to simplifications. It is critical for a model to capture the key processes but neglect (or simplify) the others. This study shows, that life- history traits are important in shaping the spatio-temporal distribution. For different species we have to know which of these processes are relevant and have to be considered in the model before the model can be used for predicting the impact of climate change. Through this process-oriented modeling study, the importance of the following aspects is highlighted:

1) The mortality was found to be a major contributor in shaping the seasonality and spatial distribution of the species. While a correlation to temperature is used as a necessary simplification for the modeling purpose, a better quantification of time- and space- specific mortality is needed.

2) Food concentration has an important influence on reproduction, especially for broadcasting species (which typically have general higher fecundity than egg-carrying copepods). The selectivity and analysis of food quality, however, has not been included in most of population models (including ours). This is also true for the inclusion of biological models in global models designed for prediction of future ecosystem scenarios, especially when copepod populations need to be included.

3) The influence of advection on the local abundance varies among species. C. finmarchicus is known to build larger overwintering stocks in the deeper slope water as well as in the Labrador Sea from where they are advected to the Scotian Shelf. Both sources influence the dynamics in the Gulf of Maine, while current tests indicate, that the upstream population has a higher influence on the deep basins in the Gulf. Summary

We have shown the importance to capture all processes relevant for a species’ life-history traits besides the generic life cycle of development, production and mortality. The uncertainty of current biological models suggests that the prediction of inter-annual variability due to climate change needs caution. The better we understand the influence of life-history traits on spatio-temporal distribution, the better we can use the model to assess the climate influence.

References

Alcaraz, M., Saiz, E., Calbet, A. 2007. Centropages behaviour: swimming and vertical migration. Progress in Oceanography, 72: 121-126.

Calbet, A., Carlotti, F., Gaudy, R. 2007. The feeding ecology of the copepod Centropages typicus (Krøyer). Progress in Oceanography, 72: 137-150.

Chen, C.S., Liu, H.D., and Beardsley, R.C. 2003. An unstructured grid, finite-volume, three-dimensional, primitive equations ocean model: application to coastal ocean and estuaries. Journal of Atmospheric and Oceanic Technology, 20:159–186.

Davis, C.S. and Alatalo, P., 1992. Effects of constant and intermittent food-supply on life-history parameters in a marine copepod. Limnology and Oceanography, 37: 1618-1639.

Davis, C.S., 1987. Zooplankton life cycles. In: Georges Bank, pp. 256–267. Ed. by R.H. Backus and D.W. Bourne. MIT Press, Cambridge, MA.

GLOBEC, 1992. Northwest Atlantic Implementation Plan. U.S. Global Ocean Ecosystem Dynamics Report, vol. 6.

Hirche, H.-J., 1996. Diapause in the marine copepod, Calanus finmarchicus – a review. Ophelia, 44: 129–143.

Hu, Q., Davis, C.S., and Petrik, C.M., 2008. A simplified age-stage model for copepod population dynamics. Marine Ecology-Progress Series, 360:179–187. IPCC, 2007. Climate Change 2007 – Impacts, Adaptation and Vulnerability. In Contribution of Working Group II to the Fourth Assessment Report of the IPCC, p. 976, Ed. by M.L. Parry, O.F. Canziani, J.P. Palutikof, P.J. van der Linden, and C.E. Hanson. Cambridge University Press, Cambridge, UK.

Ji, R., Davis, C.S., Chen, C., and Beardsley, R. 2008. Influence of local and external processes on the annual nitrogen cycle and primary productivity on Georges Bank: a 3-D biological-physical modeling study. Journal of Marine Systems, 73:31–47.

Ji, R.B., Davis, C.S., Chen, C.S., and Beardsley, R.C. 2009. Life history traits and spatiotemporal distributional patterns of copepod populations in the Gulf of Maine-Georges Bank region. Marine Ecology-Progress Series, 384: 187-205.

Ji, R., M. Edwards, D.L. Mackas, J. Runge, A.C. Thomas, 2010. Marine plankton phenology and life history in a changing climate: Current research and future directions. Journal of Plankton Research – Horizon, doi:10.1093/plankt/fbq062.

Kane, J. 2007. Zooplankton abundance trends on Georges Bank, 1977–2004. ICES Journal of Marine Science, 64:909–919.

Kiørboe, T., and Sabatini, M. 1994. Reproductive and life cycle strategies in egg-carrying cyclopoid and free-spawning calanoid copepods. Journal of Plankton Research, 16: 1353-1366.

Ohman, M.D., Runge, J.A., Durbin, E.G., Field, D.B., and Niehoff, B., 2002. On birth and death in the sea. Hydrobiologia, 480: 55-68.

Richardson, A.J. 2008. In hot water: zooplankton and climate change. ICES Journal of Marine Science, 65: 279-295.

Saumweber, W.J., and Durbin, E.G. 2006. Estimating potential diapause duration in Calanus finmarchicus. Deep-Sea Research II, 53: 2597-2617.

Stegert, C., Ji, R. and Davis, C.S., 2010. Influence of projected ocean warming on population growth potential in two North Atlantic copepod species. Progress in Oceanography, accepted. Tittensor, D.P., Deyoung, B., Tang, C.L., 2003. Modeling the distribution, sustainability, and diapause emergence timing of the copepod Calanus finmarchicus in the Labrador Sea. Fisheries Oceanography, 12 (4–5): 299–316.