Geir Huse

End-To-End Models for the Analysis of Marine Ecosystems: Challenges, Issues, and Next Steps

Abstract There is growing interest in models of marine ecosystems that deal with the effects of climate change through the higher trophic levels. Such end-to-end models combine physicochemical oceanographic descriptors and organisms ranging from microbes to higher-trophic-level (HTL) organisms, including humans, in a single modeling framework. The demand for such approaches arises from the need for quantitative tools for ecosystem-based management, particularly models that can deal with bottom-up and top-down controls that operate simultaneously and vary in time and space and that are capable of handling the multiple impacts expected under climate change. End-to-end models are now feasible because of improvements in the component submodels and the availability of sufficient computing power. We discuss nine issues related to the development of end-to-end models. These issues relate to formulation of the zooplankton submodel, melding of multiple temporal and spatial scales, acclimation and adaptation, behavioral movement, software and technology, model coupling, skill assessment, and interdisciplinary challenges. We urge restraint in using end-to-end models in a true forecasting mode until we know more about their performance. End-to-end models will challenge the available data and our ability to analyze and interpret complicated models that generate complex behavior. End-to-end modeling is in its early developmental stages and thus presents an opportunity to establish an open-access, community-based approach supported by a suite of true interdisciplinary efforts.

Artificial Evolution of Life History and Behavior

We present an individual‐based model that uses artificial evolution to predict fit behavior and life‐history traits on the basis of environmental data and organism physiology. Our main purpose is to investigate whether artificial evolution is a suitable tool for studying life history and behavior of real biological organisms. The evolutionary adaptation is founded on a genetic algorithm that searches for improved solutions to the traits under scrutiny. From the genetic algorithm’s “genetic code,” behavior is determined using an artificial neural network. The marine planktivorous fish Müller’s pearlside (Maurolicus muelleri) is used as the model organism because of the broad knowledge of its behavior and life history, by which the model’s performance is evaluated. The model adapts three traits: habitat choice, energy allocation, and spawning strategy. We present one simulation with, and one without, stochastic juvenile survival. Spawning pattern, longevity, and energy allocation are the life‐history traits most affected by stochastic juvenile survival. Predicted behavior is in good agreement with field observations and with previous modeling results, validating the usefulness of the presented model in particular and artificial evolution in ecological modeling in general. The advantages, possibilities, and limitations of this modeling approach are further discussed.

Implementing behaviour in individual-based models using neural networks and genetic algorithms

Even though individual-based models (IBMs) have become very popular in ecology during the last decade, there have been few attempts to implement behavioural aspects in IBMs. This is partly due to lack of appropriate techniques. Behavioural and life history aspects can be implemented in IBMs through adaptive models based on genetic algorithms and neural networks (individual-based-neural network-genetic algorithm, ING). To investigate the precision of the adaptation process, we present three cases where solutions can be found by optimisation. These cases include a state-dependent patch selection problem, a simple game between predators and prey, and a more complex vertical migration scenario for a planktivorous fish. In all cases, the optimal solution is calculated and compared with the solution achieved using ING. The results show that the ING method finds optimal or close to optimal solutions for the problems presented. In addition it has a wider range of potential application areas than conventional techniques in behavioural modelling. Especially the method is well suited for complex problems where other methods fail to provide answers.

Ecology in Mare Pentium: an individual-based spatio-temporal model for fish with adapted behaviour

Abstract A conceptual approach to study spatial movements of fish using an individual-based neural network genetic algorithm model is presented. Artificial neural networks, where the weights are adapted using a genetic algorithm, are applied to evolve individual movement behaviour in a spatially heterogeneous and seasonal environment. A 2D physical model (for the Barents Sea) creates monthly temperature fields, which again are used to calculate zooplankton production and predation pressure. Daily fish movement is controlled by reactive or predictive mechanisms. Reactive movement governs search for local optimal habitats, whereas predictive control enables adaptation to seasonal changes. Levels of growth and predation pressure at the time of decision are used to assess whether to apply reactive or predictive movement control. To make the model realistic on a large scale, each of the individuals are scaled up to represent a clone of one million siblings acting and growing synchronously. The fish lives for up to two years, and may reproduce in its second year. In order to spawn it has to be at the designated spawning area in the south-western part of the lattice in January. During spawning it produces a number of offspring in proportion to its body size. The “genetic constitution” of offspring (the weights of the synapses in the neural networks) is a mix of their “mothers” and a randomly picked member of the population. The model is able to solve the problem of navigating in a heterogeneous and seasonal environment. The movement of the artificial fish follows a seasonal pattern, typical for migrating pelagic fish stocks. During summer and autumn the distribution is widespread whereas during spring it is more concentrated. When trophic feedback is removed (i.e. zooplankton survival is independent of fish predation) the distribution of the fish is less dispersed which shows that the model allows for density dependent behaviour. Large-scale migration is an interplay between reactive and predictive movement control and when only one of these is allowed, the individuals are unable to adapt properly. Throughout most of its life the fish relies heavily on reactive movement, but during the spawning migration predictive movement control is applied almost exclusively.

Juvenile herring prey on Barents Sea capelin larvae

Abstract Due to a negative correlation between juvenile herring abundance and Barents Sea capelin recruitment, it has been hypothesised that predation by herring (Clupea harengus, L.) causes poor recruitment in the capelin stock (Mallotus villosus, Müller). In order to investigate whether herring are important predators of larval capelin, stomachs of herring were collected in the Barents Sea during the summers of 1992 and 1993. Capelin larvae were found in 5.6 and 3.0 % of the herring stomachs collected in the two years respectively. Herring showed preference for the larger capelin larvae. The observed predation rates correspond to estimated encounter rates between individual herring and larval capelin. The study confirms that predation takes place, but the observed rates are too low to explain the poor capelin recruitment in the two years. Increased swimming speed and selective feeding by herring in areas with high densities of capelin larvae is proposed as a mechanism that can explain the recruitment failure of the capelin stock.

Horizontal distribution and overlap of planktivorous fish stocks in the Norwegian Sea during summers 1995-2006

Abstract The Norwegian Sea harbours several large pelagic fish stocks, which use the area for feeding during the summer. The period 1995–2006 had some of the highest biomass of pelagic fish feeding in the Norwegian Sea on record. Here we address the horizontal distribution and overlap between herring, blue whiting and mackerel in this period during the summers using a combination of acoustic, trawl and LIDAR data. A newly developed temperature atlas for the Norwegian Sea is used to present the horizontal fish distributions in relation to temperature. The centre of gravity of the herring distribution changed markedly several times during the investigated period. Blue whiting feeding habitat expanded in a northwestern direction until 2003, corresponding with an increase in abundance. Strong year classes of mackerel in 2001 and 2002 and increasing temperatures throughout the period resulted in an increased amount of mackerel in the Norwegian Sea. Mackerel was generally found in waters warmer than 8°C, while herring and blue whiting were mainly found in water masses between 2 and 8°C. The horizontal overlap between herring and mackerel was low, while blue whiting had a large horizontal overlap with both herring and mackerel. The changes in horizontal distribution and overlap between the species are explained by increasing stock sizes, increasing water temperature and spatially changing zooplankton densities in the Norwegian Sea.

A comparative study of the feeding habits of herring (clupea harengus, clupeidae, 1.) and capelin (mallotus villosus, osmeridae, müller) in the barents sea

Abstract Capelin (Mallotus villosus) and adolescent Norwegian spring spawning herring (Clupea harengus) cooccur in the southern Barents Sea during early summer. The diets of both species were dominated by calanoid copepods, and the overlap in diet was large. Both for herring and capelin the proportion of copepods in the diet decreased with increasing fish size, while euphausiids and appendicularians increased in importance. The ontogenetic shift in diet was thus similar for the two species. Herring showed an increasingly deeper distribution with increasing body size. Capelin were found deeper than herring in areas of spatial overlap. Based on the similarities in diet it is concluded that the two species are potential competitors for food in times of high abundance ofplanktivores or oflow food availability in the Barents Sea.

Modelling secondary production in the Norwegian Sea with a fully coupled physical/primary production/individual-based Calanus finmarchicus model system

Abstract The copepod Calanus finmarchicus is the dominant species of the meso-zooplankton in the Norwegian Sea, and constitutes an important link between the phytoplankton and the higher trophic levels in the Norwegian Sea food chain. An individual-based model for C. finmarchicus, based on super-individuals and evolving traits for behaviour, stages, etc., is two-way coupled to the NORWegian ECOlogical Model system (NORWECOM). One year of modelled C. finmarchicus spatial distribution, production and biomass are found to represent observations reasonably well. High C. finmarchicus abundance is found along the Norwegian shelf-break in the early summer, while the overwintering population is found along the slope and in the deeper Norwegian Sea basins. The timing of the spring bloom is generally later than in the observations. Annual Norwegian Sea production is found to be 29 million tonnes of carbon and a production to biomass (P/B) ratio of 4.3 emerges. Sensitivity tests show that the modelling system is robust to initial values of behavioural traits and with regards to the number of super-individuals simulated given that this is above about 50,000 individuals. Experiments with the model system indicate that it provides a valuable tool for studies of ecosystem responses to causative forces such as prey density or overwintering population size. For example, introducing C. finmarchicus food limitations reduces the stock dramatically, but on the other hand, a reduced stock may rebuild in one year under normal conditions.

Effects of interactions between fish populations on ecosystem dynamics in the Norwegian Sea – results of the INFERNO project

The Norwegian Sea (NS) is the feeding ground for some of the largest fish stocks in the world, including Norwegian spring spawning (NSS) herring (Clupea harengus Linnaeus, 1758; Figure 1), blue whiting (Micromesistius poutassou Risso, 1827) and the Northeast Atlantic (NA) mackerel (Scomber scombrus Linnaeus, 1758). These planktivorous stocks have substantial spatial and dietary overlap (e.g. Nøttestad et al. 1997; Dalpadado et al. 2000; Kaartvedt 2000), and are often collectively referred to as the ‘pelagic complex’ in the Norwegian Sea. Due to their high abundances, they can potentially have a strong ecological impact on the ecosystem and each other (Skjoldal et al. 2004a). The NSS herring collapsed in the late 1960’s and rebuilt during the 1980’s (Dragesund et al. 1997). Following the herring collapse, high abundances of blue whiting were discovered in the Norwegian Sea (Misund et al. 1998), and it has been speculated that the blue whiting population increased concurrently with the collapse of the NSS herring (Skjoldal et al. 1993), but the evidence remains inconclusive (Daan 1980). Since the late 1980s the abundance of fish in the NS has increased steadily and this has increased the potential for interactions between the planktivorous stocks (Figure 2). This was the background for seeking funding for the INFERNO project ‘Effects of interactions between fish populations on ecosystem dynamics and fish recruitment in the Norwegian Sea’ submitted to the Research Council of Norway (RCN) in 2005. The main hypothesis to be addressed in the INFERNO project was that the planktivorous fish populations feeding in the NS have interactions that negatively affect individual growth, mediated through depletion of their common zooplankton resource. The project was funded and lasted for the period 2006 2009 and nine papers from the INFERNO project and associated research are presented in this thematic issue of Marine Biology Research. Many of the principal investigators of the project worked at the Institute of Marine Research (IMR), but the project also benefitted strongly from interactions and exchange of data and ideas with scientists from Russia (Alexander Krysov and Vladimir Zabavnikov), the Faeroe Islands (Jan Arge Jacobsen) and Iceland (Torstein Sigurdsson and Gudmundur Óskarsson). The international partners have participated actively in the project through project meetings and as co-authors of papers. During the project period the trend of a decreasing zooplankton biomass in the NS continued and the biomass now remains low (Figure 2). The fish biomass peaked in 2004 and has since decreased somewhat, but remains fairly high. The abundance of blue whiting increased until 2004, and the range of the horizontal distribution expanded in a northwesterly direction during this period. Strong year classes of mackerel from 2001 and 2002, together with increasing temperatures, resulted in an increased number of mackerel in the Norwegian Sea (Payne et al. 2012; Utne et al. 2012a). Furthermore, there were rather substantial changes in the migration pattern of herring during the study period and thus high interannual variability in horizontal overlap between the species. There was a relatively high spatial overlap between the species during the 1990s, with a southern centre of gravity (for all three species), but due to the northern displacement of Figure 1. Herring (Clupea harengus) represents an important component of the pelagic complex of the Norwegian Sea. Photographer: David Shale (www.deepseaimages.co.uk). Marine Biology Research, 2012; 8: 415 419

Modelling habitat choice in fish using adapted random walk

Abstract A new concept for modelling habitat choice, called an adapted random walk, is presented. The concept is based on the evolution of threshold values for departures and destinations using a genetic algorithm. A habitat is departed if the fitness value associated with it is below the evolved threshold value. Movement is determined probabilistically using random numbers and adapted threshold values. Different versions of this concept were tested for the ability to model horizontal distribution of the Barents Sea capelin using two different fitness criteria. When adapting both departure and destination thresholds, the model performed better than a neural network model with the same number of adapted variables. The evolved threshold values can be used interchangeably with values estimated in the field, for example preferred temperature ranges or prey abundance. The adapted random walk concept functions intuitively, and can be useful for applied purposes, such as extending models of fish distribution beyond the advected stages and studying the effects of climate change on the spatial distribution of fish.