Modeling early life history of cod
This page originated as the synthesis of my PhD thesis. In Norway it is common to write a synthesis for your dissertation that puts your findings into an international perspective within your field of research. The following pages refers to the full thesis which is available here (PDF version of Phd thesis). For more details on individual-based modeling you can also go to the “modeling ecosystems” page.
”Nogen løsning av spørsmaalet om fiskeriernes vekslinger engang for alle – ved en eller anden formel – lar seg ikke oppnaa, og enhver paastand om at ha naadd en slik løsning, vil ved nærmere prøvelse vise sig at tilhøre fantasiens verden.”
”Any solution to the variability in the fisheries once and for all – by some formula – is not achievable, and any assertion that a solution has been found, will by closer inspection, prove to belong in a world of phantasy.” (free translation)
For hundreds of years, fishermen have noted the large variability in the annual abundance of fish. Scientists have tried to explain why species of fish may explode in numbers one year, and be close to depleted the next year. So far, no single explanation has been found that solves the enigma of the variance in recruitment, however the problem is very much alive in scientific environments.
The ability to predict recruitment can have major implications for fisheries management and exploitation of commercial species, such as Atlantic cod (Gadus morhua). Spawning stock biomass was early on regarded as an indicator for the year-class strength, however with varying degree of success (Bradford & Cabana 1997). The spawner-recruit relationship assumes a correlation between recruitment biomass and the spawning stock biomass (SSB) or the biomass of eggs (E) (Marshall et al. 1998, Jennings et al. 2001). MacKenzie et al. (2003) examined the relationship between several decades of empirical data of SSB and recruitment of 20 different North Atlantic cod stocks. The authors revealed large inter-annual variability of SSB and recruitment that occurs within stocks and among stocks of different sizes. The variability that makes prediction difficult is due to mechanisms that act at the pre (i) and post-spawn (ii) stages:
i) Spawning stock biomass is not necessarily proportional to egg production since individual fecundity of females is proportional to length and condition (Marshall et al. 1998). The quality and number of eggs spawned increases with female condition and age (Kjesbu et al. 1992, Kjesbu et al. 1998), hence a population of mainly first-time spawners have less recruitment potential compared to a wide age and size structured population (Marshall et al. 1998). Spawner-recruit models try to reduce a multi-dimensional problem to 2 dimensions (SSB and recruitment), and tend to ignore variability in pre-recruit mortality. Consequently, the spawner-recruit approach fails to explain the large inter-annul variation that is caused by biological and environmental mechanisms and their interaction (Ulltang 1996).
ii) Egg quality (Marteinsdottir & Steinarsson 1998) and numbers (Marshall et al. 1998) combined with density-independent abiotic processes (e.g. Pepin et al. 1997) has consequences for larval hatching success and larval condition (Marteinsdottir & Steinarsson 1998). In addition, the environmental and biological conditions of the larval habitat may be of crucial importance for larval survival through the early life stages. In fact, the variety of processes that affect larval survival probability from the egg to the early juvenile stage is so important that the number of survivors is believed to have prognostic value for recruitment (Sundby et al. 1989). Predicting recruitment therefore requires an understanding of the early life stages of fish. What identifies survivors and what life history they have, may differ between ecosystems, between years, and between fish species.
- Survival through the early life stages
Studies of the early life history of fish focus on ontogenetic changes and the interaction between the individual and the environment (Hunter 1981, Chambers & Trippel 1997). Many important biotic and abiotic mechanisms (e.g. temperature (Otterlei et al. 1999, Ottersen & Loeng 2000), light (Blaxter 1986), turbulence (e.g. Sundby et al. 1994, MacKenzie & Kiørboe 1995), prey abundance (e.g. Hjort 1914a, Cushing 1990), predation (e.g. Øiestad 1985, Bailey & Houde 1989) of importance for survival and growth of larval and juvenile fish have been identified (ICES 1994, 2005). Still, only limited knowledge of their interaction exists (Nakken 1994, Ulltang 1996). For nearly 100 years, the early life stages of fish have received much attention from fisheries biologist that have explored possible correlations between early life stage abundance and recruitment strength and variability (Bailey & Houde 1989, Leggett & Deblois 1994, Nakken 1994). A range of hypotheses exist on explaining sources of mortality during early life stages and possible variability in survival to recruitment, and most are not mutually exclusive. Starvation was proposed by Hjort (1914a) to be the main source of mortality for first-feeding larval fish. Cushing (1990) (and references therein) extended the ideas of Hjort in his match-mismatch hypothesis, and proposed starvation as a significant mortality source for the entire pelagic phase of larval and juvenile fish. In addition to the hypothesis of Hjort (1914a) and Cushing (1990, 1996), several other theories exists that describe sources of mortality through the early life stages of fish. Physical conditions that retain early life fish in nursery habitats was highlighted by Iles and Sinclair (1982) in their member-vagrant hypothesis as a key element for survival. Physical properties of varying scale also modify the properties of a habitat such as aggregation of prey patches during calm wind events (Lasker 1975). Patches of prey increases the feeding incident while simultaneously decreases the activity level of larval and juvenile fish needed to obtain food. Reduced activity level is also believed to reduce larval conspicuousness to predators (Munk 1995, Skajaa et al. 2003). Encounter rate between predator and prey (Rothschild & Osborn 1988, MacKenzie & Kiørboe 1995) has also been recognized to increase in a dome-shape fashion (MacKenzie & Kiørboe 2000) with small-scale turbulence both in the laboratory (Kiørboe & MacKenzie 1995, MacKenzie & Kiørboe 1995) and in natural environments (Sundby & Fossum 1990, Sundby et al. 1994, Sundby 1995).
Despite good feeding conditions through the early life stages, survival is modified by mortality rates from predation (Øiestad 1985). Predation pressure varies between ecological systems both temporally and spatially (Garrison et al. 2000), and decreases with size (Folkvord & Hunter 1986, Miller et al. 1988, Bailey & Houde 1989, Leggett & Deblois 1994, Houde 1997). Thus, the probability of survival through ontogenetic stages is closely related to growth rate (Hare & Cowen 1997), size (“bigger is better hypothesis”, e.g. Leggett & Deblois 1994), and consequently the duration of ontogenetic stages (Houde 1987). When grouped this is usually termed the “growth-mortality hypothesis” (Hare & Cowen 1997).
The mechanisms that regulate recruitment of fish are many and can vary within seasons, and between years among ecosystems and among their inhabitants. By incorporating currently identified mechanisms important to early life history of fish in individual-based models (IBMs), forced by realistic environmental conditions, we are able to study their relative importance and their interaction. Individual-based models relate the individuals to the environment so that properties and dynamics on population level emerge (Grimm & Railsback 2005). This thesis focuses on modeling the early life history of cod (Gadus morhua) during the transition from newly hatched larva to early juvenile with the intention of increasing our understanding on how processes at various scales may operate and differentiate larval cod growth and survival.
- Fish species and stocks
Cod is a highly commercial fish species that lives in discrete stock units along the European coast from the Bay of Biscay to the Barents Sea, around Iceland, on the east and west coast of Greenland, and from Cape Hatteras to Ungava Bay along the east coast of USA (ICES 2005). This thesis focuses on two cod populations: the Northeast Arctic (or the Arcto-Norwegian) and the Georges Bank cod stocks. The Northeast Arctic cod (NA) is distributed throughout the Barents Sea. Mature individuals migrate to the Norwegian coastal areas as far down as 62ºN to spawn (Fig. 1 Paper 5). The majority of eggs are found in the Lofoten areas (~68 ºN) in the period from late February to early May (Pedersen 1984, Ellertsen et al. 1989). Eggs and larvae are then transported northwards with the ocean currents, away from the spawning areas and into the nursery grounds in the Barents Sea. Pelagic juveniles settle to the bottom after 5-6 months (40-70-mm) (ICES 2005).
The Georges Bank cod stock is primarily resident (ICES 2005). Spawning takes place at the Northeast Peak on Georges Bank (Fig. 1 Paper 3) during November to May, with peak spawning during February and March (ICES 2005). Eggs and larvae drift southwesterly with the currents towards the Great South Channel, before they enter the nursery grounds on Georges Bank sometime during summer (Lough & Bolz 1989). Bottom settlement occurs at the size of 40-mm (Lough & Potter 1993).
Modeling early life history of cod from the Georges Bank and Barents Sea dominates the main objectives of this thesis:
- Develop a mechanistic individual-based model for the early life history of cod based on first-principle physics and biology. Use empirical biological and environmental data to validate and force the model. Can the model reproduce observable patterns in real systems? How does larval behavior affect feeding, growth, survival, and interaction with the environment?
- Integrate the IBM with a general circulation model covering the habitat for the Northeast Arctic cod. How does the coupled model system predict larval growth under various environmental settings through the spawning season? What is the effect of behavior on survival and drift of larval cod?
Computer models provide researchers with a convenient tool for exploring and understanding the processes of an ecosystem. Or as stated by Peck (2004); “When a researcher builds a simulation model, they have created a world in which they have access to all of the laws and components of that world, and the relationships among those components.” However, how can we model complex ecosystems?
- Modeling individuals and populations
A population is a system of individuals that differ in genetics and environmental histories (Heath & Gallego 1997, Fuiman et al. 2005). Consequently, variations in phenotypical traits such as behavior, growth, size, condition, and survival arise (Paper 2, Grimm & Railsback 2005). An important step towards understanding properties at the population level is therefore to recognize that individuals have different traits. The integrated interactions and properties at the individual level, defines the dynamics and properties at the population level (Grimm & Railsback 2005). These properties are not present at the individual level, but emerge from interactions among individuals and the environment (Breckling et al. 2005, Grimm et al. 2005). Further, the differences in genetic coding between spawners of the population ensure a large gene pool and thereby genetic differences between their offspring, and consequently a wide range of life history strategies. The range of strategies between individuals may dampen variability in populations in a fluctuating environment and the best strategy may differ between years. Considering the differences that exist between individuals it therefore seems logical to model ecosystems as a collection of individuals that differ in traits using an individual-based model. Information on individual traits is obtainable from measurements, which is a major advantage for individual-based models compared to population models (Huston et al. 1988).
- Individual-based models
Complex systems are classically approached by reductionism; i.e. the system can be described by sub-elements or mechanisms (Chalmers 1978). Each mechanism represents a process (e.g. metabolic rate, gut evacuation, prey encounter) studied experimentally. The general properties of the system can be described when the elements of the system are understood. The introduction of computers greatly increased the possibilities of understanding complex systems and made it possible to perform model analyses difficult to achieve in real life (Getz 1998, Peck 2004). Models evaluated against observations may therefore represent a scientific tool that can be used to test hypothesis (Peck 2004). The use of individual-based models (Huston et al. 1988) as an experimental system in evolutionary and ecological studies has drastically
Figure 2 Gallery of drawings of larval cod raning in size from 4-35 mm. Drawings by Elisabeth A. Broughton (NOAA)
increased since the 1990’s (DeAngelis & Rose 1992, Grimm & Railsback 2005). Individuals are represented by their state variables, e.g. weight, length, stomach fullness, (i-state, DeAngelis & Rose 1992), while the environment is described theoretically or from observations. This approach is contrary to the classical population models (p-state,Caswell & John 1990) where size classes defines the structure of the population. This implies that individuals of the population are considered equal in traits. An early attempt to use IBM as a tool in ecology was DeAngelis et al. (1979) who explored the development of the size-structure of a largemouth bass cohort over a period of 50 days. The size-structure of the population was revealed from the state of the individuals, and provided results not accessible by the standard structured population models (Grimm 1999). The DeAngelis et al. model has had extreme influence on the development and use of IBMs in ecology and started a whole new family of fish population models (Grimm 1999). IBMs have since been used to explore a variety of ecological problems, such as the selection of anchovy spawning grounds in Southern Benguela (Mullon et al. 2002, Huggett et al. 2003, Parada et al. 2003), the recruitment variability of bay anchovy in Chesapeake Bay (Wang et al. 1997), growth-dependent mortality of herring (Gallego & Heath 1997), the drift and growth of Northeast Arctic cod from the spawning grounds in Lofoten to the nursery grounds in Barents Sea (Vikebø et al. 2005), how size-dependent predation controls size-distribution of larval/juvenile bloater (Coregonus hoyi) in Lake Michigan (Rice et al. 1993), artificial evolution of behavior and life history traits of Müller’s pearlside (Maurolicus muelleri) (Strand et al. 2002), and in many other applications.
An IBM represents organisms and ecosystems that can be manipulated in ways difficult to achieve in real life (Peck 2004). Still, model reliability depends on the underlying assumptions of the model and the models’ ability to reproduce observable patterns. Laboratory experiments where scientists have identified and explored the properties of a single, or a multitude of processes, provide information on relevant parts of an IBM. As the number of processes included in the model grows, the complexity and the room for error increases as well. Elements of a model should therefore be limited to the ones thoroughly studied and well documented.
This thesis presents new approaches on modeling early life stages of cod, which can be distinguished into three main components; (i) mechanistic understanding of feeding and predation, (ii) the use of environmental and biological data to run and validate the IBM, (iii) how individual behavior influences growth, and survival.
- Mechanistic modeling of feeding
The IBMs presented in this thesis is a blend of earlier model components (Fiksen et al. 1998, Fiksen & Folkvord 1999, Fiksen et al. 2002, Fiksen & MacKenzie 2002) combined with a new physiology (Paper 1) and behavioral component (Paper 1, 3, 4, 5, 6). The IBM contains a mechanistic feeding module, and the tight integration with high-resolution environmental data allows for detailed modeling of early life history of cod. The IBM is based on physical and biological parameters from observations on encounter, pursuit and attack, capture success, ingestion, assimilation, metabolism, and growth (Paper 1, 3, 4, 5, and 6). Feeding is estimated in a mechanistic fashion (Caparroy et al. 2000, Fiksen & MacKenzie 2002) where light (Skartveit & Olseth 1988), visual perception (Aksnes & Giske 1993, Aksnes & Utne 1997), and larval size are key mechanisms together with the physical and biological characteristics of the prey (contrast, visibility, size, jump angle, and escape speed) (Fiksen & MacKenzie 2002).
Modeling larval foraging
In most IBMs, capture success of larval fish is a deterministic relation between the size of the larvae and the prey, based on empirical stomach content data (Werner et al. 1996, Werner et al. 2001, Lough et al. 2005). This implies that capture of a given prey item is either a success or a failure, while the mechanistic model estimates capture success as a larval-prey size-dependent probability function. Leising and Franks (1999) modeled growth and feeding of cod and haddock on Georges Bank using an IBM based on work by Laurence (1985). In the model of Leising and Franks (1999), larval fish successfully captured prey items when the ratio of prey and predator length was above 1/70 and less than 1/10. These values are comparable to the ratio between prey and larval cod length observed in gut samples by Munk (1997). The model described here was parameterized for cod by Fiksen and MacKenzie (2002), and modeled capture success (Paper 4) agree with the observed values of Munk (1997).
The distance a predator is able to perceive and encounter prey items increases with prey size (image area), and varies with light intensity, the clarity of the water, and the eye sensitivity of the predator. Encounter between prey and predator is then followed by pursuit. The pursuit time necessary to reach attack position increases with increasing prey size (Walton et al. 1992). Small predators may therefore visually see prey items far away, but pursuit success may fail if pursuit time is long and the prey has moved out of perception area.
Combined, pursuit and capture success varies with depth, light intensity, and with the relative length ratio between prey and predator, and the result is a dynamic interaction between predator, prey, and the environment (Paper 1-6). When we compared the mechanistic model with observations of prey items larval cod were able to capture (Lough et al. 2005), only small differences were revealed (Paper 4). This suggests that the most important physical and biological characteristics of both prey and predator is included in the mechanistic model. Luo et al. (1996) modeled prey selectivity of bay anchovy in Chesapeake Bay using a mechanistic model. The model setup estimated prey-predator encounter rates in a 3D setting. The visual ability of the anchovy, light intensity, water quality, and the size of the prey determined the rate of encounter, while a size-dependent function determined prey capture. Modeled prey selectivity corresponded with stomach samples. Luo et al. (1996) conclude that understanding of fish feeding ecology is best understood by observing the environment as seen by the fish.
Turbulence has long been considered important for encounter rates between prey and predator (Rothschild & Osborn 1988, Sundby et al. 1994, MacKenzie & Kiørboe 1995, Sundby 1995), however most IBM models have ignored light (Fiksen et al. 1998). In the pelagic realm, light and water quality (turbidity level) shows large latitudinal and seasonal variations among ecosystems that influence feeding and growth of larval fish (Paper 6, Suthers & Sundby 1996). Newly hatched Northeast Arctic cod have a short growth season, but almost continuous light conditions (Pedersen et al. 1989), that enable continuous feeding and potentially high growth rates. Mechanistic models represent a method for modeling first-principle physics and biology of ecosystems as close to the real world as possible, and gives the researcher the opportunity to understand how components interact on a detailed level. Of course, our ability to correctly parameterize the model (see Fiksen and MacKenzie (2002) for discussion on uncertainty in parameter values) are critical for model reliability and validation. New laboratory experiments on visual perception of larval cod would help increase the general validity of simulation results.
- Mechanistic modeling of predation on larval cod
Rice et al. (1993) modeled size-dependent mortality of bloater fish in Lake Michigan using an IBM model. The probability of being eaten by the predator alewife (Alosa pseudoharengus) decreased with larval size, and survivors were the largest and fastest growing individuals. Similarly, Ottersen and Loeng (2000) found significant correlation between mean individual length and year-class strength of Northeast Arctic cod at the stage of 5 month-old pelagic juveniles, indicating that rapid larval and juvenile growth favors high survival. In Paper 3, 4, and 5 encounter between larval cod and predators were simulated mechanistically as a function of the visual perception of the predator, the light conditions, and the predator density (Fiksen et al. 2002). Visibility of larvae to fish predator increases with size, still, the probability of the larvae being eaten decreases as larger larvae are able to migrate deeper in the water column where predation is minimized (Paper 3-5). Many individual-based models include mortality, although the implementation differs. Some models include mortality on egg stages (Mullon et al. 2003), starvation (Letcher et al. 1996, Leising & Franks 1999, Hinrichsen et al. 2002), size-dependent functions (mortality decreases with larval size) (Bartsch & Coombs 2004, Bartsch 2005), as a growth-related function (faster growing individuals have a higher probability of survival) (Gallego & Heath 1997), or as a random function (Gallego & Heath 1997). Modeling predation is not an easy task as predation pressure varies spatially and temporally within and among ecosystems. On Georges Bank, Garrison et al. (2000) observed increased predation pressure from mackerel and herring on cod and haddock larvae in years when temperature was high. The predators followed the temperature gradient and occasionally overlapped with the distribution of larval cod and haddock and caused high predation pressure. Such mortality events may be difficult to model correctly; still, predation is of crucial importance for larval fish survival and should be included in IBMs in order to resemble real life history traits. By splitting mortality into three components (Papers 3-6), the scientists can explore how predation from invertebrates and piscivores, and starvation influences survival in size-structured populations. Explicit mechanistic simulations of predation from piscivores on larval fish enable us to understand how physical properties of the environment (e.g. phytoplankton production decreases visibility in the water column) may influence predation rates.
- Modeling the environment in IBMs
The environment (larval habitat) describes the theater where the early life histories are played out, and represents a key element of an IBM. The fluctuating and changing environment is challenging to model in a reliable way, which can result in differences between the modeled and the actual habitat of interest. Studies of advection, dispersal, and distribution of larval fish (e.g. Hare et al. 1999, Fox et al. 2006) may not require a detailed description of the prey fields, but is required when modeling feeding ecology (Paper 1, 4). In most models the habitat is an assemblage of the ocean currents, temperature, turbulence, and prey fields, where variables are theoretically based (e.g. Paper 5, Hinckley et al. 1996, Werner et al. 2001), or a combination of theoretical and observed values (Paper 1, 3, 4, 6, Lough et al. 2005). An excellent example of the latter approach is the model of Bartsch et al. (2004, 2005). Bartsch et al. (2004, 2005) derived monthly egg production of Calanus finmarchicus based on the long-term observations of the number of females from Continuous Plankton Recorder (CPR) data, and satellite data of sea surface temperature and chlorophyll a concentration. The result was weekly average spatial distribution of prey suitable for mackerel in the North Atlantic. Zooplankton models provide an efficient way of to estimate larval prey abundance in time and space (Paper 6, Huse 2005). However, such models may be difficult to validate, as observations over large areas are needed. Still, oceanographic models encounter the same problem, but model results have been verified by testing modeled data at specific stations were sampling are conducted (e.g. Vikebø et al. 2005). Most model experiments use data to validate the model (Wang et al. 1997, Voss et al. 1999), but environmental and biological observations are seldom used to force the model (Paper 1, 3, 4). The re-simulation of the macrocosm experiment (Folkvord et al. 1994) used observed values of zooplankton, temperature, wind speed, and modeled light to epitomize the habitat, and enabled a direct comparison between modeled and observed stomach content and growth (Paper 1). Paper 4 modeled growth and prey selectivity of larval cod on the Georges Bank using data on the distribution and daily variation of four species, 13 stages each, of prey. This allowed for a detailed comparison between observed and modeled stomach content, and in my opinion, extensive use of data increases the validity and viability of a model.
Scientific sampling procedures, e.g. 10m strata MOCNESS samples 6 times a day (e.g. Lough et al. 2005), do not necessarily resemble the fine scale details of the water column as experienced by larval fish (Pepin 2004). The data used in Paper 3 and 4 were obtained by following a drifter deployed at 13 m depth for several days (Lough et al. 2005). The idea was that the trajectory of the drifter and the larval cohort corresponded, and consequently water samples along the drift trajectory would resemble the environment as experienced by the members of the cohort (Lough et al. 2005). Horizontal currents differ with depth, and the drift at 13 m may not have been appropriate to describe the drift of the larval cohort. Pepin (2004) warns about the relatively low probability of sampling the same water parcel twice, thereby deducing assumptions of growth and predation by using samples from different parcels containing both zooplankton and larvae, but which essentially have experienced different ecological settings. Video plankton recorder (VPR) based studies may open up new possibilities for modeling the habitat of larval fish correctly (Davis et al. 1992, Gallager et al. 2004). The VPR is moved through the water column at the speed of larval fish and digitally records plankton composition of the water column, as experienced by larval fish. A computer scans the digital pictures of the water and automatically recognizes prey species and sizes. VPR increases our ability to understand the environmental composition as seen through the eyes of larval fish. VPR data can enforce a new, efficient method for modeling the environment more correctly, and perhaps increase our knowledge of small-scale processes effect on larval feeding and growth.
- Implementing behavior in individual-based models
Coupled bio-physical models (Paper 6) generate dispersal patterns of individual members of populations in realistic flow fields. This opens up possibilities for bridging the gap between oceanography and biology (Paper 2, Vikebø et al. 2005). Spatial and temporal variability in the distribution of prey, light, turbulence, and other environmental cues, enforce larval fish to locate habitats for feeding while simultaneously avoiding predators. During the last few years, coupled models have been used extensively to model the effect of large-scale ocean currents on population dynamics of fish species around the world. Much effort has been put into model efforts to investigate how eggs and larvae drift passively with the prevailing currents from spawning grounds to nursery habitats, and how variability in current systems may affect egg and larval survival and recruitment (e.g. Werner et al. 1993, Hare et al. 1999). Huggett et al. (2003) modeled passive drift of anchovy eggs and larvae in the southern Benguela. Based on model results, the authors concluded the eggs and larval products that reach the nursery habitats successfully originate from the areas where the majority of spawning occurs. However, they also concluded that the model reproduced observable patterns when active swimming of the larvae was included. The same conclusion was reached by Werner et al. (1993) when they modeled drift of cod and haddock larvae on Georges Bank. Number of individuals to reach the nursery grounds on Georges Bank increased when directional swimming was included. Anchovy eggs and larvae in the southern Benguela was also explored by Parada et al. (2003). Here, the authors included buoyancy of the eggs and larvae, and tested for differential transport success of eggs and larvae from the spawning grounds to the nursery habitats. Parada et al. (2003) concluded that density played a significant role in transport success. Studies like Huggett et al. (2003), Werner et al. (1993), and Parada et al. (2003) are fascinating and may help researchers understand underlying strategies of choice of spawning grounds and timing of the spawning. Still, larval fish do exhibit vertical behavior (Ellertsen et al. 1984, Lough & Potter 1993, Leis et al. 2006) which could be included in coupled bio-physical models (Paper 2, 5) and increase our understanding of dispersal and drift of eggs, larvae, and early juvenile fish. Vikebø et al. (2005) showed that large variation arise in transport trajectories of particles located just a few meters away in the vertical. Strong gradients of horizontal ocean currents can have significant impact on the drift of particles. This was illustrated by modeling efforts by Fox et al. (2006). The spawning grounds of plaice in the eastern Irish Sea are close to the nursery habitats and settlement grounds. Fox et al. (2006) revealed that when larval and juvenile plaice exhibited vertical behavior synchronized to the tides, the number of individuals to reach the nursery grounds increased. Paper 2 and 3 suggests that environmental (e.g. light, prey, predators) and physiological (e.g. size, stomach fullness) cues could be used as mechanisms to drive the behavior of larval fish. This thesis shows that flexible behavior of larval fish significantly increases the probability of larval fish survival (Paper 3, 5), growth (Paper 1, 3, 4), and dispersal (Paper 5), and represents an important element of IBMs that needs further research.
Main findings and future perspectives
- Modeled foraging of first-feeding larval cod suggests that prey density of 10 nauplii·L-1 is sufficient for the larvae to grow at their physiologic maximum rate at temperatures 6-10ºC in early May (Paper 1, 6). The implementation of stomach as a larval state variable in the IBM improved the model bioenergetics, and proved important as an energy reserve during the early hours of the night (Paper 1).
- Larval cod on Georges Bank favors Pseudocalanus spp. as prey item because of its visibility and abundance in the water column. Centropages was modeled as a potentially preferred prey item, however Centropages spp. is seldom found in stomach samples with the exception of nauplius stages (Paper 4). This indicates a negative selection for this prey item
- Flexible individual behavior motivated through environmental cues and individual states (Paper 2, 3) have significant impact on growth (Paper 1, 3, 5), and survival (Paper 3, 5), and on dispersal patterns (Paper 5).
- Day-length limit feeding and growth for newly hatched larval cod in Lofoten in Northern Norway prior to mid-April. By mid-April, day-length has increased considerably and the IBM suggests high growth at prey densities above 5 nauplii·L-1 (Paper 6). Time of season, day-length, and water temperature have major impact on growth and survival of larval cod in northern Norway (Paper 6).
- Modeling early life history of cod using a mechanistic individual-based model in combination with detailed environmental data revealed information on the interactions between biotic and abiotic processes important for larval growth and survival. Models allow us to explore how processes and mechanisms influence larval growth and survival in an ethical, reliable, and affordable way. Virtual experiments can be repeated, and each single process can be turned on and off, thereby revealing its true effect on the system as a whole.
In conclusion, understanding the functional relationships between physical and biological properties of ecosystems is a challenging task where individual-based models can suffice as a valuable tool. However, the strength of an IBM rests on our knowledge on the mechanisms and processes that compose the model in the first place. Experiments and observations provide valuable information that modelers may utilize. In future perspectives, it is tempting to hope that models, such as IBMs, can be used to predict recruitment strength of fish populations. If we understand what causes variability in year-class strength, we can take measures and precautions to manage and sustain the populations and the fisheries. The study of early life history may prove to be a viable way to understand ecosystem dynamics and for preservation and management of marine habitats. Such knowledge seems ever more important when faced with climatologic changes and increasing interest in exploration for oil in marine habitats. It is therefore necessary to strive for knowledge of marine ecosystems, not only for exploitation of resources but also for the sake of preservation.
Aksnes DL, Giske J (1993) A theoretical-model of aquatic visual feeding. Ecological Modelling 67:233-250
Aksnes DL, Utne ACW (1997) A revised model of visual range in fish. Sarsia 82:137-147
Bailey KM, Houde ED (1989) Predation on eggs and larvae of marine fishes and the recruitment problem. Advances in Marine Biology 25:1-83
Bartsch J (2005) The influence of spatio-temporal egg production variability on the modelled survival of the early life history stages of mackerel (Scomber scombrus) in the eastern North Atlantic. ICES Journal of Marine Science 62:1049-1060
Bartsch J, Coombs SH (2004) An individual-based model of the early life history of mackerel (Scomber scombrus) in the eastern North Atlantic, simulating transport, growth and mortality. Fisheries Oceanography 13:365-379
Blaxter JHS (1986) Development of sense-organs and behavior of teleost larvae with special reference to feeding and predator avoidance. Transactions of the American Fisheries Society 115:98-114
Bradford MJ, Cabana G (1997) Interannual variability in stage-specific survival rates and the causes of recruitment variation In: Chambers RC, Trippel EA Early life history and recruitment in fish populations. Chapman and Hall, London, p 597 – 630
Breckling B, Müller F, Reuter H, Hölker F, Fränzle O (2005) Emergent properties in individual-based ecological models – introducing case studies in an ecosystem research context. Ecological Modelling 186:376-388
Caparroy P, Thygesen UH, Visser AW (2000) Modelling the attack success of planktonic predators: patterns and mechanisms of prey size selectivity. Journal of Plankton Research 22:1871-1900Caswell H, John AM (1990) From the individual to the population in demographic models. In: DeAngelis DL, Gross LJ Individual-based models and approaches in ecology. Chapman and Hall, New York, p 36-61
Chalmers AF (1978) What is this thing called science? Open University Press, Berkshire, UK, p 1-266
Chambers RC, Trippel EA (1997) Early life history and recruitment in fish populations. Chapman and Hall, London, p 1-596
Cushing DH (1990) Plankton production and year-class strength in fish populations – an update of the match mismatch hypothesis. Advances in Marine Biology 26:249-293
Cushing DH (1996) Towards a science of recruitment in fish populations. Ecology Institute, D-21385 Oldendorf/Luhe, Germany, p 1-175
Davis CS, Gallager SM, Solow AR (1992) Microaggregations of oceanic plankton observed by towed video microscopy. Science 257:230-232
DeAngelis DL, Cox DK, Coutant CC (1979) Cannibalism and size dispersal in young-of-the-year largemouth bass: experiment and model. Ecological Modelling 8:133-148
DeAngelis DL, Rose KA (1992) Which individual-based approach is most approprate for a given problem? In: DeAngelis DL, Gross LJ Individual-based models in ecology. Chapman and Hall, New York, p 67-87
Ellertsen B, Fossum P, Solemdal P, Sundby S (1989) Relation between temperature and survival of eggs and first-feeding larvae of northeast Arctic cod (Gadus morhua L.). Rapports et Proces-Verbaux Des Reunions, Conseil International pour l’Exploration de la Mer 191:209-219
Ellertsen B, Fossum P, Solemdal P, Sundby S, Tilseth S (1984) A case study of the distribution of cod larvae and availability of prey organisms in relation to physical processes in Lofoten. In: Dahl E, Danielssen DS, Moskness E, Solemdal P The propagation of cod (Gadus morhua L), Vol 1. Flødevigen rapportserie 1, Arendal, Norway, p 453-477
Fiksen Ø, Aksnes DL, Flyum MH, Giske J (2002) The influence of turbidity on growth and survival of fish larvae: a numerical analysis. Hydrobiologia 484:49-59
Fiksen Ø, Folkvord A (1999) Modelling growth and ingestion processes in herring Clupea harengus larvae. Marine Ecology-Progress Series 184:273-289
Fiksen Ø, MacKenzie BR (2002) Process-based models of feeding and prey selection in larval fish. Marine Ecology Progress Series 243:151-164
Fiksen Ø, Utne ACW, Aksnes DL, Eiane K, Helvik JV, Sundby S (1998) Modelling the influence of light, turbulence and ontogeny on ingestion rates in larval cod and herring. Fisheries Oceanography 7:355-363
Folkvord A, Hunter JR (1986) size-specific vulnerability of Northern anchovy, Engraulis mordax, larvae to predation by fishes. Fishery Bulletin 84:859-869
Folkvord A, Øiestad V, Kvenseth PG (1994) Growth patterns of 3 cohorts of Atlantic cod larvae (Gadus morhua L) studied in a macrocosm. ICES Journal of Marine Science 51:325-336
Fox CJ, McCloghrie P, Young EF, Nash RDM (2006) The importance of individual behaviour for successful settlement of juvenile plaice (Pleuronectes platessa L.): a modelling and field study in the eastern Irish Sea. Fisheries Oceanography 15:301-313
Fuiman LA, Cowan JH, Smith ME, O’Neal JP (2005) Behavior and recruitment success in fish larvae: variation with growth rate and the batch effect. Canadian Journal of Fisheries and Aquatic Sciences 62:1337-1362
Gallager SM, Yamazaki H, Davis CS (2004) Contribution of fine-scale vertical structure and swimming behavior to formation of plankton layers on Georges Bank. Marine Ecology Progress Series 267:27-43
Gallego A, Heath M (1997) The effect of growth-dependent mortality, external environment and internal dynamics on larval fish otolith growth: an individual-based modelling approach. Journal of Fish Biology 51:121-134
Garrison LP, Michaels W, Link JS, Fogarty MJ (2000) Predation risk on larval gadids by pelagic fish in the Georges Bank ecosystem. I. Spatial overlap associated with hydrographic features. Canadian Journal of Fisheries and Aquatic Sciences 57:2455-2469
Getz WM (1998) An introspection on the art of modeling in population ecology. Bioscience 48:540-552
Grimm V (1999) Ten years of individual-based modelling in ecology: what have we learned and what could we learn in the future? Ecological Modelling 115:129 -148
Grimm V, Railsback SF (2005) Individual based modeling and ecology. Princeton University Press, Princeton, p 1-428
Grimm V, Revilla E, Berger U, Jeltsch F, Mooij WM, Railsback SF, Thulke H-H, Weiner J, Wiegand T, DeAngelis DL (2005) Pattern-oriented modeling of agent-based complex systems: lessons from ecology. Science 310:987-991
Hare JA, Cowen RK (1997) Size, growth, development, and survival of the planktonic larvae of Pomatomus saltatrix (pisces: Pomatomidae). Ecology 78:2415-2431
Hare JA, Quinland JA, Werner FE, Blanton BO, Govoni JJ, Forward RB, Settle LR, Hoss DE (1999) Larval transport during winter in the SABRE study area: results of a coupled vertical larval behaviour-three dimensional circulation model. Fisheries Oceanography 8:57-76
Heath M, Gallego A (1997) From the biology of the individual to the dynamics of the population: bridging the gap in fish early life studies. Journal of Fish Biology 51:1-29
Hinckley S, Hermann AJ, Megrey BA (1996) Development of a spatially explicit, individual-based model of marine fish early life history. Marine Ecology-Progress Series 139:47-68
Hinrichsen HH, Möllmann C, Voss R, Köster FW, Kornilovs G (2002) Biophysical modeling of larval Baltic cod (Gadus morhua) growth and survival. Canadian Journal of Fisheries and Aquatic Sciences 59:1858-1873
Hjort J (1914a) Fluctuations in the great fisheries of northern Europe viewed in the light of biological research. Rapports et Proces-Verbaux Des Reunions, Conseil International pour l’Exploration de la Mer 20:1-228
Hjort J (1914b) Vekslingerne i de store fiskerier. Aschehoug, Kristiania, p 245
Houde ED (1987) Fish early life dynamics and recruitment variability. American Fisheries Society Symposium 2:17-29
Houde ED (1997) Patterns and trends in larval-stage growth and mortality of teleost fish. Journal of Fish Biology 51:52-83
Huggett J, Fréon P, Mullon C, Penven P (2003) Modelling the transport success of anchovy Engraulis encrasicolus eggs and larvae in the southern Benguela: the effect of spatio-temporal spawning patterns. Marine Ecology Progress Series 250:247-262
Hunter JR (1981) Feeding ecology and predation of marine fish larvae. In: Lasker R Marine Fish Larvae: Morphology, Ecology, and Relation to Fisheries. University of Washington Press, Seattle, p 33-77
Huse G (2005) Artificial evolution of Calanus‘ life history strategies under different predation levels. GLOBEC Newsletter 11:19
Huston M, DeAngelis DL, W. P (1988) New computer models unify ecological theory. Bioscience 38:682-691
ICES (1994) Spawning and life history information for North Atlantic cod stocks. Internation Council for the Exploration of the Sea, p 1-150
ICES (2005) Spawning and life history information for North Atlantic cod stocks. International Council for the Exploration of the Sea, p 1-152
Iles TD, Sinclair M (1982) Atlantic herring: stock discreteness and abundance. Science 215:627-633
Jennings S, Kaiser MJ, Reynolds JD (2001) Marine fisheries ecology. Blackwell Publishing, Oxford, UK, p 1-417
Kiørboe T, MacKenzie B (1995) Turbulence-enhanced prey encounter rates in larval fish: Effects of spatial scale, larval behaviour and size. Journal of Plankton Research 17:2319–2331
Kjesbu OS, Kryvi H, Sundby S, Solemdal P (1992) Buoyancy variations in eggs of Atlantic cod (Gadus morhua L.) in relation to chorion thickness and egg size: theory and observations. Journal of Fish Biology 41:581-599
Kjesbu OS, Witthames PR, Solemdal P, Walker MG (1998) Temporal variations in the fecundity of Arcto-Norwegian cod (Gadus morhua) in response to natural changes in food and temperature. Journal of Sea Research 40:303-321
Lasker R (1975) Field criteria for survival of anchovy larvae: the relation between inshore chlorophyll maximum layers and successful first feeding. Fishery Bulletin 73:453-461
Laurence GC (1985) A report on the development of stochastic models of food limited growth and survival of cod and haddock larvae. In Growth and survival of larval fishes in relation to the trophodynamics of Georges Bank cod and haddock. NOAA Technical Report NMFS NMFS-F/NEC 36:83-150
Leggett WC, Deblois E (1994) Recruitment in marine fishes – Is it regulated by starvation and predation in the egg and larval stages. Netherlands Journal of Sea Research 32:119-134
Leis JM, Hay AC, Trnski T (2006) In situ ontogeny of behaviour in pelagic larvae of three temperate, marine, demersal fishes. Marine Biology 148:655-669
Leising AW, Franks PJS (1999) Larval Atlantic cod (Gadus morhua) and haddock (Melanogrammus aeglefinus) growth on Georges Bank: a model with temperature, prey size, and turbulence forcing. Canadian Journal of Fisheries and Aquatic Sciences 56:25-36
Letcher BH, Rice JA, Crowder LB, Rose KA (1996) Variability in survival of larval fish: Disentangling components with a generalized individual-based model. Canadian Journal of Fisheries and Aquatic Sciences 53:787-801
Lough RG, Bolz GR (1989) The movement of cod and haddock larvae onto the shoals of Georges Bank. Journal of Fish Biology 35:71-79
Lough RG, Buckley LJ, Werner FE, Quinlan JA, Edwards KP (2005) A general biophysical model of larval cod (Gadus morhua) growth applied to populations on Georges Bank. Fisheries Oceanography 14:241-262
Lough RG, Potter DC (1993) Vertical distribution patterns and diel migrations of larval and juvenile haddock Melanogrammus aeglefinus and Atlantic cod Gadus morhua on Georges Bank. Fishery Bulletin 91:281-303
Luo J, Brandt SB, Klebasko MJ (1996) Virtual reality of planktivores: a fish’s perspective of prey size selection. Marine Ecology Progress Series 140:271-283
MacKenzie BR, Kiørboe T (1995) Encounter rates and swimming behavior of pause-travel and cruise larval fish predators in calm and turbulent laboratory environments. Limnology and Oceanography 40:1278-1289
MacKenzie BR, Kiørboe T (2000) Larval fish feeding and turbulence: A case for the downside. Limnology and Oceanography 45:1-10
MacKenzie BR, Myers RA, Bowen KG (2003) Spawner-recruit relationships and fish stock carrying capacity in aquatic ecosystems. Marine Ecology Progress Series 248:209-220
Marshall CT, Kjesbu OS, Yaragina NA, Solemdal P, Ulltang O (1998) Is spawner biomass a sensitive measure of the reproductive and recruitment potential of Northeast Arctic cod? Canadian Journal of Fisheries and Aquatic Sciences 55:1766-1783
Marteinsdottir G, Steinarsson A (1998) Maternal influence on the size and viability of Iceland cod Gadus morhua eggs and larvae. Journal of Fish Biology 52:1241-1258
Miller TJ, Crowder LB, Rice JA, Marschall EA (1988) Larval size and recruitment mechanisms in fishes: Toward a conceptual framework. Canadian Journal of Fisheries and Aquatic Sciences 45:1657-1668
Mullon C, Cury P, Penven P (2002) Evolutionary individual-based model for the recruitment of anchovy (Engraulis capensis) in the southern Benguela. Canadian Journal of Fisheries and Aquatic Sciences 59:910-922
Mullon C, Frèon P, Parada C, van der Lingen CD, Huggett J (2003) From particles to individuals: modelling the early life stages of anchovy (Engraulis capensis/encrasicolus) in the southern Benguela. Fisheries Oceanography 12:396-406
Munk P (1995) Foraging behaviour of larval cod (Gadus morhua) influenced by prey density and hunger. Marine Biology 122:205-212
Munk P (1997) Prey size spectra and prey availability of larval and small juvenile cod. Journal of Fish Biology 51:340-351
Nakken O (1994) Causes of trends and fluctuations in the Arcto-Norwegian cod stock ICES Marine Science Symposia 198:212-228
Øiestad V (1985) Predation on fish larvae as a regulatory force, illustrated in mesocosm studies with large groups of larvae. NAFO Scientific Council Studies 8:25-32
Otterlei E, Nyhammer G, Folkvord A, Stefansson SO (1999) Temperature- and size-dependent growth of larval and early juvenile Atlantic cod (Gadus morhua): a comparative study of Norwegian coastal cod and northeast Arctic cod. Canadian Journal of Fisheries and Aquatic Sciences 56:2099-2111
Ottersen G, Loeng H (2000) Covariability in early growth and year-class strength of Barents Sea cod, haddock, and herring: the environmental link. ICES Journal of Marine Science 57:339-348
Parada C, van der Lingen CD, Mullon C, Penven P (2003) Modelling the effect of buoyancy on the transport of anchovy (Engraulis capensis) eggs from spawning to nursery grounds in the southern Benguela: an IBM approach. Fisheries Oceanography 12:170-184
Peck SL (2004) Simulation as experiment: a philosophical reassessment for biological modeling. TRENDS in Ecology and Evolution 19:530-534
Pedersen T (1984) Variation of peak spawning of Arcto-Norwegian cod (Gadus morhua L.) during the time period 1929-1982 based on indices estimated from fishery statistics. In: Dahl E, Danielssen DS, Moskness E, Solemdal P The propagation of cod (Gadus morhua L). Flødevigen rapportserie 1, Arendal, Norway, p 301-316
Pedersen T, Eliassen JE, Eilertsen HC, Tande KS, Olsen RE (1989) Feeding, growth, lipid composition, and survival of larval cod (Gadus morhua L.) in relation to environmental conditions in an enclosure at 70o in northern Norway. Rapports et Proces-Verbaux Des Reunions, Conseil International pour l’Exploration de la Mer 191:409-420
Pepin P (2004) Early life history studies of prey-predator interactions: quantifying the stochastic individual responses to environmental variability. Canadian Journal of Fisheries and Aquatic Sciences 61:659-671
Pepin P, Orr DC, Anderson JT (1997) Time to hatch and larval size in relation to temperature and egg size in Atlantic cod (Gadus morhua). Canadian Journal of Fisheries and Aquatic Sciences 54:2-10
Rice JA, Miller TJ, Crowder LB, Marschall EA, Trebitz AS, DeAngelis DL (1993) Growth rate variation and larval survival: inferences from an individual-based size-dependent predation model. Canadian Journal of Fisheries and Aquatic Sciences 50:133-141
Rothschild BJ, Osborn TR (1988) Small-scale turbulence and plankton contact rates. Journal of Plankton Research 10:465-474
Skajaa K, Fernö A, Folkvord A (2003) Swimming, feeding and predator avoidance in cod larvae (Gadus morhua L.): trade-offs between hunger and predation risk. In: Browman HI, Skiftesvik AB The Big Fish Bang Proceedings of the 26th Annual Larval Fish Conference
Skartveit A, Olseth JA (1988) Varighetstabeller for timevis belysning mot 5 flater på 16 norske stasjoner (in Norwegian), Meteorological Report Series 7, University of Bergen, Norway
Strand E, Huse G, Giske J (2002) Artificial evolution of life history and behavior. American Naturalist 159:624-644
Sundby S (1995) Wind climate and foraging of larval and juvenile Arcto-Norwegian cod (Gadus morhua). Canadian Special Publication of Fisheries and Aquatic Sciences 121:405-415
Sundby S, Bjørke H, Soldal AV, Olsen S (1989) Mortality rates during the early life stages and year-class strength of the North-East Arctic cod (Gadus morhua L.). Rapports et Proces-Verbaux Des Reunions, Conseil International pour l’Exploration de la Mer 191:351-358
Sundby S, Ellertsen B, Fossum P (1994) Encounter rates between first-feeding cod larvae and their prey during moderate to strong turbulent mixing. ICES Marine Science Symposia 198:293-405
Sundby S, Fossum P (1990) Feeding conditions of Arcto-Norwegian cod larvae compared with the Rothschild-Osborn theory on small-scale turbulence and plankton contact rates. Journal of Plankton Research 12:1153-1162
Suthers IM, Sundby S (1996) Role of the midnight sun: Comparative growth of pelagic juvenile cod (Gadus morhua) from the Arcto-Norwegian and a Nova Scotian stock. ICES Journal of Marine Science 53:827-836
Ulltang Ø (1996) Stock assessment and biological knowledge: can prediction uncertainty be reduced? ICES Journal of Marine Science 53:659-675
Vikebø F, Sundby S, Ådlandsvik B, Fiksen Ø (2005) The combined effect of transport and temperature on distribution and growth of larvae and pelagic juveniles of Arcto-Norwegian cod. ICES Journal of Marine Science 62:1375-1386
Voss R, Hinrichsen H-H, St. John M (1999) Variations in the drift of larval cod (Gadus morhua L.) in the Baltic Sea: combining fields observations and modelling. Fisheries Oceanography 8:199-211
Walton WE, Hairston NG, Wetterer JK (1992) Growth-related constraints on diet selection by Sunfish. Ecology 73:429-437
Wang SB, Cowan JH, Rose KA, Houde ED (1997) Individual-based modelling of recruitment variability and biomass production of bay anchovy in mid-Chesapeake Bay. Journal of Fish Biology 51:101-120
Werner FE, Page FH, Lynch DR, Loder JW, Lough RG, Perry RI, Greenberg DA, Sinclair MM (1993) Influences of mean advection and simple behavior on the distribution of cod and haddock early life stages on Georges Bank. Fisheries Oceanography 2:43-64
Werner FE, Perry RI, Lough RG, Naimie CE (1996) Trophodynamic and advective influences on Georges Bank larval cod and haddock. Deep-Sea Research II 43:1793-1822
Werner FE, Quinlan JA, Lough RG, Lynch DR (2001) Spatially-explicit individual based modeling of marine populations: a review of the advances in the 1990s. Sarsia 86:411-421