Modelling the transport of particles in the North Sea with reference to sandeel larvae

The transport of particles representing sandeel larvae in the North Sea is simulated with a three-dimensional circulation model for the years 1976 to 1990. A great year-to-year variability in drift patterns is demonstrated. The results may explain some of the observed differences in recruitment between the main sandeel areas in the North Sea. In the northern sandeel area it seems that strong year classes are unlikely if the retention due to unfavourable currents is poor, and that a high retention in the summer may favour a good year class. In the southern sandeel area no clear coupling between year class strength and larval drift is found, possibly because the retention always seems large enough for a potentially good year class. For more quantitative use of such models in relation to sandeel recruitment, more biological knowledge is obviously needed on larval vertical distribution and timing of sandeel hatching and settling.     

Coupling hydrodynamic and individual‐based models to simulate long‐term larval supply to coastal nursery areas

For many marine fish species, recruitment is strongly related to larval survival and dispersal to nursery areas. Simulating larval drift should help assessing the sensitivity of recruitment variability to early life history. An individual‐based model (IBM) coupled to a hydrodynamic model was used to simulate common sole larval supply from spawning areas to coastal and estuarine nursery grounds at the population scale in the eastern Channel on a 14‐yr time series, from 1991 to 2004. The IBM allowed each particle released to be transported by currents from the hydrodynamic model, to grow with temperature, to migrate vertically giving stage development, and possibly to die according to drift duration, representing the life history from spawning to metamorphosis. Despite sensitivity to the larval mortality rate, the model provided realistic simulations of cohort decline and spatio‐temporal variability of larval supply. The model outputs were analysed to explore the effects of hydrodynamics and life history on the interannual variability of settled sole larvae in coastal nurseries. Different hypotheses of the spawning spatial distribution were also tested, comparing homogeneous egg distribution to observation and potential larval survival (PLS) maps. The sensitivity analyses demonstrated that larval supply is more sensitive to the life history along larval drift than to the phenology and volume of spawning, providing explanations for the lack of significant stock–recruitment relationship. Nevertheless, larval supply is sensitive to spawning distribution. Results also suggested a very low connectivity between supposed different sub‐populations in the eastern Channel.     

Variations in the drift of larval cod (Gadus morhua L.) in the Baltic Sea: combining field observations and modelling

Coupled three-dimensional (3-D) physical oceanographic modelling and field sampling programmes were carried out in May 1988 and August 1991 to investigate the potential drift of larval cod ( Gadus morhua L.) in the Bornholm Basin of the Baltic Sea. The goals were to predict the transport of cod larvae, thus aiding the identification of physical processes influencing larval retention/dispersal. Numerical simulations were performed using a 3-D eddy-resolving baroclinic model based on the Bryan–Cox–Semtner code adapted for the Baltic Sea. Within the Bornholm Basin, the model was initialized with ground truth data of physical parameters obtained on the research cruises, and all simulations were forced with actual wind data. Outside the basin, generalized hydrographic features of the Baltic Sea were utilized by incorporation of simulated hydrographic fields from previous model runs typical for the time periods considered. Larval drift was simulated either by incorporation of passive drifters, or as the initial horizontal distribution of larvae implemented into the model. Drift model simulations of larval transport agreed relatively well with field observations. The influence of variations in the vertical distribution on a smaller scale, i.e. vertical deviations of ± 6 m from the observed mean centre of mass, on the drift was examined, revealing no significant differences in the drift of larvae depending on their vertical distribution. The different wind forcing during the investigated time periods was linked to a retention situation in May 1988 and to a dispersal situation in August 1991. Finally, observed spatial distribution patterns of 1-group cod based on Baltic Young Fish Surveys (BYFS) were compared with their predicted transport in the larval phase and examined with respect to recruitment.     

Physical transport of young pollock larvae (Theragra chalcogramma) near Shelikof Strait as inferred from a hydrodynamic model

An advective model was used to simulate the drift of larval walleye pollock ( Theragra chalcogramma ) over a 40-day period (late April through early June) near Shelikof Strait, Alaska. This model was used: (i) to assess how much of the observed change in larval positions during that period can be explained by transport at fixed depth; (ii) to demonstrate that observed change can be related to mean large-scale meteorological forcing; and (iii) to investigate accumulation of larvae in specific areas near the coast. Based on availability of larval and circulation data, three years were studied: 1988, 1989 and 1991. Velocity fields generated from a hydrodynamic model driven by winds and runoff were used to advect particles seeded in accordance with observed larval distributions in late April of each year. The modelled larvae were tracked at 40 m depth, corresponding to the mean depth of sampled larvae and the depth of neutrally buoyant drifters employed in field studies. Specific features observed in late May larval surveys were reproduced by the model, such as the accumulation of larvae in a shoal area downstream of the strait. Differences among the modelled years include extensive flushing of larvae to the south-west in 1988 and 1991, vs. limited flushing in 1989. These differences appear related to the mean large-scale atmospheric pressure patterns for April-May of those years.     

The potential for advective exchange of the early life stages between the western and eastern Baltic cod (Gadus morhua L.) stocks

In order to clarify mechanisms influencing the reproductive success of Baltic cod (Gadus morhua L.), a modelling exercise was performed to examine the effects of the wind‐driven circulation on the transport of early life stages between the western and eastern Baltic. Because the different stocks spawn in different areas and environments at different times of the year, the occurrence of variable age/length distributions of juveniles within the different potential nursery areas can be explained by the circulation pattern. A three‐dimensional circulation model of the Baltic was utilized to investigate the temporal evolution of egg and larval distributions of the western Baltic cod stock, which spawns preferentially in the Danish Straits, in Kiel Bay as well as in Mecklenburg Bay. For different scenarios (1988 and 1993), within‐ and between‐year variability of egg and larval transport showed large differences, primarily due to variations in wind forcing. In 1988, relatively low and variable wind forcing prevailed, whereas, due to sustained strong, mainly westerly, winds, in January 1993, the recent major Baltic inflow to the Baltic Sea occurred. Differences in contributions of early life stages from the western to the eastern cod stocks, depending on the physical forcing conditions, suggest that this process can be controlled by variations of atmospheric forcing conditions. The potential for early life stages from the western Baltic cod stock to drift into the Arkona Basin and the Bornholm Basin, and to contribute there to the juvenile population, has been recognized as being mainly due to strong westerly winds. During cold winters, retention of eggs, larvae and juveniles within their original spawning grounds may predominate. Transport of cod early life stages from the Øresund, as well as from the Great Belt, can occur only during periods of strong westerly winds, but significant eastwards orientated drift from Kiel Bay and Mecklenburg Bay was also evident during periods of minor westerly wind influence.     

Spawning and transport dynamics of Atlantic menhaden: inferences from characteristics of immigrating larvae and predictions of a hydrodynamic model

We examined patterns of abundance, age and spawning date distributions of Atlantic menhaden Brevoortia tyrannus larvae immigrating during two seasons through three North Carolina inlets—Oregon, Ocracoke and Beaufort—to elucidate their spawning and transport dynamics. These patterns were examined in conjunction with corresponding predictions from a three-dimensional, wind-and tide-driven hydrodynamic model. Larvae immigrating through different inlets showed consistent similarities as well as marked differences in temporal patterns of abundance, spawning dates and transport times. Intraseasonal patterns in abundance and spawning date distributions among inlets suggest that, in both study years, the spatio-temporal dynamics of menhaden immigration were driven by large-scale patterns along the Atlantic coast, rather than by localized variation in spawning activity. Interannual differences in the temporal patterns of spawning dates and larval immigration indicate interannual differences in transport dynamics and/or the spatial-temporal distribution of spawning. When the spawning locations predicted by the hydrodynamic model are interpreted in conjunction with advanced very high resolution radiometer sea-surface temperature information, the results are consistent with the limited historical information available on spatio-temporal distribution of Atlantic menhaden eggs and larvae. The transport model also predicted distributions of arrival times for immigrating larvae that were comparable in range and variability with observed patterns. Our use of data from immigrating larvae, coupled with a hydrodynamic transport model and sea-surface temperatures, allowed us to uncover relationships between spatio-temporal patterns of Atlantic menhaden spawning and transport dynamics that could not have been identified by either approach alone.     

Predicting the dispersion of sprat larvae (Sprattus sprattus (L.)) in the German Bight

Sprat (Sprattus sprattus (L.)) larval dispersal in the German Bight is predicted with a model system consisting of two three-dimensional circulation models and a three-dimensional transport model. Horizontal distribution data for specific length classes of larvae, obtained from field surveys carried out in the German Bight during summer 1991, are used as initial conditions for the starting points of the tracers in the transport model. Sprat larval distributions are predicted and are compared with distributions of the appropriate length class of larvae observed during a subsequent cruise about 3 weeks later. The predicted larval (tracer) distributions compare favourably to observed larvae distributions in the inner model areas but not in the areas close to the coast. Factors contributing to inaccurate predictions are uncertainties in horizontal and vertical larval distributions as well as the model grid resolution, which is too coarse in coastal areas.     

Larval lake sturgeon production and drift behaviour in the Menominee and Oconto Rivers,Wisconsin

Lake sturgeon (Acipenser fulvescens) have been negatively impacted by barriers to migration, pollution and overharvest. Biological data such as the timing of spawning activity and larval drift have provided a better understanding of larval production success or failure on a site-specific basis. However, many river systems that serve as reproductive locations for remnant lake sturgeon populations remain understudied. The objective of this study was to compare and contrast aspects of larval lake sturgeon drift within and between the Oconto and Menominee Rivers over multiple reproductive seasons. Differences in larval drift chronology, the size of drifting larvae during the season and the cross-sectional profile of drifting individuals as a function of river velocity were evaluated. D-frame drift nets were deployed at a single transect below spawning sites in the Menominee (2012, 2013 and 2014) and Oconto Rivers (2013, 2014 and 2015). A total of 4,442 larvae were captured across all seasons. Larvae exhibited nonrandom drift profiles related to water velocity, and larvae size was related to horizontal drift location. The seasonal increase in body size of drifting larvae also differed among drift events within and between years and across rivers. This information will provide a better understanding of lake sturgeon early life history and has practical management connections including spawning site protection/enhancement and river flow mitigation.     

Influence of larval transport and temperature on recruitment dynamics of North Sea cod (Gadus morhua) across spatial scales of observation

The survival of fish eggs and larvae, and therefore recruitment success, can be critically affected by transport in ocean currents. Combining a model of early‐life stage dispersal with statistical stock–recruitment models, we investigated the role of larval transport for recruitment variability across spatial scales for the population complex of North Sea cod (Gadus morhua). By using a coupled physical–biological model, we estimated the egg and larval transport over a 44‐year period. The oceanographic component of the model, capable of capturing the interannual variability of temperature and ocean current patterns, was coupled to the biological component, an individual‐based model (IBM) that simulated the cod eggs and larvae development and mortality. This study proposes a novel method to account for larval transport and success in stock–recruitment models: weighting the spawning stock biomass by retention rate and, in the case of multiple populations, their connectivity. Our method provides an estimate of the stock biomass contributing to recruitment and the effect of larval transport on recruitment variability. Our results indicate an effect, albeit small, in some populations at the local level. Including transport anomaly as an environmental covariate in traditional stock–recruitment models in turn captures recruitment variability at larger scales. Our study aims to quantify the role of larval transport for recruitment across spatial scales, and disentangle the roles of temperature and larval transport on effective connectivity between populations, thus informing about the potential impacts of climate change on the cod population structure in the North Sea.     

The recruitment process of the Barents Sea capelin (Mallotus villosus) stock, 2001–2003

Oceanographic and predation processes are important modulators of fish larvae survival and mortality. This study addresses the hypothesis that immature Norwegian spring‐spawning herring (Clupea harengus), when abundant in the Barents Sea, determine the capelin reproduction success through consumption of Barents Sea capelin (Mallotus villosus) larvae. Combining a hydrodynamic model and particle‐tracking individual‐based model, a realistic spatio‐temporal overlap between capelin larvae and predatory immature herring was modelled for the summer seasons of 2001–2003. Capelin larvae originating from western spawning grounds became widely dispersed during the summer season, whereas those originating from eastern spawning grounds experienced a rapid drift into the southeastern Barents Sea. Herring caused a 3% mortality of the capelin larvae population in 2001 and a 16% mortality in 2003, but the effect of predation from herring on capelin larvae was negligible in 2002. Despite a strong capelin larvae cohort and a virtual absence of predatory herring, the recruitment from the capelin 2002 year class was relatively poor from a long‐term perspective. We show that the choice of capelin spawning grounds has a major impact on the subsequent capelin larvae drift patterns, constituting an important modulator of the capelin larvae survival. Variation in drift patterns during the summer season is likely to expose the capelin larvae to a wide range of hazards, including predation from young cod, sandeel and other predators. Such alternative predators might thus have contributed to the poor capelin recruitment during 2001–2003, leading to the collapse of the capelin stock in the subsequent years.     

Connectivity of the bay scallop (Argopecten irradians) in Buzzards Bay,Massachusetts, U.S.A.

The harvest of bay scallops (Argopecten irradians) from Buzzards Bay, Massachusetts, U.S.A. undergoes large interannual fluctuations, varying by more than an order of magnitude in successive years. To investigate the extent to which these fluctuations may be due to yearly variations in the transport of scallop larvae from spawning areas to suitable juvenile habitat (settlement zones), a high‐resolution hydrodynamic model was used to drive an individual‐based model of scallop larval transport. Model results revealed that scallop spawning in Buzzards Bay occurs during a time when nearshore bay currents were principally directed up‐bay in response to a persistent southwesterly sea breeze. This nearshore flow results in the substantial transport of larvae from lower‐bay spawning areas to settlement zones further up‐bay. Averaged over the entire bay, the spawning‐to‐settlement zone connectivity exhibits little interannual variation. However, connectivities between individual spawning and settlement zones vary by up to an order of magnitude. The model results identified spawning areas that have the greatest probability of transporting larvae to juvenile habitat. Because managers may aim to increase scallop populations either locally or broadly, the high‐connectivity spawning areas were divided into: (i) high larval retention and relatively little larval transport to adjoining settlement areas, (ii) both significant larval retention and transport to more distant settlement areas, and (iii) little larval retention but significant transport to distant settlement areas.     

Predicting the occurrence of Atlantic bluefin tuna (Thunnus thynnus) larvae in the northern Gulf of Mexico: building a classification model from archival data

Although bluefin tuna are found throughout the Atlantic Ocean, spawning in the western Atlantic has been recorded predominantly in the Gulf of Mexico (GOM) in spring. Larval bluefin tuna abundances from the northern GOM are formulated into an index used to tune the adult stock assessment, and the variability of this index is currently high. This study investigated whether some of the variability in larval bluefin tuna abundances was related to environmental conditions, by defining associations between larval bluefin tuna catch locations, and a suite of environmental variables. We hypothesized that certain habitat types, as defined by environmental variables, would be more likely to contain bluefin tuna larvae. Favorable habitat for bluefin tuna larvae was defined using a classification tree approach. Habitat within the Loop Current was generally less favorable, as were warm‐core rings, and cooler waters on the continental shelf. The location and size of favorable habitat was highly variable among years, which was reflected in the locations of larval bluefin tuna catches. The model successfully placed bluefin tuna larvae in favorable habitat with nearly 90% accuracy, but many negative stations were also located within theoretically favorable habitat. The probability of collecting larval bluefin tuna in favorable habitat was nearly twice the probability of collecting bluefin tuna larvae across all habitats (35.5 versus 21.0%). This model is a useful addition to knowledge of larval bluefin tuna distributions; however, the incorporation of variables describing finer‐scale features, such as thermal fronts, may significantly improve the model’s predictive power.     

Simulating the dispersion of vertically migrating sprat larvae (Sprattus sprattus (L.)) in the German Bight with a circulation and transport model system

A model system consisting of two three-dimensional circulation models and a three-dimensional transport model is used to simulate the dispersal of sprat larvae (Sprattus sprattus (L.)) in the German Bight. A Northwest European shelf sea model simulates the currents of the North Sea and provides sea surface elevation data for the open boundaries of the German Bight circulation model. The German Bight circulation model has a horizontal grid resolution of 2.75 km and a vertical resolution of 5 m except for the bottom layer. The driving forces of the circulation models are the M2-tide, six-hourly time-dependent wind stress and air pressure fields as well as monthly climatological density fields. The simulated currents serve as input to the transport model. The model area, grid size and the vertical resolution of the transport model are in accordance with the German Bight circulation model. The transport model simulates advection and diffusion using a Monte Carlo method. Furthermore it incorporates a simulation of active vertical movement of the larvae. The vertical migration pattern is dependent on the time of day and size of the larvae and is based on field data from the German Bight. Horizontal larvae distribution data, gained from two out of five field surveys carried out in the German Bight during spring and summer 1991, are used as initial values for the starting points of the tracers in the transport model. Simulated tracer distributions are compared with larval distributions observed during subsequent cruises about 3 weeks later. Discrepancies and similarities between tracer and larval distributions are discussed.     

A synthesis of biological and physical processes affecting the feeding environment of larval walleye pollock (Theragra chalcogramma) in the eastern Bering Sea

Biological and physical phenomena that affect conditions for larval survival and eventual recruitment differ in the oceanic and shelf regions. In the oceanic region, eddies are a common feature. While their genesis is not well known, eddies have unique biophysical characteristics and occur with such regularity that they likely affect larval survival. High concentrations of larval pollock often are associated with eddies. Some eddies are transported onto the shelf, thereby providing larvae to the Outer Shelf Domain. Advection, rather than local production, dominated the observed springtime increase in chlorophyll (often a correlate of larval food) in the oceanic region. Over two-thirds of the south-eastern shelf, eddies are absent and other phenomena are important. Sea ice is a feature of the shelf region: its interannual variability (time of arrival, persistence, and areal extent) affects developmental rate of larvae, timing of the phytoplankton bloom (and potentially the match/mismatch of larvae and prey), and abundance and distribution of juvenile pollock. In the oceanic region, interannual variation in food for first-feeding pollock larvae is determined by advection; in the shelf region, it is the coupled dynamics of the atmosphere–ice–ocean system.     

Modeling the transport and survival of Japanese sardine larvae in and around the Kuroshio Current

We have numerically modeled the advection and diffusion of sardine eggs and larvae to investigate the larval transport processes of Japanese sardine from the spawning grounds by the Kuroshio.
The results indicated that the offshore drift current induced by the winter monsoon and the location of the spawning ground have significant effects on the survival of the Japanese sardine. The contribution of the drift current, the distance of the spawning ground from the Kuroshio axis, and the eddy diffusivity to the larval retention in the coastal area is approximately expressed by the following equation: where R is the retention rate in the coastal area, a the variance of initial distribution of eggs, T the time after the eggs were spawned, – V₀ the velocity of the wind-induced offshore current, y₀ the distance of the center of the spawning area from the Kuroshio axis, and K the coefficient of horizontal eddy diffusivity.
The year-to-year variation in larval survival rates stimulated by the two-dimensional model are consistent with those estimated previously by using field data of egg and larval abundance during 1978–1988.     

Modelling advection of cod eggs and larvae on the Newfoundland Shelf

We investigate interannual variations in dispersion for the drift and retention of cod eggs and larvae on the Newfoundland Shelf using a two-dimensional Lagrangian tracking model. The velocity field for the drift model is obtained from a diagnostic calculation of objectively analysed density data. Time-dependent currents are generated using an inertial-current slab model driven by observed winds. Eggs and larvae are treated as passive drifters seeded in a dispersion model of the Newfoundland Shelf region. We identify favourable and unfavourable zones of retention on the Newfoundland Shelf. We show that northerly, shelf-break spawning locations are more favourable than southerly shelf-break spawning locations for northern cod, ( Gudus morhua , in NAFO divisions 2J3KL).     

Variability of larval Baltic sprat (Sprattus sprattus L.) otolith growth: a modeling approach combining spatially and temporally resolved biotic and abiotic environmental key variables

Plankton sampling was conducted in the Baltic to obtain sprat larvae. Their individual drift patterns were back‐calculated using a hydrodynamic model. The modelled positions along the individual drift trajectories were subsequently used to provide insight into the environmental conditions experienced by the larvae. Autocorrelation analysis revealed that successive otolith increment widths of individual larvae were not independent. Otolith increment width was then modelled using two different generalized additive model (GAM) analyses (with and without autocorrelation), using environmental variables determined for each modelled individual larval position as explanatory variables. The results indicate that otolith growth was not only influenced by the density of potential prey but was controlled by a number of simultaneously acting environmental factors. The final model, not considering autocorrelation, explained more than 80% of the variance of otolith growth, with larval age as a factor variable showing the strongest significant impact on otolith growth. Otolith growth was further explained by statistically significant ambient environmental factors such as temperature, bottom depth, prey density and turbulence. The GAM analysis, taking autocorrelation into account, explained almost 98% of the variability, with the previous otolith increment showing the strongest significant effect. Larval age as well as ambient temperature and prey abundance also had a significant effect. An alternative approach applied individual‐based model (IBM) simulations on larval drift, feeding, growth and survival starting as exogenously feeding larvae at the back‐calculated positions. The IBM results revealed optimal growth conditions for more than 97% of the larvae, with a tendency for our IBM to slightly overestimate larval growth.     

Modelling the dispersal of herring and hake larvae in the Strait of Georgia for the period 2007–2009

The Strait of Georgia (SoG), between Vancouver Island and mainland British Columbia, is a larval rearing ground for both hake and herring stocks, which are commercially important. Year‐to‐year variability in larval retention within the strait is examined by simulating drift tracks of larvae for these species using an ocean circulation model and a particle‐tracking model. Larvae with different vertical swimming behaviors were tracked in the springs of 2007, 2008, and 2009. Since herring larvae mostly stay near the surface, their distribution is heavily influenced by the wind. Strong winds to the north soon after the hatching period tend to wash herring larvae out of SoG and winds to the south help retain herring larvae inside the Strait. In 2007, the model indicates a massive wind‐driven export of herring larvae which may have led to the observed failure of herring production. In contrast, hake larvae reside deeper in the water column (50–200 m). Their distribution is less sensitive to surface forcing but is shaped by a deep gyre with cross‐strait currents. This study also suggests that the northern and southern SoG are weakly connected for herring larvae dispersal, which makes both regions potentially important to recruitment.     

Modelling changes in the length–frequency distributions of fish larvae using field estimates of predator abundance and size distributions

The goal of this study is to determine if an individual-based size-dependent model can realistically simulate changes in the length–frequency distributions of several species of fish larvae collected in Conception Bay in 1993 and 1994, using field estimations of growth and predator abundance. We first model the length–frequency distribution of field samples with the best possible estimates of mean growth rate. Then, we add predation mortality given the characteristics of the predator community observed during our surveys, which was composed of macrozooplankton and adult capelin. The larval fish community is generally not affected by predation by macrozooplankton, as the average instantaneous mortality rate predicted by the model was 0.004 day^–1. Fish larvae appear to be more vulnerable to predation by the population of adult capelin. We estimate that an abundance of adult capelin ranging between 0.2 and 1.0 individuals per 1000 m^–3 may have a substantial impact on the larval fish community. The predictions of an individual-based model are directly related to the accuracy of estimates of the mean growth rates of the larval fish cohorts. We find that it is difficult to differentiate size-selective removal of individuals from random selection by analysing changes of the length–frequency distributions of the larval fish community.     

The importance of individual behaviour for successful settlement of juvenile plaice (Pleuronectes platessa L.): a modelling and field study in the eastern Irish Sea

To study the transport of plaice (Pleuronectes platessa L.) eggs and larvae in the eastern Irish Sea, we constructed a 3D‐baroclinic physical model and coupled it to a particle‐tracking scheme that allowed aspects of larval behaviour to be simulated. Starting positions for eggs were based upon data from a series of ichthyoplankton surveys and final positions were compared with results of settled plaice distributions from two beam trawl surveys conducted on beaches around the eastern Irish Sea. If simulated larval behaviour was limited to passive drift or horizontal swimming, the particles diffused away from the spawning areas but failed to reach nursery grounds in significant numbers (85–90% remaining offshore). In contrast, switching on circatidal vertical swimming significantly increased the numbers of larvae reaching the coast (only 23–30% remained offshore). Particles tended to accumulate in bays and estuaries and this pattern compared well with the distribution of settled plaice from the field surveys. Studies in the southern North Sea (where spawning and nursery grounds are widely separated) have also demonstrated the importance of selective tidal stream transport for successful recruitment of settling plaice to nursery grounds. Although our understanding of the ontogeny of this behaviour is still poor, the model results presented suggest that this aspect of behaviour is a key factor influencing plaice settlement success.