link to AFSC home page

link to AFSC home page link to NMFS home page link to NOAA home page

Resource Ecology & Fisheries Management (REFM) Division

AFSC Quarterly
Research Reports
Jan-Feb-Mar 2009
ABL Reports
FMA Reports
NMML Reports
RACE Reports
REFM Reports
All Reports (.pdf)
Quarterly Index
Quarterly Home

Status of Stocks & Multispecies Assessment Program

DisMELS: A Dispersal Model for Early Life History Stages

figure 2, see caption
Figure 2.  A screenshot of the DisMELS graphical user interface after running a dispersal simulation, showing the "spaghetti" tracks of individual simulated larvae.  Red squares encompass 5x5 grid cells (~10 km on a side) used in the NEP ROMS oceanographic model; black areas signify land.  Several hundred northern rock sole "early stage larvae" were released from five "spawning areas" (blue dots indicate release positions) along the Alaska Peninsula and tracked for 60 days.  Upon reaching ~8mm size, "early stage larvae" (tracks in blue) become "late stage larvae" (tracks in green), which undergo diel vertical migrations between 10-20m depth (nighttime) and 30-40 m depth (daytime).  Green triangles indicate positions of "late stage larvae" after 60 days.

Juveniles of many flatfish species have distinct habitat requirements (such as sediment grain size) and thus use specific areas as "nurseries." For some species, adult spawning grounds are substantially removed from potential juvenile nurseries. Consequently, the dispersal pathways of eggs and larvae via oceanographic currents are critical factors in the process of recruitment for many flatfish species, where "recruitment" is generally the number of individuals that survive the dispersive pelagic stage and reach some age as juveniles, such as age 1.

In 2002, Tom Wilderbuer, Jim Ingraham and other SSMA scientists found that annual recruitment of several flatfish species in the eastern Bering Sea was fairly well correlated with the final location of a drifter simulated using OSCURS, a simple surface wind drift current model. Their findings indicated that recruitment was potentially high for years in which the simulated drifter moved inshore from its release location at lat. 56°N, long. 165°W (near the edge of the continental shelf) into shallow water but that recruitment was always low for years in which the simulated drifter moved off the continental shelf into deep water.

Subsequently, the OSCURS model has been used to make an annual qualitative prediction of flatfish recruitment strength. This prediction is included in the Ecosystems Considerations Report provided by the AFSC to the North Pacific Fishery Management Council as part of its annual stock assessment process.

Because the OSCURS-based recruitment prediction is only qualitative in nature, it has not been incorporated directly into any of the stock assessment models used to assess flatfish stocks. Generally, these stock assessment models make their own estimates of annual recruitment strength by determining the number of individuals that are needed to recruit at age a in year x to end up with an observed number of individuals of age y in year x+y-a, given assumptions about natural and fishing mortality in intervening years.

An independent, quantitative estimate of recruitment strength for the assessment models could greatly improve these modelsí estimates of stock size and allowable catch, as well as recruitment.

In an effort to provide quantitative predictions for flatfish recruitment that can be incorporated directly into stock assessment models, William Stockhausen in REFM is developing DisMELS, a coupled individual-based biophysical dispersal model for early life history stages.

While OSCURS simulates the tracks of passive drifters that are confined to the ocean surface, the individual eggs and larvae that DisMELS simulates are not confined to the ocean's surface because it uses 3-dimensional (3-D) current fields from a 3-D oceanographic model to drive advective transport. It turns out that the depth at which individuals are tracked can make a big difference in where they end up.

The simulated individuals in DisMELS can also incorporate important aspects of the behavior of real eggs and larvae. For example, Janet Duffy-Anderson and other scientists in the joint AFSC-Pacific Marine Environmental Laboratory (PMEL) ecoFOCI research program have shown that the eggs and larvae of some flatfish species are concentrated below the surface, not at the surface (as modeled by OSCURS).

Additionally, while eggs and the youngest larvae may change depth slowly (over the course of many days) as they develop, older larvae may migrate vertically between depth zones on a daily basis. In DisMELS, it is possible to capture these ontogenetic and diel changes in depth as part of the "behavior" of simulated individuals.

In addition, DisMELS can easily track the dispersal of thousands of simulated individuals over time (Fig. 2), thus leading to a quantitative prediction of recruitment strength (as the fraction of individuals "spawned" that arrive at nursery sites when they are capable of leaving the plankton for a primarily benthic existence). This quantitative prediction of recruitment strength can then be fed into a stock assessment model to improve its own estimate of recruitment.

DisMELS consists of a series of graphical user interfaces (Fig. 2) that allow users to easily set up and run individual-based models, as well as an application programming interface (API) that allows users to create their own stage-based behavioral models for use in the dispersal simulations. Currently, three relatively simple life stages have been defined: an adult stage, a pelagic stage, and a "settler" stage.

figure 3, see caption
Figure 3.  Schematic of potential "daisy-chain" of life stages and sub-stages.  Different life stages (adult, pelagic, settler) incorporate different characteristics and potential behavior (e.g., spawning or vertical migration) whereas different sub-stages incorporate similar characteristics and behavior but with different parameter values.

Ontogenetic changes in behavior within a stage can be incorporated by "daisy-chaining" several sub-stages together (with different parameter values for successive stages, Fig.3) and setting the conditions under which individuals change from one sub-stage to the next.

For adult stages, spawning season, spawning frequency, fecundity (number of individuals spawned), and spawned life stage (e.g., egg stages with different development rates) can be specified. For pelagic stages (i.e., eggs and larvae), preferred daytime and nighttime depth ranges, vertical swimming speed, vertical diffusion, growth and mortality rates, and stage duration can be specified.

The settler stage has behavior and parameters similar to that of the pelagic stage, but the user can also specify the characteristics (as a depth range) for suitable nursery habitat; when a settler reaches suitable nursery habitat, it leaves the plankton, "settles" to the benthos, and is counted as a successful recruit. Settlers that do not reach suitable nursery habitat within a user-specified time interval are counted as dead.

Time series of precomputed 3-D oceanographic currents, temperature and salinity fields are required to drive DisMELS. At present, output from the ROMS (Regional Oceanographic Modeling System) model for the northeast Pacific (NEP ROMS), as well as other ROMS models, can be used to drive dispersal simulations.

As an initial application, Stockhausen, Wilderbuer, Duffy-Anderson, and Al Hermann (Joint Institute for the Study of the Atmosphere and Ocean/PMEL) are using DisMELS to develop recruitment predictions for northern rock sole and Alaska plaice in the eastern Bering Sea.

By William Stockhausen

<<< previous

next >>>

            Home | FOIA | Privacy | | Accessibility      doc logo