Ecosystem Model of PWS Herring

Management and Restoration Tool

Four Year Cycle

Analysis of the data from the Alaska Fish and Game’s Age Structure Analysis (ASA) Models showed that the recruitment pattern herring is characterized by large variations in the rates and a clearly defined 4 year cycle in which exceptionally high rates of recruitment are followed by 3 years of low recruitment. 

We have written a statistical subroutine for our PWS population dynamics model that provides a good description of this process. Despite the simplicity of the subroutine the fit between measured and predicted population of juveniles is good.

Further analysis using this model indicates that The process of cohort dominance within nursery grounds explains the 4 year cycle in recruitment that occurred in the 1980s. Stated most simply, the juveniles of a strong year class will greatly reduce the survival of age 0 juveniles in younger year classes as long as this strong year class resides within or regularly visits their nursery. Since most juveniles leave the nursery by age 3, the nursery is effectively closed by a strong year class for 3 years and then opened on the 4th year.

This cycle was trigged by several years of exceptional recruitment in the mid 1970s that was driven by the Pacific Decadal Oscillation. The cycle was suppressed in the early 1990s most likely because of several years of reduced zooplankton forage. This conjecture is supported by both the drop in adult survival in from mid-1991 to mid-1993 and the large drop in the recruitment of the expected strong 1992 year class. These same patterns occurred in synchrony with those in Sitka Sound.

Unrestricted Cohort Tracking Algorithm

This algorithm, developed by study participant Vince Patrick, describes the abundance of herring born in a given year as they age and mature from age 0 juveniles that enter coastal nursery grounds to age 9 adults.

It draws upon the time series of adult population size as estimated from

• Aerial surveys of milt
• coastal egg surveys
• acoustic surveys
• age distribution of fish in adult spawning schools found in sectors of PWS

We have reviewed ASA code and found that several assumptions or applications of the routine are questionable. (However they have at times been adjusted when the ASA appeared to drift from reality.) These include the assumptions that:

• Adult natural mortality rates are constant
• Juvenile maturation rates are constant.

The Unrestricted Cohort Tracking Algorithm consists of a comprehensive search routine for retrospective analysis of the ASA survey data. It is designed to reduce the number of assumptions (such as the two mentioned above) and constraints found in the ASA Model. This algorithm not only provides the means to assess annual variability in adult mortality and the apparent maturation rate of juveniles, but it also provides an interpolation scheme to assess variability in the survival probability of juvenile cohorts.

The algorithm helps provide an objective analysis of the differing hypotheses for the current low abundance state and reasons why PWS herring have not returned to a high and stable abundance. Examples of such hypotheses include:

• Viral Hemorrhagic Septicemia (VHS) and Ichthyphonus Epizootics that broke out in the early 1990s
• Whale predation
• Pink salmon fry predation upon herring larvae and metamorphs
• A residual effect of the collapse of the herring population in the early 1990s caused by the oil spill, over-fishing, and forage depravation.

It has been proposed by our consultant, Evelyn Brown and others that recovery from this collapse has been prevented by loss or disruption spawning and nursery grounds in the early 1990s. Recovery can of course be further hindered by the issues of predation and disease listed above. These alternatives are not easily tested without a flexible and objective analysis.

More information on this feature and some results obtained in our studies can be found in the "Fourth Quarterly Project Report to EVOSTC"

The Nursery Model

Dale Kiefer has created a nursery model after discovering that:

• Large annual variations in recruitment appear to be closely tied to food supply in the nurseries and the associated competition for food between juvenile cohorts for a limited supply of food.


• Patterns in the spatial distribution of nurseries and spawning grounds observed from aerial surveys suggested that
  • They are not rigidly tied to a given shoreline
  • The age composition of juveniles within the nurseries varies with location.

The model describes the dynamics of two major components, the population of zooplankton foraged by juveniles and the juveniles themselves. The juveniles consist of cohorts of age 0 through age 4.

The population of zooplankton forage within the nursery is supplied by advection and turbulent exchange between the nursery and ambient offshore waters, and the forage is depleted by predation of the population of juvenile herring (as well as other foraging species).

The rate of supply of zooplankton will depend upon the volume exchange rate between the nursery and the offshore waters as well as the gradient in zooplankton concentration between the nursery and the offshore waters.

The rate of depletion of the zooplankton will be largely influenced by the size of the juvenile cohort populations and their respective predation efficiency.

The population of a juvenile cohorts will be determined by prior seeding of the bay by metamorphs of that year class and their survival probability - presumably determined by starvation, predation, and disease.

There are two types of interactions within and between cohorts:

• They compete with each other for zooplankton forage.
  • It is reasonable to assume that the younger and smaller juveniles are at a disadvantage.
  • The younger and smaller juveniles are therefore also at a disadvantage with regard to starvation.
  
  
• They interact with each other by processing information on their rates of encounter with each other and other cohorts.

  [We have chosen to call this process quorum sensing - the means by which individuals or independent groups signal or communicate with each other to achieve a coordinated action. This behavior is observed in many species.]

  • In our model herring use quorum sensing to adjust their swimming and schooling behavior in an attempt to achieve a desired rate of encounter with other members of their cohort or with members of other cohorts that are close in age and size.
  • If the schools diverge (see next paragraph) from the desired encounter rate, we assign a cost that increases with the size of the divergence.

Finally, the model includes a bio-energetic model of herring growth and metabolism. Specifically, the system of equations describes ingestion, egestion, respiration, and growth of fish as a function of food supply, their age or size, water temperature, and swimming speed.

In our model herring schools try adjust their swimming behavior to achieve maximal growth rates. If the schools diverge from their desired encounter rate with forage, we assign a cost that increases with the size of the deficit in growth rate. One should note immediately that the goal to achieve maximal growth rates and the goal of encountering other juvenile herring at a desired rate will likely conflict over a range of environmental conditions. For example, if the supply of zooplankton forage to a nursery is low, then schools will tend to forage over a greater area in order to achieve higher encounter rates with zooplankton and thus higher growth rates.

However, such spatial expansion of the nursery will inevitably reduce herring-herring encounters, driving the schools away from their targeted encounter rate. We propose that this conflict is ultimately resolved through a cost-benefit assessment of the departure from the two goals. Such resolution will shape not only the size of the nursery but also shape the age composition of individuals in schools within the nursery and even in adult populations.

More information on this feature and some results obtained in our studies can be found in the "Fourth Quarterly Project Report to EVOSTC"