S.K. Parkinson1, J.A. Chandler
Idaho Power Company, Boise, Idaho USA
P. Goodwin
Ecohydraulics Research Group, University of Idaho, USA
1Address
for Correspondence:
Idaho Power Company, PO Box 70, Boise, Idaho 83707 USA
Tel:
(208) 388-2495. Fax (208)
388-6689. E-mail: skp2495@idahopower.com
Abstract: A program to develop a better understanding for fish response to flow management in natural rivers is outlined. The need to understand these relationships is driven by both regulatory requirements and plans for recovery of key species in the aquatic ecosystem. One particular question addressed herein, is whether rapid artificial changes in discharge due to hydropower generation influence the growth potential of juvenile white sturgeon. The methodology integrates a hydrodynamic model of the depths, velocity components and water quality parameters with field measurements of the exact location of an individual fish. These fish are monitored remotely for both position and metabolic rate (through sonic transmitters and electromyogram monitoring tags). This data will then be used to extract knowledge about the critical hydrologic conditions that cause fish to change habitat locations and the associated energy expenditure of fish due to rapidly changing flow conditions.
Keywords:
fish behavior simulation, data mining, biological modeling, white sturgeon
Determining the effects and impacts of hydro projects and their operations on aquatic species is generally a requirement in the United States by the Federal Energy Regulatory Commission (FERC) when utilities seek licenses or re-licenses of a hydro facility. Traditionally, an evaluation of the quantity of habitat has been used to address the questions and issues regarding habitat impacts. Instream Flow Assessments (IFA) can be an effective way to measure the quantity of habitat for a given flow and water quality regime (Bovee, 1982), and can give considerable insight to optimize the amount of aquatic habitat available. However, there are drawbacks to using IFA and habitat quantity as the sole decision criteria used to manage some river systems for aquatic species (Orth and Maughan 1982; Estes and Orsborn 1986; Orth 1987; Gore and Nestler 1988; Armour and Taylor 1991, Hardy 1998).
Some hydroelectric facilities are used to generate a relatively constant level of electricity over a small time scale relative to the watershed hydrology. Other hydro projects are used to ‘follow load’, and they do so by varying the project discharge such that it matches the electricity demand of customers. Discharge from the latter type of project can vary substantially within timeframes less than one hour, and the change can occur several times per day. An IFA can be used to evaluate the effect these operations have on habitat quantity under different management alternatives, but lend little understanding about the potential of the habitat to support aquatic life under this transitory flow regime. The focus of this paper is to present the development of a behavioral model being applied in concept to the Snake River for a key species – white sturgeon, Acipenser transmontanus.
The Snake River forms the border between western Idaho and Eastern Oregon in an area known as Hells Canyon (Figure 1). The Snake River above Hells Canyon is extensively developed for agriculture and hydropower. Immediately above Hells Canyon is a three dam hydro project known as the Hells Canyon Complex (HCC). The HCC provides energy for much of southern Idaho and eastern Oregon, and is used to follow variation of energy demand. As a result, the flows through the Hells Canyon reach are at times, quite variable throughout the period of a day. The Hells Canyon reach is approximately 160 km in length, and is characterized by a high gradient narrow channel that runs through a deeply incised canyon throughout much of its length. The lower approximately 70 km of the river are a lesser gradient and two large tributaries, the Salmon and Grande Ronde rivers, influence flows. The Snake River supports many native fish populations commonly associated with the Northwest, many of which are currently listed for protection under the United States Endangered Species Act (ESA) or are species of concern at the state level.
Fig. 1 Study area of the hells canyon reach of the snake river, Idaho/Oregon, USA.
The purpose of the behavioral model for white sturgeon is to provide a means of estimating the effects that various hydroelectric operating strategies have on the activity level of juvenile white sturgeon. The behavioral model is linked to operations of the upstream hydro project via fully hydrodynamic water quality models. The behavioral aspects of the model are based on movement and respiration data collected from individual fish within a study area of the Hells Canyon reach over three separate five-day monitoring periods. Data mining will be used to develop behavioral criteria that will allow simulation of movement of individual fish, and in turn, the respiration expenses associated with their movements. The behavioral model will allow the biological (respiration and activity level of individuals) implications of various river management strategies to be simulated over small special and temporal scales, and the results will provide decision support information to optimize hydro operations for white sturgeon. The behavioral model will also provide activity and respiration information to a bioenergetics model that is currently being developed for white sturgeon. By linking simulated activity and respiration information to a bioenergetics model, it will ultimately be feasible to predict the implications of various river management strategies on the growth potential of individual white sturgeon.
Overview
The overall goal of this project is to develop a modeling system that supports a comprehensive evaluation of the effect of various river management strategies on juvenile white sturgeon. The particular strategies to be evaluated are unsteady in nature (varying water depths and velocities), and variable water quality conditions (temperature, dissolved oxygen, and total dissolved gas). It is anticipated that this approach to behavioral modeling developed for juvenile white sturgeon could be applied to other fish species and river systems.
A combination of traditional approaches to aquatic modeling, detailed information about individual juvenile white sturgeon (movement and energetics), and data mining techniques will be used to fully address the goals of this project. At a fundamental level, river management strategies will be reduced to parameters that can be simulated spatially and temporally (e.g. depth, velocity, temperature, total dissolved gas, and dissolved oxygen) and represent environmental conditions available to an individual fish (or several individual fish). These parameters will be simulated for various river management strategies with a model capable of representing the complex habitat systems that sturgeon live in. The behavioral model will then use the simulated environmental components as physical boundary conditions available to individuals, and simulate how individuals may respond or behave under a set of conditions.
Compared to many other fish species, the available literature offers relatively little detailed information about white sturgeon, their habits, and energy balance within their environment. As a result, there is relatively little reference data available to directly base the development of a behavioral model upon. To provide the understanding necessary for simulation of individual juvenile white sturgeon, the movement behaviors and energetic expenditures of individual fish were monitored during three periods lasting five days each. The raw movement and energy data will be analyzed using data mining techniques to develop behavioral response functions relative to changes in environmental conditions. It should be noted that traditional modeling techniques will be used to define the processes and relationships that are well defined and understood (hydrodynamics and water quality). Data mining will be used to develop relationships for processes that are not well defined at this time and typically require that fundamental assumptions be made before an ecosystem can be simulated.
Fig. 2 Recorded locations of individual sturgeon during day and night periods, August 1999.
The behavioral response functions will allow the activities of individual fish to be simulated for flow conditions that result from various operational scenarios for the upstream dams. From the resulting movement information, corresponding energy expenditures will be determined for each individual that is simulated. The cumulative energy expenditure will be used for decision support in selecting operating scenario’s that limit adverse effects of hydro operations. In addition to the behavioral model, a bioenergetics model that will predict growth potential of individual white sturgeon is being developed. Simulated respiration and activity information from the behavioral model will be utilized by the bioenergetics model, which will simulate the growth potential of individual white sturgeon under different hydro operating scenarios.
Fig. 3 Recorded locations of individual sturgeon during a period of load following.
The foundation for the physical modeling system is a fully hydrodynamic two-dimensional finite difference model (2-d, depth averaged, Mike 21C, DHI). The model utilizes a curvilinear grid, making it suitable for riverine applications. The model is used to simulate cell depths and velocities (and water quality) throughout the study reach for steady or un-steady discharges that result from various river management strategies (Chandler and Parkinson, 2000). The boundary conditions for the 2-d model of the study area are provided by a one-dimensional hydrodynamic model (1-d, Mike 11, DHI). The 1-d model extends from the upstream dam throughout a 160 km reach downstream.
Fig. 4 Recorded locations of individual sturgeon during a period of high flow.
In order to gather information on individual fish over a wide range of environmental conditions, movements of individual sturgeon were monitored over multiple periods, each period representing a different river management strategy and flow regime. During two of the periods, the flows were held at a constant discharge, one at a low level and the other at a moderate level. During the third period, the flows resulted from an aggressive load following operation. For each flow trial and monitoring effort, several individual juvenile white sturgeon were fitted with sonic transmitters. The sonic transmitters allowed their location to be determined at a point in time. For each of the monitoring periods, the activity of each fish within the study area was monitored on a regular interval for five days. The results of the movement monitoring can be seen in Figures 2-5. Figure 2 illustrates the locations of individuals from a trial tracking effort during August 99. Figure 3, 4, and 5 show the locations of white sturgeon during the load following, high flow, and low flow periods, respectively. While it is difficult to track the specific movement of an individual from the Figures as shown, it is evident that the fish tend to show preference to some areas over others, and in a general sense, the areas of preference are similar under different flow conditions. Within those areas of preference, it is evident that individuals do move around, and some are more active than others.
Fig. 5 Recorded locations of individual sturgeon during a period of low flow.
Movement patterns alone may not allow a full assessment of behavioral responses of white sturgeon to different river management alternatives. A measure of energetic expenditure to either maintain positions or make movements may allow for the detection of a behavioral type of response that movement alone would not demonstrate. In order to evaluate energy expenditure, a portion of the sonic tagged white sturgeon in each flow period were fitted with electromyogram (EMG) radio-transmitters to determine metabolic ‘cost’ of a movement based on oxygen consumption. EMG transmitter probes are placed in red muscle tissue and detect electrical activity associated with muscle contraction. It has been shown that red muscle activity and oxygen consumption are positively correlated. Therefore the EMG signals represent an indirect measure of active metabolism (Adams and Breck 1990, Cech 1990).
EMG data were transmitted continuously from each tag. A series of underwater antenna’s and data loggers were used to collect a nearly continuous time series of energetic data from each individual as they moved throughout the study area. Due to the automated collection of the EMG data, there is much more energetic data available for each individual than location information.
In order to use the information obtained from the EMG tags, field calibration was necessary to relate EMG information to individual sturgeon at known swim speeds. Further laboratory work was conducted to establish relationship of swim speed to respiration rates. A closed respirometer was used to measure oxygen consumption rates over a range of swimming speeds and water temperatures. A model will be developed to describe the relationship between EMG pulse rates and swimming speed. This relationship will be used to assign respiration rates to the EMG readings that were measured from individuals in the field.
The position and energetic data of each individual fish will be merged with the detailed local environmental conditions (depth, velocity, temperature, dissolved oxygen, and total dissolved gas) provided from the hydrodynamic model. This combined data set will be ‘mined’ by conventional statistical methods and other more recent techniques such as genetic algorithms, genetic programming, artificial neural networks, and decision trees. The purpose is to understand the environmental inputs and critical thresholds that fish respond to. These relationships between changes in the local environmental conditions experienced by a fish and the resulting response or behavior will then be used to compare energetic expenditure associated with different river management strategies. It will also allow quantitative assessment and insight of when and why certain microhabitat areas are utilized.
The advantage of using the data mining techniques to develop behavioral relationships is that the raw data feed directly into and define the response or behavioral relationships between individual fish and their local environment. This approach will eliminate the need to make some of the broad assumptions that are required to use a more classical approach with individual energetics.
This paper describes an outline for a modeling system that will be used evaluate the effects of various hydro operation management strategies on juvenile white sturgeon. The measure of effect will be measured in respiration and its effect on activity rather than available habitat. The approach can be characterized as a hybrid of existing technology and modeling capabilities being integrated with some of the emerging data mining techniques. The data mining techniques will be used to provide relationships and understanding for processes where current ecological models require assumptions due to a lack of knowledge of the governing processes. Advantages of this hybrid approach include the ability to estimate how juvenile white sturgeon will behave and utilize habitat under various river management strategies.