2.2.2.4 Examples of temporal designs

Oregon Coast Coho Spawner Monitoring ...click here

Aniak Chum Salmon Monitoring ...click here

Oregon coast coho spawner monitoring temporal design

1. Within a year temporal design

1. Biological characteristics
• Coho spawn in Oregon coastal streams October - January

• Life span of coho salmon in spawning streams is typically about 10-12 days (Willis 1954, Perrin and Irvine 1990).

2. Structure and magnitude of variability
• A graph of within year spawning timing (see below) shows that the spawning time can be quite variable and suggests that spawning surveys should be conducted at regular intervals throughout the spawning season:
  

  

  Enlarge Image

  
  
  
3. In-season management needs
• None
4. Cost constraints
• The information on average life span of coho when spawning suggests that cost-effective and adequate area-under-the-curve estimates of spawner abundance can be achieved by repeat visits to spawning survey sites every 10 days

2. Across years

1. Balancing spatial and temporal sampling
• The goal of the Oregon coast coho monitoring project is to provide statistically rigorous information on the status and trend of natural origin coho spawners in the Oregon Coast ESU.  There is equal importance placed on the status and trend components.  Conceptually, this means that the design needs to provide a balance between sampling new sites every year and revisiting the same sites every year.  As a result, a rotating panel design was adopted - see "c" below
2. Influence of variablity
• A graph of annual estimates of spawner abundance (see below) shows that there can be a 12 fold difference difference in spawner abundance between years.  This suggests that annual estimates are needed (as is reguired for management purposes).

Enlarge Image

1. Panels
• A rotating panel temporal design (see below) consisting of 14 panels (the vertical columns); rows indicate years, with row 1 the first year of the monitoring plan. The first panel consists of a set of sites visited every year (S₀). The last panel consists of a set of new sites selected each year from the pool of sites not selected for any of the other panels (S₄). Between these "bookend" panels are three sets of panels that make up a three-year rotating design, patterned after the three-year coho spawning cycle. These three sets are grouped as blocks. S₁₀ , S₂₀, and S₃₀ consist of a set of sites that would be visited every three years, with S₁₀ sites visited the first year, S₂₀ sites the second year, and S₃₀ sites the third year, then every three years thereafter. Within each of these three year panels is an additional set of sites that would be visited on a nine-year cycle (i.e. S₁₁, S₁₂, S₁₃, S₂₁, S₂₂, S₂₃, S₃₁, S₃₂ and S₃₃). 

Enlarge Image

•  This fairly complicated looking design is flexible and meets several needs. The allocation of sampling effort can be adjusted across panels, although an initial suggestion is that equal effort be allocated to each. Only shaded panels would be visited the year indicated. The total number of sites each year is the sum allocated to each of the panels, and is used for that year’s status estimate. The sites visited every year (S₀) provide good trend detection capability; with this allocation of sampling effort, 25 sites would be visited each year. Trend detection capability is augmented by the sets of sites making up the rotating panels (three-year and nine-year cycles). Finally, the new sets of sites (S₄) allow an expansion of the sampling effort by adding sites that would not be considered in the basic fixed and rotating panel design, improving overall representation of the resource of interest, and allowing for a buffer in the event that budgets change. Sample sites could be added or deleted from S₄ without markedly disturbing the trend detection capability of the basic design.

Aniak Chum Salmon Monitoring

 From 1992 to 2001, an average of 66 thousand chum salmon were harvested annually for subsistence purposes in the Kuskokwim area, making this salmon resource an extremely important source of food to people that live along this river. Because this is such a lager river, and because this area is so remote, the strength of the annual run must be indirectly inferred by means of a few assessment projects within the entire drainage. One of the most important of these projects is the DIDSON sonar project of the Aniak River, a tributary that connects with the Kuskokwim above most of the important commercial fisheries. In 1996, the project was designed to run a 24-hours per day to passage estimates using dual-beam sonar during the salmon migration period. In 2003, the project was re-designed to use DIDSON imaging sonar, and the new design called for sampling of time, to make the project more cost effective, rather then running the project continuously. 

With this project the biological population is the Aniak River stock of chum salmon, but the statistical population is quite different. In this case, the statistical population is time, and each day defines a first-stage sampling unit. Next, each day was further divided into three 4-hour time second-stage sampling units. In actual practice, time was further subdivided, but we don't need to consider all of the complexity of the analysis to understand the important features of the temporal design. Note that each sampling unit is associated with a corresponding sampling attribute--which is the number of chum salmon that swam past the sonar during that period. The 4-hour time periods were then systematically sampled, which means that three of the six 4-hour blocks were sampled each day (0000–0400, 0800– 1200, and 1600–2000 hours). Then the passage estimate of the day can be constructed by dividing the passage for the sampled period by the fraction of the day that was sampled. Here because half of the time was in the sample, the passage count is multiplied by two.  

See, McEwen, M.S. 2007. Sonar estimation of chum salmon passage in the Aniak River, 2005. Alaska Department of Fish and Game, Fishery Data Series No. 07-86, Anchorage.