Fish sampling design in river deltas
A range of practical cost benefit considerations and trade offs frequently drive the design and effectiveness of fish sampling in delta environments
Because recover of fish stocks commonly motivates restoration funding, fish sampling design is often a critical component of effectiveness monitoring at river delta restoration sites. The high variability of fish density among habitats, across a site, between years, among seasons, and between tides, combined with the difficulty of working in a delta environment, often forces tradeoffs in the intensity, duration, and extent of sampling. Poor sampling design combined with insufficient sampling can substantially undermine the value of monitoring.
- 1 Design
- 2 Logistics
No fish sampling strategy is reliable unless the hypotheses being tested are precisely defined. A set of adaptive management objectives for delta salmon utilization and delta vertebrate biodiversity are described elsewhere. Hypotheses commonly motivating sampling design for restoration can be generally sorted into two categories:
- Observations of fish population response to changes in the delta landscape - generally demanding both more extensive sampling across habitat types, as well as sufficient baseline monitoring.
- Observations of fish use of modified habitats - generally suggesting a Before-After Control-Impact study design.
These two scales are related, yet suggest very different approaches to sampling.
Effect of Site Scale on Design
At larger sites, the relative connectivity of sampling locations within a site may affect density, increasing variance among samples, making stratified random sampling very expensive. Smaller sites may allow for representative sampling allowing stronger inference. Beamer 2014 observes patterns of fish density in the Nisqually Delta consistent with findings in the Skagit Delta that suggest fish density is strongly affected by connectivity to the mainstem channel.
Representative vs. Fixed Site Sampling
Because of sampling costs, many programs rely on making comparisons of fixed sample sites, or 'index sites', rather than completing a randomized and therefore representative sampling of habitats. Specific assumptions are made about how these 'index sites' compare to one another and represent the delta landscape.
Fixed sample sites are often selected because they are accessable. Crews travelling by boat, and fish entering the system through main channels may have a similar perception of accessabiity, potentially biasing fixed site sampling toward more accessible portions of the delta. Because of this, fixed sites may provide a overestimate of the density of delta habitat utilization.
Continuous vs. Episodic Sampilng
Under limited budgets, it may be tempting to reduce sampling costs by conducting more intensive sampline but not every year. In this way the continuity of data from year to year is sacrificed to obtain a larger number of samples within a given year. If monitoring funding is unreliable, this episodic sampling may result unintentionally.
An advantage of continuous sampling is that you can measure salmonid density during both large and small juvenile outmigration years. The size of the outmigration affects fish density as well as distribution in the system [cite]. Comparison of large migration years to small migration years can indicate habitat preferences. Continuous sampling allows for observation of patterns from year to year. Episodic sampling, where samples are not taken consistently every year reduce the likelihood of viewing distribution patterns among different outmigrations, or under different river conditions. So if observing landscape use patterns is not an objective, the risk of episodic sampling is less. However, budget constraints are the only real motive for episodic sampling.
Consideration of Independant Variables
State and tribal agencies sometimes maintain smolt traps to estimate total salmonid outmigration on rivers. This provides an additional independant variable to support analysis of density samples within the delta, allowing for more robust comparison among years.
Enumeration of the Sample
Typically count by species is used as the dependant variable, either by completing an inventory of the catch, or by sub-sampling (by volume??) if the quantity of fish is overwhelming. The investigator them must decide how much effort to apply to describing the size structure of the population. Understanding of the size structure of samples from multiple sites can provide insight in how fish populations differentially use habitat based on size.
Labor costs of sampling vary drammatically based on travel time between sites. Beach seine sampling, which is the most common approach for sampling fish density within delta wetlands, requires a crew of three. At the Stillaguamish Delta a crew was able to sample three sites (with three net sets per site) within a work day. Sampling strategies (Rice 2011 DRAFT) recommend sampling every two weeks over the anticipated period of outmigration, which for Chinook Salmon may extend seven months from February to August. Thus an annual sampling effort may require approximately 15 person/days per sampling location, or at a mean rate of $25/hour, $3000 per sampling location per year, not including fixed costs, data management, or analysis.
Anticipating Budget Fluctuation
A delta monitoring strategy can anticipate budget fluctuation by designing a sampling scheme that can be increased when budgets allow for additional sampling. There are different approaches for expanding sampling.
- Expanding duration -
- Expanding frequency
- Expanding number of sites - increasing the number of sites can allow a program to switch from a fixed site approach to a representative site approach.
The following references provide the best regional foundation for designing fish sampling in deltas: