Fy 2007 Innovative Project Solicitation Application of Innovative Acoustic Telemetry Technology to Underpin Statistically-Valid Survival Estimates for Chinook Salmon in the Nearshore Ocean Off the Mouth of the Columbia River

FINAL, 5/18/07

Section 10: Narrative

The purpose of this study is to establish innovative methods that can enhance and expand into the nearshore ocean estimates of juvenile salmonid survival and migration patterns, especially for subyearling Chinook salmon (Oncorhynchus tshawytscha). The study objectives are to: 1) extend the geographic range for estimating smolt survival through the estuary to near the mouth of the Columbia River; 2) investigate key assumptions and logistical issues to support future studies designed to obtain robust survival estimates during early ocean residence; and 3) describe smolt migration characteristics, including residence time in and exit points from the study area within 20 km off the mouth of the Columbia River (MCR). The innovation is the use of acoustic telemetry technology using the smallest acoustic tag available (0.35 g in water) and large sample sizes (>20,000 tagged fish total) capable of reliably monitoring the migration patterns of subyearling and yearling Chinook salmon in the nearshore ocean. This is not being done for subyearlings by others because their acoustic tags are too large to implant in subyearling Chinook salmon without adverse tag effects. The JSATS (Juvenile Salmon Acoustic Telemetry System) technology we propose can successfully be used to tag fish as small as 95 mm without deleterious impacts to behavior, growth, and survival. When coupled with quantitative survival data for transported and in-river migrants through the hydrosystem and estuary, smolt survival data off the mouth of the Columbia River from studies using this one as a springboard will lead to improved fish protection strategies in the Federal Columbia River Power System that will result in direct improvements in smolt survival.

The Independent Scientific Advisory Board recently recommended extending the geographic range over which smolt survival can be estimated from the hydrosystem downstream and into the ocean. The implication is that survival through the estuary and early ocean residence are important indices contributing to the formulation of any life cycle-type model. This proposal addresses this aspect directly to support scientifically defensible estimates of smolt survival in the ocean.

Twenty autonomous acoustic nodes will be placed in the ocean in three 20-km arrays configured to box the core of the Columbia River plume and outer areas of the MCR to sample tagged fish during early ocean entry from May to September 2008. An immediate, direct benefit from this project will be to extend the current downstream limit for smolt survival estimation from near the East Sand Island to the Columbia River bar between the jetties at the mouth of the river. Future statistically rigorous estimates of smolt survival in the ocean and management application of these estimates will benefit from this innovative project. The project will rely on cooperation and support of the U.S. Army Corps of Engineers, Portland District. The project team will be led by Drs. Carlson (acoustician, Pacific Northwest National Laboratory) and Skalski (statistician, University of Washington).

After over 15 years of attention in the Northwest Power and Conservation Council’s Fish and Wildlife (F&W) Program and the National Marine Fisheries Service’s Biological Opinions on operation of the Federal Columbia River Power System (FCRPS), the survival of subyearling and yearling Chinook salmon through the mainstem dams is still a major concern (National Marine Fisheries Service 2004). Measures, such as flow augmentation, voluntary spill, barging, turbine intake screens, and surface flow outlets, to protect juvenile salmonids during their downstream migration have improved survival rates in the hydrosystem, but there is still substantial mortality downstream of Bonneville Dam as indicated by the number of returning adults relative to the number of exiting smolts. One hypothesis is that a portion of this mortality is due to latent effects from passage through the hydrosystem, as opposed to natural abiotic and biotic environmental conditions in tidal freshwater and saltwater (Independent Scientific Advisory Board 2007). Efforts to understand and model latent mortality are underway for the new FCRPS Biological Opinion, and will undoubtedly be part of the Council’s forthcoming F&W Program amendment process. The problem is these management decisions lack basic data on the survival of juvenile salmonids in the lower river, estuary, and nearshore ocean¹.

In their recent treatise on latent mortality (Independent Scientific Advisory Board 2007), the ISAB recommended that investigators focus on extending the geographic range over which smolt survival can be estimated. They said, “Future monitoring and research is needed to further quantify biological factors that contribute to variability in estimated post-Bonneville mortality. In particular, the ISAB recommends that acoustic tags continue to be developed and used to assess and partition mortality in the lower river, the estuary, and the Pacific Ocean shelf.” The implication being that survival through the estuary and early ocean residence are important indices contributing to the formulation of any life cycle-type model. This proposal addresses this aspect directly.

B.2 Background

B.2.1 The Opportunity: In 2008, in excess of 20,000 fish tagged with JSATS transmitters will be released in the Columbia basin. This proposed innovative project would take advantage of that large pool of tagged fish. Species coverage would include both yearling and subyearling Chinook salmon. Currently survival estimates are reported for the spring- and summer-migrating fish, but only through a portion of the lower Columbia River and estuary to East Sand Island (river kilometer 8.3). Our study would extend those estimates through the entire lower river and estuary to the Columbia River bar (river kilometer 2.8). Furthermore, the inclusion of subyearling Chinook salmon will provide the first opportunity to study migration of these fish during early ocean entry. Migration characteristics of smolts once they enter the marine environment will dictate our ability to obtain robust survival estimates in nearshore and offshore waters. Identifying relationships between migration patterns and oceanographic conditions will inform decisions on locations for future acoustic telemetry receiving arrays.

B.2.2 Previous Work on Juvenile Salmon Migration near the MCR: Numerous articles and reports support study of juvenile salmon in the nearshore ocean and application of the data to fisheries management in the Columbia basin. Schiewe et al. (1989) and Francis et al. (1989) designed studies to examine the behavior and survival, respectively, of juvenile salmonids in the estuary and nearshore ocean. Casillas (1999) advocated consideration of ocean conditions and survival in fisheries management decision-making for the Columbia basin. Brodeur et al. (2000) recommended research on the distribution and movement patterns of juvenile salmon in marine waters, and noted that lack of understanding of survival in the marine phase hampers fisheries management. Researchers are currently studying smolt migration characteristics and ecological relationships in the nearshore ocean (see section on Relationships to Other Projects), but none are estimating smolt survival using statistically-valid methods.

Juvenile salmonid density in the nearshore ocean off the MCR was variable among and within years 1998-2003 (Emmett et al. 2006). These authors noted that subyearling Chinook salmon abundance tended to peak in July and August. The spatial distribution of juvenile salmonid horizontally was also variable. Although catch numbers were highest at stations closest to shore, distance from land was not a significant independent variable in their analyses after year was accounted for (Emmett et al. 2006). Juvenile salmon off the Oregon coast apparently do not actively migrate vertically (Pearcy and Fisher 1988). They are typically distributed in the upper 12 m of the water column, maintaining their distribution in the upper water column after entering the ocean (Emmett et al. 2004). Fisher and Pearcy (1988) concluded that environmental conditions during early ocean entry for coho salmon strongly affected their survival. Emmett et al. (2006; p.17) stated that “…marine distribution [of juvenile salmon migrating from Columbia River] was influenced by the shape, direction, and strength of the Columbia River plume.” Brodeur et al. (2003) summarize research conducted on juvenile Pacific salmon in the ocean.

Several researchers have applied telemetry techniques to study juvenile salmonid movements in the Columbia River estuary and MCR. Using hydrophones arrayed in two 5-km lines directly off the north and south jetties at the MCR during 2004, Schreck et al. (2005) studied the early ocean migration of acoustically-tagged spring/summer (yearling) Chinook salmon. They used a mooring apparatus to deploy acoustic receivers, as described by Clements et al. (2005). Of the total number of yearling fish detected at both arrays combined, 88% (275 of 313) were last detected on the north jetty array. These researchers also studied steelhead during 2001-2003 and observed a similar pattern. Schreck et al. (2005) caution that these results should be used carefully as it is not known where the fish went after they were last detected. McMichael et al. (2006) deployed acoustic receivers near the north jetty repair site in 2005. They found that subyearling Chinook salmon were more likely than steelhead or yearling Chinook salmon to enter areas near the jetty; the latter fish were generally oriented to the main river channel. In another study, radio telemetry data indicated that smolt mortality is greater in estuarine waters than tidal freshwater of the Columbia River (Schreck et al. 2006).

B.2.3 Survival Estimation: Estimation of salmonid survival in the Columbia River basin is usually performed with the Single Release (SR) model (Skalski et al. 1998), based on the Cormack-Jolly-Seber (CJS; Cormack 1964; Jolly 1965; Seber 1965) model. These models are statistical release-recapture models that estimate survival between tag detection sites, using the assumption that all smolts travel in the same direction and have common survival and detection probabilities. In the past, most detection data for these models have been collected at dams from PIT tags.

There are several key differences between analyzing river survival with PIT-tag data and analyzing ocean survival with acoustic-tag data. First, acoustic-tag models must adjust for tag life. This can be done by incorporating a tag-life parameter into detection probabilities, as in Townsend et al. (2006). A second difference between river and ocean survival statistical models is that the river survival models are designed for the mostly unidirectional migrations within the river systems, while the ocean survival model must account for less restricted movement in the ocean (Skalski 2006). If it were known that all smolts follow the same migration path once they enter the ocean, then the SR or CJS model could be easily adapted to estimate ocean survival from acoustic-tag data by incorporating tag life. If some smolts migrate along the shore while others migrate directly out to the open ocean, however, then the SR and CJS models will produce negatively biased estimates of ocean survival. The appropriate ocean survival model will need to take into account the choice of direction (i.e., north, south, or directly out) faced by smolts when they enter the ocean environment. Single acoustic arrays monitoring the three direction choices will provide estimates of the joint probability of direction choice, survival, and detection. It will be possible to monitor relative movement and survival over time. These results will assist in developing future ocean survival statistical models.

B.3 Study Area Description

The study area extends from the mouth of the Columbia River 10 km to the north and south and 20 km off the MCR (Figure 1). Water depth in the area ranges from 20 to 100 m. This area is within the California Current System and is strongly influenced by large-scale oceanic processes including baroclinic and tidal circulation (Hickey and Banas 2003). Winds from the north tend to occur at variable levels from April through August along the Pacific Northwest coast. Northerly winds cause southward currents and upwelling of shelf water while southerly winds cause northward currents and downwelling. The change in April or so from southerly to northerly winds and associated oceanographic conditions is called the spring transition.

The Columbia River plume is the dominant hydrologic feature of the study area. The depth of the plume with its low salinity water is generally less than 20 m, depending on prevailing oceanographic conditions and the amount of Columbia River discharge. The plume is typically oriented to the south and disassociated from the coastline during summer (Thomas and Weatherbee 2006), whereas the opposite is typically the case during winter. Transitions during spring and fall can be relatively sudden (Thomas and Weatherbee 2006). Local winds can also affect the shape and location of the plume. The study area is, to say the least, a very dynamic environment.

C. Rationale and Significance to the Council’s Fish and Wildlife Program

The Independent Scientific Review Panel (ISRP) stated that “effective adaptive management requires the existence of monitoring data…based on mathematical and statistical procedures” (ISRP 2005). The ISRP also noted “…that it is difficult to imagine…adaptive management without consistent, unbiased monitoring…” Successful management of salmonid and hydro resources and the recovery of salmonid stocks in the Pacific Northwest depends on reliable information. Adaptive management can only be as successful as the information used in decision making is accurate and precise. In the Columbia River Basin, much of that information comes from salmonid tagging studies. In addition, assessing compliance with F&W Program and Biological Opinion goals depends heavily on salmonid survival information from tagging studies. This innovative project is addressing this need for the nearshore ocean 20 km off the MCR.

The 2000 Fish and Wildlife Program (Council Document 2000-19) included a primary strategy called Ocean Conditions: “Identify the effects of ocean conditions on anadromous fish and use this information to evaluate and adjust inland actions.” Based on text in the 2000 Program accompanying this strategy, the Council clearly understands that knowledge of salmonid survival rates and associated environmental conditions during their initial phase in the nearshore ocean, which this innovative project will lead to, will help them assess the relative value and prioritization of measures taken in freshwater habitats. This project will also help meet the Council’s desire to “separate the effects of ocean-related mortality from that caused in the freshwater part of the life cycle” because the project will support the effort to estimate survival in the early ocean entry component of the life cycle.

The 2003 Mainstem Amendments to the Council’s Fish and Wildlife Program (Council Document 2003-11) included a specific objective to “…evaluate the impact of ocean conditions on…smolt-to-adult survival rate objectives.” The proposed project will address this objective by leading to an analytical method to separate the component of survival in the nearshore ocean immediately off the Columbia River mouth.

The 2000 and 2004 Biological Opinions on FCRPS Operations called for research on juvenile salmonid ecology and survival in the plume. The upcoming 2007 FCRPS Biological Opinion will also likely include plume and ocean mandates. The proposed project applies to the FCRPS Biological Opinions because the project addresses survival in the nearshore ocean.

There are several salmon ecology and acoustic telemetry projects that relate directly to our innovative project. The National Marine Fisheries Service (NMFS) is undertaking a USACE project (EST-P-02-03) to examine the relationship between conditions in the nearshore ocean off the MCR during initial entry by juvenile salmon and subsequent adult returns (Muir and Emmett 2004). These researchers are waiting for additional adult return data before completing their analyses. The results from this complementary study will be used to inform our study. The NMFS and collaborators are performing a major study (BPA 199801400) of juvenile salmon and associated environmental conditions in the Columbia River plume that includes surface trawling stations in the general vicinity of our study area. Species composition and catch data from the trawls will be useful to interpret our data. Vice versa, our data on the relationship between migration patterns and ocean conditions will be applicable to ecological modeling efforts in NMFS’s plume study. An acoustic telemetry study (BPA 200311400) of juvenile salmonids is ongoing with receiving arrays in the main stem Columbia River, estuary, and nearshore ocean from California to Alaska. This study will provide data on migration characteristics for yearling fish; however, it is not designed to estimate survival in a quantitative manner, nor is it investigating subyearling Chinook salmon.

This innovative proposal is intertwined with other JSATS acoustic telemetry research in the Columbia River and estuary. Tagged fish for our study will be available because of the objectives of USACE studies within the Anadromous Fish Evaluation Program to investigate acoustic tag effects (SPE-P-06-02), estimate smolt survival at the Bonneville Dam spillway (SPE-P-02-02), and estimate smolt survival in the lower river and estuary, i.e., post-Federal Columbia River Power System (EST-P-02-01). In fact, our acoustic deployment off the MCR will extend the survival estimation capability from East Sand Island to the Columbia River bar when the data are integrated with data from the acoustic receiving arrays deployed for the post-FCRPS study.

The innovative project we propose is dependent on in-kind contributions of acoustic receiving nodes from the USACE, in addition to tagged fish as explained above. This project benefits the USACE because of their responsibilities for jetty repair and shipping channel maintenance. Thus, in reality, the BPA and USACE (in-kind support) would jointly fund this innovative project.

The purpose of this study is to establish innovative methods that can enhance and expand estimates of juvenile salmonid survival, especially for subyearling Chinook salmon (Oncorhynchus tshawytscha). The study objectives are to: 1) extend the geographic range for estimating smolt survival through the estuary to near the mouth of the Columbia River; 2) investigate key assumptions and logistical issues to support future studies designed to obtain robust survival estimates during early ocean residence; and 3) describe smolt migration characteristics, including residence time in and exit points from the study area within 20 km off the mouth of the Columbia River.

Methods are described below for five work elements: collect/validate data, analyze/interpret data, prepare an annual report, produce PISCES status reports, and manage the project.

E.1. Work Element: Collect/Validate Data (WE157)

E.1.1 General Approach: This study proposes to apply state-of-the-art acoustic telemetry technology and statistical techniques for quantitative survival estimation. The ability to extend survival estimates into early ocean residence requires that sampling capabilities in the nearshore environs be clearly understood. Thus, logistical constraints and opportunities must be explored and tested in the field, and detection capabilities of acoustic systems demonstrated. This is a key, underlying component of this study. Given the opportunity of tens of thousands of tagged fish from other studies that will be migrating out of the MCR, the strategy is to collect a focused set of acoustic telemetry data on fish migration that will support statistical design of future survival studies for the nearshore ocean. That is, the general approach is the extend into the ocean the statistically-valid survival estimates for salmonid smolts that are ongoing using JSATS in the lower river and estuary.

E.1.2 Description of JSATS: A basic acoustic telemetry system consists of a tag (the transmitter), a hydrophone (the receiving transducer), a signal processor (the receiver), and processing and analysis software. This simple system can be used to detect the presence of a tagged animal in an area of interest. The micro-acoustic tags used in this study transmit 417 kHz sound. The equipment we propose to use is called JSATS (Juvenile Salmon Acoustic Telemetry System; for more information, see Section G, Facilities and Equipment).

E.1.3 Tag Effects: Extensive study of the effects of JSATS tags on juvenile salmon has occurred in the last two years and is ongoing to thoroughly examine this critical issue (USACE project SPE-P-06-02). Hockersmith et al. (2007; p. v) stated, “The minimum fish length at which surgical implantation of a JSATS transmitter and a PIT tag did not negatively influence the growth (change in weight) of juvenile Chinook salmon was 88.3 mm (95% confidence intervals 80 to 97 mm). The minimum fish length at which surgical implantation of a JSATS transmitter and a PIT tag did not negatively influence survival was 95 mm fork length. For fish used during this study, this is the equivalent to a burden (combined burden of acoustic transmitter and PIT tag) of approximately 7.6% for a 95 mm fish which would weigh approximately 9.2 g. Acoustic transmitter expulsion was only observed in subyearling fish under 108 mm (tag burden > 4.8%).” Tag effects research is ongoing in 2007; the work proposed herein will incorporate the findings from this critical research.

Figure 2. Time to failure for tag-life studies for 30-day (top) and 60-d (bottom) tags. A survivoirship curve (blue line) was fitted to empirical measurements (black dots). (From Ploskey et al. 2006.)

To deploy the autonomous nodes, all rigging and equipment components will be assembled and loaded onto the 31-ft deployment vessel. Deployment locations will be plotted on an electronic chart, and navigated to using GPS. Just prior to deployment the assembly will be attached to an anchor, and pertinent information will be recorded on a data sheet (node serial number, acoustic release code, water depth, date, and time of deployment.). Once the boat is in position, two people will hoist the anchor to the gunwale and lower it over the side, along with the rest of the apparatus. At this time, the GPS point will be recorded.

In the deeper near ocean environment of the Columbia River plume for message lengths on the order of 0.8 ms, multipath will not going to be the main limit for transmitter detection range. Range will be limited in the plume by transmission losses which are a function of sound frequency, range, and water salinity and temperature (Figure 4). Detection range performance in the plume is expected to be similar to that observed for JSATS deployments in previous years in the mouth of the Columbia River. At lower frequencies while transmission loss resulting from absorption are less background noise levels are higher.

In addition to the factors given above the detection performance of JSATS nodes may also be somewhat affected by the presence of a halocline depending upon the gradient through the halocline in the transition from an upper layer of fresher water to salt water. At angles of incidence of about 45° or less a sound signal will pass through the halocline. At greater angles of incidence the probability that sound incident on the halocline will be reflected in part or totally increases. Thus the detectability of a juvenile salmon will in part be a function of its location relative to a halocline as well as the characteristics of the halocline.

We expect to deploy JSATS nodes in salt water below any halocline. Tagged juveniles that are within or below the halocline will have detectability very similar to that observed for tagged juveniles observed using JSATS nodes in the mouth of the Columbia River during flood tides. Fish in a fresher water layer above a halocline will have higher detectability when passing through the upper segment a cone shaped region centered on a JSATS node with a full angle of approximately 90º.

E.1.9 Sampling Schedule: Juvenile salmonids will be tagged and released at Bonneville Dam and other sites in the Columbia River from late April through July. Accordingly, we will deploy the nearshore ocean acoustic nodes during late April or early May 2008. We will remove the nodes in late Sept. 2008.

E.1.10 Node Servicing: To recover the equipment, we will navigate to the GPS position and trigger the acoustic release. When the equipment comes to the surface, we will hook the line with a boat hook and bring the equipment on board. The autonomous nodes be serviced about every 30 days, weather permitting. During servicing, batteries will be replaced and data downloaded. The nodes will have a 60-d battery life.

E.1.11 Ancillary Data: Various ancillary data will be collected from external sources. 1) upwelling data at 45^oN, 125^oW from the NOAA/NMFS Pacific Fisheries Environmental Laboratory (www.pfeg.noaa.gov); 2) temperature and salinity data from the CORIE ocean bouy; 3) Columbia River discharge data from the USGS from their Beaver Terminal monitoring station at rkm 87 (waterdata.usgs.gov/nwis/nwis/).

E.2. Work Element: Analyze/Interpret Data (WE162)

E.2.1 Data Processing: Data collected by the autonomous nodes will be recorded as text files on Compact Flash cards. Physical data (date, time, pressure, water temperature, tilt, and battery voltage) are written to file every 15 seconds. Valid detection data will be recorded on the Flash media as they are received. Detection data will include individual transmitter code, time stamp, receive signal strength indicator (RSSI), and a calculated measure of background noise (RxThreshold). The data file will be transferred to a laptop computer following recovery from the node following servicing or retrieval events.

Data files from all nodes will be coded with the node location and stored in a database developed specifically for storing and processing acoustic telemetry data. To filter out ‘false positives’ (detections of TagIDs that did not meet criteria to be considered a valid detection), a post-processing program will be implemented. This program is comprised of a sequence of steps that included comparing each detection to a list of tags actually released (only tags that were released will be kept), then comparing the detection date to the release date (only tags detected after they were released will be kept). Finally, a minimum of four detections in a time-window dictated by the pulse rate interval of the tag will be required. The time spacing between detections will be analyzed and only detection events with the correct time spacing will be kept in the valid detection file.

From the valid detection file, a detection history will be created for each fish. Detection histories will be analyzed to determine the relationships between detections and tides, nearshore study area exit distribution, and travel time from point of release to point of detection for each release group. In addition, detection histories will be used as a secondary array for the JSATS array deployed between the jetties in the MCR, allowing for survival estimation further into seawater than is currently possible.

To evaluate relationships between detections and tides, a count of detections will be made over five minute intervals. Using the tide generating software WXTIDE32 (http://www.wxtide32.com/), we will produce tide elevation plots for periods during which tagged fish are migrating past the primary detection array. Counts of detections will be then plotted against the change in tide.

Array exit distribution will be determined separately for yearling and subyearling fish by plotting valid tag observations at each node location. From this, the number of valid codes observed at each location will be calculated by year class for all release groups combined. These distribution data sets will also be analyzed for temporal trends and relations with oceanographic conditions.

Arrival times will be defined as the first observation (detection) of each fish observed on an array. A count of fish for each hour (independent of day or night) will be then plotted. Day will be considered to begin half an hour before sunrise and end half an hour after sunset.