Early Marine Survival, Movement, and Habitat Use of Juvenile Steelhead Trout (Oncorhynchus mykiss) as Determined by an Acoustic Array: Quatsino Sound, British Columbia David W. Welch*1,3, Bruce R. Ward2, and Sonia D. Batten3
2 B.C. Fisheries Research and Development, 2204 Main Mall, University of British Columbia, Vancouver, B.C. V6T 1Z4 Canada
3 Kintama Research Inc., 321-2815 Departure Bay Rd, Nanaimo, B.C. V9S 5P4
LaDouceur & Jacobs?
* Corresponding author
A pilot study was conducted to assess the performance of an acoustic tracking array in a large marine system (45x42 km). An array consisting of six separate acoustic detection lines was placed on the seabed of Quatsino Sound, for two months. Forty steelhead smolts were surgically implanted with long-lived acoustic tags in the Waukwass River, and 31 subsequently entered the ocean from the river. Tracking success of the individual acoustic lines forming the array approached 100%. The array was successful in describing the movements, habitat use, and fate of all tagged smolts entering the ocean. Twenty-seven smolts moved through the fjord system to the open shelf using a variety of routes, and did not return. Four smolts remained within Quatsino Sound, two of which were detected moving over the inner lines of the array 1-2 days before the array was recovered, and two were detected moving up a long blind inlet, but did not return. Early marine survival in Quatsino Sound was either 94% or 100%, contradicting frequent assumptions that this period is a critical bottleneck determining recruitment. A permanent west coast array would provide large benefits to our understanding of marine animals occupying the continental shelf and slope regions.
The early ocean life history of Pacific salmon smolts is the least understood phase of the complex life history of salmon, chiefly because of the difficulty of studying the movement patterns of individual smolts. Although there are now several research programs that have begun to study the ocean biology of Pacific salmon based on newly-developed large epi-pelagic trawls (review by Brodeur et al 2000?), there are significant limitations on the interpretation of data collected from research programs that capture (and kill) fish at sea.
Because capture almost invariably involves the death of the animal, the detailed behaviour that allowed the animal to move to a specific geographic position in the ocean is missing. Thus, although sampling of fish caught at sea provides important information on the size, energetic status, and feeding success of the collected animals as a group, interpretation of such data is necessarily limited by the need to make some sort of assumption that the animals were static, and did not make significant movements between study regions. A second common assumption is that the stock composition of a sample of fish does not vary over time, so that differences evident over time can be interpreted as temporal changes in the characteristics of the same group of fish, rather than changes brought about by differential movements of different populations of the same species.
Such assumptions may be unwarranted. Given the remarkable ability of Pacific salmon to return precisely to the vicinity of their own birthplace (e.g. Quinn 1993; Quinn et al 1999), the marine feeding grounds and ocean migration pathways taken to get to them may be as population-specific as their rivers of origin, but as yet undiscovered. Although there is, as yet, little specific evidence for sharply defined and population specific movements in the ocean, the evidence for remarkable population-specific migration behaviours is known in both birds (e.g. xxx) and insects (e.g. Monarch butterflies xxx). It seems unlikely that marine fish, with their evolutionarily ancient lineage and the vast size of the oceans, would not have developed migratory abilities at least as sophisticated as those of birds and insects.
Accepting the possibility that marine animals such as Pacific salmon may shuttle between two postal addresses– their well-defined and long accepted freshwater spawning grounds, and their as yet undiscovered ocean migration pathways and marine feeding grounds, project "POST"– the Pacific Ocean Salmon Tracking project, was initiated.
Where do salmon go? What do they do when they get there? How do they return to spawn in their home rivers? How do changes in the ocean environment affect their survival? Underlying these questions is the belief that salmon may have "two zip codes"– or postal addresses– homing not only back to their rivers of origin, but also to specific feeding grounds in the ocean.
The marine ecology of the North Pacific can be broken down into three fundamental ecological zones– the pelagic offshore overlying the abyssal plains (water depths of 3-4 kms), the continental shelf (depths 200m), and the narrow continental slope region separating the two (depths ranging from 200m at the shelf edge down to the abyssal plain). As water depths increase rapidly in the slope region, with an average gradient of just over 4º in most regions of the world (Emery 1980, Khan 2000, Wiseman and Ovey 1953), the 1,000m isobath is typically found only some 10-11 kms seaward of the edge of the continental shelf. Shoreward of the shelf edge, the shallow shelf region can be very wide in many parts of the world’s oceans. However, off the West Coast of North and South America the shelf is frequently only 15-30 kms wide, making this area one of the narrowest (and longest) continental margins in the world. Because most marine animals consistently occupy specific depth zones, efforts to monitor the movements of animals remaining in the shelf or slope water regions are perhaps simplest to implement off the West Coast of the Americas. Partly for this reason, the shelf tracking component of POST is focussed in the Pacific.
Pacific salmon provide an excellent prototype organism for studying marine movements because there is great social and economic interest in these animals and they occupy both shelf and offshore regions of the North Pacific for extended periods of time. The oceanic phase of the life history is also vastly understudied relative to the great body of freshwater research that has been undertaken in the past. In general, the migratory movements of all species of juvenile Pacific salmon (excluding steelhead) are confined to the shelf-slope region regions for many months, and some stocks eventually migrate to the open ocean after reaching the Aleutians, while other stocks appear to take up permanent residence on the shelf (Hartt and Dell, 1986; Welch, in prep.). Once past their first year of life in the ocean, most species of Pacific salmon take up a pelagic life style in the offshore, while two species (coho and chinook) appear to have both shelf-resident and offshore variants (Groot and Margolis 1991).
As the marine movements of individual salmon are poorly understood, a concerted research program using new electronic tagging technology offers the opportunity to make major breakthroughs in our understanding of how salmon use the ocean: where they go, how they use the structure of the ocean environment to accomplish these migrations, and what they experience when they reach their marine feeding grounds. POST is intended to address a number of major research questions, whose eventual resolution would likely contribute significantly to the improved management and conservation of Pacific salmon:
Understand how Pacific salmon use the ocean environment
Identify distribution and habitat use by key species and life stages.
Identify important oceanic features & critical habitats
Do salmon respond to sharp thermal boundaries?
Do salmon depend upon specific ocean structural features?
Do salmon use common migration pathways?
Examine the coupling between biology and the physical environment– How do different species respond to changing ocean conditions?
Establish latitudinal patterns of movement and habitat utilization by steelhead
Establish whether Pacific salmon have "two zip codes", adding to the marine life history information that is already widely known and accepted about the freshwater phase of the life history
Determine how longer-term changes in ocean conditions relate to changes in fish condition, growth, survival, and distribution
Materials and Methods Study Location
Quatsino Sound (50.47ºN; 127.94ºW) is an extensive marine fjord located in northwest Vancouver Island (Fig. 1). In addition to a long (32 km) east-west oriented fjord providing access to the narrow continental shelf lying off the west coast of Vancouver Island, and then the open ocean, the fjord system contains two blind arms, Holberg Inlet (45 kms) at the interior end, and Neroutsos Inlet (21 kms) lying just west and south of Quatsino Narrows, a narrowing of the fjord where the semi-diurnal (?) tidal currents reach xx km/hr (Coastal Pilot or Thompson’s book?).
The Waukwass River (50.58ºN; 127.41ºW) drains into the southern end of Holberg Inlet, at the farthest interior end of the fjord system. (The interior inlet south of Quatsino Narrows is formally known as Rupert Inlet, but we have referred to the entire interior area of the fjord as Holberg Inlet for simplicity). The Waukwass River drains an area of ?? kms, and forms the adjacent watershed to the Keogh River. Both watersheds are located on the same mountain, and lie in the same bioclimatic zone (ref?).
In common with all west coast Vancouver Island steelhead stocks, the marine survival of Waukwass River steelhead is high, and the status of this population is excellent. Despite the near-proximity of the Keogh River steelhead population to the Waukwass River—the river mouths lie only 11 kms away apart, but on opposite sides of the northern tip of Vancouver Island, marine survival of all steelhead stocks on the east coast of Vancouver Island dropped dramatically in the 1990s, reducing the number of adults returning to these systems to a few dozen adults, as compared to several thousand adults in the late 1980s (Ward et al ??; Welch et al 2001). A secondary objective of the Census of Marine Life project was to compare the initial marine movements of these two populations because of their radically different marine survivals. In this paper we report on the early marine movements of Quatsino Sound steelhead smolts.
A total of six acoustic detection lines were placed in Quatsino Sound on May 22nd, prior to the surgical implantation and release of steelhead smolts with acoustic tags in the Waukwass River (Fig. 1). Each line consisted of either two or three Vemco VR-2 acoustic receivers placed on the seabed in a line perpendicular to the local axis of the inlet. The paired acoustic lines E-F and G-H were spaced approximately 10 kms apart in order to establish speed and direction of movement for any tagged smolts present in upper Holberg Inlet or the outer reach of Quatsino Sound. Paired listening lines were not placed in Neroutsos Inlet because of limitated amount of equipment available.
Acoustic receivers were deployed by attaching them to a groundline, using technology similar to that used in commercial fishing. Receivers were held about 0.5 m above the seabed with the hydrophone oriented upwards by attaching them to a vertical line connected to the groundline and with pressure-resistant floatation attached to the upper end to maintain the receiver with the hydrophone pointed towards the surface. The flotation was located approximately 4 m above the hydrophone in order to form only a small acoustic shadow (ca. 3 º). In cases where water depths exceeded 200 m (the maximum rated depth of the receivers) the length of the vertical risers was increased to keep the receiver above 200m.
The entire array was placed on the seabed, with no floats or equipment visible from the surface. Individual lines were recovered either by triggering an acoustic release on one end of the line, or by grappling for the groundline. This was done partly to ensure that equipment would not be molested during the nearly two month deployment, but also because an important design principle for the continental-scale array to be built is that is must be placed on the seabed, where it will not constitute a hazard to navigation or to commercial fishing operations. Although the eventual design of the full array will use different deployment methodologies, the current Census of Marine Life pilot project was intended to adhere to this design philosophy.
Spacing between receivers varied with the local geography and the number of receivers used. In all cases spacing between adjacent VR2 receivers was 850m or less. The greatest measured separation from a receiver at the end of an acoustic listening line to the shoreline was 328 m.
The Vemco VR-2 acoustic receiver is an autonomous battery powered underwater unit capable of long-term placement in the ocean (up to 15 months using a single lithium “D”-cell battery). Receivers were programmed with a special purpose internal code map which would detect acoustically coded tags manufactured to our specification to ensure that they were unique to our study. Acoustic tags from studies other than our own would not be recorded by our receivers; conversely, our acoustic tags would not be recorded by any other investigation’s receivers. We chose this approach because the continental-scale array planned for future deployment can potentially monitor 256,000 unique codes at the same time. As modern acoustic tags can have lifespans of months or years, it is critical for the integrity of future scientific studies that the uniqueness of a specific acoustic code be protected. Our study was designed to adhere to this principle, by using a unique code map which we could be certain was not in use elsewhere.
Steelhead smolts were surgically implanted with either a Vemco V8SC-2L or –6L acoustic tag (28 mm and 20 mm long, by 9 mm diameter). Each tag was programmed to broadcast its code randomly at an average time interval of 60 seconds, and actual transmission intervals distributed uniformly in the 30-90 second range. Projected lifespans for this programming choice was 134 and 465 days for the 6L and 2L tags, respectively. The larger V8SC-2L tags with their 15 month lifespans were superfluous for the purposes of this two month study, but we chose to use these tags as well to examine whether that these very long-lived tags could be used with the size of wild smolts encountered in the field study.
Wild smolts were collected from two rotary screw traps placed in the Waukwass River, several hundred meters upstream from the river mouth and 3.1 kms from the first acoustic detection line (RM) lying off the river mouth. (All distances quoted in this paper represent minimum feasible straight line paths between two points). Surgical protocols are described in more detail elsewhere (Welch et al In Press). Briefly, smolts were anaesthetised using clove oil and surgically implanted on the river bank at the collection point using a specially built portable battery powered surgical table. The main deviation from our previous protocol was the use of oxytetracycline, a broad spectrum antibiotic, at a dosage of 20 mg/g body weight, which was injected into the body cavity through the incision prior to insertion of the tag. Smolts were released in groups of 5 or 6 back into the river at the surgery site during daylight hours after holding them in a darkened and aerated recovery tank for periods ranging from 10 minutes to approximately one hour. All smolts swam normally at release and showed no evidence of visually abnormal behaviour.
A total of 40 steelhead smolts were collected and surgically implanted without respect for size, with tagging chosen to match the time of the peak of the emigration. Smolt migration was delayed in 2002 by several weeks from normal because of low water levels caused by an unusually dry spring. As the Waukwass River is a wild river with little human influence other than past forestry operations, smolts were all wild, and likely constituted a mixture of 2 and 3 year old smolts at the time of tagging (scales were not collected). Eighteen smolts were present in two rotary screw traps and implanted on 22 May, 2002 (mean fork length: 17.6 cm, range: 15.9-20.7 cm). On 26 May an additional 22 smolts were recovered from the traps and surgically implanted (mean fork length: 17.45 cm, range: 15.5-21.4 cm), bringing the study group to the target 40 animals.
Results Two of the initially tagged 18 smolts were recaptured in the traps on 26 May, two days after release; their abdominal incisions were visually inspected before being released again. The smolts appeared to be in generally healthy condition, inflammation was not evident, and all sutures were still secure.
All receivers making up the six detection lines were recovered (Table 1). One receiver, the southernmost of 3 receivers on the outer detection line (E) failed as a result of flooding with seawater and contained no data. As a result, the outermost detection line did not provide complete coverage of smolts passing on the southern side of the fjord.
Detection Success A total of 31 of the 40 smolts initially implanted in the river were subsequently detected on the first detection line (RM). The location of the RM line was placed 2,700m from the river mouth (Fig. 1; Table 1) to ensure that any detected smolts were present in the estuary, and not just detected in freshwater at the river mouth. We report our results in the context of these 31 animals forming the study population, and comment on the fate of the 9 animals not entering the ocean later. Timing of movements are based on either time since release in freshwater or the time since first detection in the ocean, as appropriate, with times between receiver pairs calculated as the last time of detection at one receiver to the time of first detection at the next receiver.
All 31 smolts detected entering the ocean at the RM line were subsequently identified moving over the remainder of the array. Conversely, no smolts were detected on outer lines that were not detected first detected on the RM line. No tag codes not used in the study were recorded by the tracking array. Thus there was a 100% success rate for re-detection on smolts entering the ocean, and zero false positives (detection of a code known not to be present). In addition, no steelhead tagged in the simultaneous Queen Charlotte Strait study on the east coast of Vancouver Island (Welch et al., Submitted) were detected entering Quatsino Sound, although these tags used a common coding scheme.
The two outer lines in Quatsino Sound provide some information on the detection rate for the array lines, although the failure of one of the three outer receivers limits the comparison somewhat. The innermost line (F) detected qqq
When the array was recovered on July 2nd, 2002, 37-39 days after tagging, all but 4 of the smolts (87%) had passed the outer lines leading to the open shelf (Lines E and F; Figure 1). All four of these residual 4 smolts apparently remained in Holberg Inlet, but their fates were different; 2 smolts were heard on line H on the 1st and 2nd July. The other 2 smolts were last heard on line G on 27th May & 12th June, subsequent to being first detected on Line H. As Holberg Inlet is very long, extending 20 kms northwest of Line G, the fate of these animals is uncertain; they may have either remained resident further up the inlet or died.
The 27 smolts which reached the open ocean (passing over Lines F and E in the process) took a variety of routes to get there (Fig. 2). These routes can be roughly classified in three groups:
I. Rapid Directed Movements (N=20)
From RM straight to lines F/E – 2 smolts
From RM to Holberg Inlet then to lines F/E – 16 smolts (13 were detected on line G as well as line H)
From RM to Neroutsos Inlet (line I) then to lines E/F – 2 smolts
II. Less Directed Movements (N=7)
From RM to Holberg Inlet, then to Neroutsos Inlet then to lines F/E – 2 smolts (1 heard on line G as well as line H)
From RM to Holberg Inlet, back to RM then to lines F/E – 4 smolts (2 heard on line G as well as line H)
From RM to line F, then to Neroutsos Inlet then back to line F – 1 smolt
Estuary Resident (N=4)
Repeated detection in Holberg Inlet, with no evidence of departure from the interior fjord.
With the exception of the last smolt pathway, all smolts that reached line F were either not detected again, or detected on line E (the outermost line) and then not detected again. Once these animals reach the entrance to Quatsino Sound they thus appear to leave Quatsino Sound for the open sea and do not return. Apart from the 2 smolts still in Holberg Inlet when the array was removed, the last detections occurred before June 21st. If the smolts had re-entered Quatsino Sound it seems likely that they would have been detected.
Timing of Movements
The time taken from release in freshwater after surgical implantation to detection at the RM line (Fig. 2) was comparatively rapid, with a mean time of 3.2 days. The time taken from release in freshwater to reaching line F varied from 7 to >30 days (Fig. 3). >30 days represents those smolts still in Holberg Inlet at the end of the study.
A comparison of the relationship between the length of time to reach RM and the length of time to reach Line F provided little evidence for a correlation between initial speed of movement and subsequent speed of movement within Quatsino Sound (Fig. xx). Thus those smolts slow to reach the river mouth were not necessarily also the slowest to reach the mouth of Quatsino Sound. The 4 smolts that did not reach the mouth of Quatsino Sound during the study had some of the fastest times from release to RM (2-4 days).
The speed of movement between lines was calculated using the shortest possible route between lines and the time from the last record at one line to the first record at the next line. Speed was calculated in terms of body lengths per second using the fork length measurements made of each fish when the tag was implanted.
The frequency histogram (Fig. Xx) shows the speeds of all measured movements between lines (a total of 158 movements, and an average of 5 movements per fish). Most fish moved at about 1 BL sec-1. A few records of speeds over 3 BL sec-1 may reflect the movements of a predator that has swallowed a tagged fish. (Sonia– can you check and see if the tag with these rapid movements subsequently made it to the open sea, or were they for the tags that remained within the inlet system?) Its also possible to examine the speed of movement between lines according to the behaviours of the fish. Three groups can be identified (Fig 5):
Rapid Emigration. Fish moved quickly out to the open shelf. Four fish either went from RM straight to lines F and E or were heard very briefly on lines H or I before being heard at F/E.
Extensive estuarine residence, followed by emigration to the open sea. Fish spending sometime in Holberg Inlet before migrating out to the shelf. This was the dominant pattern of behaviour with 15 fish taking this route.
“Permanent” Estuarine Residence. Four fish remained in Holberg Inlet for the 37-39 day duration of the study. (Note that the array was deployed for two months, but that we delayed tagging until the peak of the run, which was delayed by several weeks in 2002). Movement of these four smolts after the array was lifted is unknown, so the duration of estuarine residence is unknown.
The remaining 7 fish either also made extensive use of Neroutsos Inlet or went from Holberg inlet back to RM before moving out to the open sea.
The figure shows the mean speed (error bars are standard deviation) for each behaviour type. Note though that the ‘Fast Migrants’ are both few in number and, because they headed more or less straight out, there are limited detected movements between lines. Number of movements between lines (n) for the 3 groups respectively were 12, 79, and 30.
Although differences in mean speed are not significant, one explanation for the apparently slower speed of the ‘Fast Migrants’ is that they swim slowly and steadily out of Quatsino Sound while the fish that remain within system for a while spend days in one inlet and then move relatively rapidly from line to line. This is borne out if the mean speed per line movement is examined. The figure below shows the mean speed (and standard deviation) for the most common line movements (note that some are uni-directional movements, some bi-directional).
Movements from RM to F were the slowest with movements within Holberg inlet (between G and H) being some of the fastest. The fastest movement of all was from line F to E, suggesting that once fish head out of Quatsino Sound the movement is rapid. Some further support for this view comes from the average number of detections made on the two outer lines; individual detections on the inner line of each tagged fish averaged 17.7 times greater than the detections on the outer line, just 10 kms to the west (range: 1.7-79.0; N=17 fish).
(Comparison of Two Outer Line Detections: A total of 17 & 18 fish tags detected on the two lines respectively, 9 of the tagged smolts detected on inner line only, and 1 tagged smolt detected on outer line only;<- Check this; these numbers don’t seem to make sense).
A similar higher “dwell time” was evident when comparing the ratio of detections over the inner and outer detection lines in Holberg Inlet north of the Quatsino Narrows (Lines G & H); detections over the innermost line were 23.4 times higher than over the northern line (range: 0.1-409.0). Here, however, the results are less clear-cut because the majority of animals that made extensive movements up Holberg Inlet also moved back down; if they happened to reverse direction over the northern-most line a greater number of detections would be expected over the southern line as the smolts would pass over this line twice. (Investigate whether the timing of detection peaks on these two lines indicates whether each smolt moved over G then H, then H (again) then G; could do an interesting graph of movements for each smolt, reconstructing dwell time, since the y-axis will be the number of detections; would need to correct for simultaneous detections on multiple receivers on each line (time differences of <30 seconds)). In contrast, the ratio of detections over the RM line and the inner Holberg line were very similar (1.2; range 0.02-5.7), suggesting some degree of orientation so that smolts moving in the upper reaches of Holberg Inlet may on average increase their swim speeds as they search for the exit from the inner fjord.
In summary, the slowest average speeds were measured when calculated for the longest distances considered, and the fastest speeds measured for the shortest distances. This suggests that movement is not strongly uni-directional and that considerable milling occurs. Mean speed drops as would be expected if the animals spend significant time and effort at exploring their local environment. However, none of the animals reaching the outer reach of Quatsino Sound returned to the fjord (check this), suggesting that once the open shelf was reached the steelhead were able to orient to prevent entering Quatsino Sound again.
Survival and Detection Efficiency (Comparison of Two Outer Line Detections: A total of 17 & 18 fish tags detected on the two lines respectively, 9 of the tagged smolts detected on inner line only, and 1 tagged smolt detected on outer line only;).
At 1 BL/sec, it would take a 15 cm smolt 11.6 days to cover the 150 kms distance between the mouth of the Keogh River and the outermost acoustic line in Quatsino Sound. The lack of detection of either the ecvi or west coast of Vancouver Island smolts in the other area is perhaps not surprising, but does indicate that the two study areas are likely not visited by smolts from the other region.
Discussion Mention failure of one VR2.
Make point that given high detection rates and demonstrated long-term survival and tag retention by implanted smolts (Welch et al In Press), the optimal study design should involve placing large numbers of acoustic receivers out for detection, and using relatively few tagged fish. In other words, in considering the funding of such studies, it makes more sense to place most effort in placing the array and letting the tagged fish move over it; the more geographically extensive the array, and the longer that it is maintained in the ocean, the greater the information that is returned on what the tagged animals are doing. This is in direct contrast with traditional tagging studies, which must rely on releasing large numbers of tags in order to recover a few tagged fish at a later date. For example, if we had replaced the array after initially retrieving it in early July, it is conceivable that we would have eventually detected the two smolts last heard on the inner array (Lines ??) one and two (?) days before retrieval of the equipment. Similarly, the only two tagged smolts whose movements are unaccounted for were last detected moving into the far northern extension of Holberg Inlet. It is conceivable that they might eventually have moved south and then out to the environment of the open shelf, which a permanent array would have detected.
The lack of acoustic listening lines farther north in Holberg Inlet illustrates how a more widely spaced array would provide further valuable information. The current method of array deployment, which involved large quantities of anchors, rope, and additional weights was costly and labour intensive. The major goals of the current experiment were, however, to demonstrate (1) that deployment of a large-scale acoustic tracking array would result in substantial new biological environment on the movements and survival of small fish in a technically demanding environment and (2) that it is technically feasible to design and run a large geographic array (many lines, multiple receivers per line), and to design it so that it would be possible to recover all of the equipment after several months at sea.
Both goals were clearly met. We surgically implanted a combination of acoustic tags with life-spans of either 4.5 months and 20(?) months into wild steelhead smolts, demonstrating that available tracking equipment can be used to track even quite small smolts (15-18 cm) with tags that would allow tracking their movements for substantially more than one year at sea. (In fact, the major difficulty with the surgical implantation of these smolts revolved around the fact that their guts were packed with large amount of food that they had fed upon within the screw traps, which apparently also acted as a collector for their food. In future, we might consider holding the smolts in a large enclosure using filtered river water to allow their guts to clear prior to surgery).
Given that important new biological information could clearly be collected from a much more geographically wide-spread array, the next issue involves the engineering of a permanent continental-scale acoustic tracking array for the West Coast of North America (Welch et al Submitted and In Prep.). As all of the hardware components needed to produce this array now exist, the major remaining obstacle to a permanent seabed tracking array primarily involve some engineering issues. The major uncertainties with the construction of a widespread, modular, distributed array involve questions concerning the expected life-span of the individual seabed nodes making up the array. Such questions will not be fully answered until large numbers of nodes are deployed and statistics on overall “survivability” are accumulated. However, we have demonstrated with the current experiment that the building of such an array is feasible, since it primarily involves a scaling up from existing technology, in both geographic scale and duration of deployment. More importantly, however, we have demonstrated with the current experiment that marine research can pay major dividends in terms of an improved understanding of the ocean life history of Pacific salmon (or other marine organisms). Our results indicate that a properly deployed array can provide detailed understanding of marine habitat use (areas of residence or non-residence), plus a wealth of information on direction and rates of movement using relatively few animals. Finally, it appears that we have been able to provide detailed information on ocean survival over the first few weeks of life in the ocean. Early ocean survival in Quatsino Sound is high (94%), and contradicts the common assumption that mortality during this period plays the major role in determining marine survival of salmon.
Our current observations could be strengthened by the development of a modified acoustic tag which would ensure that a tagged animal consumed by a predator would either stop transmitting (and thus die along with the host) or transmit a modified signal that would indicate that the host animal had been eaten.
STUFF FROM OTHER PAPER (OCEANOLOGICA ACTA) Acoustic tags small enough to be carried by salmon cannot store environmental information for later transmission, so a picture of the movement of tagged Pacific salmon and their environmental conditions needs to be built up by having the position and current environmental conditions for individual salmon logged by multiple receivers. Acoustic tags send out information on the identity of the tagged animal and (for some types of tags) currently experienced environmental conditions (depth or temperature). Self-contained submersible acoustic receivers capable of detecting and logging the movements of a tagged juvenile are now available that can operate autonomously for over a year underwater (e.g. Voegeli et al 1998; Lacroix and Voegeli 2000). This then allows the movement patterns of individual animals to be reconstructed, essentially by “connecting the dots”: knowing the date and time an individual is heard in the vicinity of each acoustic receiver, speed and distance between each pair of receivers can be determined.
Recently developed technology offers the prospect of putting out many long-lived low cost acoustic receivers on the seafloor, in a series of detection lines which can act as a grid over which thousands of individually identifiable tagged animals might move at will, with their movements passively recorded by the seabed array (Fig. 2). Because the detection technology is relatively low-cost, the possibility of a continental scale acoustic tracking array for the shelf and slope regions is potentially quite economic.
Because acoustic tags offer lifetimes of 4.5 months to several years (e.g. Lacroix and Voegeli 2000) and experimental studies demonstrate that they can be successfully implanted into salmon smolts as small as 10.5 cms and retained long-term (Welch et al in review), the possibility now exists to tag and track the movements of many animals for most of their life cycle. By building up a dense array of low-cost receivers sited on the seabed, it is potentially possible to reconstruct the movement patterns in great detail. This is perhaps the most ambitious aspect of the POST project; to help demonstrate the feasibility of a continental scale acoustic tracking array on the entire continental shelf. Curiously, despite considerable recent interest in the possibility of seafloor observatories by the oceanographic community (e.g. National Research Council 2000), almost all of the proposals to date have involved highly expensive networks in the deep ocean consisting of a relatively few nodes cabled to land. While such a network would provide a high bandwidth but relatively sparse network, project POST would provide the complement: a broadly distributed but low cost and low bandwidth acoustic tracking array that could be run off long-lived lithium batteries. In this paper we focus on describing the fish tracking component, and defer most of the technical details of how the array would be constructed for later consideration.
Continental Shelf Program
Because the continental shelf off the West Coast of North America is very narrow (less than 20 kms wide in most places), the narrowness of the shelf lends itself to a broad-scale monitoring program using acoustic tags for animals such as juvenile salmon that remain on the shelf. Recent developments in acoustic technology (e.g. Lacroix and Voegeli 2000) allows reliably detecting uniquely identifiable sonic tags using low-cost passive receivers ($1,000 per receiver). These receivers can detect sonic tags within an ca. 1-2 km diameter circle centred on the receiver (depending on the acoustic power of the tag), recording the date and time that individual tags are detected for a year or more, and have a recording capacity per receiver of 300,000 or more detections (about 8,000 per day, on average). In principle then, a series of autonomous receivers laid in a line across the shelf perpendicular to the long-shelf migration path of animals such as Pacific salmon would then be capable of detecting and recording the movements of each individual passing over the detection line. By placing separate lines of cross-shelf receivers at appropriate spacing on the shelf, a detailed picture of the movement patterns of tagged animals would be possible (Fig. 3).
As the shelf on the West Coast is usually less than 20 kms wide, a string of 30 receivers laid across the shelf and down the slope region should be capable of detecting most tagged shelf and slope animals crossing each detection line. The costs of developing a broad scale acoustic array are surprisingly modest. The current generation of receivers are commercially available for $1,000 so, taking into account the need for additional infrastructure to place the receivers and recover the data, it may be possible to place and maintain each listening node on the seabed for perhaps $5,000. As a result, this leads to an approximate cost for a single monitoring line on the order of $150,000; thus for roughly 3 million dollars a network of 20 or so acoustic listening lines could be deployed that would stretch from California to the Aleutian islands (Fig. 2). Such a line would be capable of detecting individual animals as they crossed the monitoring lines. The smallest uniquely identifiable acoustic tags operating at frequencies useful for detection at hundreds of meters are capable of being surgically implanted into salmon smolts as small as 10.5 cms and retained for months or years (Welch et al in press). As these tags have operational lifetimes under continuous operation of ca. 4.5 months, and slightly larger tags have lifespans of years, it is potentially possible to expect to tag young salmon in-river or on the shelf and be able to follow their movements for the rest of their life history. The shelf component of project POST will focus on defining a “proof of principle” experiment to demonstrate the validity of putting into place such a network. Such a monitoring framework would provide the basis for tracking any animal present on the continental shelf that was tagged with a uniquely identifiable sonic tag: smolts, immature shelf-resident salmon stocks in their second or third year of life, or even whales and other marine mammals. (A wide complement of other marine fish could also be detected by such a monitoring network, bringing in the possibility of building broad support for the monitoring network and spreading costs by monitoring the movements of other high-valued fish such as halibut, black cod, Pacific hake, and Pacific mackerel). The advantage of a fixed coastal array is that the geographic position of each array element is known with precision, so the detection of uniquely tagged animals by the array elements allows reconstruction of the coastal movements (direction and rate of movement between array elements, and time of occurrence in specific regions of the coast). Alternative systems, such as a system where the tag records the time that an acoustically active fishing vessels is nearby, requires the recovery of the tag from the animal; this means that many more tags must be used than will be recovered. The great advantage of a fixed array is that recovery of tagged animals is not required, and many species might be simultaneously studied by the same array.
The British Columbia study in 2002 will be focussed on steelhead from two adjacent river watersheds (Keogh & Waukwaas), whose rivers empty into the eastern and western coasts of Vancouver Island (Fig. 5). Despite their geographic proximity (the watersheds are adjacent to one another on the same mountain), the marine survival of smolts from all rivers on the eastern side of Vancouver Island is much lower than that of steelhead smolts exiting to the west coast of Vancouver Island (Welch et al 2000). This difference is believed to have a genetic basis, with steelhead from the Keogh & Waukwaas taking different marine migration routes and therefore being exposed to different oceanographic regions.
An important principle in the development of a continental scale array is that all of the equipment must eventually sit on the seabed and not involve surface floats that are vulnerable to vessel traffic or fishing activities. The deployment was thus planned to gain some experience with this principle. Both detection lines were placed on the seabed using a horizontal groundline, and vertical floats spaced 350m apart were used to position the receivers about 0.5 m off the seabed. On the outer edge of the northern detection line, where a deep channel reached ca 350 m depth, the length of the vertical float lines was increased to position the receivers at or above 200 m. All equipment was placed on the bottom, out of sight, and was retrieved by triggering acoustic releases which brought the floats and the end of the groundlines up to the surface. The groundline was then retrieved using a chartered seine vessel. A commercial fishing vessel was used for the deployment both to develop involvement with the industry and to maintain the principle of using low cost vessels for deploying the array.
Conclusions There is no prior information on the inshore movements of steelhead smolts after entering the ocean, so it is difficult to put the biological findings from the current very limited study into a broader context at this time. From tagging work using conventional numbered tags there is evidence for steelhead moving north along the continental shelf migration path that the other species of Pacific salmon all seem to use as juveniles; however, there is also some evidence that at least some steelhead may move directly off the continental shelf to the open ocean (Hartt and Dell 1986). The resolution to such questions awaits the development of a permanent continental-scale acoustic tracking array.
Acknowledgements We thank the Census of Marine Life for support, inspiration, and funding, and a number of individuals within our institutions for their enthusiastic support. We would especially like to acknowledge Reg Bigham and Hugh MacLean for technical support with the moorings, and Lloyd Burroughs, Melinda Jacobs, and Adrian LaDouceur for assistance with the surgical implantation (and not complaining too much when left alone to complete them with a cougar in the vicinity!). Captain Robert Howitch of Quatsino provided us with much useful information on the area and made the recovery of the array look simple.
References Arnold, G.P., and Dewar, H. (2002) Electronic tags in marine fisheries research: A 30-year perspective. in: J.R. Sibert and J.L. Nielsen (eds.), Electronic Tagging and Tracking in Marine Fisheries, p. 7-64.
Emery, K.O. (1980) Continental margins: Classification and petroleum prospects. Bull. Amer. Assoc. Pert. Geol. 64:297-315.
Groot, C., and Margolis, L., (eds) (1991) Pacific salmon life histories. UBC Press, Vancouver. 564 p.
Hartt, A.C., and Dell, M.B. (1986) Early Oceanic Migrations and Growth of Juvenile Pacific Salmon and Steelhead Trout. Int. North Pacific Fish. Comm. 46:1-105.
Khan, K. (2000) Maritime Boundaries. Globe. (August Issue; See http://www.defencejournal.com/globe/2000/aug/maritime.htm)
Lacroix, G.L., and Voegeli, F.A. Development of automated monitoring systems for ultrasonic transmitters. In: Fish Telemetry: Proceedings of the 3rd Conference on Fish Telemetry in Europe. (Eds: Moore, A; Russell, I) CEFAS, Lowestoft, UK, 37-50.
National Research Council (2000) Illuminating the Hidden Planet. The Future of Seafloor Observatory Science. National Academy Press, Washington, D.C. 135 pages.
Quinn, T.P. (1993) A review of homing and straying of wild and hatchery-produced salmon. Fisheries Research 18: 29-44.
Quinn, T.P., Volk, E.C. and Hendry, A.P. (1999) Natural otolith microstructure patterns reveal precise homing to natal incubation sites by sockeye salmon (Oncorhynchus nerka). Canadian Journal of Zoology 77: 766-775.
Voegeli, F.A., Lacroix, G.L., and Anderson, J.M. (1998) Development of miniature pingers for tracking Atlantic salmon smolts at sea. Hydrobiologia 371/372, 35-46.
Welch, D.W., and J.P. Eveson. (1999) An assessment of the geoposition accuracy of data storage (archival) tags using light. Can. J. Fish. Aquat. Sci. 56(7):1317-1327
Welch, D.W. and Eveson, J.P. (2001) Recent Progress in Estimating Geoposition using Daylight. In: Electronic Tagging and Tracking In Marine Fisheries” In: Methods and Technologies in Fish Biology and Fisheries, Vol. 1. J. Sibert and J. Nielsen (eds.) Kluwer Academic Press, Dordrecht, The Netherlands. p. 369-384.
Welch, D.W., Batten, S.D., and Ward, B.R. (in review) Growth, Survival, and Rates of Tag Retention for Surgically Implanted Acoustic Tags in Steelhead Trout (O. mykiss). Trans. Am. Fish. Soc.
Welch, D.W., Ward, B.R., Smith, B.D., and Eveson, J.P. 2000. Temporal and Spatial Responses of British Columbia Steelhead (Oncorhynchus mykiss) Populations to Ocean Climate Shifts. Fisheries Oceanography 9:17-32.
Wiseman, J.D.H., and Ovey, C.D. (1953) Definitions of features on the deep-sea floor. Deep-Sea Research 1:11-16.
List of Figures
Fig. 1. Geography of Quatsino Sound and layout of the acoustic array. Holberg Inlet extends 20 kms to the northwest of Line G. The scale bar is 10 kms.
Fig. 2. Three examples of movement patterns within Quatsino Sound: (A) Rapid emigration to the open sea from freshwater; (B) Extensive use of an inlet before leaving for the open sea; (C) Long-term estuarine residence. Note that for clarity straight line connecting sequential detection points have been plotted, however, distances used in speed calculations used the shortest feasible movement path through water.
Fig. 3. Time from release to first detection on the RM line for the 31 steelhead smolts detected entering the ocean.
Fig. 5. Frequency histogram of measured rates of movement for all recorded path segments.