NOAA logo JAS 2000 Quarterly Rpt. sidebar

Survival and Straying of Pink Salmon
Using Recoveries of Coded-Wire Tags and
Thermally-Induced Otolith Marks

(Quarterly Report for July-Aug-Sept 2000)

Donald Mortensen, Alex Wertheimer, Jacek Maselko, and Sidney Taylor

picture of thermally induced otolith marks
Figure 1. Thermally induced otolith marks is an alternative
method of marking pink salmon.

Anadromous salmonids are known for their ability to home to their natal streams to spawn. The ability to home is thought to be due to olfactory memory of trace chemicals in the natal stream and perhaps, also, population-specific odors (pheromones). Although most salmon home to natal streams, some stray to others. The degree of straying varies among years, species, stocks, ages, and genders and can change with the number of fish returning. Stream type, water source, and physical location may also play an important part in the extent of straying. Natural straying may be essential for colonizing new habitat, recolonizing degraded spawning habitats and also may reduce inbreeding in small populations.

picture of coded-wire tag
Figure 2. Typical half-length (0.5 mm) coded-wire tag, a device
commonly used in marking pink salmon.

Pink salmon, Oncorhynchus gorbuscha, have a particular reputation as strayers. Studies conducted in Prince William Sound, Alaska, (Sharr et al., 1995 and  Habicht et al., 1998) indicate that coded-wire tags (CWT) (Figure 2 above) inserted in juvenile pink salmon snouts may exacerbate straying of those surviving to maturity. As olfaction is thought to play an important part in the ability of the salmon to find its home stream, the injection of a CWT into sensitive olfactory tissue in the  fish’s nasal area could negatively influence homing ability.  Studies conducted at Auke Creek, Alaska, in the 1980s noted that return rates of unmarked adults have been as high as seven times that of fry tagged with CWTs (tagged fish) concurrently, suggesting a higher mortality or straying rate for tagged fish. To provide an external cue of the CWT, the adipose fin is usually clipped.  It has also been shown that fin-clipping of wild pink salmon fry substantially reduces return rates. The apparent handicap of tagged fry may be due in part to the trauma of fin-clipping as well as tag injection, which affects survival or homing. Greater mortality rate of tagged fish in relation to that of untagged fish could result in errors in the estimation of fishery contributions, survival rates, and straying rates of hatchery or wild fish that the tagged fish are assumed to represent.

The relatively new method of thermally inducing marks on the otoliths of fish (Figure 1 at top) is considered to be a noninvasive way of marking pink salmon fry and can be used to measure survival, straying, and the effect of applying CWTs on survival and straying.  Thermal marking is not known to cause trauma to olfactory or other nerve tissue, and an entire hatchery production can be marked, greatly increasing sampling resolution. About 14 km away from Auke Creek, the Gastineau Hatchery, operated by Douglas Island Pink and Chum Corporation, has routinely thermally marked portions or all of their pink and chum salmon production since 1990  with no apparent effect on survival.  In 1994, the hatchery thermally marked about 33% of their 1994 brood (1996 adult ) pink salmon.

Our study objectives were to compare the survival and straying rates of thermally marked  and coded-wire-tagged Auke Creek pink salmon and thermally marked Gastineau Hatchery pink salmon.


Auke Creek is a small, lake-fed stream, which empties into Auke Bay in northern Southeast Alaska (Figure 3).  The National Marine Fisheries Service maintains a permanent two-way counting weir and small research hatchery at the confluence of Auke Creek with Auke Bay.  We spawned a portion of the early- and late-run pink salmon returning to Auke Creek in 1994 to provide eggs for thermal marking and seeded them into gravel incubators plumbed with water from Auke Creek.  A small water heating system at the Auke Creek hatchery provided warm water, which allowed us to thermally induce unique five-band patterns in the otoliths of prehatch early- and late-run pink salmon embryos.  To induce a thermal mark on the otoliths of salmon, the ambient water temperature surrounding the incubating salmon eggs or alevins is raised or lowered by at least 3 degrees for 16 to 24 hours.  The water temperature is then quickly returned to ambient which results in a dark increment or stress check being entrained in the otolith of the incubating salmon.  This procedure is repeated several times to put a series of checks (much like a bar-code) on the otolith. As these marked pink salmon fry emerged from the gravel incubators in spring 1995, we adipose-clipped and injected CWTs into representative samples.  We also adipose-clipped and injected CWTs into wild Auke Creek fry throughout their emergence (Table 1).

Upon their return to Auke Creek and Gastineau Hatchery in 1996, adult pink salmon were examined for adipose clips (denoting CWTs) daily.  The adipose clipped fish were killed and counted and the heads removed in order to recover the CWTs and otoliths.  We also conducted foot surveys twice a week at Waydelich, Fish, and Auke Nu Creeks.  At Salmon Creek, the water source for Gastineau Hatchery, we conducted foot surveys once a week. During the foot surveys otoliths were removed from spawned-out pink salmon carcasses and the carcasses were also examined for adipose clips (denoting stray, coded-wire tagged Auke Creek pink salmon).

When an adipose clipped carcass was found, the head was taken back to the laboratory where the CWT was removed and decoded.  We removed the otoliths of all carcasses found without adipose clips to provide estimates of thermally marked pink salmon strays from Auke Creek and Gastineau Hatchery. At the laboratory, CWTs were extracted from the heads of adipose-clipped fish and decoded using a dissection microscope.  The sagittal otoliths from the carcasses were glued to microscope slides and ground to the central lateral plane using 1,000-grit sandpaper.  The ground otoliths were then examined for thermal marks under a compound microscope at magnifications between 200 and 400 X.

Adult survival of tagged (CWT) fish was computed as the proportion of fry released from a tagged group that returned to the Auke Creek weir.  We computed the survival of thermally marked fish returning to Auke Creek as the expanded number of marks divided by the estimated number of thermally marked only fry.  The run composition of Auke Creek fish was then estimated by allocating unmarked fish based on representative tagged or marked groups. The estimates of unmarked pink salmon from Gastineau Hatchery and the wild Auke Creek component were subtracted from the total estimate of unmarked fish which spawned in Auke Creek.  The remainder of the unmarked fish in Auke Creek were assumed to be strays from other streams or from intertidal spawning below the Auke Creek weir.

Our estimate of the number of unmarked, wild Auke Creek adults was derived from CWT wild fry survival and a “handicap ratio,” defined as the ratio of the survival of marked only fry to tagged and marked fry.  The corrected survival rate was then applied to the number of unmarked fry released at the weir.

Spawning populations in the three streams within 10 km of Auke Creek (Figure 3) were estimated using a mark-recapture method. We estimated the proportion of carcasses in the stream and the number of strays in each creek by dividing the number of otolith marks or CWTs found in the sampling fraction.  We also examined pink salmon carcasses in Salmon Creek, which is approximately14 km from Auke Creek, for adipose-fin clips and thermal marks but did not make a population estimate.  All pink salmon adults returning to Gastineau Hatchery were examined for the presence of adipose fin clips (CWT), and about 12% were sampled for thermal marks.

We computed the straying rate of marked Auke Creek fish to the index creeks by summing the estimates of the strays to each site, and dividing by the number of fish that returned to the sample area.

Results and Discussion

Tagging small pink salmon with CWTs and removing the adipose fin reduced their rate of survival; fry marked only with thermal marks had consistently greater survival rates. Survival rates for the three CWT groups, 0.20%, 0.17% and 0.16% for the early-, late- and wild run pink salmon, respectively (Table 1), were not significantly different (P > 0.5).  Survival rates for the groups marked only with thermal marks were 0.41% and 0.47% for the early- and late-run pink salmon,  respectively; these rates were also not significantly different (P >0.4).  Estimated survival of fry marked only with thermal marks was 2.1 times greater for early-run and 2.8 times greater for late-run than for the respective CWT early and late run groups.  We took the average of these ratios (2.4) to provided a mean handicap to estimate a survival rate for unmarked wild pink salmon fry (0.39%).

Thermally inducing marks on salmon otoliths is commonly assumed to have no effect on survival, but there has been little or no explicit testing of this assumption.  Our results showed  no significant difference in survival between thermally-marked fry from the early- and late-run groups that were also tagged with CWTs and wild fry tagged with CWTs but with no thermal marks, leading to the conclusion that, indeed, thermal marking had no effect on survival.

Pink salmon returning to Auke Creek originated from the various groups of Auke Creek pink salmon and from strays from other streams and  nearby Gastineau Hatchery (Figure 3).  Auke Creek coded-wire-tagged and thermally-marked pink salmon accounted for 3,516 of the total return.  The unmarked portion of the return from Auke Creek wild fry was estimated at 175.  We calculated this by multiplying the survival rate for unmarked wild fry (0.39%) by the number of unmarked wild fry released (43,967). Expanding for sampling fraction and marking fraction, the number of Gastineau Hatchery strays in Auke Creek was 462 fish.  This left about 605 of the Auke Creek run, which was composed of strays from other streams or from intertidal spawning below the weir.

We recovered strays from five of the six Auke Creek marked groups in four streams and the Gastineau Hatchery (Table 3). Most of these fish were from thermally-marked groups: 49 Auke Creek  thermal marks were observed in the carcasses sampled, while only three CWTs were recovered.  The frequency of Auke Creek strays in the escapement varied substantially among the streams that were sampled.  Salmon and Waydelich Creeks had the greatest proportions of Auke Creek strays, 7.6% and 6.1%, respectively, while Auke Nu Creek and Fish Creek were smaller, at 2.1% and 2.6%, respectively.  Gastineau Hatchery had the lowest frequency of Auke Creek strays (0.3%), significantly less than the proportion in Salmon Creek, the watershed for the Gastineau Hatchery.  Crowding in the hatchery ladder may have influenced more of the Auke Creek strays to enter the less crowded stream bed at nearby Salmon Creek.

For coded-wire-tagged and thermally-marked groups, more late-run strays were observed than early-run strays (Table 4).  We recovered two late-run tagged strays compared to no early-run tagged strays and 31 late-run thermally-marked strays vs. 18 early-run thermally-marked strays.  There were not enough CWT recoveries to test for statistical differences, but we did find that thermally-marked early-run fish strayed significantly less than the thermally-marked late-run fish.  This may be the result of the early return timing and stream characteristics.  These early-run fish may experience unique stream characteristics such as odor, water temperatures, and flows which would result in greater homing specificity.  Additionally, by virtue of its early return,  the early-run fish are not as susceptible to attraction by aggregations of pink salmon near other streams as are the late-run fish.

We estimated straying rates for Auke Creek pink salmon to streams within 10-km of Auke Creek (Auke Nu, Waydelich and Fish Creek); and within 14-km of Auke Creek (Salmon Creek and Gastineau Hatchery).  Within 10-km the straying rate for  fish tagged with CWTs was based on just one recovery; estimated rates were zero for early- and wild-run fish and 5.6% for the late-run fish (Table 4).  Estimated straying rates increased to 1.4% and 6.6% for wild and late-run coded-wire-tagged fish respectively, when strays to14 km are included.  Within 10 km of Auke Creek we estimated straying rates for thermally-marked Auke Creek pink salmon of 2.5 % and 3.2 % for the early- and late-run fish,  respectively.  Estimates of straying rates increased to 4.4% and 6.7% for the early- and late-run respectively with the addition of strays to 14 km.  These estimates are biased low because there was no expansion of strays due to sampling fraction in Salmon Creek.

To increase the precision of the straying rate estimates, the data for coded-wire-tagged and thermally marked pink salmon were pooled across run types.  Pooling was justified because both treatments marked fish over the entire emigration of the Auke Creek population.  We calculated straying rates between tagged and marked pinks as a group by estimating the total stray rate for each group and observing the overlap of the confidence intervals (Table 4).  For streams within 10 km and 14 km of Auke Creek, the point estimate of straying for the coded-wire-tagged fish was 1.9% and 2.4%, respectively.  The point estimates of straying of the thermally marked fish were higher: 2.8% and 5.7%, respectively, for streams within 10 km and 14 km of Auke Creek (Table 4).  The extensive overlap of the confidence intervals from each group indicates that adipose fin clipping and marking with CWTs did not affect straying rates in this study.  Although tagging with CWTs did not increase the straying rate of pink salmon in our study, other researchers (Habicht et al.,1998 and Thedinga et al., 1999) have shown circumstantial and direct evidence that CWTs can affect homing and straying in pink salmon.

graph of pink salmon return
Figure 4.  Origin of pink salmon returning to Auke Creek, expanded for marking and sampling fraction. Vertical bars denote the standard deviation around the estimate.

Only 19% of the pink salmon sampled for thermal marks at Gastineau Hatchery had the hatchery  marks (Figure 4 above). This proportion is significantly less than the 33% expected  based on marking rate.  Assuming there was no mortality or differential straying due to thermal marking, then large numbers of unmarked salmon must have strayed to the hatchery to reduce the mark rate. Movement of fish from other streams into the hatchery is demonstrated by the presence of Auke Creek thermal marks (2%) in the return.  At Salmon Creek, 15% of the carcasses sampled had Gastineau Hatchery thermal marks.  This frequency was not significantly different than that at the hatchery itself, indicating that Salmon Creek pink salmon and the Gastineau Hatchery pink salmon were returning in similar proportions to the two sites. The suspected decrease in the proportion of sampled fish with the Gastineau Hatchery mark at the hatchery may have been largely due to Salmon Creek fish returning to the hatchery.  Because the hatchery and Salmon Creek are adjacent and share the same water shed, we considered the hatchery fish returning to either the hatchery or Salmon Creek to have homed.

graph of pink salmon carcasses
Figure 5.  Percent of Auke Creek and Gastineau Hatchery pink salmon found in samples of carcasses taken from local creeks and the Gastineau hatchery.

Gastineau Hatchery fish also strayed to the other four streams sampled.  The percentage of  thermal marks in carcasses sampled in these streams ranged from 4% in Auke Creek to 8% in Waydelich Creek (Figure 4 above).  The estimated straying rate of Gastineau Hatchery pink salmon was 6.9% (Table 4), if Gastineau Hatchery marks recovered in Salmon Creek and at Gastineau Hatchery are assumed to have homed.  This estimate is biased high because we did not estimate the population of Salmon Creek and were unable to adjust the number of homing fish in the sampling fraction.

Our results indicate substantial straying among wild and hatchery pink salmon populations in the Auke Creek area.  This is consistent with the concept of spatial or geographic structure within populations of pink salmon, where the lack of multiyear age structure provides no buffering to compensate for poor survival years. The genetic homogeneity of pink salmon found in Southeast Alaska also indicates a relatively high degree of straying especially between local systems.  An earlier study using a genetic marker in Auke Creek pink salmon (Gharrett, 1985) showed low gene flow to Waydelich Creek. Thus while we did find strays from all Auke Creek population components in local creeks, they may have enjoyed limited reproductive success.  The large degree of mixing does however provide the mechanism for gene flow and colonization.

That local adaptations and differences persist among pink salmon populations exposed to large potential introgression underscores the importance of environmental selection for fitness traits, such as migration timing or developmental rates, in the adaptive strategy of pink salmon.  Further straying and survival studies on Auke Creek pink salmon are being considered, perhaps combining genetic, thermal, and CWT marking to directly address the question of straying and resultant gene flow to other pink salmon populations in the vicinity of Auke Creek.  A higher level of CWT- and thermal-mark recovery will provide better estimates of pink salmon populations in the streams as well as better estimates of straying and survival