Weitkamp et al.: Seasonal and interannual variation in juvenile salmonids in the lower Columbia River 
445 
low water with a beach seine in 2007-08 (53.2% and 
73.7%, respectively; Roegner et al., in press). This find- 
ing suggests that relatively more unclipped (presumably 
wild) Chinook salmon were closer to shore. 
This overwhelming hatchery origin is consistent with 
the low variation we observed in the timing of juvenile 
salmon presence and abundance in the estuary (Fig. 4) 
and fish size (Fig. 6), given that most hatchery fish are 
released at a target size and date, with little varia- 
tion from year to year (FPC database). Accordingly, we 
would expect to see greater variation in both timing 
and size of salmon if more fish were of wild origin, 
because of the greater fine-scale life history diversity 
observed in other wild-dominated systems and his- 
torically within the Columbia River (Rich, 1920; Myers 
and Horton, 1982; Burke, 2004, Bottom et al., 2005; 
Campbell, 2010). 
For our calculation of the percentage of hatchery 
fish, we assumed that the survival of all hatchery fish 
migrating from the hatchery to the estuary is equal. 
However, this assumption is unlikely to be true because 
fish that travel longer distances and pass through more 
mainstem dams have greater mortality (Williams et al., 
2001). This differential mortality undoubtedly contrib- 
uted to the inflated percentages of hatchery subyearling 
Chinook and coho salmon we estimated (>100% in mul- 
tiple years because percentages of clipped fish in our 
catches exceeded percentages of clipped fish reported 
at hatcheries). For these 2 groups, survival rates were 
likely to differ by distance from the estuary, and the 
percentage of clipped fish varied as well. The percent- 
age of clipped subyearling Chinook and coho salmon 
released in the upper Columbia and Snake rivers were 
much lower (43-46% for subyearling Chinook, 7-36% 
for coho) than the percentage of fish released lower in 
the basin (90% for subyearling Chinook, 76% for coho). 
Although these releases in the upper basin are a small 
fraction (<29%) of the total releases every year, higher 
expected mortality rates would inflate our estimates 
of hatchery influence, likely contributing to our ex- 
tremely high estimates for both species. Consequently, 
the percentage of clipped fish we observed in the estu- 
ary for these 2 species (76-77%, Table 4) represent the 
minimum hatchery influence; the true level of hatchery 
influence cannot be estimated without methods of as- 
signing origin for unclipped fish (e.g., using scales, 
otoliths, or other means). This bias should not influence 
estimates for yearling Chinook salmon or steelhead, 
however, because mark rates are more consistent across 
the Columbia basin (FPC database). Clearly, clipping 
all hatchery fish would greatly increase our ability to 
identify hatchery fish and result in better estimates of 
hatchery influence and detection of potential ecological 
interactions. 
We also observed that clipped (hatchery) coho salmon 
and steelhead were larger than their unclipped (hatch- 
ery and wild) counterparts. In stream environments, 
such size discrepancies have been shown to have nega- 
tive consequences for wild fish (Weber and Fausch, 
2005), although whether this finding applies to estua- 
rine environments has not been addressed (Naish et al., 
2008). Two recent studies in which juvenile (age 1.0) 
Chinook and coho salmon were examined in marine en- 
vironments revealed no evidence for competitive advan- 
tage (in the form of feeding intensity or prey selection) 
of larger hatchery individuals over their unclipped (and 
smaller) counterparts (Sweeting and Beamish, 2009; 
Daly et al., 2011). 
We did not observe a size difference between clipped 
and undipped subyearling and yearling Chinook salmon. 
However, actual size variation between clipped and 
unclipped Chinook salmon may have been confounded 
by our length-based age assignment; small, unclipped 
yearling Chinook salmon may have been incorrectly 
assigned to the subyearling age category while large, 
clipped subyearling Chinook salmon may have been as- 
signed to the yearling category. In addition, the size of 
Chinook smolts released from different hatcheries can 
vary greatly (FPC database), adding to the confusion. 
Restriction of such size comparisons to specific salmon 
stocks (e.g., Daly et al., 2011) would help resolve this 
issue. 
Historical comparisons of fish communities 
in the lower Columbia River estuary 
The fish assemblage we observed in 2007-10 generally 
was similar to that documented 3 or more decades ear- 
lier (Haertel and Osterberg, 1967; McCabe et al., 1983; 
Bottom and Jones, 1990). For example, all the species 
that we consistently caught (Tables 2, 3) were listed as 
commonly occurring in lower estuary pelagic habitats 
by earlier researchers (e.g., McCabe et al., 1983; Dawley 
et al. 7 ; Bottom and Jones, 1990). As in these earlier 
studies, we did not catch any marine fish that might 
be predatory on juvenile salmon (e.g., adult lingcod, 
Pacific hake [Merluccius productus], and Pacific chub 
mackerel [Scomber japonicus ]), with the exception of 
adult salmon. Haertel and Osterberg (1967) report that 
Pacific tomcod were a common species in their survey 
of estuarine fish in the 1960s (both juveniles and older 
fish), and Dawley et al. 7 report occasionally catching 
eulachon ( Thaleichthys pad ficus) and redtail surfperch 
(Amphistichus rhodoterus), neither of which we have 
encountered. Although the absence of eulachon in our 
catches is likely due to their recent population decline 
(Gustafson et al., 2011), the reason for the absence of 
other fishes is unknown. 
The difference in abundance of American shad (which 
were primarily juveniles) in the estuary in 1978-80 and 
in 2007-10 (in our study) is consistent with changes in 
counts of adults across Bonneville Dam (FPS database). 
Specifically, data from Dawley et al. 7 indicates that 
American shad contributed 1%, 2%, and 10% to the 
fish assemblage in April, May, and June, respectively, 
in 1978-80, when dam counts averaged 947,000 adults. 
The relative abundance of American shad we observed 
in our study was roughly twice as high (2%, 5%, and 
20% in April, May, and June, respectively), consistent 
with twice as many adults crossing Bonneville Dam 
