822 
Fishery Bulletin 95(4), 1997 
and survival, mid-October to November, coincides 
with the period when fingerlings first become abun- 
dant in the lower river and begin their transition to 
oceanic waters. The period centered on mid-Septem- 
ber, when there is some evidence of a negative corre- 
lation between flow variability and survival, corre- 
sponds to the earlier time when fry are still migrat- 
ing downriver and have yet to grow to the point where 
they are able to withstand the transition to salt wa- 
ter. Survival is not correlated, either positively or 
negatively, with flow variability at the beginning of 
the season (August, when most fry have yet to hatch) 
or at the end of the season (January, by which time 
most fingerlings have left the river). The correlation 
also tends to disappear when flow variability is av- 
eraged over more than about 100 days, a period that 
is consistent with the 90-day freshwater residence 
period of Rakaia fingerlings. 
The tendency for survival to be positively corre- 
lated with flow variability rather than flow volume 
(as measured by Q), and the short duration of each 
individual flood peak, suggest that these flood pulses 
must be an integral part of any plausible linking 
mechanism. Maximum survival appears to result 
when stable flows prior to mid-October are followed 
by a few (perhaps two or three) large floods over the 
next four to six weeks. By contrast, seasons when 
there are no major floods during October and No- 
vember seem to result in poor survival irrespective 
of flows earlier in the season. Although any discus- 
sion based solely on the present results is specula- 
tive, I suggest that sudden increases in Rakaia dis- 
charge coinciding with peak emigration of fingerlings 
from the river mouth may increase survival by buff- 
ering the transition from fresh to saline waters in 
the vicinity of the offshore plume. If so, these pulses 
may help to compensate for the lack of an estuary at 
the Rakaia mouth and the low importance of the la- 
goon as a salmon rearing habitat, 2 one of the key 
features distinguishing the Rakaia (and other New 
Zealand salmon-producing rivers) from the extensive 
tidal basins below the Sacramento River mouth 
(Kjelson et al., 1982). The effect may be compounded 
by the well-documented tendency for downstream 
migration rates to increase with flow, both in New 
Zealand (Irvine, 1986) and North American popula- 
tions (e.g. Kjelson et al., 1982; Berggren and Filardo, 
1993), so that each flood pulse increases both the 
number of fingerlings leaving the river and their 
prospects for successful acclimation to salt water. 
Outflow of turbid flood waters may also increase sur- 
vival by reducing visibility, and hence decreasing 
losses to inshore marine predators such as kahawai 
(. Arripis trutta), although reduction in visibility is 
likely to be no more than a secondary effect (cf. St. 
John et al., 1992). A positive correlation between 
survival of hatchery-reared Atlantic salmon ( Salmo 
salar ) and maximum river discharge during the first 
seven days after release has been attributed to re- 
duced predation at higher flows (Hvidsten and Han- 
sen, 1988). 
A distinguishing feature of chinook salmon com- 
pared with other species of Oncorhynchus is their 
gradual acquisition of seawater tolerance while still 
in fresh water, without the sudden transition associ- 
ated with smoltification in species such as coho (O. 
kisutch) or steelhead (O. mykiss) (Hoar, 1976). By 
early November, Rakaia fingerlings are of an age and 
size close to the generally accepted minima for suc- 
cessful transfer to seawater (Clarke and Shelbourn, 
1985; Franklin et al., 1992), and water temperatures 
in the Rakaia River (12-15°C; Unwin, 1986) and at 
sea ( 12-14°C ) 8 appear to be within the optimal range 
for juvenile chinook salmon reported by these stud- 
ies. However, acclimation to seawater also depends 
on the rate of transition. A gradual transfer to saline 
waters allows even very young fish to acclimatize 
successfully (Hoar, 1976), as evidenced by the abun- 
dance of chinook fry in low salinity estuarine waters 
in North America (Reimers, 1973; Healey, 1980; Levy 
and Northcote, 1981), including the lower Sacra- 
mento River (Kjelson et al., 1982). Changes in es- 
tuarine ecology during low-flow seasons in the Snake- 
Columbia River system have been suggested as a 
factor contributing to reduced survival of Snake River 
chinook (Williams and Matthews, 1995). In the Strait 
of Georgia, where the Fraser River plume forms a 
well-developed halocline at a depth of 5-10 m, juve- 
nile salmonids showed a preference for surface wa- 
ters of low salinity (10-15 ppt) in the plume, com- 
pared with the more saline waters (25-30 ppt) in 
other regions of the Strait (St. John et al., 1992). The 
Georgia Strait study also reported a tendency for 
plankton and larval fish to align with the plume 
boundary, providing enhanced feeding opportunities. 
There have been no studies on salinity gradients 
off the Rakaia mouth, but nearshore salinity off 
Otago Harbour (on the east coast of the South Is- 
land 220 km south of the Rakaia) is inversely corre- 
lated with discharge from the Clutha River a fur- 
ther 100 km to the south (Jillett, 1969). The coastal 
shelf off the Rakaia River has a very gentle slope, 
with the 20-m isobath 5-10 km offshore (see Fig. 1). 
Consequently, peak Rakaia outflows should have a 
substantial impact on inshore salinity. For example, 
a daily mean discharge of 1 200 m 3 /s over 24 h (which 
is less than the mean annual spring flood) represents 
a total volume of fresh water of 0. 1 km 3 , equivalent 
8 1997. NIWA, Christchurch, New Zealand. Unpubl. data. 
