164 
Fishery Bulletin 117(3) 
schedule of anadromous rainbow smelt cannot be deter¬ 
mined because of the lack of previous ogive estimates, 
sampling differences, and limited historical information 
on cohort dynamics and raw length data. Our work high¬ 
lights important data gaps, especially the lack of under¬ 
standing of the environmental and phenotypic triggers 
on maturity of rainbow smelt and of the underlying influ¬ 
ences that caused the increase of age-1 rainbow smelt in 
spawning runs. 
Eggs per recruit 
The estimates of Z and fecundity at age generated in 
this study were applied to an eggs-per-recruit (EPR) 
analysis that used the Fishmethods package in R and 
compared inputs from previous studies of rainbow smelt 
in the Parker River (Murawski and Cole, 1978), Jones 
River (Lawton et al. 8 ), and Miramichi River (McKenzie, 
1964). The purpose of the EPR analysis was to deter¬ 
mine how lifetime fecundity responds to changes in fish¬ 
ing mortality (F). We introduce the EPR analysis here, 
in the “Discussion” section, because of its dependency 
on historical references for model input parameters and 
exploratory approach. The model assumes that fecundity 
has not changed over time and requires assigned values 
for F and natural mortality (M); however, no empiri¬ 
cal measurements exist for rainbow smelt in our study 
area. Murawski and Cole (1978) used McKenzie’s (1964) 
age data for rainbow smelt from the Miramichi River to 
derive Z (1.043) and tagging data to derive F (0.062) and 
used the equation Z=M+F to estimate M as 0.981; then 
they applied M to derive an F of 0.268 for the population 
of rainbow smelt in the Parker River during 1974-1975. 
These values were used in our study as starting points for 
a sensitivity analysis on assumed levels of M and F given 
the estimates of Z. 
The most influential inputs for the EPR model are M 
and the proportion offish mature at each age (PMAT). The 
only value available for the proportion of mature age-1 
anadromous rainbow smelt is the 0.6 derived from the 
trawl survey data used in this study. The first run of the 
EPR model was made with PMAT fixed at 0.6 for all data 
sets except the data set from the study of rainbow smelt 
from the Miramichi River where slower growth resulted 
in no age-1 rainbow smelt participating in the spawn¬ 
ing run. The results shown in Figure 8A indicate that 
samples from the Parker and Jones Rivers in the 1970s 
and 1980s had the highest EPR despite having higher 
Z than those from the Miramichi River. The Miramichi 
River sample had the strongest presence of older rain¬ 
bow smelt; however, with no age-1 rainbow smelt in the 
spawning run at that location, lifetime EPR was lower. 
Results of the EPR analysis indicate that the rainbow 
smelt populations in rivers in our study had relatively 
low and stable EPR and that reducing F caused little 
improvement to EPR because Z was dominated by M. 
The model estimated that EPR for rainbow smelt in the 
Jones River declined 38% from that reported in Lawton 
et al. 8 and that EPR for those in the Parker River declined 
24% from that reported in Murawski and Cole (1978) (at 
F=0.200 on the x-axis, Fig. 8A). 
In the interest of investigating the role of age-1 matu¬ 
rity, PMAT was reduced to 0.3 for the data sets based 
on samples from the Parker and Jones Rivers for a 
second run of the model. The decline in EPR from the 
model run with a PMAT of 0.6 to the model run with 
a PMAT of 0.3 for hypothetical spawning runs reflects 
the importance of age-1 participation in the spawning 
runs sampled for this study: with PMAT reduction, the 
EPR for samples from the Jones River declined 32% and 
the EPR for samples from the Parker River declined 
29% (at F=0.200 on the x-axis, Fig. 8B). The third run 
of the EPR model was used to analyze the potential for 
reduced F in samples collected during spawning runs for 
this study (Fig. 8C). For the first run of the EPR model, 
F was set at 0.268, the Murawski and Cole (1978) esti¬ 
mate for rainbow smelt from the Parker River, for all 
samples of spawning runs except those from the Mira¬ 
michi River for which an F estimate of 0.062 was avail¬ 
able. However, the fishery for rainbow smelt in the 1970s 
occurred during winter from ice shacks set in the inter¬ 
tidal Parker River. That ice-shack fishery is essentially 
gone because it is uncommon in recent winters for estu¬ 
ary ice to be thick enough to safely support shack fishing. 
Therefore, the third EPR analysis set F at 0.000, with 
Z equal to M (Fig. 80. Under this scenario of high M, 
the decline in EPR estimated in our study from the EPR 
reported for earlier studies was 44% for samples from the 
Jones River and 33% for samples from the Parker River 
(at F=0.200 on the x-axis, Fig. 80. Changes in growth 
and maturity are possible influences that caused the 
proportion of age-1 rainbow smelt found in spawning 
runs sampled in our study to increase. However, this 
influence cannot be separated from cohort dynamics 
without further studies. 
Evidence from our study and previous studies 
(Murawski and Cole, 1978; O’Malley et al., 2017) sup¬ 
ports the premise that age-1 anadromous rainbow smelt 
are partially recruited to spring spawning runs. Further, 
our data indicate that the proportion of age-1 rainbow 
smelt in spawning runs in the study area has increased 
from the level observed several decades ago and that the 
presence of rainbow smelt older than age 2 has declined. 
The decline in the proportion of older rainbow smelt in 
the spawning run is the primary influence that results in 
higher Z estimates in this study. The higher proportion of 
age-1 rainbow smelt does not directly affect Z estimates 
because this cohort is not fully recruited and excluded 
from survival estimates. It has been noted that more 
males than females are precocious at age 1 in spawning 
runs of anadromous rainbow smelt (Murawski and Cole, 
1978; Sutter, 1980). In this study, a higher proportion of 
males than females was found for mature age-1 rainbow 
smelt at each river sampled with a fyke net, at ratios 
similar to the sex ratios for all ages combined from fyke 
net sampling; however, the bias created by a higher rate 
of repeat spawning in males limits the quantification of 
the true sex ratio. 
