Chase et al.: Reproductive life history of anadromous Osmerus mordax in Massachusetts 
161 
Parker River 
■ Age 5+ 
o Age 4 
s Age 3 
□ Age 2 
□ Age 1 
■ Age 5+ 
□ Age 4 
n Age 3 
□ Age 2 
□ Age 1 
Jones River 
■ Age 5+ 
□ Age 4 
s Age 3 
□ Age 2 
□ Age 1 
Fore River Saugus River 
Figure 7 
Age-frequency distribution for rainbow smelt (Osmerus 
mordax) sampled in 4 coastal rivers in Massachusetts 
during spawning runs in the spring of 2005-2007 as part 
of this study and for rainbow smelt caught during previous 
sampling in the Parker River (Murawski and Cole, 1978) 
and the Jones River (Lawton et ah, 1990). Age-frequency 
distribution for fish from the Parker and Jones Rivers 
were significantly different between years (chi-square 
test: df-3, P<0.01), except for rainbow smelt caught at the 
Parker River in 1974 and 2006 (P=0.444) and at the Jones 
River in 2006 and 2007 (P=0.220). 
5 annual samples in those studies than in each of the 
annual samples of our study (Fig. 7). The change is most 
pronounced in the Jones River, with an average annual 
frequency of 6% for age-1 rainbow smelt from 1979 
through 1981 and 55% from 2005 through 2007. Addi¬ 
tional evidence of changes in age structure was found in 
the Jones River: the average annual frequency of age-3 
fish decreased from 23% during 1979-1981 to 3% during 
2005-2007. These comparisons indicate a shift in age 
composition with a larger role for age-1 rainbow smelt in 
spawning runs. 
The Chapman-Robson age-based survival estimator 
used in this study produced lower estimates of S and 
higher values of Z for rainbow smelt caught in the Jones 
and Parker Rivers than the estimates reported for the 
previous studies (Table 6). For all data sets of ages of 
rainbow smelt, Z estimates were calculated by using the 
Chapman-Robson estimator with years and sexes com¬ 
bined. Samples from the Fore River, with the highest rep¬ 
resentation of rainbow smelt older than age 2, had S and Z 
estimates closest to those of samples from the Parker and 
Jones Rivers in the earlier studies. Rainbow smelt from 
the Jones and Saugus Rivers had lower S and higher Z 
than samples from the Fore River and those examined in 
Murawski and Cole (1978) and Lawton et al. 8 , as well as in 
another earlier study for which age data were reported for 
rainbow smelt in the Miramichi River in New Brunswick, 
Canada (McKenzie, 1964). 
Length-weight analysis Length and weight data for the 
rainbow smelt from fyke net sampling and the trawl sur¬ 
vey were evaluated as a measure of fish condition and to 
determine if differences occurred among sampling locations 
(Table 7). All regressions of the relationship of length and 
weight exhibited allometric growth with minor apparent 
differences between the trawl samples collected in winter 
and fyke net samples collected in spring. For all compar¬ 
isons, the regression slopes were larger for females than 
males. There were no differences in the same-sex regression 
slopes for rainbow smelt collected at the Fore and Saugus 
Rivers. However, for both sexes, the regression slopes for 
the fyke net samples collected in coastal rivers did differ 
from those for the trawl samples collected in marine waters. 
Female rainbow smelt were significantly heavier at length 
than males for all locations, and the marine samples were 
heavier at length than the river samples. 
Sex ratio The trawl samples collected in winter included 
significantly more females than males (1.3:1.0 female-to- 
male ratio; chi-square test: x 2 =7.13, df=l, P=0.008), and the 
fyke net samples collected in spring included significantly 
more males than females (male-to-female ratios: Jones 
River, 1.8:1.0; Fore River, 3.3:1.0; Saugus River, 10.6:1.0; 
and Parker River, 5.7:1.0; chi-square test: x 2 =88> 1092, 
1653, and 533, respectively, df=l, PcO.OOl). The higher 
number of males captured with the fyke net is a sampling 
bias that resulted from males spawning more frequently 
than females (Marcotte and Tremblay, 1964; McKenzie, 
1964; Murawski et al., 1980). It is possible that the unbal¬ 
anced sex ratio found in trawl samples is biased by a higher 
proportion of males than females for the 60 immature rain¬ 
bow smelt that were excluded from the sex ratio estimate. 
Condition indices Fulton’s condition factor for the trawl 
samples from marine waters and for the fyke net sam¬ 
ples from the Fore and Saugus Rivers ranged between 
