Chase et al.: Reproductive life history of anadromous Osmerus mordax in Massachusetts 
163 
PcO.OOl). The GSI values ranged from 3.5% to 38.1% and 
increased with length and age (Table 2). 
Discussion 
The comparison of rainbow smelt caught during winter in 
coastal waters off Massachusetts to rainbow smelt caught 
during spring spawning runs in coastal rivers contributes 
information about sparsely described aspects of the repro¬ 
ductive life history of anadromous rainbow smelt. Data 
from this comparison provide quantitative relationships 
of fecundity with size and age, a maturity ogive, the first 
ogive available for anadromous rainbow smelt, and bio¬ 
logical references necessary to investigate influences on 
recruitment of rainbow smelt. Age and length data indi¬ 
cate that the role of age-1 rainbow smelt in spawning 
populations may be increasing. During the study period, 
the proportion of age-1 rainbow smelt in spawning runs 
was higher than proportions previously found in Massa¬ 
chusetts (Murawski and Cole, 1978; Lawton et al. 8 ), and 
differences in the size of age-1 rainbow smelt and their 
proportion of the run age composition were observed 
between proximate spawning runs. The point of increas¬ 
ing participation of age-1 rainbow smelt in spawning runs 
is supported by our data; however, future investigations 
should consider that the collection methods were not 
identical between studies, raw length and age data were 
not available for samples in the other studies, and age 
composition data in all studies were derived from pooled 
age keys. 
The cause of differences in life history characteristics 
among populations of rainbow smelt is not well described. 
Our findings of differences in length and age imply 
that maturation and growth rates vary over a modest 
geographical range. Kovach et al. (2013) conducted an 
analysis of genetic population structure for anadromous 
rainbow smelt from the Gulf of Maine, including sam¬ 
ples from the 4 fyke net stations in Massachusetts, and 
found differences within this small geographic range but 
no river-level genetic variation. The populations of the 
Fore River and Saugus River were most similar geneti¬ 
cally (and were also examined as a single Boston Harbor 
group). The populations of the Parker River and Jones 
River were related to the Boston Harbor group, but a 
majority of their genetic distinctiveness was associated 
with other populations north of the Parker River and 
south of the Jones River. 
Fecundity and maturation 
The small sample sizes of previous studies of fecundity 
in rainbow smelt limited the quantitative analyses of 
fecundity data and the interpretation of life history rel¬ 
evance. The number of samples collected from the trawl 
survey and used in fecundity analysis in our study (n=110) 
more than doubled that of the largest previous study of 
anadromous rainbow smelt (n=48; Chen, 1970), and anal¬ 
yses with those samples included fecundity regression 
relationships with 4 size and age predictor variables. Our 
collection of late-winter, marine rainbow smelt avoided 
biases associated with sampling during active spawning 
runs. The sample size of 110 was reached by pooling trawl 
tows and years, and pooling data does not allow inferences 
on annual or geographic differences. The general similar¬ 
ities in length distributions, length-weight relationships, 
and age structure among the sample locations indicate 
that pooling samples from the trawl survey over the study 
range (approximately 120 km) was suitable for regression 
analysis. A similar conclusion can be made on the pool¬ 
ing of age samples by year and sex and the application 
of age keys to sample rivers where rainbow smelt were 
not aged. Our analysis would be improved with the ability 
to consider annual differences in age composition by river 
and sex. However, the suspected influences of differences 
in annual growth rates and cohort strength on changes in 
mean length do not appear to translate to large changes 
in length-at-age data and age modes for this fast growing, 
short-lived species. 
Our finding of increasing total fecundity and generally 
stable relative fecundity with age is interesting because 
of the implication of the higher reproductive potential of 
the uncommon older rainbow smelt due to higher fecun¬ 
dity and large oocyte size (Hixon et ah, 2014). However, 
neither our study nor past reproductive studies of rain¬ 
bow smelt have quantitatively addressed this point. 
Further investigation is needed on the role of oocyte 
size and fecundity in anadromous smelt reproductive 
potential. 
The 2 previous studies with higher numbers of samples 
for fecundity analysis were focused on landlocked popu¬ 
lations of rainbow smelt (n=172, Chen, 1970; n=208, Luey 
and Adelman, 1984). Chen (1970) found that fecundity 
for 3 landlocked populations and 1 anadromous popula¬ 
tion of rainbow smelt was significantly related to fork 
length, BW, OW, and age. His estimates of relative fecun¬ 
dity for the 4 stocks indicate that the anadromous stock 
was the most productive and had the highest mean Ful¬ 
ton’s condition factor for both sexes. Luey and Adelman 
(1984) found significant differences in growth, fecun¬ 
dity-length relationship, and length distribution among 
3 regions in western Lake Superior over a range of 220 
km; yet the differences were slight. They suspected that 
these population differences were related to selective 
pressures (environmental and predatory) from the rela¬ 
tively recent introduction of rainbow smelt to the Great 
Lakes (Luey and Adelman, 1984). They also found few 
significant differences in length frequency among gear 
types, years, and sampling locations but consistent differ¬ 
ences in length distribution between males and females 
(Luey and Adelman, 1984). 
A comparison of the maturity ogive derived from 
marine samples to the length-at-age data from river 
samples indicates that not all age-1 rainbow smelt are 
recruited to annual spawning runs. Our study also pro¬ 
vides evidence of a higher proportion of age-1 rainbow 
smelt in the spawning runs than has been found in previ¬ 
ous studies. The actual extent of changes in the maturity 
