Chase et al.: Reproductive life history of anadromous Osmerus mordax in Massachusetts 
155 
fish (Ricker, 1975). Differences in condition factor among 
sample locations and sex were tested with the Kruskal- 
Wallis rank sum test and the Kruskal-Wallis multi¬ 
ple comparison test. The gonadosomatic index (GSI), 
calculated as (OW/BW)xlOO, and the fecundity index, 
calculated as total fecundity/TL 3 , were determined for 
marine samples and the Fulton’s condition factor was 
calculated for both the river and marine samples. Sex 
ratios of rainbow smelt sampled at all locations were 
tested for significant differences by using the chi-square 
test (a=0.05). 
Size and age analyses 
The Massachusetts Division of Marine Fisheries conducts 
annual fyke net sampling of rainbow smelt at spawning 
runs in coastal rivers to monitor abundance and size and 
age composition. Fyke nets were set in the Jones, Fore, 
Saugus, and Parker Rivers (Fig. 1) at mid-channel, inter¬ 
tidal locations downstream of spawning riffles to intercept 
spawning rainbow smelt 3 days a week over an 11-week 
spawning run (Chase et ah, 2009; Enterline et al., 2012). 
Fyke nets were set in 2004-2007; however, the catch data 
for 2004 were not used because of changes in methods fol¬ 
lowing that pilot season. The fyke nets have square wings 
that are 1.2 x 1.2 m and lead to a square entrance, also 1.2 x 
1.2 m, that connects to 7 hoops with a terminal codend 
after 2 throats (all mesh sizes were 7 mm). At all locations, 
rainbow smelt were counted and measured for TL, and 
their sex was determined. Total length was measured for 
100 male and female rainbow smelt from each haul, with 
all remaining rainbow smelt enumerated by sex. Spawn¬ 
ing male rainbow smelt are readily identified by the pres¬ 
ence of nuptial tubercles. At the Fore and Saugus Rivers, 
samples were collected randomly from each haul to addi¬ 
tionally measure TW and to collect scales for aging. The 
subsamples for aging were collected with a weekly tally of 
5 samples per 10-mm bin per sex. 
Length and weight correction Measurements of the sam¬ 
ples collected with a fyke net were made when fish were 
fresh, but the marine samples were frozen prior to pro¬ 
cessing. To allow direct comparison of length and weight 
data for the 2 sets of collections, a sample of rainbow smelt 
caught with a fyke net in the Fore River were measured 
for length and weight before and after they were frozen, 
and changes in length and weight were compared with 
regression analysis. 
Length analysis The TL of rainbow smelt collected at the 
4 stations where fyke net sampling occurred and caught 
during the trawl survey were evaluated for differences 
among locations and sexes by using the Kruskal-Wallis 
rank sum test and the Kruskal-Wallis multiple compar¬ 
ison test. Given limitations of sample sizes, all samples 
from the trawl survey were pooled for the study period. 
Length and length-at-age analyses (only for samples from 
the Fore River) were conducted on pooled samples for 
the study period when sample sizes allowed meaningful 
comparisons. The interpretation of length data among 
sample locations must be made cautiously because of 
the within-season repeat-spawning behavior of males 
(Murawski and Cole, 1978) and changes in size and age 
of rainbow smelt that occur temporally during a spawning 
run (Marcotte and Trembley, 1948; McKenzie, 1964). 
Age analysis Age keys were prepared for Fore River 
samples and for the data pooled for samples from the 
Fore and Saugus Rivers. Age keys combined sexes and 
annual data for the study period. The pooled age key was 
applied to length frequencies of samples from the Saugus, 
Jones, and Parker Rivers to estimate age composition for 
those rivers. The age compositions were applied to the 
Chapman-Robson age-based survival estimator, by using 
the Fishmethods package, vers. 1.10-4 (Nelson, 2017) in 
R to estimate annual survival (S) and instantaneous total 
mortality (Z) rates. Annual age data for each sampling 
location were combined for the study periods to smooth the 
effect of year-class variation and as a practical approach 
to allow a comparison of sparsely reported population 
demographics for anadromous rainbow smelt. The annual 
age proportions estimated for rainbow smelt from fyke net 
sampling were compared with historical data (Murawski 
and Cole, 1978; Lawton et al. 8 ) by using the chi-square 
test (a=0.05). 
Length-weight analysis Length-weight data from the 
trawl survey samples were compared with length-weight 
data from the fyke net samples to examine seasonal differ¬ 
ences in condition of rainbow smelt, and as an alternative 
means to consider if the pooled samples of rainbow smelt 
from the trawl survey composed a representative sample 
for fecundity analysis. We used linear regression on natu¬ 
ral log-transformed TL and TW data for rainbow smelt col¬ 
lected for fecundity analysis and fyke net monitoring. The 
homogeneity of slopes was tested by analysis of covariance 
to determine if length or weight differed significantly by 
location, sex, or age class (P<0.05). Length was the inde¬ 
pendent variable, weight was the dependent variable, and 
location, sex, and age class were the group variables. 
Results 
Reproductive analyses 
During the trawl survey in 2004-2007, 634 rainbow 
smelt were collected from 13 tows of the trawl between 
11 January and 5 March (Fig. 1). Rainbow smelt were 
captured in nearshore, marine waters off Massachusetts 
at an average depth of 38 m (range: 19-89 m) with an 
8 Lawton, R., P. Brady, C. Sheehan, S. Correia, and M. Bor- 
gatti. 1990. Final report on spawning sea-run rainbow smelt 
(Osmerus mordax ) in the Jones River and impact assessment of 
Pilgrim Station on the population, 1979-1981. Pilgrim Nuclear 
Power Station Mar. Environ. Monit. Program Rep. Ser. 4, 72 p. 
[Available from Mass. Div. Mar. Fish., 251 Causeway St., Ste. 
400, Boston, MA 02114.] 
