Wood et al.: A comparison between warm-water fish assemblages of Narragansett Bay and Long Island Sound waters 
91 
Table 1 
Description of data used from Narragansett Bay and Long Island Sound for this study, including the location of sampling site 
(RI=Rhode Island data from Narragansett Bay and the surrounding salt ponds, LIS=Long Island Sound), number of stations 
sampled, years when sampling occurred, sampling frequency, gear type used for sampling, and the source for the data. 
Site 
Number of 
stations sampled 
Years 
Sampling frequency 
Gear type 
Source 
RI 
2 
1959 to present 
Year round 
weekly 
Trawl 
10.4-m headrope 
5.1 cm codend (sic) 
30 min at 2 knots 
University of Rhode 
Island Graduate School 
of Oceanography 
37 
1987 to present 
Year round 
monthly 
Trawl 
12.2-m headrope 
0.95 cm codend (sic) 
20 min at 2.5 knots 
Rhode Island Department 
of Environmental 
Management 
5 
1987 to present 
April- 
November 
monthly 
Seine 
61-m net 
6.4-mm mesh 
LIS 
Random 
(about 80) 
1984 to present 
Fall 
Trawl 
9.1-m headrope 
5.1 cm codend (sic) 
30 min at 3.5 knots 
Connecticut Department 
of Environmental 
Protection 
7 
1984 to present 
September 
Seine 
7.6-m net 
6.4-mm mesh 
6 
1976 to present 
Bi-weekly 
Trawl 
9.1 m headrope 
6.4 mm codend liner (sic) 
20 mins at 1.2-1. 4 knots 
Milford Laboratories 
5 
1969 to present 
May- 
November 
Seine 
9.14 m net 
6.4-mm mesh 
that are not present in the databases represented in 
this study. In the jackknife estimation, the number 
of species present (S ; ) was calculated with each of the 
years (i) removed in turn. The jackknife estimate was 
calculated by averaging together pseudovalues (Yj), 
which represent the likely number of species for each 
year, by using the equation 
Y- = nS - (n-l)S,, 
where n = the number of years. 
The jackknife mean and variance were calculated 
from the resulting pseudovalues (Y i ). The jackknife 
estimate was determined to obtain more likely values 
of species richness for each area. These estimates were 
compared between estuaries with a two-sample t-test 
(Johnson and Bhattacharyya, 2001) to determine if 
there was a significant difference in species richness. In 
addition, species accumulation curves were calculated 
for both populations and graphed together to assess 
whether the curves were similar for the two estuaries 
for the years 1987-2000. 
A multidimensional scaling (MDS) analysis was 
conducted in the statistical analysis program PRIM- 
ER (PRIMER-E Ltd., Ivy Bridge, U.K.) to determine 
whether species composition of the annual abundances 
differed between Narragansett Bay and Long Island 
Sound. A Bray-Curtis measure of similarity was used 
for developing the similarity matrix (Krebs, 1999). Data 
were standardized, and a fourth root transformation 
was applied to the data to give less weight to abun- 
dant species. An MDS plot was computed for all of 
the years by species abundances in each location. A 
similarity percentage (SIMPER) analysis (in PRIMER) 
was then conducted to determine which of the species 
were driving the similarity and dissimilarity between 
the areas. 
Tinning of occurrence of warm-water fishes 
Five aspects of the timing of the warm-water fauna were 
examined. First, the relationship between temperatures 
in the two estuaries was examined. Correlations were 
calculated between weekly mean temperatures in Nar- 
ragansett Bay and Long Island Sound. Correlations 
were also calculated between the means of summer 
temperatures (June, July, and August combined) and 
semi-annual temperatures (May, June, July, August, 
