Kane: Spatial and temporal abundance patterns of Centropages hamatus 
87 
Statistical analysis 
Estimates of Centropages hamatus and predator 
abundance were log transformed [log 10 (no./10Gm 3 
+ 1)] prior to contouring and data analysis. Contoured 
C. hamatus distribution maps were made by using 
Surface III software (Sampson, 1988) on station 
abundance data from the 11-yr data set grouped by 
monthly intervals. 
Evaluation of species interannual abundance vari- 
ability was facilitated by subdividing the ecosystem 
into four subareas: Middle Atlantic Bight (MAB), 
Southern New England (SNE), Georges Bank (GBK), 
and Gulf of Maine (GOM) (Fig. 1). Each subarea is 
characterized by distinct patterns of circulation and 
bathymetry (Sherman et a!., 1983). The average an- 
nual cycle of abundance and its variation was por- 
trayed for each subarea by plotting the monthly mean 
abundance of all samples with its 95% confidence 
interval bar. Individual survey mean abundance and 
its 95% confidence interval bar were then superim- 
posed on the latter plot. Surveys where the error bar 
did not overlap the one from the average cycle were 
judged to be situations where abundance departed 
substantially from the average cycle. Only surveys, 
except the one noted below, that covered 75% or more 
of a subarea were included in the analysis of 
interannual variability. Statistical analyses, compar- 
ing individual survey means with the time series 
monthly mean were not undertaken because they re- 
quire the assumption of independence. 
Several surveys (see Table 1) prior to 1981 were 
conducted by foreign vessels that did not have per- 
mission to sample east of the U.S. -Canada maritime 
boundary line in the GBK and GOM subareas. Al- 
though areal coverage in these surveys was reduced 
approximately 40% in relation to complete surveys, 
I included them in the analysis of this study because 
the area undersampled was consistent and our sur- 
veys still provided adequate coverage of the depth 
strata found within the two subareas. 
Spearman’s rank correlation coefficients were cal- 
culated for monthly subsets of station data to mea- 
sure the strength of the relationship within indi- 
vidual months between Centropages hamatus abun- 
dance and the following variables: surface tempera- 
ture, bottom depth, and the average water-column 
values of temperature, salinity, and total chlorophyll. 
Initial distribution plots of C. hamatus revealed that 
species abundance has a strong onshore-offshore 
gradient. Thus, to control the effect of depth on the 
calculation, Spearman’s partial correlation coeffi- 
cients were calculated for monthly subsets where 
both abundance and the other variable were signifi- 
cantly (,P<0.05) correlated to depth. 
Results 
Distribution and abundance 
The time-series mean distribution charts by month 
for Centropages hamatus are presented in Figure 2, 
A and B. Immediately apparent is the persistent on- 
shore-offshore abundance gradient throughout the 
study area. There are high concentrations of the cope- 
pod inshore and within the shoal waters of GBK. 
Abundance in offshore waters is always much lower. 
Centropages hamatus is found throughout most of 
the ecosystem at some time during the year, the only 
exception being certain areas of the eastern offshore 
waters of the GOM where it is absent year round. 
The timing of the annual abundance cycle of 
Centropages hamatus was not consistent through- 
out the ecosystem. The population in southern 
reaches of the study area declines through the sum- 
mer, nearly disappearing from the water column 
during early autumn (Fig. 3). In December dense 
concentrations of C. hamatus begin to appear close 
to shore in the MAB subarea. These inshore centers 
of abundance slowly enlarge and expand along the 
coast and over the central shoals of GBK with the 
northward progression of spring (Fig. 2, A and B). 
Thus, peak times of abundance in the designated 
subareas vary with latitude (Fig. 3): May in the MAB, 
June in SNE, July on GBK, and September in the 
GOM. The population becomes distributed over 
nearly the entire shelf of each subarea during the 
annual peak period of abundance. The distribution 
and monthly abundance figures clearly show that 
GBK is the area of highest abundance for C. hamatus 
within the northeast shelf ecosystem. 
Distribution begins to constrict towards the shore 
in each subarea during the months approaching the 
annual period of low abundance (Fig. 2). Abundance 
estimates in the SNE, GBK, and GOM subareas de- 
cline slowly through the autumn and, unlike the 
MAB region, do not reach the annual low until win- 
ter (Fig. 3). 
Interannual variation in abundance of Centropages 
hamatus is shown in Figure 4 and individual survey 
statistics are given in Table 1. Although no long-term 
temporal trends in abundance were evident within 
any of the subareas, population estimates in certain 
years were exceptional. For example, the copepod’s 
abundance in both the MAB and SNE subareas was 
high for an extended period in 1984 (Fig. 4). Both of 
these areas also had high abundance during the 
spring of 1987 and low population estimates in 1982. 
Departures from the average annual cycle of abun- 
dance were not always continuous across these sub- 
areas; C. hamatus density was low during early 
