34 



Fishery Bulletin 89(1), 1991 



(Grant 1977a, 1979, 1988). Routine collections at each 

 station included paired 60-cm bongo net samplers (202 

 and 505 ^m mesh nets), towed from just below the sur- 

 face to near, but safely off, the bottom, then back to 

 the surface (so-called double oblique tows; 220 samples), 

 and eight surface layer (upper 10 cm) collections ob- 

 tained at 3-hour intervals over a 24-hour period (496 

 samples), using a 1-m wide hyponeuston net (505 ^m 

 mesh). Of the 716 collections (Table 1), onjy 1 bongo 

 and 79 hyponeuston (mostly daytime) collections lacked 

 chaetognaths. 



Laboratory processing 



Collections were divided into successively smaller ali- 

 quots for the more numerous taxa, using a sample- 

 splitting device of proven design (Burrell et al. 1974). 

 However, chaetognaths were generally obtained from 

 whole or half samples, unless very abundant. 



Data analysis 



Collection data were sorted by species, stations, and 

 collection methods. Analysis of the relationship of 

 species abundance to hydrography was limited to 

 hyponeuston collections because subsurface tows were 

 oblique, often traversing multiple layers of different 

 water types. Mean temperatures and salinities of cap- 

 ture in surface-layer collections were calculated for 

 each common chaetognath species, weighting each 

 observed temperature and salinity by the size of catch 

 (log N + 1), where N = total catch in a standard 20-min- 

 ute hyponeuston net tow at 2.5 knots. Thus, 



tiflog n t + 1) + t 2 (log n 2 + 1) + . . . + tn(log n n + 1) 

 (log n x + 1) + (log n 2 + 1) + . . . + (log n n + 1) 



Figure 1 



Location of Middle Atlantic Bight plankton stations ( • ) sam- 

 pled quarterly, October 1975-August 1977. Those in transect 

 off New Jersey (Cl-Jl, o) were sampled for 2 years. Other 

 stations were added for the second year. Depth contours in 

 meters. 



s = 



s^log n t + 1) + s 2 (log n 2 + 1) + . . . + s n (log n n + 1) 

 (log n, + 1) + (log n 2 + 1) + . . . + (log n n + 1) 



where tj, s i; and n ; are the surface temperature, sur- 

 face salinity, and total catch in each positive collection, 

 respectively. 



Presence, absence, and joint occurrences of the 15 

 most frequent species from both bongo and hyponeus- 

 ton collections were used in an analysis of association 

 between and among species. As recommended by Lud- 

 wig and Reynolds (1988), the significance of associa- 

 tion among all 15 species and 716 collections was first 

 tested simultaneously using a variance ratio (VR) 



derived from a null association model (Schluter 1984). 

 The expected value of VR under the null hypothesis 

 of independence is 1. When VR>1, a positive associa- 

 tion of species is indicated; VR< 1 indicates a negative 

 association. A test statistic W = (N)(VR) provides a test 

 of significance for deviations from 1. If species are not 

 associated there is a 90% probability that W lies be- 

 tween the chi-square limits: 



X 2 0.05.N < W < X": 



ii.'.ir,,N - 



Because of the large degrees of freedom in this study 

 (where N = 716), critical values of x 2 were approx- 

 imated (see Zar 1984, p. 482). 



