140 
Fishery Bulletin 95( 1 ), 1997 
dance was compared with the volume of each source’s 
discharge (Table 1) divided by the distance separat- 
ing the video samples from the nearest source of dis- 
charge. No attempt was made to sample the nutri- 
ents or suspended materials of these discharges. El- 
evated organics associated with these water masses 
are documented in the literature (Bromwell, 1992; 
City and County of Honolulu 4 ; Laws and Allen 5 ). 
East Oahu statistical analysis 
The distribution of the data and the categorical na- 
ture of the habitat variables required the use of non- 
parametric analysis (Siegel and Castellan, 1988). The 
type-I error for statistical significance was set at 0.05 
(2-tailed test). Kruskal-Wallis ANOVA (K-W) was 
used to assess station effects in both “multicanyon 
and multiyear” analyses and to assess the effect of 
substrate type. 
Differences in substrate by depth were tested with 
chi-square analysis. Replicate bottom grabs were 
compared by using Wilcoxon matched pairs sign 
ranks (MPSR), and Spearman’s correlation was used 
for association of snappers with slope, sediment frac- 
tions, and influence of drainage. 
Spatial variation of ranked snapper abundance 
was related to all habitat variables together by us- 
ing logistic regression. Snapper abundance was 
grouped into two categories, aggregation pi'esent 
(n>5) and aggregation absent (n< 5), and assessed 
relative to the habitat variables that significantly 
influenced snapper abundance in the previously de- 
scribed univariate analyses (Norusis, 1992). Models 
of the variables and their plausible interaction ef- 
fects were explored with the simple logistic regres- 
sion model (Kleinbaum, 1992): 
where n is the probability of detecting snappers with 
the linear combination of the habitat variables X t in 
a given location. The coefficients estimated with the 
nonlinear regression by using maximum likelihood 
are represented by B r The base of the natural loga- 
rithm is e. The P-value for retention of independent 
variables in the model was set at 0.01. 
5 Laws, E. A., and C. B. Allen. 1993. Impact of land runoff on 
water quality in Kaneohe Bay, a subtropical Hawaiian 
estuary. Proceedings of the first biennial symposium for main 
Hawaiian islands marine resources investigation, November 17- 
18. Hawaii Department of Land and Natural Resource Tech- 
nical Report 95-01, p. 232—248. Hawaii Dep. Land Natl. Re- 
sources, 1151 Punchbowl, Honolulu, HI 96813. 
Survey of the archipelago 
Conventional fishing gear (e.g. trawls, longlines, 
traps, handlines) was used to survey a total of 332 
km of longshore habitat dispersed over seven islands 
of the archipelago (1989-94). The effectiveness of 
each gear at catching juvenile snappers was tested 
at the east Oahu study site. Sites surveyed included 
areas outside of embayments, places with large shelf 
areas at snapper depths, and sites of previous re- 
search fishing where juveniles had been documented 
incidentally (Struhsaker, 1973). Sites where snap- 
pers were found were then reassessed with longshore 
baited video surveys (range 5.5-42.6 km) to permit 
comparison with snapper abundance at the east 
Oahu study site. Numbers of juvenile snappers ob- 
served at each site were standardized by effort. The 
distance of each video drop from the coastal reef edge 
(15-m isobath) and the type of substrate seen in the 
video image were tabulated for each site; these vari- 
ables were then compared with the respective video 
index of juvenile snapper abundance. Catch-per-unit- 
of-effort (CPUE) data from sets of conventional fish- 
ing gear at these coastlines were included to provide 
an independent index of snapper abundance. 
In comparing video data from other coastlines with 
those of east Oahu, data for the two Kaneohe areas 
were pooled. Coastlines with point sources of drain- 
age were identified, and the distance between sources 
of discharge and the video drops (weighted for maxi- 
mum depth of discharge) were calculated. Importance 
of proximity to drainage sources to snapper abun- 
dance was then evaluated for these archipelago sites. 
Snapper production estimates 
To assess the importance of the contribution of juve- 
niles from a site with premium habitat (e.g. Kaneohe) 
to the adult fishery, the adequacy of recruit produc- 
tion from other habitat areas was estimated. The 
density of snappers at habitat without snapper ag- 
gregations was compared with the density of snap- 
pers needed to explain the catch from the main Ha- 
waiian Islands (MHI) commercial snapper fishery. 
Derived from mandatory reporting from the commer- 
cial fishery, the estimate is based on the catch of -3- 
year-old snappers (termed “immature”) just enter- 
ing the MHI adult snapper fishery (Ralston, 1981; 
DeMartini et al., 1994). Based on the years 1989- 
92, the estimated mean annual catch, C (i.e. the com- 
mercial catch [WPRFMC 2 ]) was -22,000 immature 
(1.3 kg) snapper/year. Recreational fishing produces 
a significant additional catch in Hawaii, but it is 
poorly documented and was not considered in this 
estimate. 
