12 
Fishery Bulletin 110(1) 
Mexico did reasonably well in predicting presence or 
absence of grouper in a subregion for a year under 
a given set of environmental conditions. The percent 
deviance from the null model explained by the best 
of these models ranged from 14.3% to 94.5%. The 
TX-LA shelf models examined explained very little 
of the deviance in the data, and several models fit 
the data almost equally (low AIC c weights; Table 5). 
Thus, grouper occurrence in this subregion is not 
well explained by any combination of surface tem- 
perature, surface salinity, water depth, or year. The 
west Florida shelf model based on the fall survey 
data and the Mississippi-Alabama-north Florida 
shelf model based on the spring survey data were 
the most successful in predicting the occurrence of 
grouper larvae, describing 94.5% and 84% of the 
deviances, respectively. The significant covariates in 
each subregion by season GAM revealed changes in 
frequency of occurrence over time and regionally spe- 
cific influences of water depth, surface salinity, and 
surface temperature (Table 5). Annual frequencies 
of grouper collections were sufficient for generating 
GAMs for the three subregions characterized by the 
presence of the smallest larvae for the fall season 
(Fig. 2, subregions b-d). The only spring data set with 
larvae collected in enough years to warrant modeling 
was the MS-AL-nFL shelf subregion. Models of Gulf 
subregions east of 90° W longitude (wFL shelf and 
MS-AL-nFL shelf) were positively correlated with 
year, with higher occurrence since the early to mid 
1990s (Fig. 5). The west Florida shelf model was also 
significantly influenced by mean surface temperature 
(>29°C), mean surface salinity (>35.5), and water 
depth (>129 m) (Table 5, Fig. 5). The occurrence of 
grouper larvae in the gulf west of 90° W longitude 
(TX-LA shelf) was significantly influenced by mean 
surface temperature (>28°C; Table 5, Fig. 5), although 
this relationship was weak. The occurrence of grou- 
per larvae in the north central Gulf (MS-AL-nFL) 
increased from 1990 to 2000, but was highest after 
1995, in midrange surface salinities (34-35), and in 
mean water depth <350 m (Table 5, Fig. 6). 
Change in occurrence over time 
The Gulf of Mexico subregion by season GAMs 
revealed a change in grouper occurrence over the 
SEAMAP survey time series, with %FO highest after 
the mid 1990s. This shift was evident in the patterns 
of occurrence of the more abundant grouper species 
and species groups (Fig. 7). Before 1995, grouper 
occurrences were higher in the spring than in the fall. 
Since 1995, higher occurrences have been observed 
in the fall than in the spring. No Mycteroperca spp. 
(three size groups combined) were collected in the 
fall before 1995, but since 1995, these larvae have 
occurred in fall survey samples. Similarly, larval 
graysby were rarely collected before 1995 (occurring 
in 2 of the 10 years between 1986 and 1995), but they 
have become more common in samples during recent 
