Marancik et a!.: Spatial and temporal distribution of grouper larvae in the Gulf of Mexico and Straits of Florida 
7 
qI_1 . . j ■ ■ — ■ * J 
1980 1985 1990 1995 2000 2005 1980 1985 1990 1995 2000 2005 
Year 
Year 
Figure 2 
Spatial distribution of the most recently spawned (least developed) grouper 
larvae with standard pigment collected during Southeast Area Monitor- 
ing and Assessment Program plankton surveys, 1982-2005. (A) The five 
subregions (a-e) in the northern Gulf of Mexico based on the presence of 
small larvae and orientation of the coastline in relation to bathymetry. 
The southern Gulf of Mexico was sampled only during one year and was 
not included in analyses. (Bi Bar graph of the percentage of recently 
spawned grouper larvae collected in each region by year. Lowercase let- 
ters in each subpanel correspond to the subregion letters in panel A: a) 
Mexico-Texas shelf to 90°W longitude, b) Texas, Louisiana shelf, c) Mis- 
sissippi-Alabama-north Florida shelf, d) west Florida shelf, and e) open 
water east of 90°W longitude. 
y r ,s = a r.s + gi(y ear rj + g 2 (stemp rs ) + g 3 (ssal rs ) + 
g 4 (wdep r s ) + e rs , 
where a = the subregion by season intercept; 
g = the nonparametric smoothing function for 
each term; 
stemp, ssal, = the mean surface temperature, mean sur- 
and wdep face salinity, and mean water depth for each 
subregion by season for each year, respec- 
tively; and 
e = a normally distributed random error term 
with a mean of zero and finite variance. 
The model was run with all combinations of covari- 
ates to find the best subset of covariates (best fit) re- 
quired to explain %FO for each subregion by season. 
Two evaluation techniques were used to select the best 
model. The generalized cross-validation (GCV) score is 
a measure of the predictive squared error of the model 
(Wood, 2006). Akaike’s information criterion with a 
