248 
Fishery Bulletin 112(4) 
The number of angler fishing trips, by state, that targeted red snapper 
(Lutjanus campechanus ) from private boats in the Gulf of Mexico in 2009. 
Data are from the National Marine Fisheries Service’s Access Point Angler 
Intercept Survey. Trips are distributed across coastal counties in Florida, 
Alabama, Mississippi, and Louisiana according to locations of intercept sites 
on shore where anglers were interviewed for the survey. 
trips during which anglers targeted red snapper (Table 
2). With application of the sampling weights, these ob- 
servations correspond to an estimated 716,701 trips. 
Anglers that targeted groupers traveled an average of 
53 mi (85 km) round trip between their permanent res- 
idence and the fishing site, whereas those anglers that 
targeted red snapper averaged 72 mi (116 km) round 
trip. Anglers caught an average of 0.39 groupers per 
trip in the previous 5 years and an average of 1.38 red 
snapper per trip. 
The estimation results of the site-choice model show 
that regardless of model version, anglers preferred 
sites that had lower travel costs than other sites in 
their choice set (Table 3). The parameter <P C ) related to 
the travel cost variable was statistically significant and 
similar in magnitude across all versions of the model. 
In the unweighted version of the model, the estimated 
travel cost parameter was -0.060, versus -0.056 for 
the version of the WESML model weighted with the 
mean site frequencies (Table 3). The results also in- 
dicate that anglers were relatively more likely to be 
observed fishing from zones with a relatively higher 
number of fishing sites. The size of the parameter for 
the site aggregation variable was fairly consistent be- 
tween the unweighted version of the model (0.251) and 
the WESML model weighted with 
the mean site frequencies (0.242). 
The results indicate that an- 
glers preferred sites with higher 
average catch-and-keep rates re- 
gardless of target species. The 
parameter for the grouper catch 
rate is positive and significant 
across all model versions (Table 
3). Compared with the grouper 
catch-and-keep parameter in the 
unweighted version of the model 
(£>1=0.784), the catch-and-keep 
parameter with the WESML ver- 
sion of the model with the mean 
site frequencies was 4% larger 
(£>1=0.819). For red snapper, the 
comparisons across versions of 
the model had results similar to 
those comparisons for groupers. 
The estimated parameter value 
for the catch-and-keep rate was 
statistically significant in all ver- 
sions of the model. For the red 
snapper catch-and-keep rate, the 
unweighted version of the model 
resulted in a parameter estimate 
of 0.471. In contrast, as with the 
grouper catch-and-keep rate, the 
parameter estimate for the red 
snapper catch-and-keep rate from 
the WESML version of the model 
weighted with the mean site fre- 
quencies was higher (£>i=0.603). 
For the travel cost variable, in comparison with the 
unweighted version of the model, weighting with the 
lower-limit site frequencies resulted in lower estimated 
parameter values whereas the use of the upper-limit 
site frequencies resulted in higher estimated param- 
eter values. The same pattern was observed with the 
parameters of the harvest rate variables. Ultimately, 
however, we were interested in the net effect that sam- 
pling weights had on the estimates of mean marginal 
WTP for changes in harvest rates. 
The mean marginal WTP for groupers varied be- 
tween $8 and $15 and was significant in all model 
versions (Table 4). The mean WTP for an additional 
grouper estimated with the model version that used 
the weighted catch rates plus the correction for choice- 
based sampling at the mean frequencies was $14.67, a 
value 12.8% higher than the estimate of $13.01 from 
the unweighted version of the model. For red snap- 
per, the mean marginal WTP varied between $4.50 
and $13.81 across the different versions. For this spe- 
cies, the model that used weighted catch rates plus 
the correction for choice-based sampling at the mean 
frequencies had a WTP estimate of $10.81. This value 
was 38.6%' higher than the estimate of $7.81 from the 
unweighted version of the model. These results indi- 
