Powers and Anson: Change in fishing effort in response to changes in length of fishing season for Lutjanus campechanus 
3 
because effort probably is not similar between weekend 
days and holidays and weekdays. This issue applies 
only to the 2012 and 2013 seasons because in 2017 the 
season was opened only on Friday-Sunday (considered 
weekend days) and major holidays (e.g., the Fourth 
of July). For the seasons in 2012 and 2013, we used 
the modeled estimates directly from Powers and An¬ 
son (2016) that included a weekend and holiday factor 
(for further details, see Powers and Anson, 2016). For 
the other seasons in the time series (2014, 2015, 2016, 
and the short season of 2017), fishing days were con¬ 
secutive from 0000 on 1 June to 1159 on the last day. 
Finally, for the seasons in 2014-2017, the frequency of 
5-min intervals analyzed for each hour from 0500 to 
1000 was increased from 1 interval to 5 intervals to 
better capture the increased boat traffic during these 
hours. 
When watching the videos, analysts recorded the 
number of boats launched and the number of anglers 
on each boat. For a boat launch to have been counted, 
the analyst had to have observed the boat coming off 
the trailer during the 5-min interval. To count the total 
number of potential anglers, the analyst was allowed 
to observe the boat outside of the 5-min interval to ac¬ 
count for the time anglers might have taken to return 
from parking the towing vehicle. Analysts classified 
boats into 1 of 4 categories: offshore fishing, inshore 
fishing, non-fishing, and unknown. Categorization was 
based on the size and design of fishing vessel as well as 
on the presence and type of fishing gear. All analysts 
were trained on videos recorded prior to the study pe¬ 
riod by an experienced staff member before they began 
analyzing video recordings from the study period. All 
videos were viewed by a second analyst, and the 2 sets 
of observations were averaged to estimate the number 
of boat launches per 5-min interval per ramp and the 
number of potential anglers per 5-min interval. The av¬ 
eraged observations were then transformed to hourly 
estimates depending on the number of 5-min intervals 
observed within the designated hour. Daily estimates 
were made by summing the hourly estimates for each 
day. Finally, the daily estimates for each of the 6 boat 
launches were summed across all days to calculate the 
potential numbers of anglers and offshore boats during 
the short seasons. For the long seasons, a daily aver¬ 
age of potential numbers of anglers and offshore boats 
was calculated by multiplying the average of each met¬ 
ric from observed days at each boat ramp by the total 
number of days in the season and then summing esti¬ 
mates for all boat launches to provide an estimate for 
the season (see Powers and Anson, 2016). 
Because weather may affect angler effort (Fraiden- 
burg and Bargmann, 1982; Powers and Anson, 2016), 
observations of weather conditions were gathered by 
the analyst watching the videos as well as from near¬ 
by monitoring stations. Analysts recorded precipitation 
events and cloud cover as binary responses (1 for rain, 
0 for no rain; 1 for dark clouds present, 0 for dark 
clouds not present). Maximum nearshore hourly wind 
speed and daily precipitation were obtained from the 
weather station located near the public boat ramp at 
Billy Goat Hole on Dauphin Island (Mobile Bay Estu¬ 
ary Program, website). Offshore hourly maximum wind 
speed and sea height were obtained from the sea buoy 
located 22 km offshore from Orange Beach, Alabama 
(station 42012; National Data Buoy Center, website). 
Analyses 
We used linear and nonlinear regression analyses of 
the estimates for mean number of anglers per day and 
of boat launches per day during each season to exam¬ 
ine the relationship between angler effort and season 
duration. Mean daily numbers of boat launches and 
anglers were normally distributed; hence, no transfor¬ 
mation was applied (Shapiro-Wilk test for both depen¬ 
dent variables: P>0.90). Next, multiple linear regres¬ 
sions were used to determine if the model prediction 
could be improved by the addition of weather variables. 
Because of the limited number of years in our study, 
interactions of main effects could not be included in the 
models. The corrected Akaike’s information criterion 
(AICc) was used to compare model fits and determine 
the most parsimonious model. Because nearshore and 
offshore winds were expected to be strongly correlated 
and nearshore winds may influence anglers’ decisions 
to fish for reef fishes more than offshore wind speeds, 
only wind data from the nearshore weather station at 
Dauphin Island were used in calculating the daily av¬ 
erage wind speed. Previous work has indicated that the 
nearshore wind measurements were the better predic¬ 
tor of daily effort (Powers and Anson, 2016). Both wind 
speeds and offshore sea heights from Dauphin Island 
were explored in the models because the 2 conditions 
were not found to be significantly correlated. 
To estimate total harvest of red snapper in the seg¬ 
ment of the recreational sector that used public boat 
launches, seasonal estimates of the number of anglers 
were multiplied by the observed average number of red 
snapper harvested per angler and average weight of 
red snapper. Estimates of average number of red snap¬ 
per caught per angler and average weight were derived 
from dockside intercepts of anglers interviewed at the 
6 boat ramps included in this study. In 2012 and 2013, 
data necessary to produce these averages for each year 
were collected as part of the Marine Recreational Infor¬ 
mation Program of the National Marine Fisheries Ser¬ 
vice. For 2014-2017, data collected as part of dockside 
surveys conducted by the Marine Resources Division of 
the Alabama Department of Conservation and Natural 
Resources to complement a new mandatory reporting 
program were used to calculate the average number 
of red snapper harvested per angler and the average 
weight of red snapper for each year. Once these season¬ 
al estimates of harvest in number and weight were cal¬ 
culated, the relationship between these estimates and 
season duration were examined by linear regression. 
The seasonal estimates of number and weight of red 
snapper were normally distributed (Shapiro-Wilk test 
for both dependent variables: P>0.70), and no trans- 
