80 
Fishery Bulletin 108(1 ) 
Gulf of Mexico on the Louisiana coast directly west of 
the Mississippi River. The region is both humid and 
subtropical and is dominated by saltmarsh vegetation 
(Day et al., 1989). Barataria and Caminada bays lie 
in the lower saline portion of Barataria Basin,; they 
are turbid bays of less than 2 m in depth (on average), 
have limited tidal amplitude (less than 0.32 m), and are 
separated from the Gulf of Mexico by a series of barrier 
islands (Baltz et al., 1993). 
Survey method 
Monthly surveys were initiated in Barataria Basin in 
June 1999 and continued until May 2002. Two or more 
independent observers used a small 3.5-m (17-ft) motor- 
boat to survey six designated strata in random order 
and sequence (Miller, 2003). When a bottlenose dolphin 
group was sighted, standard photo-identification tech- 
niques were used to document individuals (Wiirsig and 
Wiirsig, 1977). Microhabitat data were also collected. 
More specifically, an onboard Hydrolab (Environmental 
Data Systems, model CR2-SU, Austin, TX) was used to 
record sea-surface temperature (°C), salinity (psu), and 
dissolved oxygen (mg/L). Furthermore, a weighted line 
measured shallow depths (m), and nautical charts and 
mapping software were used for deeper water measure- 
ments and to determine distance to the nearest shore 
(m). Lastly, water samples were collected for laboratory 
assessment of turbidity (nephelometric turbidity units 
[NTU]) using a Hach 2100N Turbidimeter (Loveland, 
CO). Bottlenose dolphin group size and composition were 
recorded, as well as the presence of juveniles and calves 
(Miller, 2003). Behaviors were categorized as foraging, 
social, rest, and travel behaviors (Allen and Read, 2000). 
However, for this article only foraging activity defined 
by one or more of the following behaviors was used: fish 
in mouth, numerous steep dives in rapid succession, 
quick circling behavior at the water surface, or direct 
pursuit of a prey item (Allen et al., 2001). All sightings 
were made during daylight hours in Beaufort Sea state 
conditions of 3 or less. 
Statistical methods 
Environmental variables were assessed for univari- 
ate and bivariate normality and when necessary were 
transformed to meet normality requirements. Seasonal 
(Fall: September-November, Winter: December-Febru- 
ary, Spring: March-May, and Summer: June-August) 
differences among dissolved oxygen, turbidity, salinity, 
water depth, water temperature, and distance from 
shore were assessed by using a multivariate analysis 
of variance (MANOVA) approach and pair-wise com- 
parisons. Least-square means with Tukey’s adjustment 
were performed on variables and produced a significant 
Shapiro-Wilks result. 
A factor analysis (FA) of dissolved oxygen, turbidity, 
salinity, water depth, water temperature, and distance 
from shore was employed to examine the pattern of 
habitat use by bottlenose dolphins and also to highlight 
relationships among individual variables. The number 
of orthogonal components retained for interpretation 
was chosen after examining the scree plot for Eigen 
values >1, and interpretation was aided by a varimax 
rotation. 
A logistic regression and habitat suitability curves 
(Saucier and Baltz, 1993) were used to investigate 
whether particular environmental variables were use- 
ful in describing foraging activity. The forward step- 
wise logistic regression with maximum likelihood esti- 
mation (with a Wald chi-square statistic) was used to 
select variables that were most strongly related to the 
observed activity. Variables investigated in this analy- 
sis were dissolved oxygen, turbidity, salinity, water 
depth, water temperature, distance from shore, time 
of day, season, and the minimum number of individu- 
als present in a group. Variable multicollinearity was 
checked before the analysis by using variance inflation 
factor (VIF) values and the final model was evalu- 
ated by a Hosmer and Lemeshow test. Least-square 
means (with associated standard errors) of foraging 
and nonforaging observations were computed for all 
significant variables. To describe seasonal foraging 
activity, seasons in which the highest proportion of 
foraging and nonforaging observations took place were 
used for modeling. 
Habitat suitability curves (Baltz et al., 1993) were 
constructed to characterize the patterns of dissolved 
oxygen, turbidity, salinity, water depth, water tem- 
perature, distance from shore, and group size at sites 
where foraging was observed. For this approach, we 
considered the proportional frequency of foraging and 
nonforaging activity at defined intervals for each en- 
vironmental variable along its range. Specifically, a 
foraging habitat suitability value was calculated by 
dividing the probability of observing foraging in each 
interval by the total number of observations in each. 
Habitat suitability values were normalized to a scale 
of 0 (nonforaging) to 1 (high probability of foraging) 
by dividing each habitat suitability interval value by 
the highest habitat suitability interval value of each 
environmental variable. 
Results 
Number of survey days and frequency of bottlenose dol- 
phin groups sighted were relatively even across seasons 
as was the proportion of observations during which forag- 
ing activity was observed. Nevertheless there was some 
variability in the number of individuals seen throughout 
the year (Table 1). Variability in the environmental 
characteristics of the study area was also evident. Sig- 
nificant seasonal differences in dissolved oxygen, turbid- 
ity, salinity, and water temperature were detected by a 
MANOVA. Posterior pair-wise comparisons were used on 
significant variables (Table 2). Water temperatures were 
significantly different across all four seasons and, as 
expected, were lowest in winter and highest in summer. 
Minimum observed dissolved oxygen levels were found 
