230 
Fishery Bulletin 99(2) 
Q- 
D 
O 
a 
5 - 
3r 
2 -. 
0 J 
200 400 600 800 1000 1200 
Depth (m) 
20 22 24 26 28 
Surface temperature (°C) 
30 
F6 
- 4 
•2 
0.2 0 3 0.4 1.1 1.2 
Surface temperature SD (°C) 
Figure 7 
Sighting rate distributions of bottlenose dolphins with respect to (A) depth, (B) surface temperature and (C) surface temperature 
variability computed by using equal-effort, class sizes. 
again in the MANOVA (Wilks’ A=0.675, PcO.0001). The 
first two canonical variables in the canonical LDF analysis 
accounted for 94.2% of the total variability, and likelihood 
ratio tests indicated that only these first two canonical 
variables were significant (P<0.0001 for the first, P<0.05 
for the second). The correlation structure suggested that 
high values of zooplankton biomass and deep occurrences 
of the depth of the 15°C isotherm were associated with 
positive values of the first canonical variable, whereas 
high values for depth and surface chlorophyll were associ- 
ated with positive values of the second canonical variable 
(Fig. 6B). Although there seems to be considerable overlap 
between the Risso’s dolphin, pantropical spotted dolphin, 
and Kogia spp. in the canonical space, the sperm whale is 
separated from the other species primarily along canoni- 
cal axis 1 (Fig. 6B). 
Bottlenose dolphin 
Because the bottlenose dolphin was never encountered 
seaward of the 750-m isobath, only the surveyed transect 
sections shallower than 1000 m were used in the compari- 
sons between the sightings and the effort. The distribu- 
tions of this species with respect to depth, depth gradient, 
surface temperature, and surface temperature variability 
were significantly different from a uniform distribution 
(Table 5). Monte Carlo tests suggested that the distribu- 
tion with respect to depth gradient may have been an 
artifact of the distribution with respect to depth (P>0.05; 
Table 5). The sighting rate distribution of the bottlenose 
dolphin with respect to depth (Fig. 7A) was bimodal as 
indicated by the peak in the sighting rate at the shal- 
lowest depth class ( <75 m) and another peak just sea- 
ward of the shelf break. Although no coherent pattern was 
apparent in the sighting rate distribution with respect 
to surface temperature (Fig. 7B), group sighting rates 
increased with increasing surface temperature variabil- 
ity (Fig. 70. 
Interpretation of the sighting rate distributions for sur- 
face temperature and surface temperature variability was 
confounded by this species’ bimodal distribution with re- 
spect to depth. To address this, the sightings were sepa- 
rated into a shelf group (<150 m) and a shelf break group 
(>150 m) by using the local minimum in the sighting rates 
with respect to depth as the separation criterion (Fig. 7A). 
The shelf dolphins (z? =24 ) were found in cooler surface 
waters in relation to that observed during the sighting 
effort ( Mann-Whitney test, 17=2.23, P<0.05), whereas the 
distribution of the shelf break dolphins {n= 33) with re- 
spect to surface temperature was not significantly differ- 
ent from the effort (Mann-Whitney test, [7=1.03, P>0.05; x 2 
test, ^ 2 =9.7, df=5, P>0.05). Both the shelf 0? = 16, [7=3.23, 
P<0.01) and shelf break bottlenose dolphins (/i=14, [7=2.93, 
P<0.01) were encountered in regions of significantly higher 
surface temperature variability in relation to the effort. 
It should be noted that the bottlenose dolphin appears 
to have a distribution with respect to zooplankton biomass 
that is significantly different from the effort for all depths 
(Fig. 5G). In fact, a Mann-Whitney test supports this asser- 
tion ([7=5.42, P<0.0001). Once the analysis is restricted to 
the continental shelf and upper continental slope (0-1000 
m), however, the distribution of the bottlenose dolphin with 
respect to zooplankton biomass is not significantly different 
from the effort ( [7=0.19,P>0.05). This apparent discrepancy 
is due to higher zooplankton biomass over the continental 
shelf than anywhere else in the northern Gulf of Mexico. 
Risso's dolphin 
The distribution of the Risso’s dolphin was significantly 
different from a uniform distribution for both depth and 
depth gradient (Table 5) and there was strong evidence 
