Bradford et al.: Abundance estimates of cetaceans within the U.S. Hawaiian Islands EEZ 
139 
for rough-toothed dolphins) than the ^(0) estimates of 
Barlow (2006). The estimates of g(0) for 2010 are even 
more reduced than the values from 2002 for sightings 
with more than 20 individuals because g(0) previously 
was assumed to be 1 for larger groups of most species 
(Barlow, 2006). 
The lower g'(O) estimates for 2010, in combination 
with group sizes numbering in the tens to hundreds of 
individuals, are responsible for the relatively large es¬ 
timates of abundance for the small and medium delphi- 
nids (Table 3)—values that are strikingly higher than 
the estimates determined by Barlow (2006). Point esti¬ 
mates of abundance in Barlow (2006) are larger than 
those of the present study for only 4 species: the killer 
and sperm whales, Blainville’s beaked whale {Meso- 
plodon densirostris), and Cuvier’s beaked whale. The 
estimates for killer and sperm whales are of the same 
magnitude in both studies and indicate that random 
variation in other aspects of the estimation (e.g., the 
number of encounters and group size for killer whales 
and the mean ESW for sperm whales) likely countered 
the effects of the slightly lower g(0) estimates for the 
HICEAS in 2010. 
The encounter rate for beaked whales was much 
lower for the HICEAS in 2010 because survey effort 
in Beaufort sea states 0-6 was used in the abundance 
estimation, but only effort in Beaufort sea states 0-2 
was used in the analysis for the HICEAS in 2002 (Bar- 
low, 2006). The corresponding decrease in g{0) for the 
HICEAS in 2010 was not enough to reduce the effect 
of the decreased encounter rate for Cuvier’s beaked 
whales, and random variation did not mitigate the ef¬ 
fect, as the larger group size of the sighting in 2010 
did for Blainville’s beaked whales. As a result, the 
abundance estimate for Cuvier’s beaked whales was 
more than 20 times larger for 2002 than for 2010. Re¬ 
sults of an analysis in which habitat associations were 
used to estimate the densities and abundances of a 
subset of species encountered during the HICEAS in 
2002 and 2010 (Forney et al., 2015) are also not di¬ 
rectly comparable with results from the present study 
because Forney et al. (2015) used g{0) estimates of the 
same order of magnitude as those in Barlow (2006). 
A major assumption with cetacean line-transect 
analyses that was challenged by the estimation of g^(O) 
by Barlow (2015) is that g(0) is equal to 1 for large 
groups of dolphins (Brandon et ah'*; Gerrodette and 
Forcada, 2005). However, the model used to infer the 
relative values of g(0) in different sighting conditions 
did not specifically test for the effect of group size 
on g(0) or allow for potential interactions between 
group size and sighting conditions. The analysis did 
determine that group sizes decreased with increasing 
* Brandon, J., T. Gerrodette, W. Perryman, and K. Cram¬ 
er. 2002. Responsive movement and g(0) for target spe¬ 
cies of research vessel surveys in the eastern tropical Pacific 
Ocean. Southwest Fish. Sci. Cent. Admin. Rep. LJ-02-02, 
28 p. [Available from Southwest Fisheries Science Center, 
National Marine Fisheries Service, 8901 La Jolla Shores Dr., 
La Jolla, CA 92037.] 
Beaufort sea state for many of the species considered 
(Barlow, 2015). If individuals of some species do form 
smaller groups in rougher sea conditions, abundance 
estimates based on observations of these groups would 
be positively biased. However, Barlow (2015) suggested 
that the decrease in group sizes at higher Beaufort 
sea states is more likely due to the underestimation of 
group size in rougher sea conditions. 
Although more testing is needed, there is no evi¬ 
dence that actual group size changes as a function of 
Beaufort sea state (Barlow, 2015). Further, the Barlow 
(2015) g(0) model not explicitly incorporating group 
size is presumably not an issue for the estimation 
in the present study unless the distribution of group 
sizes in the data subset from the HICEAS in 2010 is 
different from that of the full data set of the Barlow 
(2015) model. This question is difficult to assess quali¬ 
tatively because summaries of mean group sizes from 
other study locales represented in the full data set 
(e.g., Ferguson et al., 2006; Barlow and Forney, 2007) 
do not reflect the underlying distribution of group siz¬ 
es overall or by Beaufort sea state. Additional analy¬ 
ses are needed to quantitatively evaluate the effect of 
group size on the Beaufort-specific estimates of g(0) 
and, therefore, to confirm that the estimates can be 
applied to all group sizes in the study locations cov¬ 
ered by Barlow (2015). Validation of the actual giO) es¬ 
timates (e.g., by comparisons with acoustic detections) 
would also be valuable. 
For the species that were sighted during the HI¬ 
CEAS in 2010 (Table 3), but were not included in the 
analysis of Barlow (2015) (i.e., the Fraser’s dolphin, 
melon-headed and pygmy killer whales, and Long¬ 
man’s beaked whale), use or averages of the g(0) es¬ 
timates of associated species in the detection function 
species pools (Table 2) may not have been appropriate 
and could have biased the estimation of abundance for 
these species. Future efforts to estimate g{0) for these 
species when sufficient sample sizes are available 
would resolve this issue and are recommended. 
The rough-toothed dolphin was noted as an outlier 
in the estimation ofg(O) by Barlow (2015), showing the 
most rapid decline in ^(0) with increasing Beaufort 
sea state of all the species. The impact of this effect 
is clear in the abundance estimation for the HICEAS 
in 2010 in that the value of g(0) for the rough-toothed 
dolphin is the lowest of all the species and the result¬ 
ing abundance estimate is the highest (Table 3). Given 
their relatively small group sizes and subtle surfacing 
behavior (i.e., surfacing without conspicuous splashes), 
rough-toothed dolphins have been described by expe¬ 
rienced observers as difficult to detect (Yin^), but this 
characterization has not been explicitly quantified and 
is not readily apparent from qualitative comparisons 
of multispecies data. For example, the mean group size 
and ESW for rough-toothed dolphins in this study are 
^ Yin, S. 2015. Personal commun. Hawaii Marine Mammal 
Consortium, Kamuela, HI 96743. 
