Byrd et al.: Effects of commercial fishing regulations on stranding rates of bottlenose dolphin ( Tursiops truncatus) 
79 
albeit insignificant, in the mean strandings per month 
for the Hi-other and HI-CBD categories between TP1 
and TP2. However, stranding rates for these categories 
could have been influenced by bycatch reductions. All 
stranded dolphins categorized as Hi-other were found 
either with missing appendages, cuts on the abdomen, 
or both, but they were too decomposed for a determi- 
nation of whether or not entanglement lesions were 
present. Fishermen occasionally cut appendages from 
a marine mammal to aid in the removal of the animal 
from their nets, or they slit the abdomen to aid in the 
sinking of the carcass, or do both (Kuiken et al., 1994; 
Read and Murray, 2000). It is likely, therefore, that a 
portion of the Hi-other stranded dolphins were indeed 
entangled in fishing gear because of the mutilations 
they exhibited. Of the HI-CBD strandings, an unknown 
proportion was likely caused by fishing interaction, 
but decomposition obscured evidence of entanglement 
lesions. Reductions in bycatch would decrease rates of 
strandings categorized as Hi-other and HI-CBD be- 
cause of the portion of them that were really HI-FI but 
could not be identified as such. 
Because the rate of HI-FI strandings is proportional 
to the number of bycaught animals, stranding rates 
may be used as a proxy to detect increases in bycatch 
mortality and to determine a threshold for triggering 
a management response. The Cl values calculated with 
the regression analyses were too broad for this method 
to be useful for setting threshold values even though 
the r-squared (coefficient of determination) value was 
high (0.79). The UME threshold method proved more 
useful; the stranding threshold value for the NC win- 
ter mixed MU adequately identified elevated stranding 
levels during months when the spiny dogfish fishery 
was most active and bycatch levels were highest. Fur- 
thermore, with the UME threshold method we were able 
to identify other periods of elevated strandings during 
winter and summer. 
The stranding thresholds calculated by the UME 
threshold method, while informative, have some limita- 
tions. First, stranding data between October 2002 and 
April 2005 were included in the calculations, but the 
corresponding bycatch estimates were not yet avail- 
able. If new bycatch estimates for that time period 
exceeded PBR levels, then stranding thresholds would 
need to be recalculated after eliminating the corre- 
sponding stranding data. New thresholds may result in 
the identification of other months that exceed the re- 
vised stranding threshold because the current thresh- 
old would be biased upward. Another limitation of a 
stranding threshold is the level at which increased 
bycatch is apparent as HI-FI strandings. The bycatch 
estimate for years with an active spiny dogfish fishery 
was more than 7.5 times greater than bycatch esti- 
mates for years after an active fishery (Rossman and 
Palka 1 ); our method of calculating a threshold may be 
too conservative in that it may not detect increases in 
strandings soon enough. Alternative methods to deter- 
mine thresholds may be more sensitive and could be 
investigated by using this or a similar data set that 
c 
No. of HI-FI strandinas oer season 
Figure 5 
Predicted bycatch (solid line) of bottlenose dolphin ( Tur- 
siops truncatus) in ocean gill nets, with 95% confidence 
intervals (CIs; dashed lines) and 68% CIs (dotted lines) 
by using HI-FI strandings (i.e., having evidence of fish- 
ery interaction) per season (winter: November-April, 
summer: May-October) and their corresponding bycatch 
estimates from observer data (dots). 
has periods of estimated bycatch that are greater and 
lesser than the PBR level. 
Once an appropriate threshold is established and ex- 
ceeded, a series of response actions can be triggered, as 
is done for UMEs (Wilkinson, 1996). Most importantly, 
active fisheries in the area would need to be identi- 
fied. The possibilities for response actions then would 
vary from immediately increasing observer coverage 
in these fisheries to implementing emergency fish- 
ing regulations to reduce mortality (MMPA 16 U.S.C. 
1387 [ 1 18] ) such as gear modifications, time and area 
closures, or limited soak durations. The advantage of 
an increase in observer coverage is to not only increase 
the precision of the bycatch estimate but to also docu- 
ment fishing practices and determine if practices have 
changed in a way that may be affecting the level of 
bycatch. 
Stranding data provide valuable information about 
fisheries bycatch if there is consistent, thorough de- 
termination of human interaction and comprehensive 
coverage of shorelines to establish baseline data and 
to detect changes. For example, stranding data have 
served as indicators of bycatch in fisheries that do not 
have federal observer coverage, such as crab-pot, stop- 
net, pound-net, and inshore gillnet fisheries (Steve et 
al., 2001; Waring et al., 2002). 
Additionally, stranding data provide additional in- 
formation about bycatch in gillnet fisheries that have 
low observer coverage. Although observers have not 
documented a bottlenose dolphin entanglement in the 
gillnet fishery for spot (FAO name: spot croaker; Leios- 
tomus xanthurus), stranding data indicated that bycatch 
had occurred. In 1997 and 1998, more than 50% of 
