222 
Fishery Bulletin 115(2) 
Arm diameter 
5.0 mm (SD 0.01) 
Total length 
87.2 mm (SD 1.11) 
Front length 
40.4 mm (SD 1.10) 
Gap 
33.2 mm (SD 0.59) 
Maximum length 
43.3 mm (SD 0.64) 
Figure 2 
Details of the J hook used in this study on the possible effects 
of leader material on catches of pelagic longline fisheries in 
the southwest Indian Ocean. Standard deviations (SDs) are 
given in parentheses. Hook measurements are shown in the 
illustration. The offset angle is 10° and refers to the angle of 
sideways bending of the end of the hook inward in relation to 
the hook shank. 
accidentally caught but not retained were considered 
discards, consisting mostly of teleost and elasmobranch 
species with low commercial value or shark species 
whose retention is currently forbidden by the lOTC, or 
a combination of both (i.e., bigeye thresher [Alopias 
superciliosus] and oceanic whitetip shark [Carcharhi- 
nus longimanus]). Other more occasional discards were 
small specimens of commercial species (e.g., occasional 
captures of small swordfish) or depredated individuals. 
Data analysis 
Catch rates were expressed as catch per unit of effort 
in number of specimens (no. of specimens /lOOO hooks 
(CPUEn). Mortality per unit of effort (MPUE) was also 
calculated as the number of dead specimens (at haul- 
back) per 1000 hooks (see, for example, Afonso et ah, 
2012, who used this measure). For the retained species, 
catch per unit of effort in weight (CPUEb) was also es¬ 
timated as the weight (in kilograms) per 1000 hooks. 
Conversion equations were used because the retained 
catches were processed and frozen onboard and there¬ 
fore weighing was difficult. For billfishes and tunas, 
these rates were calculated with the lOTC conversion 
equations. However, in the case of the remaining spe¬ 
cies, conversion equations from the Institute Portugues 
do Mar e da Atmosfera (IPMA'^) were used. The value 
** IPMA (Instituto Portugues do Mar e da Atmosfera). 2014. 
Unpubl. data. Instituto Portugues do Mar e da Atmosfera, 
Rua C do Aeroporto, 1749-077 Lisboa, Portugal. 
of the catch per unit of effort (VPUE), measured 
in euros per 1000 hooks (and given in both U.S. 
dollars and euros per 1000 hooks), was estimated 
for the retained species. The reference (price) val¬ 
ues used for each species for the VPUE calcula¬ 
tions, were those registered for frozen products at 
the Vigo (Spain) auction in April 2014. The U.S. 
dollars-to-euro exchange rates were also consid¬ 
ered for April 2014. These values were chosen 
because most European flagged pelagic longlin- 
ers ship their frozen products to the Vigo market 
(Amorim et ah, 2015). 
Kolmogorov-Smirnov tests with Lilliefors cor¬ 
rection (Lilliefors, 1969) were used for testing 
the CPUEn, CPUEb, and MPUE for normality, 
whereas Levene tests (Levene, 1960) were used 
for testing the homogeneity of variances. Because 
of the general lack of normality and homogene¬ 
ity of variances, the differences between different 
leader types were tested with randomization tests 
to determine whether the observed differences be¬ 
tween different leader types were significant or 
whether they were occurring owing to random¬ 
ness in the sampling (Manly, 2007). For the ran¬ 
domization tests, a Monte Carlo approach was 
used; the data were randomized and resampled 
9999 times to build the expected distribution of 
the differences under a random distribution and 
the result was then compared and used to deter¬ 
mine the significance of the differences observed in the 
sample. 
For all captured species, the mean length and respec¬ 
tive SDs were calculated and for the 2 most abundant 
species caught (swordfish and blue shark), the size fre¬ 
quency distributions were plotted with histograms. For 
those 2 species, the mean sizes for the 2 leader types 
were compared with randomization tests. 
The bite-off rates were calculated for each fishing 
set as the number of missing hooks owing to a cut in 
the gangion line, per 1000 hooks, within each leader 
type. The mean and SDs were calculated and plotted, 
and the differences between leader types were tested 
with randomization tests. 
The relationship between hooking location (mouth 
or jaw) and hooking mode (deeply ingested, and exter¬ 
nally hooked) with leader type material was assessed 
with plots and with contingency table analysis and chi- 
square proportion tests. This analysis was performed 
only for species that numbered >30. In some cases, 
because of the small sample sizes, particularly for the 
externally hooked, the analysis was simplified to com¬ 
pare only mouth or jaw and deeply ingested hooking 
locations. 
Generalized linear models with a binomial error 
distribution and a logit link function were created for 
determining the influence of changing between the 2 
different leader types in the various species or com¬ 
bined taxonomic groups. This was tested both for the 
catches in number and for the mortalities caused by 
hooking and was applied only to species or other taxa 
