Loher and Hobden: Length and sex effects on the spatial structure of Hippoglossus stenolepis 
47 
-172° -170° -1 68°W -166° -164° 
Figure 1 
International Pacific Halibut Commission survey stations that were 
sampled to test for spatial structure in Pacific halibut ( Hippoglossus 
stenolepis ) longline catches. Full 600-hook inventories of all halibut, 
bycatch, and empty hooks were conducted at numbered stations, 
and the halibut data were subsequently subjected to tests of spatial 
aggregation with nearest neighbor analysis. At stations indicated 
by letters, the position of each halibut on the longline was recorded, 
but not of empty hooks and bycatch; for these stations, and for all 
numbered stations, runs rests were performed to test for sequential 
segregation within the halibut population. 
dependent feeding hierarchies have been 
demonstrated empirically (Stoner and 
Ottmar, 2004), the possibility for intraco- 
hort spatial structure deserves additional 
attention. 
The aim of the present study was to 
examine whether significant levels of spa- 
tial structure were detectable within a 
small subsample of IPHC longline sur- 
vey catches. Specifically, we sought to ex- 
amine whether halibut catch is spatially 
structured relative to body size, sex, or 
the other species present. Although the 
longline survey has been conducted with 
standardized gear since 1984, data have 
not been collected with the spatial reso- 
lution required to test these hypotheses. 
Historical survey data are at the “skate 
level” (i.e. , all fish collected on a single 
skate of gear (100 hooks) are pooled on 
deck and subsequently processed in hap- 
hazard order). In the present study, se- 
quential hook-by-hook censuses were taken and 
subjected to one-dimensional spatial analysis. 
Materials and methods 
Sampling 
Field sampling was conducted during the 
IPHC’s standardized setline survey in June 
2006. Longline gear consisted of six skates of 
groundline tied end-to-end, with each skate measur- 
ing 549 m and having one-hundred 16/0 circle-hooks 
secured by 0.6— 1.2 m gangions spaced 5.5 m apart. 
Each hook was baited with semibright chum salmon 
( Oncorhynchus keta). Longline sets were conducted 
at pre-established stations located along the eastern 
Aleutian Islands, Alaska (Fig. 1). Gear was never set 
before 0500 hours and was allowed to soak for a mini- 
mum of five hours before being hauled back. Upon 
haul-back, a sequential hook-by-hook inventory (i.e., 
600 hooks per set) was attempted, during which each 
hook was designated as either empty or containing a 
halibut or other species. For every halibut captured, 
length was determined and its position on the longline 
was recorded. Sex was determined by dissection for all 
commercially caught legal-size ( >82 cm fork length [FL] ) 
individuals. For sublegal-size fish, sex was determined 
for only 36% of the individuals on each set because we 
were not granted approval to sacrifice sublegal-size hali- 
but outside of the standard survey protocol. Hereafter, 
“known” sex will refer to the proportion of the sampled 
population for which sex identification by dissection 
was performed. 
A total of 31 stations were sampled between 1 June 
and 24 June 2006. Five stations at which fewer than 25 
halibut were captured were eliminated from analyses. 
A sixth station was eliminated because all but one fish 
were sublegal-size and therefore also predominantly 
of unknown sex. Twenty five stations (Fig. 1; Table 1) 
remained that were amenable to statistical analysis; 
runs rests (RT; described subsequently) were conducted 
for all of these stations. Rapid hauling rate or other 
logistical constraints precluded full 600-hook invento- 
ries at 16 of the aforementioned stations. Thus, for only 
nine stations (Fig. 1; Table 1) were both RT and near- 
est neighbor analysis (NNA; described subsequently) 
possible. 
Statistical analyses 
Spatial structuring of halibut on longlines was exam- 
ined by using two one-dimensional statistical analyses, 
treating each longline as a transect and hook-status as 
events. Here, we make an explicit distinction between 
“aggregation” and “segregation.” Aggregation will refer 
to significant physical clustering of an event type with 
respect to linear distance, irrespective of whether other 
event types were also observed with those clusters. 
Segregation will indicate sequential occurrence of an 
event type in nonrandom sequence, irrespective of the 
distance between observations. Segregation indicates 
that the demographic is “undiluted” by other popula- 
tion segments, but does not necessarily imply that 
individuals occur in close proximity. Note that the 
