I 02 
Fishery Bulletin 95( 1 ), 1997 
cial samples collected from January through August 
(see below). These criteria (described in the Results 
section) were then used by an independent reader to 
classify May 1992 and May 1993 female hoki as 
prespawners or nonspawners. 
Monthly samples were also collected from commer- 
cial vessels to monitor hoki development between 
January and September. Up to 40 ovary samples from 
a range of adult-size fish were preserved in 8% buff- 
ered formalin at the time of capture. Samples were 
processed in the laboratory, sectioned, and stained 
with standard haemotoxylin and eosin preparations. 
Samples were collected mostly from Chatham Rise 
(Jan-May) and the west coast of the South Island 
(Jun-Jul) because these were the areas where com- 
mercial vessels were operating at that time. 
Each fish was staged histologically (as described 
above) to determine the earliest month in which de- 
velopment began. Five of the most developed fish 
from each month were then selected for oocyte mea- 
surement to confirm that New Zealand hoki are syn- 
chronous or group synchronous like their Australian 
counterparts (Gunn et al., 1989). The mean diam- 
eter of 200 oocytes was measured (after Foucher and 
Beamish, 1980) from each of the five fish. 
Ageing 
It is unknown whether recruitment to the spawning 
fisheries is length-driven or age-driven. Observers 
from the west coast hoki fishery have found spawn- 
ing hoki as young as 2 years and as small as 42 cm 
total length, in some years (Sullivan et al. 2 ). Because 
the model used for stock assessment is an age-struc- 
tured one (Sullivan et al. 2 ), fish were aged as part of 
this study, and all analyses were carried out by us- 
ing age data. It was also important to age fish so 
that comparisons of the numbers of fish in each co- 
hort in December and May could be made. 
Otoliths from each fish in the histological samples 
from the May surveys and from the biological samples 
in December were aged by using the validated age- 
ing method described by Horn and Sullivan (1996). 
These data were also used to develop age-length keys. 
Where there were no fish in the sample of a given 
length, the age-length key was interpolated with 
nearby values of age. 
Proportion of each age class developing to 
spawn 
The proportion of fish in each age class that were 
classifed as prespawners was estimated for each age 
stratum from the aged histological samples. The 
number of fish at each age in each stratum was esti- 
mated from the age-length key and the length-fre- 
quency distribution in the stratum. For each age class 
the total number of prespawners was therefore esti- 
mated from the proportion of fish that were develop- 
ing to spawn and from the total number of fish in 
each stratum. The standard error of these estimates 
was estimated by using a resampling technique, 
whereby in each stratum, a sample, the same size as 
the original sample, was selected (with replacement) 
from the original sample. The age-length key and 
proportion of prespawners were calculated from the 
combined sample of the 15 strata. This process was 
repeated 1,000 times. The standard error of the esti- 
mates was estimated from the standard deviation of 
the values in 1,000 replicates. 
Proportion of adult spawning fish 
An estimate p + of the proportion of prespawners in 
the plus group of adult fish (p + ) can be obtained by 
using the method described above but by consider- 
ing all adult fish as a single plus group. However, if 
some fish had already left the Southern Plateau to 
spawn before the survey in May, they also should be 
counted as prespawning fish. This means that the 
proportion of adult prespawners present on the 
Southern Plateau in May ( p + ) is an underestimate of 
the total proportion of fish that will spawn (p) as 
s + ns s + ns + g 
where s is the number of prespawners on the South- 
ern Plateau in May, ns is the number of nonspawners 
on the Southern Plateau in May, and g is the num- 
ber of fish that had left the Southern Plateau before 
May, presumably to spawn. 
If we define a migration ratio, 
x = — — — > (2) 
s + ns 
to be the ratio of the number of fish which have gone 
to spawn to the number of fish (both prespawners 
and nonspawners) that are still in the survey area 
when the second survey is done, then 
p= S + § =P±±±. ( 3 ) 
s + ns + g l + x 
Thus p equals p + when x is zero and increases to- 
wards an asymptote of 1 when x is very large. 
