Livingston et a I.: Estimating the annual proportion of nonspawning adult Macruronus novaezelandiae 
would leave the Southern Plateau after May would 
be underestimated. Third, the number of developing 
fish in May that could have resorbed their eggs and 
not gone on to spawn after all could lead to an over- 
estimation of the total proportion spawning. Other 
sources of error concern changes in catchability and 
vulnerability between surveys and the difficulty of 
detecting any bias or size selectivity when sampling 
a population with the trawl. 
With regard to the first source of error above, it 
was encouraging that hoki collected in April showed 
significant signs of development compared with those 
collected in February (Table 2). The most likely fish 
to be affected by late development are the younger, 
smaller fish because they spawn later in the season 
compared with the older, larger fish which spawn at 
the beginning of the season (Langley, 1993). Because 
we estimated the proportion of hoki age 7 years and 
above that were spawning, the problem was minimized. 
With regard to the second source of potential er- 
ror, undeveloped hoki of age 4 and greater are not 
caught on the spawning grounds during the spawn- 
ing season (Langley, 1993), suggesting that there may 
be 100% spawning among hoki that migrate to the 
west coast spawning grounds. We have no data on 
the number of fish that could resorb their eggs be- 
fore the spawning season, but none were found in 
the May samples in this study. 
With regard to the trawl survey technique, it is 
possible that there are systematic changes in 
catchability or vulnerability between December and 
May. We considered it more likely, however, that the 
changes in fish numbers were real, and that some 
fish had already migrated away from the Southern 
Plateau before the May survey, particularly in 1993 
when the survey took place later than in 1992. 
The ratio of the number of missing fish to the num- 
ber of fish present on the Southern Plateau (the mi- 
gration ratio, x) can also be expressed in terms of 
the proportion of fish that have already migrated (p ) 
x = -P *- 
!-Pm 
or, equivalently 
x 
Both x and p m change as fish leave the Southern Pla- 
teau. In December, when no fish have migrated, both 
x = 0 and p m = 0. If a survey were to be done at a 
point when 33% of the fish had migrated, or p m = 
0.33, there would be one fish missing for every two 
fish still in the survey area, and x would be 0.5. By 
July 1993, when an estimated 82% of fish have gone 
to spawn, p m is 0.82 and x has increased to 4.6. 
Therefore, in May 1993, when x - 0.92 and P m = 
0.48, we can estimate that more than half (58.5%) of 
the fish that were going to spawn had already gone. 
This is, of course, poorly estimated, as is x, but it 
is higher than was expected before the surveys were 
done. If the survey results are correct, they suggest 
that in May 1993, fish had already started to mi- 
grate in large numbers. The survey in May 1993 be- 
gan two weeks later in the year than that in 1992, 
indicating that May is a critical time for the spawn- 
ing migration of hoki. This interpretation is sup- 
ported by the change in distribution of fish from the 
east to the west between December and May in 1993, 
but not in May 1992. If, however, our estimates of 
the numbers of fish that have migrated away from 
the Southern Plateau by May are incorrect (e.g. be- 
cause of changes in catchability or vertical availabil- 
ity between December and May, or because not all 
fish have begun to develop by May), then the propor- 
tion of prespawners on the Southern Plateau in May 
could be used as a lower limit of the proportion 
spawning of the total population. Given that a stan- 
dardized trawl survey technique was the best method 
available to us to sample the adult hoki population, 
we believe that it would be difficult to improve on 
the estimates of proportion spawning obtained. 
The 4-6 yr age classes show different proportions 
of prespawners in each year, with more 4 year olds 
but with fewer 5 and 6 year olds developing to spawn 
in 1992 than those in 1993 (Fig. 6). Differences in 
the proportion of spawning fish in the younger age 
classes could also relate to the preceding spawning 
history of a particular age class. 
Although every year many hoki spawn on the west 
coast of the South Island, it is clear that a large num- 
ber of individuals do not. Species that exhibit such 
behavior usually have a major accessory activity that 
requires a significant amount of energy in addition 
to spawning itself (Bull and Shine, 1979). Hoki mi- 
grate over vast distances of about 1,500 km from the 
Southern Plateau to the west coast spawning 
grounds. The energy cost incurred during migration 
may be so high that there is not enough left for egg 
production the following year. 
Lack of food and migration distance have been 
suggested as reasons for lack of spawning among 
orange roughy (Bell et al., 1992) and yellow fin bream 
(Pollock, 1984). However, Pulliainen and Korhonen 
(1990) found that nonspawning burbot maintained 
a condition similar to that of spawning burbot and 
ruled out low food supplies as an explanation for 
nonspawning adults. 
Within species where different populations show 
different levels of nonspawning, it has been found 
