Francis et al : Age and growth estimates for Polyprlon oxygeneios 



231 



Tag recapture 



1623 hapuku were tagged at three main sites: Poor 

 Knights Islands (July 1987-August 1989, ;! = 106), 

 Cook Strait (August 1979-June 1984, 7i=599) and 

 Oamaru (March 1988-April 1990, «=918) (Fig. 2). 

 Fish were caught by longline or drop (dahn) line in 

 depths less than 160 m. Only lively fish without 

 everted stomachs or bulging eyes were tagged. Dis- 

 tended swimbladders were vented by means of a hol- 

 low needle inserted through the wall of the body cav- 

 ity. Hapuku were measured (TL), tagged with a loop, 

 dart, or spaghetti tag, and released. Oamaru hapuku 

 were double tagged with both a loop and a dart tag. 

 Loop and spaghetti tags were inserted through the 

 muscle just anterior to the dorsal fin and clipped or 

 tied together, and dart tags were inserted into the 

 muscle directly below the first dorsal-fin ray. Each 

 tag was printed with a serial number, a return ad- 

 dress and notice of a reward. Recapture data and most 

 measurements of recaptured hapuku were provided to 

 the Ministry of Agriculture and Fisheries by the fish- 

 ermen who caught them, but occasionally whole fish 

 were returned for measurement. 



Von Bertalanffy growth estimates were obtained 

 from the tagging data by using the maximum likeli- 

 hood method and the computer program GROTAG 

 (Francis, 1988). GROTAG also estimates ^^^ and ^p, 

 the mean annual growth of fish of lengths a and (i 

 respectively. The reference lengths a and (i were cho- 

 sen to lie within the range of lengths of tagged 

 hapuku. Francis (1988) showed that these param- 

 eters describe the growth information in tagging data 

 better than the more conventional von Bertalanffy 

 growth parameters L^ and /C do. The expected length 

 increment, AL, for a fish of initial length L^ at lib- 

 erty for time t^T is given by 



AL = 



§a -§13 



Ll 



1 + 





.AT 



Also estimated were m and s (the mean and stan- 

 dard deviation of the measurement error), v (the co- 

 efficient of variation of growth variability), andp (the 

 proportion of outliers) (Francis, 1988). Preliminary 

 GROTAG fits suggested that the von Bertalanffy 

 growth model was not appropriate for tagged hapuku; 

 therefore a linear GROTAG function with only one 

 growth parameter ig^^) was fitted to the data. A sea- 

 sonally varying growth function was also tested. Dif- 

 ferences in growth rate between the two sexes in Cook 

 Strait (the sex of recaptured fish was not determined 

 for the other two areas) and among the three tag- 

 ging areas were also investigated. 



The approach used was to fit a simple GROTAG 

 model with few parameters to the data. The com- 

 plexity of the model was then gradually increased 

 by introducing additional parameters (for example, 

 parameters to allow for seasonal variation in growth). 

 At each stage, new parameter estimates were made 

 and likelihood ratio tests for significant improvement 

 in model fit were carried out as described above for 

 MULTIFAN models. A significance level of 0.05 was 

 used for testing whether there was any gain from 

 introducing additional parameters. 



Results 



Otolith ageing 



Otolith structure The otoliths of the hapuku we ex- 

 amined had an opaque core (dark in transmitted 

 light) and an opaque line that radiated along the 

 dorsoventral axis (Fig. 3). The line was usually split 

 by 1-3 hyaline bands (light in transmitted light). The 

 sulcus was also mostly opaque. The remainder of the 

 otolith consisted of alternating opaque and hyaline 

 bands. In young fish, the hyaline bands were wider 

 than the opaque bands, but difference in width de- 

 clined in older fish. Opaque bands usually consisted 

 of multiple, fine, wavy opaque and hyaline zones 

 (Figs. 3 and 4A). The demarcation between each com- 

 pound opaque band and the adjacent hyaline bands 

 was often indistinct, making the bands difficult to 

 count. The banding pattern was usually clearest in the 

 dorsal half of transverse otolith sections. We used counts 

 of the compound opaque bands as age estimates. 



Five otolith sections were considered unreadable, 

 and a further four otoliths were lost or damaged dur- 

 ing preparation. Of the remaining 241 otoliths, 78% 

 were scored as poor or moderate (readability 2 or 3); 

 only 22% were good or exceptionally clear (readabil- 

 ity 4 or 5). 



Validation Otolith marginal state was difficult to 

 assess because of the presence of split opaque rings. 

 There was no apparent seasonal cycle in marginal 

 composition, and sample sizes in most two-month 

 periods were inadequate to provide good estimates 

 of the percentage of the population with opaque mar- 

 gins ( Jan-Feb percent opaque=50.0, n = 14; Mar-Apr 

 37.5, 8; May-Jun 36.4, 33; Jul-Aug 42.9, 7; Sep-Oct 

 50.0, 10; Nov-Dec 25.0, 4). The same conclusion was 

 reached for the central data alone («=50). 



Five OTC-injected hapuku were recaptured after 

 0.20-2.66 years at liberty (Table 1). All had distinct, 

 bright OTC bands when viewed under UV light (Fig. 

 4B). There was no apparent relation between OTC 



