Francis et al / Age and growth estimates for Polyprion oxygeneios 



233 



Figure 4 



Photomicrographs of the dorsal end of a transverse section from a hapuku otolith illuminated by (A) transmit- 

 ted white light, and iB) reflected W light. The hapuku had been tagged, injected with oxytetracycline, and 

 recaptured after 2.66 years at liberty (see Table 1, tag number SGO306, for further details). Arrowheads indi- 

 cate the position of the inner edge of the oxytetracycline mark, which lies within an opaque band. Outside that 

 band are three hyaline (Hi and two opaque (O) bands. Scale bar = 0.1 mm. 



been at liberty for the shortest time (SGO340, 0.20 

 years) had the OTC mark near the margin (postin- 

 jection increment width along the dorsal axis=0.01 

 mm) (Table 1). Three hapuku with intermediate pe- 

 riods at liberty (0.37-0.45 years) had moderate 

 postinjection otolith increments of 0.05-0.16 mm. 

 SG0328, which was the smallest of the three hapuku 

 at tagging, had the largest increment. SG0322 and 

 SG0328 each had an incomplete opaque band and 

 an incomplete hyaline band outside the OTC mark. 

 SG0329 had a complete hyaline band and possibly 

 an opaque band beginning to form at the margin near 

 the dorsal tip of the otolith. A hapuku that had been 

 at liberty for 2.66 years (SGO306) had a wide 

 postinjection increment (0.61 mm) consisting of a 

 partial opaque band followed by three hyaline and 

 two opaque bands (Fig. 4; Table 1). 



Age estimates and growth rates R2 j and R., .^ were 

 strongly positively correlated (coefficient of multiple 



determination [i?'^]=94.6), although they differed 

 markedly for some individuals (Fig. 5A). Some of the 

 larger differences were for otoliths that were judged 

 difficult to count ( readability 2 ), but large differences 

 were also found for clearer otoliths (readability 3-5). 

 R,2 J tended to be higher than R.^ •> for ^^h aged 3-18 

 years in R.^.^ (mean difference^l.O years) (Fig. 6A). 

 Sample sizes were small for hapuku older than 18, 

 but it appears that the same bias occurred for them. 



Age estimates made by both readers (Rj and R2 2' 

 were also highly correlated (i?2=92.4) but there were 

 some substantial differences (Fig. 5B). There was no 

 apparent bias between the two sets of estimates for 

 hapuku aged 3-12 in R2 2, but there was a slight ten- 

 dency for R2 2 to exceed Rj over the range 13-18 years 

 (Fig. 6B). 



Ageing precision was low and variable, both within 

 and between readers, for hapuku aged 3-18 years 

 (Fig. 7). The CV of the age estimates was slightly 

 lower between the two readings of reader 2 (R,, j ver- 



