504 



Fishery Bulletin 104(4) 



The MANOVA of the four PCs identifie(i significant 

 differences in the PC scores of otolith shape among 

 regions, but were not consistent between cohorts (1995 

 and 1999; cohortxregion interaction. Table 4). Uni- 

 variate ANOVAs indicated that these differences were 

 due to variation in PC II among regions and in PC IV 

 among reefs within regions — both spatial patterns vary- 

 ing depending on the cohort in consideration (Table 5). 

 Differences in PC II were due to differences between 

 Lizard Island and Mackay regions in 1995 and between 

 cohorts in Mackay (Fig. 3A; HSD, P<0.05). The apparent 

 inconsistency between the significant reef within region 

 effect in the ANOVA (PC IV) and no such reef effect 

 detected in the MANOVA is most likely explained by 



the difference between two reefs in the Lizard Island re- 

 gion, for which there was a significantly lower mean PC 

 IV score for Reef 14132a than for Reef 14147 in 1995, 

 whereas the reverse was observed in 1999 (Fig. 3B; 

 HSD, P<0.05). There was also a significant difference 

 among cohorts on Reef 14132a (Lizard Island) and Reef 

 20296 (Mackay Region; Fig. 3B; HSD, P<0.05). 



Although only explaining 18% (PC II and PC IV com- 

 bined) of the variation in our morphological data, the 

 PC analysis indicated that the presence of at least two 

 stocks of P. leopardus on the GBR in 1995, one in the 

 northern part of the GBR (Lizard Island) and one in 

 the southern part (Mackay), whereas one homogeneous 

 stock was indicated in 1999. 



