Fitzgerald et al.: Elemental signatures in otoliths of larval Theragra chakogramma 



609 



o 



4-2024 

 D ft ft o 



ft M° 



ft 

 ft 



* o 



< * a 



° * 



ft #A *° ^o 



ft o° 



> *>■* *° *o * * 



_•' " O . " A 4 .»-* 



II ' t o ° *4 » • 



•4-20246 



Canonical variate 1 



Figure 2 



Plot of first two canonical variates con- 

 trasting multivariate elemental sig- 

 natures in otoliths of walleye pollock 

 (Theragra chalcogramma) determined 

 by using electron probe microanaly- 

 sis, at 0-20 (im from the nucleus (A) 

 and 20-40 Jim from the nucleus (B). 

 Larvae were collected from the North 

 Bering Sea (▲), southeast Bering Sea 

 (O), Bristol Bay IB), Shelikof Strait 

 !♦), Prince Wiiliam Sound lO>. and 

 Yakutat (ft). 



 »\ 



J 8 o - 



° . ° 



OA 



* * i 



* t Acf 



H 



-  A. ° 



. ^ A V 



-3-113 



Canonical variate 1 



Figure 3 



Plot of first two canonical variates 

 contrasting multivariate elemental 

 signatures in otoliths of walleye pol- 

 lock (Theragra chalcogramma) deter- 

 mined by electron probe microanalysis 

 at 0-20 /im from the nucleus iAi and 

 20-40 jim from the nucleus IB). Larvae 

 were collected from the North Bering 

 Sea (A), southeast Bering Sea (O). and 

 Bristol Bay  



We used CDA to visualize multivariate differences 

 among locations in reduced dimensional space. Three 

 groups were readily discernible in a plot of the first two 

 canonical variates (Fig. 2). Samples from the North Ber- 

 ing Sea, the southeast Bering Sea, and Shelikof Strait 

 formed one group separated from Yakutat, Bristol Bay, 

 and Prince William Sound samples along the first ca- 

 nonical variate. The second canonical variate separated 

 Yakutat samples from Bristol Bay and Prince William 

 Sound individuals. Elemental signatures at 20-45 jum 

 from the otolith nucleus were distributed similarly in 

 canonical space to samples from the otolith nucleus 

 (Fig. 2). Three groupings were apparent in the canoni- 

 cal plot, and Bering Sea larvae were separated from 

 Bristol Bay and Prince William samples on canonical 

 variate one, and Yakutat samples were separated from 

 all other locations on canonical variate two. We then 



conducted a similar analysis with only samples from 

 the southeast and North Bering Sea and Bristol Bay. 

 Elemental signatures of larvae from the Bering Sea 

 separated from Bristol Bay on canonical variate one. 

 The southeast Bering Sea samples separated from the 

 North Bering Sea along canonical variate two, although 

 not as clearly as with the elemental signatures from the 

 Bering Sea and Bristol Bay (Fig. 3). 



Elemental profiles across otoliths varied significantly, 

 as determined by repeated measures ANOVA, among 

 the five locations for Na, P, S, and Sr (Table 4). Both 

 S and Sr concentrations declined from high values at 

 the nucleus to significantly lower values towards the 

 edge of the otolith (Fig. 4). Repeated measures ANOVA 

 also provided a test of the differences among locations 

 when data were averaged over the otolith profiles. Sig- 

 nificant differences among locations were detected for 



