MAGNUSON and HEITZ: GILL RAKER APPARATUS 



Table 5. — Percent crustaceans by volume in the stomachs, median fork length, mean gill raker gap, and species 

 of fish. (Ranks are from smallest to largest and ordered by the percentage of crustaceans in the stomachs.) 



analyses. The galatheids and portunids domi- 

 nating the crustaceans found in T. albacares 

 stomachs in the eastern tropical Pacific (Alver- 

 son, 1963) are much larger (Longhurst, 1967; 

 Jerde, 1967b) than the tyijical crustaceans from 

 the stomachs of central Pacific scombrids given 

 in Table 3. Also, data were not used if fewer 

 than 25 stomachs had been examined. None of 

 the 238 A. solanderi contained crustaceans and 

 O'^f crustaceans in the stomach was considered 

 a reasonable estimate for any \&Yger A. solanderi. 

 The median or midrange fork length of fish 

 was determined for each set of stomach data. 

 Then mean gill raker gaps for fish of those spe- 

 cies and length were estimated with the regres- 

 sions from Table 1. Data on median fork length, 

 mean gill raker gap, and percent crustaceans 

 by volume in the stomach are presented in nu- 

 merical and ranked form in Table 5. 



Percent volumes that crustaceans contributed 

 to the stomach content were inversely related 

 to mean gill raker gap (Figure 4a) (Kendall 

 rank correlation coefficient, r = — 0.59; n = 16; 

 P <0.001) and to fork length (Figure 4b) 

 (Kendall rank correlation coefficient, t = — 0.45; 

 n = 16; P <0.01). Several notable exceptions 

 occurred in the relation with fork length 

 (Table 5, Figure 4b). C. hippurus, 81 cm long, 

 contained 2% crustaceans while T. albacares, 

 80 cm long, contained 45 "^f crustaceans. T. 

 albacares, 135 cm long, also contained 2% crus- 



taceans. Not unexpectedly, C. hippurus, 81 cm 

 long, and T. albacares, 135 cm long, both had 

 mean gill raker gaps near 5 mm whereas the 

 81-cm T. albacares had a smaller mean gill raker 

 gap near 3 mm. The somewhat closer corres- 

 pondence of percentage of crustaceans to gill 

 raker gap than to fork length can be observed 

 by comparing Figures 4a and 4b or by comparing 

 the associated probabilities of no correlation 

 (<.01 versus <.001). 



Kendall partial rank correlation coefficients 

 were computed to determine the association be- 

 tween percent crustaceans in the stomach and 

 gill raker gap, with the effect of fork length 

 held constant. The partial correlation coefficient 

 between percent crustaceans and gap, indepen- 

 dent of variation in fork length, was — 0.43 

 while the partial correlation between percent 

 crustaceans and fork length independent of var- 

 iations in gap, was only — 0.05. Thus, although 

 fork length was correlated with the percent 

 crustaceans, this correlation resulted from the 

 association between gill raker gap and fork 

 length. Gill raker gap was the important var- 

 iable correlated to percent crustaceans in the 

 diet. 



Data on percent crustaceans in the stomach 

 by volume were also presented for K. pelamis 

 and T. albacares of various size by Alverson 

 (1963) and for K. pelamis and blackfin tuna, 

 Thunnus atlanticus (Lesson), by Suarez Caabro 



367 



