THEILACKER and WATANABE: NUTRITIONAL STATUS OF NORTHERN ANCHOVY 



(S4/F) where P = 0.8, indicating that delaying 

 feeding affected the final fish size at a given age 

 (ANOVA; Table 2). 



Progi'essively increasing the periods of star- 

 vation before feeding also caused a decrement 

 in the gi'owth rate after feeding resumed (Fig. 

 2; Table 2). Larvae were unable to compensate 

 for a delay in feeding at the onset, and fish 

 length at an age where all fish had been feeding 

 for 6 days differed for each treatment [P = < 0.01 

 for all diet combinations except starved 1 d/fed 

 (Sl/F) vs. 2 d delay in feeding (S2/F) where P 

 = 0.08 (ANOVA; Table 2)]. Growth rates for 

 the 6 d period ranged from 0.41 mm/d for the 

 fed group, 0.39 mm/d for 1 d delay in feeding 

 (Sl/F), 0.34 mm/d for 2 d delay (S2/F), 0.31 

 mm/d for 3 d delay (S3/F). Length was trans- 

 formed to natural logs for the ANOVAs. The 

 test for the equality of slopes showed that all 

 slopes differed (P = <0.0001 for all combina- 

 tions). Thus, delaying feeding affected the final 

 fish size and caused a decrement in the growth 

 rate after feeding resumed. 



Evaluation of Midgut Cell Height 

 Criterion 



For a subset of northern anchovy larvae taken 

 from the treatments where larvae were fed, 

 starved, and feeding was delayed 1 day (Sl/F), 

 we compared the number of larvae correctly 

 classified to feeding treatment by the midgut cell 



height and by the traditional histological index 

 (O'Connell 1976). To define the midgut cell 

 height interval for each feeding treatment, we 

 selected the midpoint between successive cell 

 height means (Table 3) as the interval break- 

 point for predicting feeding history from cell 

 height. The midgut cell height was correlated 

 with the histological score (Fig. 3, n = 38, F < 

 0.001, ^ratio = 25.85). Both the cell height and 

 the histological score correctly classified all of 

 the subset larvae that were fed to the correct 

 group and 78 and 79% of the starved larvae to 

 the correct group. On the other hand, for the 

 delayed feeding group (Sl/F), the midgut cell 

 height was a better predictor of feeding history 

 than the histological score. The midgut measure- 

 ment classified 78% correctly whereas the histo- 

 logical score classified only 50% correctly. 



After the apparent success of this evaluation, 

 we used the midgut cell height data set (Tkble 1; 

 Fig. 4) to establish the criteria needed to define 

 and cahbrate the midgut cell height measure- 

 ment for predicting feeding history and for calcu- 

 lating starvation rates of anchovy larvae in the 

 sea. 



Diet and Midgut Cell Height Categories 

 for Larvae <4.0 mm 



We defined the diet and cell height categories 

 for first-feeding larvae (<4.00 mm SL) which are 

 vulnerable to starvation. The height of the 



8.0 



3.0 



STARVED 1d/FED 



STARVED 2d/FED 

 / 



^ STARVED 3d/FED 



 STARVED 4d/FED 



_L 



10 15 



AGE (d) 



20 



25 



Figure 2. — Growth of EngrauUs mordax larvae under different 

 feeding conditions. 



461 



