is not a feasible approach for small larvae already 

 preserved as part of plankton samples at sea. A 

 histochemical approach is feasible, however, and 

 we elected the periodic acid-Schiff (PAS) proce- 

 dure, which has largely superseded other histo- 

 chemical tests for glycogen (Davenport 1960). 



Cardell et al. (1973) showed good correlation of 

 PAS staining reactions with biochemical deter- 

 minations of liver glycogen for the rat. Glycogen 

 decreased from about 9% of liver wet weight to 

 0.77c after 1 d of fasting and to 0.4% after 3 d. At 

 the start of the experiment all hepatocytes had 

 dense masses of intensely stained glycogen. As the 

 fasting period lengthened to 2 and 3 or more days, 

 the masses decreased in size, number, density, and 

 stain intensity, and the number of cells showing 

 glycogen decreased markedly. Presumably such 

 differences in staining reaction in livers of north- 

 ern anchovy larvae would be an indication of 

 differences in glycogen reserves. 



Materials and Methods 



In the laboratory, 49 larvae in the size range 

 3.5-11 mm SL were selected by random dipping 

 from about half of the 37 nearshore net tow 

 samples that showed larvae of generally good 

 histological condition from the March 1977 cruise 

 (O'Connell 1980). Forty-one larvae were selected, 

 also by random dipping, from three of the four 

 nearshore tows containing abundant larvae 

 in generally poor histological condition. An addi- 

 tional dozen larvae were selected from three of the 

 offshore tows. 



All larvae had been fixed at time of capture in 

 Bouin's fluid and stored in 70*7^ ethyl alcohol. They 

 were subsequently dehydrated in n -butyl alcohol, 

 embedded in paraffin, sectioned sagittally and 

 stained by the PAS method (Preece 1965). We did 

 not subject the material used in this study to 

 diastase digestion tests but feel confident that the 

 red coloration in the livers indicates glycogen. 

 Glycogen gives one of the stronger reactions to the 

 PAS stain (Lillie and Fullmer 1976), and Cardell 

 et al. (1973) established with diastase controls 

 that the PAS-positive material in the livers of rats 

 was glycogen. The diastase test essentially dis- 

 tinguishes between glycogen and mucins (Preece 

 1965), some of which show strong PAS reactions, 

 but these occur primarily in the integument and 

 epithelia of the digestive tract and various glands 

 of animals (Lillie and Fullmer 1976). 



After staining, slides were randomized with 



their identities concealed and then rated by micro- 

 scope examination. Each of two observers rated 

 each .specimen as High, Medium, or Low, depend- 

 ing on the degree and extent of red coloration 

 in the liver. The Low grade was assigned when 

 livers showed virtually no red color, the High 

 grade when color was strong and widespread. The 

 Medium grade was assigned when color was light 

 and scattered, or irregular. The two readers dis- 

 agreed on a little >12'% of the larvae but never by 

 more than one grading step. These differences 

 were reconciled by reexamination and discussion. 



No attempt was made to characterize the speci- 

 mens stained by the PAS procedure as robust or 

 emaciated on the basis of histological factors per 

 se. There was the possibility that the PAS proce- 

 dure would be less precise and consistent than the 

 previously used hematoxylin and eosin stain in 

 demonstrating cell and tissue components other 

 than polysaccharides. 



Study of material from the sea samples was 

 preceded by analysis of 99 specimens from groups 

 of larvae that were fed or starved in the labora- 

 tory. Fixation, staining, and microscope analysis 

 were exactly as outlined above, except that the 

 laboratory material was held in 70% ethyl alcohol 

 for only a few days before dehydration and embed- 

 ding. Reader disagreement was 9% on the labo- 

 ratory material. 



The larvae obtained from laboratory containers 

 ranged from 5- to 26-d-old. The rotifer Brachionus 

 plicatilis was introduced into containers as food at 

 a density of 40 to 60 ml when the larvae were 3-d- 

 old and maintained above 30;ml by additions as 

 needed. The smaller Gymnodinium splendens was 

 included as a starting food at the outset, but was 

 not afterwards maintained. Northern anchovy 

 larvae require Brachionus at densities of at least 

 10 to 20 /ml to survive and grow well in laboratory 

 containers for the first weeks of life (Theilacker 

 and McMaster 1971). 



Food was withheld from two containers, and 

 specimens from these were sacrificed on the first 

 and second day after yolk exhaustion. Starvation 

 at more advanced ages was accomplished by 

 removing the food from selected containers 2 to 4 d 

 before the larvae were sacrificed. Food was re- 

 moved with a siphon filter devised by P Paloma 

 for the purpose. A typical air-driven aquarium 

 siphon was enclosed in an 8.89 cm <3.5 in> diam- 

 eter perforated plastic cylinder covered by nylon 

 netting with 0.333 mm mesh openings, which 

 allowed food organisms but not fish larvae to pass 



807 



