FISHERY BULLETIN: VOL. 76, NO. 2 



conditions were most suitable at this time. Hand 

 nets were used in the intertidal pools and multi- 

 prong spear guns in the subtidal area. Hook and 

 line, poison, traps, and gill nets were not used as 

 further biases may be induced to the feeding data 

 (Randall 1967). Fish collected with hand nets were 

 immediately placed in a 10% Formalin^ solution, 

 whereas this procedure was delayed for up to iy2 h 

 in the case of those taken by spear. It was con- 

 cluded that death stops or greatly slows digestion, 

 as the stomach contents were found to be in an 

 equally digested state in both groups on later 

 analysis. This has also been observed by Hobson 

 (1974). 



The fish were left in Formalin for 10 to 14 days. 

 This time period was maintained throughout to 

 standardize any length and weight changes in- 

 duced by the fixative (up to 5%, Royce 1972). 

 About 10 scales were removed from under the pec- 

 toral fin and cleaned with a camel hair brush after 

 having been soaked overnight in water with a 

 trace of carbolic acid (Pinkas 1966). They were 

 mounted dry and examined over a white 

 background using a low-power binocular micro- 

 scope. Standard lengths to the nearest millimeter 

 were taken and the stomach removed and placed 

 in 45% n-propyl alcohol. 



The stomach is here defined as that part of the 

 gut between the last gill arch and the gut caecae. 

 The intestines were not examined as some food 

 items are more resistant to digestion than others, 

 with resultant biases as one moves along the gut 

 (Randall 1967; Kionka and Windell 1972; Gannon 

 1976). Food items were identified to species where 

 possible. 



Numerous methods have been employed in 

 analyzing the food habits of fishes and volumetric 

 and gravimetric techniques are being more widely 

 used today with the current trend towards greater 

 accuracy (Windell 1971). Both suffer from the 

 same limitation in that digestion of the food both 

 reduces its volume and weight. This has resulted 

 in the use of reconstructed weights and volumes 

 where the live weight and/or volume is back- 

 calculated from a measureable parameter, e.g., 

 carapace length (M. Bruton, pers. commun.). In 

 this particular study some of the fish had fed on 

 diatoms and it is not feasible to determine the 

 volume of such small items (Windell 1971). Simi- 

 larly, the reconstructed weight could not be de- 



termined as a sample of monospecific, uncontami- 

 nated diatoms is impracticable to obtain and 

 contains an indeterminate number of dead frus- 

 tules which varies from sample to sample (Round 

 1971). 



In such cases, the points (Swynnerton and 

 Worthington 1940) and ranking index methods 

 (Hobson and Chess 1973) would appear to be more 

 satisfactory and were initially used in the present 

 study. The points system was modified by Frost 

 (1943) and subsequently by Hynes (1950) to take 

 into account gut fullness, 30 points being allotted 

 when the stomach was distended, 20 when full, 10 

 when half full, and so on. One, two, four, eight, or 

 sixteen points were assigned to each food item 

 rather than fractions of the total allotted to each 

 stomach in proportion to their volumes. This is an 

 artificial situation and the method was revised as 

 described below. 



After removal, the stomach is allotted between 

 and 30 points in proportion to its fullness. This is 

 very subjective but overcome to some extent when 

 large numbers of guts are handled. The contents 

 are then sorted, identified, and the percentage 

 volume estimated for each food item with the aid of 

 Data Sheet No. 6 of Geotimes.'' All estimations are 

 made with the organisms spread out to an even 

 depth throughout the microscope field or equiva- 

 lent surface. The total number of points allocated 

 to that stomach is then subdivided amongst the 

 food items in proportion to their percentage vol- 

 umes. The points gained by each food item are 

 summed for the total sample offish and the mean 

 calculated. The values are then scaled down to a 

 percentage to give the dietary composition of the 

 fish examined. In the case of the ranking index 

 (RI), the volume is estimated as above and the 

 mean calculated for each food item per fish. The 

 mean volume is then multiplied by the ratio of the 

 number of fish containing that item to the total 

 sampled. 



The points method, however, places too much 

 weight on single food items that have been fed on 

 to distension by a few fish, whereas the RI method 

 fails to consider stomach fullness. It is therefore 

 suggested that an alternative, here termed the 

 comparative feeding index (CFI), would be more 

 suitable as it takes into account all three factors, 

 i.e., the volume, fullness, and frequency of occur- 

 rence of each food item. The method involves the 



^Reference to trade names does not imply endorsement by the 

 National Marine Fisheries Service, NOAA. 



390 



'Available from the American Geological Institute, 2101 Con- 

 stitution Avenue, N.W., Washing^n, D.C. 



