FISHERY BULLETIN: VOL. 72, NO. 4 



and included data on food habits; however, he in- 

 cluded little information on activity. The 1970 

 United States Tektite II program provided many 

 scientists with the opportunity to make direct ob- 

 servations on a Virgin Island reef, and reports 

 concerning the fishes have been published in one 

 volume (Collette and Earle, 1972). Many other 

 reports of limited scope are scattered through the 

 literature, most of them being fragmented data on 

 food habits; nevertheless, accounts of activity 

 based on direct observations are sparse, especially 

 of nocturnal activity. 



The great variety of feeding mechanisms for 

 which teleostean fishes are so well known occur 

 among coral-reef fishes far more so than among 

 the fishes of any other habitat. I take advantage of 

 this circumstance in the discussion that concludes 

 the present report and consider the feeding rela- 

 tionships among fishes on Kona reefs in the con- 

 text of teleostean evolution. 



METHODS 



Direct Observations 



I observed activity of the fishes during 632 h 

 underwater at all periods of day and night using 

 scuba and by snorkeling. Except when collecting 

 specimens, I tried not to influence the fishes or 

 their environment, hoping that events were tak- 

 ing a natural course. Fishes considered in this 

 report are those that can be seen by an underwater 

 observer at some time during day or night. Al- 

 though this includes by far most of the reef fishes, 

 some abundant species are not included because 

 they remain secreted in the reef at all hours. 



Food Habits 



The gut contents of 1,547 fish specimens of 102 

 species were analyzed. With a few isolated excep- 

 tions, noted below, all the specimens were col- 

 lected with spears. I find spearing the most effec- 

 tive way to collect fishes for study of food habits. 

 Using this method, specimens were collected in 

 specific locations at the times of day and night that 

 best define diurnal-nocturnal activity patterns. 

 Because I speared all the specimens myself, I 

 know what each individual was doing when cap- 

 tured, and this knowledge significantly influenced 

 analysis of the data. Even the response of the 

 various fishes to being stalked and speared (or 

 missed) provided certain behavioral insights. 



Food habits change over the life of at least most 

 fishes, usually along wdth recognizable changes in 

 behavior and morphology. Unless otherwise indi- 

 cated, specimens selected for this study showed 

 behavior and morphology judged typical of adults. 



The collections were spread over time and space, 

 so that possible effects of transient localized, 

 perhaps atypical, situations were reduced. Gener- 

 ally, only a single individual of any one species 

 was collected during a single period of observa- 

 tions; thus, for a given species, most individuals 

 each represent a separate collecting station. For 

 these reasons, I judge the data from the food habit 

 analysis to accurately represent the situation ex- 

 isting on Kona reefs over the 15 mo of this study. 



The collections were spaced throughout day and 

 night, so that relative digestion of gut contents 

 supplements direct observations of activity in de- 

 termining specific feeding times. All specimens 

 were sealed in individual plastic bags im- 

 mediately after being speared, most while still 

 underwater. Gut contents of specimens collected 

 while snorkeling were preserved immediately by 

 injecting a concentrated formaldehyde solution 

 directly into the gut cavity, whereas gut contents 

 of fishes taken by scuba were preserved as soon as 

 possible after emerging from the water. I was un- 

 able to see a difference in the digestion undergone 

 by material collected in each of these two ways, 

 suggesting that digestion is sharply curtailed by 

 the death of the fish. Where practical, 

 identifications of items in the guts were carried far 

 enough to establish such general prey characteris- 

 tics as habitat and mode of life. 



Quantifying Food Habits 



For those species represented by enough num- 

 bers in the analysis of gut contents, I state: 1) the 

 number offish of that species containing each food 

 item, and 2) the mean percent of that item in the 

 diet volume, which is the total volume of gut con- 

 tents in all specimens of that species. This second 

 figure was calculated from estimates of the per- 

 cent each item taken by the species contributed to 

 the gut contents of each individual fish (0 to 

 100%). The food items are listed in order of a 

 ranking index, which is computed by multiplying 

 the ratio offish containing the item to the number 

 of fish sampled, by the mean percent that item 

 represented of the diet volume. Thus, for example, 

 for Holocentrus sammara (Table 10), the number 

 one prey, xanthid crabs, has a ranking index of 



916 



