BRAY and EBELING: THREE "PICKER-TYPE" FISHES 



any 24-h experimental trial, probably 

 approximated that encountered naturally by the 

 species in nearby shallow inshore waters. 



A grid of 10-cm squares was painted on the tank 

 bottom, so that the activity of a fish could be 

 measured as the number of grid lines it crossed 

 during 5-min sample periods, one for each of 

 twelve 2-h intervals (cf. Bortone 1971). On the 

 bottom of the tank were arranged small patches of 

 concrete blocks, rubble, freshly cut kelp, and sand 

 to simulate a natural reef area as much as possible. 

 After each fish had acclimated for a day, it was 

 observed from a platform directly overhead or 

 through the peep holes in the plastic shroud. 



Habitat 



Habitat occurrences of the three species were 

 determined from previous observations of 414, 

 2.5-min Super-8 mm underwater movie strips in 

 color (cinetransects) filmed by scuba divers. Cine- 

 transects were made between 1000 and 1500 h 

 during all seasons of 1969-70 and during the fall of 

 1971 and 1972 in a variety of kelp-bed habitats off 

 the Santa Barbara mainland and Santa Cruz 

 Island (Ebeling, Larson, and Alevizon in prep.). 

 Each cinetransect recorded most fish in a 5-m wide 

 path through or over a particular habitat type and 

 so was classified into scored categories of bottom 

 relief, kelp density, density of low algae on the 

 bottom, and position in the water column 

 ("canopy" or near the bottom). ("Kelp" herein 

 refers to the giant kelp Macrocystis; "algae" 

 refers to smaller plants other than giant kelp; and 

 "canopy" refers to the zone of the upper spreading 

 layer of the kelp bed— in our study, within 3 m of 

 the surface.) Fish were tabulated by species as the 

 film was projected in slow motion, reverse, and 

 stop action. Later the frequency of occurrence of 

 each species among all cinetransects was recorded 

 for every habitat parameter. 



Concordance, Breadth, and Overlap 



For each species, dietary arrays were compared 

 among four "seasons" that correspond roughly to 

 different hydrographic periods off Santa Barbara 

 (cf. Brown 1974): December-February, March- 

 May, June-August, and September-November. 

 Kendall's W coefficient of concordance (Tate and 

 Clelland 1957) was used to measure differences in 

 rank orders of volumes of food items pooled by 

 season. To estimate dietary variability among in- 



dividuals, concordance was also calculated for each 

 of five or six samples of four to nine fish that were 

 speared at the same general time and place. Thus 

 we assumed that food items must have been about 

 equally available to all the fish in a particular 

 sample. 



Breadth and overlap of resource use were com- 

 puted from values of p, , the proportion of item i in 

 the sample (fish), or pooled sample of fish (species). 

 For food breadth, p, is the proportionate volume of 

 any of the species total (S) of different food items; 

 for activity breadth, it is for any of six 2-h daytime 

 intervals, either the proportionate number of bites 

 per minutes by fish in the field or the proportionate 

 number of grid-line crossing per 5-min period by 

 fish in a tank; and for habitat breadth, it is the 

 proportionate occurrence of a species in any of 12 

 scored categories of habitat: 5 each of kelp density 

 and bottom relief and 2 of depth. Sample values 

 derive only from those resources or activities for 

 which an entire array of items are observable for 

 individual fish, i.e., from food and swimming ac- 

 tivity only. Only the habitat parameters that 

 seemed to be mutually independent were included 

 in the analysis. "Density of low algae on the bot- 

 tom," which was highly correlated with "bottom 

 relief," was excluded. Yet "kelp density" was 

 included. In the study area off Santa Barbara, kelp 

 density is more or less independent of bottom 

 relief because here kelp can anchor to flat as well 

 as high relief bottoms. Because cinetransects were 

 not evenly distributed among habitat 

 categories— most films were taken in relatively 

 dense kelp and high reef— the proportionate 

 frequencies were standardized by total films taken 

 per category. 



The sum of p^^ over 5 items in the array equals 

 the probability that any two units selected at ran- 

 dom will be of the same item (see Simpson 1949). 



s 

 Thus, the reciprocal, B = 1/2 pf, measures 



i = 1 



breadth or diversity (Levins 1968; MacArthur 

 1972). Or, B is the theoretical number of equally 

 used resources. For example, if all items are in 

 equal proportions, B equals the total items in the 

 array, S. It follows that the value of S determines 

 the maximum value of B. B was computed for fish 

 (individuals within a species) and for pooled sets of 

 fish (individuals pooled by species) as unsealed 

 values and as scaled values. The unsealed values 

 incorporate two contributions to breadth: that of 

 richness (S), and that of evenness of the distribu- 



817 



