May 1980 to October 1981 at a rock reef at Arrow- 

 head Point in Stillwater Cove, Carmel Bay, CA (about 

 lat. 36°33.6'N, long. 121°56.3'W). Bottom depths at 

 the study site averaged about 12.5 m below MLLW 

 (mean low low water). The reef was roughly tri- 

 angular, bounded by sand flats which sloped into 

 deeper water on all but the nearshore side (Fig. 1). 

 The reef substrate was composed of both high and 

 low relief basaltic rock interspersed with areas of 

 coarse, granitic sand and a few patches of worm 

 tubes (Diopatra sp.). Stillwater Cove is protected 

 from the prevailing north and northwest swells and 

 winds. However, it is exposed to the most severe 

 southerly storms which commonly occur from Oc- 

 tober to February. 



The rock substrate was covered with an extensive 

 mat of coralline algae and sessile invertebrates. The 

 dominant algal form was the giant kelp, Macrocystis 

 pyrifera, which became very thick in the summer 

 months and represented a major structural compo- 

 nent of the reef. Except for corallines, understory 

 algae (Pterogophora californica, Cystoseira osmun- 

 dacea, and seasonally dense patches of Desmarestia 

 ligulata) were relatively sparse and patchy. Des- 

 marestia abundance increased in the fall as the sur- 

 face kelp canopy decreased. Drift algae of several 

 types entered the reef at times and often became 

 a major microhabitat. It usually occurred along the 

 sand/rock interface or in low patches in the interior 

 of the reef. 



Field Collections 



All fish used for stomach content analysis were 

 collected with almxlmxl.Sm opening-closing, 

 diver-held net. The net was constructed of 1/8-in 

 stretch-mesh nylon netting on a 1/4-in PVC frame 

 mouth. The mouth of the net was hinged in the mid- 

 dle with tygon tubing. This allowed easy operation 

 by a single diver in close spaces. Collected fish were 

 brought to the surface and the stomachs injected 

 with 10% Formalin^. The fish were then preserved 

 whole in 10% Formalin, then washed in freshwater, 

 and stored in 70% ethanol. 



The majority of fish collected for gut analysis were 

 taken from June to August of both 1980 and 1981 

 with some supplemental collections occurring in Oc- 

 tober and November of 1980. Collections were made 

 during both day and night. 



^Reference to trade names does not imply endorsement by the 

 National Marine Fisheries Service, NOAA. 



FISHERY BULLETIN: VOL. 83, NO. 4 



Laboratory Methods 



Each fish was blotted dry and weighed to the 

 nearest 0.01 g. The entire visceral mass was re- 

 moved, and the stomach and intestine were then 

 separated from the remaining viscera. The contents 

 of each stomach were emptied into a small dish and 

 examined under a binocular dissecting scopa Food 

 items were separated by general taxonomic category 

 (eg., calanoid copepods, zoea larvae, mysid shrimp, 

 etc). These general categories were used because not 

 all prey items found were in good enough condition 

 to identify to species, thus identifying some items 

 further than others could be misleading. Each 

 category was enumerated by number and percent 

 volume The percent of the total stomach volume of 

 each prey type was estimated by spreading the con- 

 tents to a uniform thickness over a background grid 

 and then estimating the area represented by each 

 type Since contents were spread to an equal thick- 

 ness, area and volume were considered directly pro- 

 portional. Digestive state of contents was estimated 

 on a subjective one to five scale (DeWitt and Cailliet 

 1972). In addition, a subset of each prey type was 

 taken from stomach contents and measured along 

 its longest axis to the nearest 0.005 mm using an 

 ocular micrometer for later estimation of mean sizes 

 of each prey type 



Analytical Methods 



A plot of cumulative number of prey types against 

 randomly pooled number of stomachs was con- 

 structed for each species in order to assess adequacy 

 of sample sizes. An asymptotic leveling of this type 

 of plot indicates a sufficient sample size The mean 

 percent by estimated volume (%V) and by number 

 (%N) of each prey type were calculated for each fish 

 species as an average of all values for individual 

 specimens. The mean frequency of occurence (%F0) 

 was also calculated for each prey type in each 

 predator as the number of times the prey type 

 was seen divided by the total of stomachs exam- 

 ined. The importance of each prey type was 

 calculated using the index of relative importance 

 (IRI) described by Pinkas et al. (1971). The IRI 

 used the proportion by amount (%N), volume 

 (%V), and occurrence (%F0) of each prey type 

 (IRI = (%N + %V) X %F0). The diets of the 

 seven species studied were compared using the 

 percent similarity index (PSI) (Whittaker 

 1952): PSI = 1 - 0.5 I pih - pjh. Where pi and 

 pj are the proportions by IRI of each prey type (h) 

 in the two predators being compared. In calculating 



532 



