FISHERY BULLETIN: VOL. 81, NO. 3 



merits of hermit crabs, gastropods, and sea stars pro- 

 duced distinct traces in surface sediments. Because 

 the bottom was undisturbed around the isolated pits 

 made in extracting the visually conspicuous prey, the 

 snout and vibrissae were probably unimportant in 

 locating these species. 



The feeding excavations of walrus clearly indicate 

 that clams are not excavated with the tusks. Fay 

 (1982) gave a convincing argument based on anat- 

 omy and tusk abrasion patterns that the tusks are not 

 used to excavate prey. Their main function apparent- 

 ly involves aggressive interactions, especially among 

 the males (Miller 1975). None of the furrows or pits 

 we discovered could be produced by plowing or dig- 

 ging with the tusks. As suggested by Vibe (1950) and 

 Fay (1982), most excavations probably involve "root- 

 ing" with the snout and vibrissae. According to Fay 3 , 

 snout widths of subadult and adult walrus range from 

 29 to 41 cm for males and 23 to 35 cm for females. 

 These sizes correspond exactly to the diameter of the 

 upper portion of Mya pits and the width of furrows if 

 the snout is swung in a narrow arc during excavation 

 (Table 2). 



We hypothesize that in addition to "rooting," a puls- 

 ing jet of water also was used to excavate prey. The 

 walrus' mouth and tongue are well adapted for suck- 

 ing and expelling water (Fay 1982) (a well-known fact 

 to visitors who are sprayed regularly at Sea World 

 Park in San Diego). Hydraulic jetting is the only fea- 

 sible mechanism for producing the deep (30 cm) 

 central shafts of Mya pits. These hydraulic pulses 

 also may be used to produce furrows and other pits, 

 probably in conjunction with snout and vibrissae 

 movements. This idea was tested by constructing a 

 suction-jet similar to the clam guns used to extract 

 bait from intertidal mudflats. By manipulating the 

 nozzle diameter and the volume of water exchanged 

 per stroke, divers have produced excavations similar 

 to the pits and furrows made by walrus. 4 A similar jet- 

 ting process was observed in bat rays by Gregory et 

 al. (1979), who suggested that it was used to excavate 

 infaunal prey. 



All clams were excavated prior to consumption. 

 Shells were found on the surface of the sediment in a 

 nonliving orientation. There was no evidence that 

 biting (Vibe 1950) or suction was used to remove the 

 soft parts of the clam while the shell was held in the 

 sediment. Soft parts were clearly consumed near the 

 sea floor, because discarded shells were closely as- 



3 F. H. Fay, Institute of Marine Science, University of Alaska, Fair- 

 banks, AK 99701, pers. commun. May 1982. 



4 Oliver, J. S., and E. F. O'Connor. Hydraulic excavation of bivalve 

 prey by walrus. Unpubl. manuscr. Moss Landing Marine Labora- 

 tories, Moss Landing, CA 95039. 



sociated with pits and furrows. The soft parts of 

 clams probably were sucked from between the two 

 shells (Vibe 1950; Fay 1982). 



Perhaps the most exciting potential of the benthic 

 feeding record is to quantify the activities of a single 

 dive. The continuous pit- furrow system we dis- 

 covered showed the location, excavation, and con- 

 sumption of 34 clams along >60 m of the bottom. 

 Over half of these clams (19) lived 30 cm deep in the 

 sediment. At this water depth average dive times are 

 about 5 min (Fay 1982), which suggests that one 

 walrus ate more than six clams per minute. Divers can 

 locate a number of long, continuous pit-furrow and 

 furrow systems where the species, size, and number 

 of prey can be measured. These may be the most ac- 

 curate records of the diving and foraging activities of 

 any marine mammal. 



Effects of Bottom Disturbance 



Walrus have an obvious impact on their large bi- 

 valve prey, but they also displace many of the small 

 and abundant infauna that are not walrus prey. All 

 the furrows and pits we observed were probably < 1 

 or 2 mo, and probably <1 mo old (see section on 

 Study Area). While there were dramatic differences 

 between the structures of nonprey communities at 

 the major feeding sites, the abundances of most small 

 infauna were significantly lower inside all of the re- 

 cent excavations (Figs. 6,7). The few exceptions were 

 either immobile species that were passively concen- 

 trated in the excavations (e.g., Myriochele oculata 

 and Rhizomogula sp.), or more motile species that 

 may be attracted to the excavations or to scavenging 

 events inside excavations (e.g., lysianassid amphi- 

 pods). Walrus disturbance clearly produces new 

 habitats, opens considerable space, and modifies 

 resources that influence subsequent patterns of 

 colonization. The tissue that remains attached to dis- 

 carded shells may be an important source of food for 

 several benthic scavengers, including asteroids, 

 ophuiroids, and crustaceans. 



Interactions Among Marine Mammals 



Walrus may interact trophically with a number of 

 other bottom-feeding marine mammals (Lowry et al. 

 1980; Lowry and Frost 1981). Gray whales and beard- 

 ed seals share the walrus' feeding grounds in the 

 Bering and Chukchi Seas, while the sea otter and 

 walrus overlap in the southeastern Bering Sea. Be- 

 cause these other large predators produce a benthic 

 feeding record that is distinct from the walrus, poten- 

 tial interactions can be examined by comparing 



510 



