FISHERY BULLETIN: VOL. 80, NO. 2 



low tides of the vernal equinox. In April 1981, A. 

 rhodorus eggs were found in the same area, but 

 about half were dead while the remainder all 

 hatched upon return to the laboratory. Perhaps 

 the protracted exposure to air in April killed 

 many of the earlier embryonic stages; an in- 

 crease in embryonic temperature tolerance with 

 development has been documented for another 

 intertidal cottid, Clinocottus acuticeps (Marliave 

 1981a). Considering a relatively dense spawning 

 of about one mass/3 m 2 found in March 1979, and 

 the lack of egg masses at other times, this species 

 might be characterized by a brief spawning sea- 

 son. 



Superficially, A. rhodorus eggs resembled 

 Hexagrammos spp. eggs in size (about 2.0 mm) 

 and color, although there were far fewer eggs 

 per mass. As with Hexagrammos spp., newly 

 spawned eggs were a semitranslucent blue to 

 purple, grading toward opaque white toward the 

 egg center (personal observation by J. B. Marli- 

 ave). Eggs with advanced embryos, showing 

 guanine eye pigment, appeared brown overall, 

 due to melanophores overlying dark olive yolk 

 material. All egg masses were incubated and 

 hatched in the laboratory; none were used for egg 

 counts but some egg diameter measurements 

 were taken. Hatching occurred on March 23 (1 

 mass) and 26 (1 mass) and on April 9 (3 masses) 

 and 17 (4 masses). This range of hatching dates 

 indicated a minimum spawning period of 25 d. 

 The collection and final hatch dates indicated a 

 minimum egg incubation period of 24 d at 10°C. 



During the planktonic larval stage, larvae of 

 A. rhodorus displayed relatively high-speed 

 schooling behavior and a marked tendency to- 

 ward startle responses. It is of note that both this 

 species and another common northeastern Pa- 

 cific Ocean fish, Trichodon trichodon (Marliave 

 1981b), are rare or unknown from plankton sam- 

 ples and school soon after hatching in the con- 

 fines of a tank. Unlike T. trichodon, however, A. 

 rhodorus larvae do not school immediately upon 

 hatching but develop schooling behavior within 

 2 wk of hatching. Ascelichthys rhodorus larvae do 

 not swim as fast as those of T. trichodon; A. rho- 

 dorus cruised at 2.5-7.5 body lengths/s at 2 wk of 

 age, at 3-10 body lengths/s at 4 wk, and at 2.5-9.0 

 body lengths/s at 6 wk, with usual speeds close to 

 5 body lengths/s. From hatching onward, the A. 

 rhodorus larvae were very easily disturbed, 

 either by physical interference from other types 

 of zooplankton, by movements of observers, or by 

 abrupt changes in lighting. Startle responses 



were characterized by rapid bursts of undirected 

 swimming which, in older larvae, effected the 

 breakdown of schools. 



After 2 wk, larval A. rhodorus schooled near 

 the surface at all ages except for those larvae in 

 tanks with larval shrimp, Pandalus danae. The 

 P. danae occupied the surface layers and A. rho- 

 dorus schooled off the tank bottom until the P. 

 danae settled from the plankton, after which A. 

 rhodorus schooled near the surface. This pattern 

 occurred successively in two separate tanks; no 

 shrimp were present in the third tank. Thus, the 

 vertical distribution of the A. rhodorus larvae 

 was modified by the presence of other planktonic 

 organisms. 



Settlement to the bottom started at 55-60 d of 

 age (14-18 mm SL) and schooling generally 

 ceased. However, for unknown reasons all larvae 

 in a tank would temporarily resume schooling 

 from time to time. Between 60 and 90 d, there 

 was a gradual increase in the proportion of set- 

 tled larvae with no observed difference in feed- 

 ing behavior between settled and schooling fish. 

 By 90 d of age, the majority of juveniles were per- 

 manently settled and no further schooling was 

 noted. Among cottids, protracted ambivalence 

 about settlement from the plankton has been 

 observed in Gilbertidia sigalutes (Marliave 

 1981c). 



After initial settling was observed, substrate 

 trays containing sand, gravel, and pebbles were 

 placed in the tanks to determine substrate pref- 

 erences of the larvae (Marliave 1977), but the 

 trays were avoided. Larval A. rhodorus never 

 settled against vertical surfaces, as is typical of a 

 variety of other cottids which lack discreet sub- 

 strate preferences (personal observation by J. B. 

 Marliave). Settlement was typically on open bot- 

 tom throughout the month of ambivalence be- 

 tween settlement and reentry to the plankton. 



ACKNOWLEDGMENTS 



Laboratory rearing was conducted at the Van- 

 couver Public Aquarium (Vancouver, B.C., Can- 

 ada). We thank the following at the Northwest 

 and Alaska Fisheries Center Seattle Laboratory 

 of the National Marine Fisheries Service: Bev- 

 erly Vinter who assisted in identifying and illus- 

 trating the larvae; Bernie Goiney for technical 

 assistance; Arthur Kendall for review of the 

 manuscript; and Jean Dunn for helpful discus- 

 sions on caudal fin development and review of 

 the manuscript. Kevin Howe, College of Fisher- 



354 



