by stream, were completed for the Kvichak sys- 

 tem (Demory, Orrell, and Heinle, 1964) and the 

 Wood system (Marriott, 1964). A catalog for 

 the Naknek system is being prepared. 



The amount of potential spawning grounds 

 for streams was estimated as follows: (1) The 

 stream was divided into sections from the mouth 

 to the upper limit of possible spawning, i.e., to 

 the source or to obstructions to the passage of 

 fish, e.g., falls and beaver dams. (2) The total 

 area of each section was determined. (3) The 

 percentage of each section not usable for spawn- 

 ing was estimated on the basis of bottom com- 

 position, gradient, and water velocity. Bottom 

 types defined as not usable included silt, bed- 

 rock, and large rock. Pools and torrential water 

 were considered unsuitable for spawning. (4) 

 The amount of potential spawning grounds in 

 each section was estimated by subtracting the 

 area unsuitable for spawning from the total 

 area. Spring-fed ponds were treated as streams. 

 Wood system. — The estimates of the amount 

 of potential spawning grounds in streams of the 

 Wood system (table 12) apply only to streams in 

 which spawning had been observed. 



Size of the gravel and depth of the water were 

 used in the lakes of the Wood system as criteria 

 in estimating the area of beach suitable for 

 spawning. The area of lake surface from the 

 shore to 10-m. depth contours (over bottom 

 judged by field observations to be suitable for 

 spawning) was measured on bathymetric maps. 

 Because other factors, such as upwelling ground 

 water, also appear to be important in beach 

 spawning, the estimates based only on size of 

 the gravel and depth of the water exceed the 

 amount of potential beach spawning grounds. 

 Kvichak system. — The estimates of the 

 amount of spawning grounds in the Kvichak 

 system are minimal. For stream spawning, es- 

 timates were made only for streams in which 

 spawning had been seen. For beach spawning 

 in Iliamna Lake, estimates were made only for 

 beaches on which a small escapement of sockeye 

 salmon had spawned (1961) . Beaches in the fol- 

 lowing areas were not examined: Flat Island, 

 Southeast Beaches, Tommy Beach, Lonesome 

 Bay, and Finger Beach. 



Observations in Iliamna Lake during the 

 spawning of a large escapement (1960) indi- 

 cated that beach spawning was much more ex- 

 tensive than was formerly thought, particularly 



around islands, where it extended to depths of 

 30 m. 



Naknek system. — Estimates of the amount of 

 potential sockeye salmon spawning grounds are 

 available for all streams in the Naknek system, 

 even those in which spawning has never been 

 seen. The reliability of the estimates varies from 

 stream to stream, and the estimates will be modi- 

 fied as more information is obtained. 



Sockeye salmon spawn on beaches in Gros- 

 venor Lake and in a lesser amount in other lakes 

 in the Naknek system, but the extent of these 

 spawning grounds has not been estimated. 



Karluk system. — Data on the amount of 

 spawning grounds in the Karluk system are 

 from Owen et al. (footnote 9) and are for 

 grounds on which sockeye salmon are known to 

 spawn. 



Estimates of Space Requirements of Spawning Females 



A spawning female sockeye salmon requires a 

 certain minimum area of gravel bed in which to 

 deposit her eggs. Eggs are buried in about five 

 pockets in an area of disturbed gravel known as 

 the redd. During construction of the redd and 

 spawning and postspawning activities, both the 

 female and male defend a certain territory sur- 

 rounding the redd. 



Wood and Kvichak systems. — Estimates of 

 the minimal space requirements for spawning 

 females in the Wood and Kvichak systems were 

 based on the egg-carrying capacity of the gravel 

 independent of observed spawning density. 



Three principal sources of loss of eggs were 

 considered: (1) increased retention of the eggs 

 by the female because of increased density of 

 spawners, (2) dislodgment of eggs from the 

 redds because of repeated excavations of the 

 .same gravel by subsequent females (superim- 

 position), and (3) increased egg mortality be- 

 cause of crowding of eggs in the gravel. Data 

 from the Wood and Kvichak systems indicate 

 that as spawning density increases, superimpo- 

 sition of redds limits survival of the eggs. We 

 estimated egg loss from superimposition to be 

 less than 10 percent with a spawning density of 

 one female per 2 m.- This density is low enough 

 to avoid significant egg loss by increased i-eten- 

 tion or crowding of eggs in the gravel. Calcula- 

 tion of the capacity of streams in the Wood and 

 Kvichak systems to support spawning sockeye 

 salmon is based, therefore, on a density of one 

 female per 2 m.- 



SOCKEYE SALMON IN MAJOR RIVER SYSTEMS IN SOUTHWESTERN ALASKA 



423 



