Gall et al : Geographic variation in population genetics of Oncorhynchus tshawytscha 



85 



hatcheries. Although egg and fingerHng transfers be- 

 tween areas have been reduced recently, a considerable 

 amount of historical mixing of the hatchery stocks has 

 occurred (Alan Baracco, Calif. Dep. Fish Game, 

 Sacramento, CA 95616, pers. commun. Dec. 1986). Ad- 

 ditionally, many salmon from the San Joaquin River 

 stray into the Sacramento River on their spawning 

 migration due to easier access and better water qual- 

 ity in the Sacramento River (Alan Baracco and For- 

 rest Reynolds, Calif. Dep. Fish Game, Sacramento, CA 

 95616, pers. commun. Dec. 1986). 



Independent estimates of straying based on coded- 

 wire tagged fish indicate that chinook salmon in the 

 Sacramento River do stray within the system. Rough 

 estimates are that 2-5% of the Sacramento fall-run fish 

 are from hatcheries in the San Joaquin River system. 

 Approximately 1% of the fall-run chinook salmon 

 returning to the Feather River Hatchery is composed 

 of stray fish from the Nimbus (American River), Moku- 

 lumne, and Coleman Hatcheries. Straying also occurs 

 in northern streams because chinook salmon marked 

 on the Rogue River are recovered in the Klamath- 

 Trinity drainage (Fred Meyer, Calif. Dep. Fish Game, 

 Rancho Cordova, CA 95670, pers. commun. Feb. 1991). 

 Therefore, it is not surprising that gene flow esti- 

 mates for the Sacramento-San Joaquin drainage were 

 high and that southern coastal populations from 

 Oregon should resemble northern California coastal 

 populations. 



Stability of allele frequencies over time is often 

 assumed in the methodology of genetic stock identifica- 

 tion. Although the present study was not intended to 

 uncover temporal variation of allele frequencies, some 

 samples we examined also had been analyzed earlier. 

 Eighteen locations from the present study were sam- 

 pled in 1984-86 by Bartley and Gall (1990). For the 

 interstudy comparison, loci chosen had to have a fre- 

 quency of less than 0.95 for the common allele in at 

 least two populations reported by Bartley and Gall 

 (1990); isoloci were not used. Twelve loci fit the cri- 

 terion: AH-1, DPEP-1, PDPEP-2, TAPER, GPI-2, 

 IDDH-2, IDH-2, MPI, PGDH, PGK-2, PGM-2, and 

 SOD-1. 



We found 18 instances of significant change in allele 

 frequencies for seven hatchery samples (21.4%), 16 

 significant results for seven wild populations (19.0%), 

 and five instances of significant change for a pond rear- 

 ing project (41.7%) based on the G-statistic (Sokal and 

 Rohlf 1981). Interstudy comparisons of the samples 

 from Bogus Creek ( = Bogas Creek in Bartley and Gall 

 1990), Shasta Creek, and the Feather River Fish 

 Hatchery revealed no significant differences in allele 

 frequencies. 



Six hatcheries sampled in the present study also had 

 been sampled by Utter et al. (1989). Loci selected to 



compare allele frequencies for these studies had to have 

 a common allele frequency of less than 0.95 in one of 

 the studies. Eight loci met the frequency criterion: 

 AH, DPEP-1, TAPER, GPI-2, GR, MPI, PGK-2, and 

 SOD-1. Five of the six hatchery samples displayed 

 significant changes in allele frequency between the two 

 studies. Waples and Teel (1990) also reported signifi- 

 cant changes in allele frequencies in hatcheries sam- 

 pled in different years. 



Although we observed differences in allele frequen- 

 cies between this and earlier studies, we do not know 

 if this represents temporal variation. It is tempting to 

 make statements on the temporal stability or instability 

 of allele frequencies in samples of chinook salmon from 

 a given area, but without estimates of sampling vari- 

 ability for a given year, it is not possible to separate 

 intrasample variation, random sampling error, and 

 temporal variation. Nevertheless, given the presumed 

 constancy of allele frequency data (Allendorf and Utter 

 1979), the number of significant G statistics uncovered 

 in comparisons between samples in this study and those 

 of Utter et al. (1989) and Bartley and Gall (1990) re- 

 quires some explanation. 



Waples and Teel (1990:149) stated, "tests of the 

 equality of allele frequencies in temporally spaced 

 samples must be interpreted with caution." In addition, 

 Waples and Teel (1990) list inaccurate or artifactual 

 genetic data, nonrandom sampling of fish for genetic 

 analysis, selection, and migration as possible causes of 

 significant change in allele frequencies. For example, 

 large differences in allele frequencies at IDH-3 and 

 IDH-4 between the present study and Bartley and Gall 

 (1990) may be due to banding artifacts associated with 

 tissue breakdown. One of us (Bentley) has observed the 

 increased appearance of variant "alleles" at these loci 

 in samples that were not properly frozen and stored. 

 Therefore, the data for these two loci presented in 

 Bartley and Gall (1990) may be artifactual. In addition, 

 the analyses of Utter et al. (1989), Bartley and Gall 

 (1990), and the present study were done by different 

 personnel in different laboratories. Although standar- 

 dization was attempted, scoring of gel banding patterns 

 may have been inconsistent. 



The level of temporal instability of allele frequencies 

 is an important issue in the use of GSI to manage and 

 conserve chinook salmon populations (Waples 1990, 

 Waples and Teel 1990). However, sampling design 

 should specifically address this question before one 

 draws conclusions concerning wild or hatchery popula- 

 tions. Although we documented differences in allele fre- 

 quencies between this and earlier studies, the overall 

 association between genetic similarity and geographic 

 location remains constant for populations of chinook 

 salmon in California and Oregon. 



