Temporal Stability of Genetic Variation within ... - Wiley Online Library

Transactions of the American Fisheries Society 138:1052–1064, 2009 Ó Copyright by the American Fisheries Society 2009 DOI: 10.1577/T08-165.1

[Article]

Temporal Stability of Genetic Variation within Natural Populations of Summer Steelhead Receiving Mitigation Hatchery Fish SCOTT M. BLANKENSHIP*

AND

MAUREEN P. SMALL

Washington Department Fish and Wildlife, Genetics Section, 600 Capitol Way North, Olympia, Washington 98501-1091, USA

JOSEPH D. BUMGARNER Washington Department Fish and Wildlife, Snake River Laboratory, 401 South Cottonwood Street, Dayton, Washington 99328, USA Abstract.—Stochastic fluctuation in allele frequencies may reflect microevolutionary processes responsible for genetic change, such as small effective population size (Ne), which is a legitimate concern for imperiled populations affected by environmental or anthropogenic factors. In salmonids, recent empirical studies have provided conflicting results regarding the consistency of within-population genetic variation over time. In the present study, we surveyed the genetic variation at 14 microsatellite loci in two endemic populations of summer steelhead Oncorhynchus mykiss (Tucannon and Touchet) for seven consecutive years and the Lyons Ferry Hatchery stock (LFH) for 4 years. The LFH (mitigation) stock is used to enhance fishing opportunities in southeastern Washington State and not to aid in the recovery of natural populations. We observed statistically significant differences in allele frequencies between replicated collections from the same location. One of twenty-one genic tests for the Tucannon River collections, 9 of 21 genic tests for the Touchet River collections, and all genic tests regarding LFH collections were statistically significant. Temporal variation was larger than spatial variation (0.86% and 0.44%, respectively). We also used genetic data to infer effective population sizes for natural steelhead populations and the LFH mitigation hatchery stock, with point estimates of 729.7, 599.4, and 266.7 for the Tucannon, Touchet, and LFH populations, respectively. Despite the temporal variation observed, relative genetic differentiation was stable over time, replicated collections tending to cluster together in factorial correspondence analyses of allele frequency data. However, the genetic data were consistent with the potential for gene flow between the Tucannon River and LFH populations. The potential benefits of stocking genetically differentiated hatchery fish to enhance fishing opportunities in the Tucannon River may be offset by the negative effects of hatchery introgression and the small Ne in hatchery fish.

Temporal variation in the genetic composition of a population has long been of fundamental interest to evolutionary biologists, since changes in gene frequency over time are a signal of micro-evolutionary processes and may elucidate the agents responsible for genetic changes in populations (Lessios et al. 1994 and references therein). Although until recently there was a lack of empirical studies reporting genetic diversity estimates based on temporally replicated sampling, there is now an increasing trend in the literature of studies using samples collected over multiple years. This trend is being driven by two factors: (1) concern that biased estimates of population differentiation are being inferred from ‘‘snapshot’’ genetic heterogeneity studies, where samples are collected at single time-points (Waples 1998), and (2) interest in using temporal data to estimate the effective * Corresponding author: [email protected] Received August 11, 2008; accepted April 7, 2009 Published online July 30, 2009

population size (Ne) (Waples 1989), a key parameter in conservation and population biology (Hedgecock et al. 1992). In salmonids, recent empirical studies have provided conflicting results regarding the consistency of withinpopulation genetic variation over time, with both longterm temporal stability and temporal variability observed. Evidence exists to suggest that allele frequencies are stable over time, with reports concluding that the temporal variation within a population is minor compared with differences among populations (Tessier and Bernatchez 1999; Banks et al. 2000; Carlsson and Nilsson 2000; Hansen et al. 2002). Yet, several recent studies have reported inconsistencies in allele frequency estimates taken from a single population at multiple time periods, with some temporal variation high enough in magnitude to cause erroneous conclusions about population differentiation (Heath et al. 2002; Østergaard et al. 2003; Palm et al. 2003; Jensen et al. 2005). Analyzing collections from multiple generations to reliably estimate genetic diversities is especially

1052

TEMPORAL STABILITY OF GENETIC VARIATION IN STEELHEAD

important within a conservation setting, where critical population management decisions are made using population genetic information and population sizes are usually reduced. Populations with small Ne are more prone to temporal instabilities in gene frequencies and genetic erosion than are populations with large Ne (Frankham et al. 2002), since Ne is the main factor mediating any changes in neutral genetic diversity over time caused by genetic drift. Even though temporal variation in gene frequency may not have direct biological significance, stochastic fluctuations in allele frequencies may signify a small Ne, which is a legitimate concern for imperiled populations being affected by environmental or anthropogenic factors. There are two observations suggesting a potential for reduced Ne in endemic populations of summer steelhead Oncorhynchus mykiss in western North America. One factor is that census sizes are drastically reduced from historic levels for many steelhead populations (Busby et al. 1997), and Ne is thought to be between 0.10 and 0.33 of the estimated census size in salmon (Bartley et al. 1992; R. S. Waples, National Oceanic and Atmospheric Administration, personal communication). Another factor pertains to the common use of hatcheries to culture salmonids for a variety of purposes. Hatchery programs have the potential to alter the Ne of small populations by overrepresenting certain segments of the population in a subsequent generation, thereby stochastically altering the genetic constitution of the total population (Ryman and Laikre 1991; Busack et al. 1997). A static census size coupled with hatchery production has the potential to lower Ne, which then increases the influence of genetic drift and thus temporal fluctuations in allele frequencies. This temporal instability may undermine efforts to document the genetic characteristics of populations and lower the accuracy of inferred genetic relationships between populations. Moreover, the common assumptions in surveys of genetic diversity that allele frequency estimates are stable over time and do not require temporal study become dubious. The Ne depends on a variety of demographic factors and is a difficult quantity to estimate. For salmonids, which exhibit a life history strategy for differential age at maturity, each generation of juveniles is produced from multiple cohorts of adults from several previous years. If age data are available, then adult population samples can be sorted into their relevant cohorts, facilitating the inference of Ne (Waples 1990a). Here we describe a temporal analysis of allele frequencies at 14 microsatellite loci for three populations of summer steelhead, with sample collections replicated over a period of 7 years. We compared two,

1053

spatially separated, summer steelhead populations (Tucannon and Touchet rivers) with a single hatchery population (Lyons Ferry Hatchery [LFH]) that is used for harvest augmentation in both rivers. Specifically, the LFH stock is produced to enhance fishing opportunities by providing compensation for abundance losses arising from development of hydroelectric facilities in the lower Snake River (USACE 1975). A mitigation hatchery such as LFH considers the mitigation stock a different entity than populations present in the river, and steps have been taken over time to prevent interaction between the mitigation stock and naturally spawning fish (use of acclimation ponds or smolt-release locations are below prime natural spawning areas) (USACE 1975; WDFW 2005). While the magnitude of gene flow between the natural and hatchery steelhead stocks are unknown, Waples et al. (1993) reported that the Tucannon River and LFH summer steelhead were differentiated based on allozyme data, which is not surprising given the LFH stock was founded with steelhead genetically divergent from both Tucannon River and Touchet River steelhead. Hatcheries can affect natural populations by reducing overall genetic diversity (Ryman and Laikre 1991), reducing the fitness of the natural populations through relaxation of selection or inadvertent positive selection of traits advantageous in the hatchery (Lynch and O’Hely 2001; Ford 2002), and by reducing the reproductive success of natural populations (McLean et al. 2003). Our primary objective was to determine whether allele frequencies within, and genetic distances among, geographic samples were temporally consistent. These data will reliably determine population differentiation and allow subsequent investigation of interactions between mitigation hatchery stocking and endemic natural summer steelhead. A secondary objective is to determine the effective population sizes (Ne) relative to the estimated census sizes for natural populations and the mitigation hatchery stock. These objectives will provide a basis for investigating potential risks of stocking genetically differentiated hatchery fish to enhance harvest opportunity. Methods Tissue Collection and DNA Extraction Summer steelhead individuals produced by natural reproduction used in the temporal stability analysis were collected from 1998 to 2005 from two localities: Tucannon River, a tributary of the Snake River, and Touchet River, a tributary of the Walla Walla River (Figure 1). Steelhead adults from the Tucannon River (n ¼ 458) were collected at either the lower Tucannon River adult trap at river kilometer (rkm) 17.7 or from the Tucannon Fish Hatchery (TFH) adult trap at rkm 59

1054

BLANKENSHIP ET AL.

FIGURE 1.—Collection locations (Xs) for natural Tucannon River and Touchet River summer steelhead and Lyons Ferry Hatchery (LFH) stock. The abbreviation TFH stands for Tucannon Fish Hatchery.

(Figure 1). Steelhead adults from the Touchet River (n ¼ 508) were collected at the Dayton adult trap, located within the city of Dayton, Washington (rkm 87.4). The mitigation hatchery sample included in the temporal analysis was from LFH (Figure 1; see Table 1). The LFH stock was developed primarily from Wells Hatchery stock (upper Columbia River) and the

Wallowa River stock. The Wallowa stock was developed from trapping adult summer steelhead at the lower Snake River dams and reared at Wallowa Hatchery by the Oregon Department of Fish and Wildlife. Wells and Wallowa stock fish that returned to LFH during the 1980s and used for broodstock were eventually termed the LFH stock. In 1999 48 adults and

TABLE 1.—Description of steelhead collections used for genetic analysis and summary of within-collection genetic diversity results. Abbreviations are as follows: N is the number of sampled individuals, A and J refer to the adult and juvenile life stages, GD is unbiased gene diversity, HO is observed heterozygosity, FIS is the quantified deviation from random association of alleles within loci, and LD is the proportion of pairwise locus combinations showing significant (a ¼ 0.05) correlation of alleles across loci. Values in bold italics are statistically significant. Sample Tucannon River 1998–1999 2000 2001 2002 2003 2004 2005 Lyons Ferry Hatchery 1998–1999 1999 2003 2004 2005 Touchet River 1999 2000 2001 2002 2003 2004 2005

N

Life stage

GD

HO

Richness

FIS

LD

36 45 51 45 85 69 127

A A A A A A A

0.809 0.817 0.817 0.826 0.811 0.813 0.815

0.764 0.757 0.78 0.786 0.727 0.767 0.774

13.65 14.56 13.99 14.58 14.07 14.39 14.68

0.056 0.074 0.045 0.049 0.048 0.056 0.049

0.00 0.00 0.00 0.00 0.00 0.00 0.00

45 48 100 100 100

A J A A A

0.824 0.752 0.803 0.806 0.814

0.796 0.702 0.753 0.774 0.793

14.31 11.58 12.53 13.05 12.78

0.034 0.068 0.063 0.040 0.026

0.00 0.00 0.01 0.03 0.00

33 30 116 85 73 96 75

A A A A A A A

0.819 0.812 0.811 0.811 0.803 0.816 0.823

0.765 0.77 0.769 0.778 0.748 0.779 0.79

14.32 13.88 13.45 13.39 13.54 13.69 13.95

0.067 0.052 0.052 0.042 0.069 0.046 0.040

0.00 0.01 0.00 0.00 0.00 0.00 0.00


45 juveniles were collected, and 100 adults were collected each year from 2003 to 2005, for a total of 393 individuals. Only steelhead with the adipose fin absent were included as mitigation broodstock. Coded wire tag (CWT) recovery data showed that over 99% of these fish with adipose fins clipped (ad-clipped) were of LFH origin; therefore, any ad-clipped fish was assumed to be from LFH. Age and CWT data were used to reorganize population samples into cohorts for effective population size (Ne) analysis. Tissue collections were either made as fin clips or operculum punches and were stored immediately in ethanol after collection. The DNA was extracted from stored tissue using Nucleospin 96 Tissue following the manufacturer’s standard protocol (Macherey-Nagel, Easton, Pennsylvania). Laboratory Analysis Polymerase chain reaction (PCR) amplification was performed using 14 fluorescently end-labeled microsatellite marker loci: One101, -102, -108, and -114 (Olsen et al. 2000), Omy77 (Morris et al. 1996), Omy1001 and -1011 (Spies et al. 2005), Omm1070, -1128, and -1130 (Rexroad et al. 2001), Ots1 and Ots3M (Banks et al. 1999), Ots100 (Nelson and Beacham 1999), and Ots103 (Small et al. 1998). The PCR reaction volumes were 10 lL and contained 1 lL 103 PCR buffer (Promega), 1.0 lL MgCl2 (1.5 mM final) (Promega), 0.2 lL 10 mM dNTP mix (Promega), and 0.1 lL Taq DNA polymerase (Promega). Loci were amplified as part of multiplexed sets, so primer molarities and annealing temperatures varied. Multiplex 1 had an annealing temperature of 558C, and used 0.08 M One102, 0.07 M One114, and 0.04 M Ots100. Multiplex 2 had an annealing temperature of 628C and used 0.06 M Omm1130, 0.03 M Omm1070, and 0.04 M Omy1011. Multiplex 3 had an annealing temperature of 558C and used 0.04 M One108, 0.011 M Ots103, and 0.021 M One101. Multiplex 4 had an annealing temperature of 528C and used 0.03 M Omy1001 and 0.025 M Omm1128. Multiplex 5 had an annealing temperature of 498C and used 0.035 M Ots1, 0.03 M Omy77, and 0.02 M Ots3M. All thermal cycling was conducted on a PTC200 thermal cycler (MJ Research) as follows: 958C for 2 min, 30 cycles of 958C for 30 s each, 30 s annealing, 728C for 30 s, a final 728C extension, and then a 108C hold. The PCR products were visualized by denaturing polyacrylamide gel electrophoresis on an ABI 3730 automated capillary analyzer (Applied Biosystems). Fragment analysis was completed using GeneMapper 3.0 (Applied Biosystems). Genetic Data Analysis Assessing within-locus and within-population genetic diversity.—MICRO-CHECKER (Van Oosterhout et

1055

al. 2004) was used to determine whether excess homozygosity observed per locus was consistent with the presence of null alleles or genotyping errors (e.g., large-allele dropout). Genetic diversity was quantified using Nei’s (1987) unbiased gene diversity formula (i.e., expected heterozygosity) and Hedrick’s (1983) formula for observed heterozygosity. For each locus and collection GENEPOP 4.3 (Raymond and Rousset 1995) was used to assess Hardy–Weinberg equilibrium, where deviations from the neutral expectation of random associations among alleles (FIS) were calculated using an exact test. The rejection zone for the null hypothesis was generated using a Markov chain method (Raymond and Rousset 1995). If the exacttest null hypothesis (i.e., random association) was rejected, a second test was performed to explicitly test the alternative hypothesis (i.e., heterozygote excess or deficit) (Raymond and Rousset 1995). The significance level for all Hardy–Weinberg equilibrium test P-values was adjusted for multiple comparisons using the Bonferroni correction (a ¼ 0.05; Rice 1989). Correlations between alleles across loci, commonly termed linkage disequilibrium (or gametic-phase disequilibrium), was calculated following Weir (1979) using GENETIX version 4.05 (Belkhir et al. 2001). Statistical significance of linkage disequilibrium (LD) was assessed using a permutation procedure implemented in GENETIX for each locus by locus combination within each collection, where results are reported as the proportion of pairwise combinations significant at a ¼ 0.05 (P-values were adjusted using the Bonferroni correction). The models underlying the genetic tests employed in this study assume the genetic markers used are selectively neutral. This assumption is especially critical for the estimation of Ne (see the following section), where changes in genetic variance are attributed to the stochastic processes of genetic drift. To determine whether observed genetic variation at each microsatellite locus was consistent with neutrality, we used the neutrality test of Beaumont and Nichols (1996). This test compares the observed genetic differentiation (FST) of each locus given its allelic diversity (heterozygosity) to that expected under neutrality. The null hypothesis of neutrality is rejected if a locus has an unlikely FST value given the range of FST expected from neutrality simulations. An additional neutrality test of Kauer et al. (2003, equation 2) was performed for locus Omy1001, pairwise for Tucannon, Touchet, and LFH. Effective population size (Ne).—Estimates of the Ne were obtained using a multisample temporal method (Waples 1990a) on consecutive cohorts of steelhead. We used age from scale analysis and coded wire tag data to sort populations into cohort samples. Only

1056

BLANKENSHIP ET AL.

cohort samples with more than 20 individuals were used in the temporal method analysis. Comparing samples from years i and j, Waples’ (1990a) temporal method estimates the effective number of breeders (Nˆ b(i,j)) according to the equation Nˆ bði;jÞ ¼

b : ˆ ˜ 2ðF 1=Si;j þ 1=NÞ

ˆ is The standardized variance in allele frequency (F) calculated according to Pollack (1983). The parameter b is calculated analytically from age-structure information and the number of years between samples (Tajima 1992). The age-at-maturity information required to calculate b was obtained from the cohort data. ˆ One parameter is Two ‘‘corrections’’ are applied to F. ˜S , the harmonic mean of sample sizes from years i i,j and j. The Waples’ (1990a) multisample temporal method assumes juvenile sampling, so the additional parameter 1/N, where N is the number of spawners subject to ‘‘resampling process,’’ was a correction suggested by Waples’ (1990a) to account for the sampling of adults before spawning. We defined N as the mean census size over the cohort years analyzed. SALMONNb (Waples et al. 2007) was used to ˆ b, and S˜ , with the harmonic mean over calculate F, i,j all pairwise estimates of Nˆ b(i,j) being N˜ b. As suggested by Waples et al. (2007), alleles with a frequency below 0.05 were excluded from the analysis to reduce potential bias. Confidence intervals were calculated for N˜ b estimates according to Waples (1989) chi-square approximation. The lower bound used for N˜ b was the lower bound calculated for the smallest observed Nˆ b(i,j) within a population. For example, the smallest Nˆ b(i,j) observed for Tucannon River was between brood years 1996 and 2000. The lower bound calculated for this Nˆ b(i,j) was used to bound Tucannon River N˜ b. The upper bound used for N˜ b was the upper bound for the largest Nˆ b(i,j) observed within a population. To obtain an estimate of Ne, N˜ b must be multiplied by the mean generation length. Mean generation length (GL) was calculated for each population using age-composition information from cohort data. Among-population genetic differentiation.—The temporal stability of allele frequencies was assessed by the randomization chi-square test implemented in FSTAT version 2.9.3.2 (Goudet 1995). Alleles were randomized between samples (i.e., genic test) and the significance level was adjusted for multiple comparisons using the Bonferroni correction (a ¼ 0.05) (Rice 1989). Population structure among samples was investigated using a hierarchical analysis of molecular variance (AMOVA) that partitioned the total variance into covariance components due to intra-individual,

interindividual, intercollection, and intergroup differences (Weir and Cockerham 1984). The structure of the AMOVA was the same as the temporal analysis of allele frequencies, where Tucannon River, Touchet River, and LFH were defined as groups, and the temporally replicated samples (i.e., years) were defined as collections. The statistical significance of the covariance components was determined using a nonparametric permutation approach described in Excoffier et al. (1992). ARLEQUIN 3.01 (Excoffier et al. 2005) was used to conduct the AMOVA. Population differentiation was also investigated using pairwise estimates of FST and factorial correspondence analysis (FC) on allele frequencies. Multilocus estimates of pairwise FST, estimated by a ‘‘weighted’’ analysis of variance (ANOVA) (Weir and Cockerham 1984), were calculated using GENETIX version 4.05 (Belkhir et al. 2001). To determine whether the FST estimates were statistically different from panmixia, 1,000 permutations were implemented in GENETIX version 4.05 (Belkhir et al. 2001). Population differentiation was assessed further using FC on allele frequencies. In brief, genetic data are transformed into a contingency table, where each individual is described by its multilocus genotype (i.e., contingency table is individual’s X alleles). The relationship between any two individuals in n-dimensional space (n ¼ number of alleles) is represented by their v2 distance. Specifically, the plot represents the ordination of individuals along three orthogonal vectors that represent the three largest eigenvalues derived from the weighted contingency table. Additionally, instead of showing each individual, we showed the collection centroids, which are the ‘‘centers of mass’’ for each collection. Mahalanobis distances were calculated (over all individuals) using MATLAB R2007a (Mathworks 2006) to determine whether the relative genetic distances between collection centroids were statistically significant. Results Microsatellite Diversity within Loci and Populations Substantial genetic diversity was observed within collections (i.e., years), with unbiased heterozygosity estimates, over all loci ranging from a low of 0.752 (1999 LFH) to 0.826 (2002 Tucannon) (Table 1). Mean allele richness over all collections and loci was 13.50, with allele richness ranging from a low of 11.58 (1999 LFH juveniles) to a high of 14.68 (2005 Tucannon) (Table 1). The number of alleles sampled per locus was standardized to the smallest sample size of complete genotypes (N ¼ 28, 56 alleles) using a rarefaction method, although the mean sample size was much larger (N ¼ 73). Statistically significant deviations from

1057


TABLE 2.—Derivation of effective population sizes (Ne) for steelhead. Variable N˜ b is the harmonic mean of the pairwise estimates of the effective number of breeders, and Age is the age composition used to calculate the b parameter and the mean generation length (GL). Confidence intervals are shown in parentheses. See text for details. Population

N˜ b

Age

GL

Ne

Tucannon River Touchet River Lyons Ferry Hatchery

164.9 (34.7–‘) 129.6 (31.3–‘) 81.1 (45.2–187.8)

0.06, 0.48, 0.43, 0.03, 0.00 0.01, 0.45, 0.46, 0.07, 0.01 0.71, 0.28, 0.01, 0.00, 0.00

4.4 4.6 3.3

729.7 (153.5–‘) 599.4 (144.8–‘) 266.7 (148.6–617.5)

Hardy–Weinberg expectations were observed for most steelhead collections, with 5 of 7 Tucannon River, 4 of 5 Lyons Ferry Hatchery, and 6 of 7 Touchet River collections showing heterozygote deficits (FIS; Table 1). The FIS values ranged from a high of 7.4% deviation from expectation for the 2000 Tucannon River collection, to a low of 2.6% deviation for the 2005 LFH collection (Table 1). MICRO-CHECKER (Van Oosterhout et al. 2004) analysis did not find any consistency in the observed reduced heterozygosity that could be attributed to genotyping error or the

presence of null alleles (data not shown). The LD was minimal, with no steelhead collections showing statistically significant correlations between alleles across microsatellite loci. To test neutrality for each locus, the observed FST was plotted against heterozygosity. Values for all loci fell within the bounds expected under neutrality, given the Beaumont and Nichols (1996) simulation (data not shown). While the value for locus Omy1001 was consistent with neutral expectations over all populations, it was close to the Beaumont and Nichols (1996) upper bound (i.e., large

FIGURE 2.—Factorial correspondence plot of collection centroids for temporally stratified steelhead collections. The Lyons Ferry Hatchery, Tucannon River, and Touchet River collections are identified by squares, filled circles, and open circles, respectively. The numerical labels correspond to collection years. The 1998–1999 and 2001 Tucannon River collections are obscured by the 2002 collection. Collections that are not statistically different, as measured by Mahalanobis distances, are joined by horizontal lines (or a minimum convex polygon). Note that only the 1999 LFH samples were from juvenile fish.

1058

BLANKENSHIP ET AL.

TABLE 3.—Values for within-population pairwise tests for the Tucannon River, Touchet River, and Lyons Ferry Hatchery (LFH) collections. Above the diagonal are P-values for pairwise tests of allelic differentiation, below the diagonal pairwise estimates of FST. Values in bold italics are statistically significant. Tucannon Sample Tucannon 1998–1999 2000 2001 2002 2003 2004 2005 LFH 1998–1999 1999 2002 2004 2005 Touchet 1999 2000 2001 2002 2003 2004 2005

1998–1999

0.001 0.000 0.003 0.001 0.004 0.004

LFH

2000

2001

2002

2003

2004

2005

0.408

0.529 0.234

0.161 0.733 0.438

0.236 0.167 0.045 0.532

0.043 0.060 0.261 0.666 0.031

0.002 0.013 0.037 0.022 0.000 0.085

0.001 0.001 0.001 0.001 0.004

0.000 0.000 0.001 0.001

0.001 0.001 0.003

0.001 0.002

FST given observed heterozygosity). An additional neutrality test was performed for this locus. Following Kauer et al. (2003, equation 2), we calculated log10 (RH) pairwise for Tucannon, Touchet, and LFH samples. The only pairwise test that was statistically significant was Tucannon versus LFH stock, which had a Z value of 1.99. Effective Population Size Estimates of Ne derived from Waples (1990a, 1990b) are shown in Table 2. For the Tucannon River analysis, cohorts from 1996 to 2001 were used. The harmonic mean of all pairwise estimates of Nˆ b(i,j)(N˜ b) was 164.9. Following Waples (1989), the confidence interval (CI) for this Nb estimate derived from the underlying N˜ b was 34.7–‘. Waples (1990b) showed that the effective population size is approximately equal to GLN˜ b, where GL is the mean generation length (i.e., average age of parents). The mean GL for Tucannon River was 4.4, resulting in Ne ¼ 729.7 (Table 2). The same method was used to derive the CI bounding the Tucannon Ne estimate, 153.5–‘, where 4.4 was multiplied by 34.7 and infinity (‘) to calculate the lower and upper bound, respectively. For the Touchet River analysis, cohorts from 1996 to 2001 were used. The harmonic mean of all pairwise estimates of Nˆ b(i,j)(N˜ b) was 129.6. The CI bounding the Touchet estimate of N˜ b was 31.3–‘. Using a mean GL of 4.6, the Touchet River Ne was estimated as 599.4 (CI ¼ 144.8–‘). For LFH, cohorts

1998–1999

1999

2002

2004

2005

0.000

0.000 0.000

0.000 0.000 0.000

0.000 0.000 0.000 0.000

0.002

0.037 0.004 0.002 0.002

0.040 0.039 0.036

0.006 0.008

0.007

from 1999 to 2001 were used. The harmonic mean of all pairwise estimates of Nˆ b(i,j)(N˜ b) was 81.1 (CI ¼ 45.2–187.8). Using a mean GL of 3.3, the LFH Ne was estimated as 266.7 (CI ¼ 148.6–617.5). Among-Population Genetic Differentiation The P-values for the genic differentiation tests are shown in Table 3, with pairwise comparisons of allele frequencies conducted separately for temporally replicated collections. Based on the permutation procedure in FSTAT, the adjusted level of statistical significance (a ¼ 0.05) was 0.0001. For Tucannon River collections, 1 of 21 pairwise tests was statistically significant (Table 3). All LFH collections were differentiated (Table 3). Additionally, the 1998–1999 LFH collection was statistically equivalent to all the Tucannon River collections (data not shown). For the Touchet River collections, 9 of 21 pairwise tests were statistically significant (Table 3). For the AMOVA analysis, the proportion of variance among collections within groups was 0.86%, approximately twice as large as that between groups (0.44%) (Table 4). Correspondingly, the FST estimates for temporally replicated collections were 0.002 for Tucannon River, 0.006 for Touchet River (0.002 excluding the 2000 Touchet sample), and 0.015 for LFH (0.005 when 1999 LFH juvenile sample was excluded). Comparing locations, the FST estimates were 0.006 Tucannon River versus Touchet River (0.004 excluding the 2000 Touchet sample), 0.008


TABLE 3.—Extended.

Touchet Sample

1999

Tucannon 1998–1999 2000 2001 2002 2003 2004 2005 LFH 1998–1999 1999 2002 2004 2005 Touchet 1999 2000 2001 2002 2003 2004 2005

0.026 0.001 0.001 0.003 0.001 0.002

2000

2001

2002

2003

2004

2005

0.002

0.005 0.000

0.000 0.000 0.011

0.015 0.000 0.147 0.528

0.005 0.000 0.000 0.000 0.000

0.005 0.000 0.005 0.022 0.016 0.002

0.027 0.032 0.029 0.029 0.030

0.002 0.002 0.002 0.001

0 0.002 0.003

0.003 0.004

0.002

Tucannon River versus LFH (0.004 with 1999 LFH juvenile sample excluded), and 0.015 LFH versus Touchet River (0.009 with 1999 LFH juvenile and 2000 Touchet samples excluded). The global FST value was 0.013 (60.004). All FST estimates discussed above were significantly different from panmixia. The evaluation of genetic data using factorial correspondence analysis (FC) revealed three distinct clusters of steelhead collections with no temporal trend observed in genetic distances among collections (Figure 2). The Touchet River collection centroids (Figure 2, open circles) connected by a polygon were all undifferentiated from each other based on Mahalanobis distances, except the 2001 and 2005 centroids were statistically different from each other. The 1999 Touchet centroid was statistically different from the TABLE 4.—Global analysis of molecular variance results as weighted averages over loci. Groups are defined as the Tucannon River, Touchet River, and Lyons Ferry Hatchery populations of steelhead; collections are defined as the temporally replicated samples. Source of variation

Variance components

Percentage variation

Among groups Among collections within groups Among individuals within collections Within individuals

0.026 0.050 0.492 5.244

0.44% 0.86% 8.47% 90.23%

1059

other Touchet centroids and was located centrally between the Touchet and Tucannon river clusters. The 2000 Touchet centroid appeared to be an outlier. The Tucannon River collection centroids, including the 1998–1999 LFH centroid, connected by a polygon formed another genetically undifferentiated cluster based on Mahalanobis distances (Figure 2). Note that the 1998–1999 and 2001 Tucannon centroids are obscured behind the 2002 centroid in Figure 2. Additionally, the 2005 centroid was statistically different from the 1998–1999 LFH centroid, and the 2000, 2003, and 2004 Tucannon centroids. The 2003 LFH collection centroid was undifferentiated from the 2004 and 2005 LFH centroids based on Mahalanobis distances (Figure 2, black lines); however, the 2004 and 2005 LFH centroids were statistically different from each other. The 1998–1999 LFH centroid was statistically different from the other LFH centroids, as it was undifferentiated from the Tucannon River centroids. The 1999 LFH centroid appeared to be an outlier. Discussion Comparing replicated collections from the same location for seven consecutive years of two endemic summer steelhead populations and 4 years of an associated mitigation hatchery stock, we have observed temporal instability in allele frequencies. Statistically significant genic tests were observed for both natural steelhead populations, although we observed fewer significant tests for the Tucannon River population than for the Touchet River. While the allele frequency differences were slight, the observed differences between replicated LFH collections were larger than those observed for either natural population. Despite the temporal variation observed and the close genetic relationships, relative genetic differentiation appeared stable. In general, there was a greater genetic affinity among multiple collections from a single location, than among collections from the same year for different localities. The factorial correspondence analysis (FC) of allele frequency data showed LFH collections were genetically more similar to Tucannon River than to Touchet River, even though the hatchery stock was founded by steelhead genetically divergent from both Tucannon and Touchet natural steelhead. Additionally, the LFH stock is presently less differentiated from the Tucannon River stock than it is from a progenitor stock (Wallowa; S. M. Blankenship, unpublished data). These genetic distance data suggest that there has been gene flow between Tucannon River and LFH, and this gene flow was probably due to different release practices in the two watersheds, as described in Conservation Implications. Since hatcheries can alter

1060

BLANKENSHIP ET AL.

the fitness of natural populations in many ways (Lynch and O’Hely 2001; Ford 2002; McLean et al. 2003), the benefits of increasing fishing opportunities will need to be weighed against any negative effects on the recovery of the depressed natural Tucannon summer steelhead population. Distribution of Genetic Variation within Populations We concluded from our analyses of genetic data for microsatellite loci from steelhead population samples that the distribution of genetic variation was consistent with neutral expectations; therefore, the assumption of Waples (1990a) temporal method that genetic drift was the source of allele frequency variance was not violated. We did not observe heterozygote deficits consistent with null alleles or laboratory artifacts (e.g., large-allele dropout), and the minimal LD observed was not associated with certain locus combinations. Additionally, the magnitude of the heterozygote deficit we observed was similar to other steelhead studies reporting microsatellite data (Heath et al. 2002; Arden and Kapuscinski 2003; Narum et al. 2004). The deviations from Hardy–Weinberg equilibrium and positive FIS observed could have resulted from nonrandom mortality, gene flow from resident trout, or salmonid biology (i.e., differential age at maturity). We also compared the FST for each locus versus its heterozygosity (Beaumont and Nichols 1996), and found all loci were consistent with neutral expectations, although some observations suggested locus Omy1001 deviated from neutral expectations. Perhaps Omy1001 is in linkage disequilibrium with a region of the genome experiencing differential selection in Tucannon River and the LFH stock. We observed temporal instability in allele frequencies for replication collections from a single location. Studies that report temporal instability, which have analyzed consecutive years or samples collected a few years apart, attribute the observed instability to a small Ne. Jensen et al. (2005) observed temporal heterogeneity among bottlenecked population samples of brown trout Salmo trutta from the same location. Palm et al. (2003) analyzed multiple consecutive cohorts of brown trout and observed temporal heterogeneity among cohorts from a location, and their estimates of Ne were generally below 50. There are no comparable studies specifically for steelhead, although Heath et al. (2002) observed temporal heterogeneity among temporally replicated samples for steelhead from British Columbia. The AMOVA results are also relevant to the issue of temporal stability, as the amount of variance attributed to temporal differences among collections is twice that attributed to spatial differences (0.86% and 0.44%, respectively). This result is contrary to that of

Palm et al. (2003), who reported spatial variance greater than temporal variance, although the magnitude of FST was substantially larger in Palm et al. (2003) then in our study (0.144 and 0.013, respectively). In steelhead Heath et al. (2002) reported equal spatial and temporal components of variance, each accounting for approximately 2% of the total variance. Effective Population Size The observation that natural steelhead populations show greater temporal stability of allele frequencies than LFH suggests that a smaller Ne may exist for the hatchery. The point estimate of Ne calculated using genetic data from the LFH samples was lower than those for natural steelhead populations (Table 2). In contrast, when the best available census data are considered for natural and LFH populations, a comparable Ne is expected for natural steelhead and the LFH stock, as all populations are estimated to be of similar size. Yet, the Tucannon and Touchet river census values are probably underestimates for several reasons, so we are not confident in the available values. For example: (1) the stream sections surveyed to estimate abundance were different before and after 1998, (2) a census estimate was not available in 1997 for Tucannon River, and (3) the census size for Tucannon River in 1998 was extrapolated from trapping data, not the usual method used for spawningground surveys. For reference, the arithmetic mean census size for Tucannon (1996–2001) was N ¼ 402 (CI ¼ 123–680), Touchet (1996–2001) was N ¼ 344 (CI ¼ 144–544), and LFH (1999–2001) was N ¼ 415 (CI ¼ 361–468). Additionally, Tucannon, Touchet, and LFH combined was N ¼ 1,166 (CI ¼ 595–1,737). In general, these estimated census data would generate ratios of the Ne to census size consistent with published Ne /N ratio values. Because the natural population census data may be underestimates, the Ne /N ratios could be similar between the natural and LFH populations despite the LFH having a lower estimated Ne. Arden and Kapuscinki (2003) found that for 18 brood years of Snow Creek steelhead surveyed, the Ne : N ratio ranged from 0.41 to 0.67, and had an overall harmonic mean of 0.61. Heath et al. (2002) reported Ne estimates ranging from 0.09 to 0.29 for samples of steelhead from British Columbia. Gene flow can bias estimation of Ne, and factorial correspondence analysis of allele frequencies suggested gene flow was likely among the study populations. Additionally, Narum et al. (2004) suggested gene flow was occurring between resident (rainbow trout) and anadromous (steelhead) forms of O. mykiss in the Walla Walla River watershed, one region under study here. Given that temporal methods estimate Ne by


measuring the rate of allele frequency change, gene flow could cause a population to behave as if strong drift was changing allele frequencies very rapidly, resulting in an estimate of Ne that is too low (Wang and Whitlock 2003). Wang and Whitlock (2003) showed in simulations that the true Ne of a population (Ne ¼ 100) was underestimated slightly if that population was receiving substantial numbers of migrants from another population (migration rate, 10%) when the two populations are at migration–drift equilibrium and FST ’ 0.02. In this scenario, ignoring migration only causes limited bias, because migrants have a small effect on allele frequency changes since the two populations are similar genetically. While the Wang and Whitlock (2003) model has not been evaluated for salmonid life history, it does provide a means to evaluate possible bias. In our study, the approximate magnitude of FST observed was 0.01, which coincides with Nem ’ 25 (where m is migration rate). If we assume an Ne ¼ 1,000, the approximate Ne observed in this study for natural populations, m is estimated to be 2.5%. Given our observed values of FST and Ne, the assumption of no migration by Waples (1990a) temporal method probably resulted in limited bias of the reported Ne estimates. Another potential source of bias for Ne estimation could occur through gene flow between resident and anadromous O. mykiss, whereby the mean generation length (GL) is changed. Resident rainbow trout mature at 2–4 years of age, and adults are believed to suffer heavy mortality after first spawning (Behnke 1992). Therefore, if residents contribute to reproduction and are not accounted for, the age composition of the steelhead population would probably be biased upward. Waples (1990a) did state his model was robust to imperfect knowledge of age composition; however, GL remains an important parameter because it modulates the estimation of Ne (i.e., multiplier to Nb). While the reproductive contribution of residents in our study was unknown, we provide an example to illustrate potential bias. If a steelhead population under study had a GL ¼ 4.5, a 10% contribution from residents with a GL ¼ 3.0 would reduce the Ne estimated for the total population by 3%. We have not specifically investigated the possible causes of lower Ne observed for LFH; however, Ne is generally lower than N (census size) because of fluctuations in population size, unequal sex ratios, and variance in reproductive success (i.e., the number of offspring produced per individual). Given the best available census data, LFH is of similar size as the natural populations from year to year, so any reduction in Ne due to population size fluctuation may be similar for all populations. Yet, this LFH census does not

1061

include returning LFH adults not used as broodstock. An unequal sex ratio may not be an important factor regarding reductions of Ne, as the sex ratio for the LFH population is probably more consistent than those of natural populations (J. D. Bumgarner, unpublished data). For brood years between 1997 and 2005, the spawning population at LFH averaged (arithmetic mean) 182 females and 246 males, and the mating scheme within LFH was either one male per female or two males per female. In contrast, the number of steelhead males observed in the natural populations seems highly variable from year to year based on adult trapping data; however, resident male O. mykiss could have contributed to the gene pool (Narum et al. 2004). Variance in reproductive success is unknown, but is generally thought to be higher at spawning for a natural population than for those in hatcheries, as some wild fish do not reproduce. Therefore, the slightly lower Ne of LFH does not appear to be from changes in population size, unequal sex ratios, or variance in reproductive success. Our conjecture is that the lower Ne of LFH probably results from nonrandom postspawning mortality, such as family-correlated survival (Moyer et al. 2007). Genetic Differences among Populations Among-population genetic analyses provide mixed results, but the general conclusion is that while the differentiation among the Tucannon, Touchet, and LFH populations is slight, they appear genetically distinct, despite allele frequency instability, and the genetic relationships among locations remain consistent across the study period. Pairwise FST estimates comparing replicated steelhead collections within a location were smaller than estimates derived by comparing collections from different locations. When the temporally stratified samples were analyzed using FC analysis, replicated collections were generally not statistically different based on Mahalanobis distances (Figure 2). Additionally, all the Tucannon River centroids clustered together and all but two Touchet River centroids clustered together, so there was statistical support for the conclusion that the genetic relationships among populations was consistent from year to year. Yet, there were some observations that conflicted with the general conclusion. In Figure 2, the 1999 and 2000 Touchet River collection centroids did not cluster with the other Touchet centroids, and the 2000 collection appeared to be a genetic outlier. It is possible the small sample sizes for the 1999 and 2000 collections imprecisely estimated the actual allele frequencies for these Touchet River collections. Yet, evidence suggests sample size is not the only issue, as levels of allelic

1062

BLANKENSHIP ET AL.

diversity were not atypical for the 1999 and 2000 collections (Table 1). Since the sample sizes capture genetic information similar to the other collections, the genetic constitution appears slightly different between the early and late Touchet samples. The 1998–1999 LFH collection centroid was statistically different from the other LFH collections, and not statistically different from Tucannon collections (Figure 2). While this observation was not surprising given that the 1998– 1999 LFH collection was indistinguishable from the Tucannon River natural collections based on genic differentiation tests (data not shown), it is perplexing, as the 1998–1999 LFH adult sample consisted of marked hatchery fish collected at the TFH trap. To be clear, two types of steelhead could be captured at TFH during this time period: (1) steelhead with adipose fin removed and coded-wire-tagged (i.e., LFH mitigation stock), and (2) unmarked adults produced from natural spawning in the river. The ‘‘origin’’ of the 1998–1999 LFH collection would not have been questionable; therefore, the unmarked collections had to contain many steelhead with hatchery ancestry. This is a realistic explanation, as LFH fish are not barred from the Tucannon River and were the dominant returns in the parental generation of the Tucannon River collections in this study (J.D.B., unpublished data). Conservation Implications Since the Tucannon and Touchet river summer steelhead are populations of concern, there are conservation implications to the observations reported. Even though population differentiation was low, the Tucannon and Touchet river populations were significantly differentiated, and both groups were differentiated from LFH stock. This was the expected result, as the Tucannon and Touchet rivers are geographically separated watersheds. Additionally, the LFH mitigation stock was developed from genetically differentiated steelhead, primarily Wells Hatchery stock (upper Columbia River) and Wallowa River stock (Snake River composite), and management practices are supposed to limit the potential for gene flow between the mitigation stock and natural steelhead stocks (USACE 1975; WDFW 2005). Yet, natural Tucannon River collections were closely related to the LFH collections, and this was probably from 20 years of interbreeding between hatchery and natural fish and not from shared ancestry. Gene flow is clearly possible, as juvenile steelhead from LFH stock are released in both the Tucannon and Touchet rivers, and LFH adult returns are present in river. Historically, there has been more opportunity for gene flow in the Tucannon River, since hatchery juveniles were released between 1984 and 1997 in the same area as the prime spawning

habitat for natural Tucannon River steelhead and thus returned to the same spawning area used by natural steelhead (Bumgarner et al. 2003). Furthermore, hatchery-origin adults that are not brought into the hatchery for spawning are left in the Snake River system to increase sport-fishing opportunities within the Snake and Tucannon rivers. The reproductive success of LFH stock adults in the wild is unknown, but at the very least, successfully reproducing LFH adults would have offspring visually indistinguishable from endemic steelhead. If these individuals returned as adults to the Tucannon River, they could inadvertently be used as broodstock (i.e., intact adipose fin and no coded wire tag). In contrast to the Tucannon River, LFH stock juveniles are released below the prime spawning areas for natural steelhead in the Touchet River (Bumgarner et al. 2003). The greater potential for gene flow between Tucannon River and LFH than for Touchet River and LFH is consistent with the genetic distance results (Figure 2). Evidence for introgression between Tucannon River and LFH stocks is perhaps not surprising, given that in the Tucannon River the LFH stock steelhead are essentially allowed to traverse the Tucannon River up to the TFH where they are removed, a section of river that comprises much of the spawning area. In the Touchet River, hatchery fish tend to come back to an acclimation pond area, so there is less overlap with the majority of the natural spawning area. The use of a genetically divergent hatchery stock to enhance fishing opportunities in a river is not necessarily troubling if the mitigation stock does not interact with the naturalorigin fish in the area. This appears to be the case for Touchet River steelhead, which are reasonably distinct from the LFH stock. We conclude that the temporal instability in allele frequencies observed for Touchet collections was not from the interactions with hatchery fish, but more likely the result of interactions with residents, as suggested by Narum et al. (2004), or from salmonid biology (e.g., differential age at maturity). In contrast, data suggest gene flow was occurring between the natural Tucannon River steelhead and LFH mitigation stock. Yet, we observed fewer significant genic tests for the Tucannon collections than for the Touchet collections. It is unclear whether the temporal stability observed for Tucannon collections is the result of a sufficiently large Ne to resist stochastic changes or gene flow was somehow promoting allele frequency stability. The former seems more likely, as the substantial inclusion of genetically divergent hatchery fish within the Tucannon River should alter allele frequencies. Nevertheless, even low levels of gene flow could affect natural Tucannon River steelhead, since the LFH stock originated outside of the basin, have


temporally unstable allele frequencies along with smaller estimated Ne, have less age structure diversity, and hatcheries are known to affect the fitness of natural populations (Lynch and O’Hely 2001; Ford 2002; McLean et al. 2003). Any fishery enhancement benefits derived from the use of mitigation hatchery fish in the Tucannon River may be offset by the degradation of recovery efforts for natural summer steelhead. Acknowledgments We thank Cherril Bowman, Norm Switzler, and Judy Morgan for laboratory analysis. We also thank Denise Hawkins, Mark Schuck, Ken Warheit, Glen Mendel, and anonymous reviewers for providing helpful comments. The Lower Snake River Compensation Plan Program, the Bonneville Power Association, and the Washington State General Fund funded this project. References Arden, W. R., and A. R. Kapuscinski. 2003. Demographic and genetic estimates of effective population size (Ne) reveal genetic compensation in steelhead trout. Molecular Ecology 12:35–49. Banks, M. A., M. S. Blouin, B. A. Baldwin, V. K. Rashbrook, H. A. Fitzgerald, S. M. Blakenship, and D. Hedgecock. 1999. Isolation and inheritance of novel microsatellites in Chinook salmon (Oncorhynchus tshawytscha). Journal of Heredity 90:281–288. Banks, M. A., V. K. Rashbrook, M. J. Calavetta, C. A. Dean, and D. Hedgecock. 2000. Analysis of microsatellite DNA resolves genetic structure and diversity of chinook salmon (Oncorhynchus tshawytscha) in California’s Central Valley. Canadian Journal of Fisheries and Aquatic Sciences 57:915–927. Bartley, D., B. Bentley, J. Brodziak, R. Gomulkiewicz, and M. Mangol. 1992. Geographic variation in population genetic structure of Chinook salmon from California and Oregon. U.S. National Marine Fisheries Service Fishery Bulletin 90:77–100. Beaumont, M. A., and R. A. Nichols. 1996. Evaluating loci for use in the genetic analysis of population structure. Proceedings of the Royal Society of London B 263:1619–1626. Behnke, R. J. 1992. Native trout of western North America. American Fisheries Society, Monograph 6, Bethesda, Maryland. Belkhir, K., P. Borsa, and L. Chikhi et al. 2001. GENETIX 4.02: logiciel sous Windows TM pour la ge´ne´tique des populations. [GENETIX 4.02: Windows software for population genetics.] Laboratoire Ge´nome, Populations, Interactions, Universite´ de Montpellier II, Montpellier, France. Bumgarner, J., M. P. Small, L. Ross, and J. Dedloff. 2003. Lyons Ferry Complex hatchery evaluation: summer steelhead and trout report 2001 and 2002 run years. Washington Department Fish and Wildlife, Annual Report FPA3-15, Olympia. Busack, C., T. Pearsons, C. Knudsen, S. Phelps, B. Watson,

1063

and M. Johnston. 1997. Yakima Fisheries Project spring Chinook supplementation monitoring plan. Bonneville Power Administration, BPA Report DOE/BP-64878-1, Portland, Oregon. Busby, P. J., T. C. Wainwright, G. J. Bryant, L. J. Lierheimer, R. S. Waples, F. W. Waknitz, and I. V. Lagomarsino. 1997. Status review of West Coast steelhead from Washington, Idaho, Oregon, and California. U.S. Department of Commerce, NMFS-NWFSC-27, Springfield, Virginia. Carlsson, J., and J. Nilsson. 2000. Population genetic structure of brown trout (Salmo trutta L.) within a northern boreal forest stream. Hereditas 132:173–181. Excoffier, L., P. Smouse, and J. Quattro. 1992. Analysis of molecular variance inferred from metric distances among DNA haplotypes: application to human mitochondrial DNA restriction data. Genetics 131:479–491. Excoffier, L., G. Laval, and S. Schneider. 2005. Arlequin version 3.0: an integrated software package for population genetics data analysis. Evolutionary Bioinformatics Online 1:47–50. Ford, M. J. 2002. Selection in captivity during supportive breeding may reduce fitness in the wild. Conservation Biology 16(3):815–825. Frankham, R., J. D. Ballou, and D. A. Briscoe. 2002. Introduction to conservation genetics. Cambridge University Press, Cambridge, UK. Goudet, J. 1995. FSTAT (version 1.2): a computer program to calculate F-statistics. Journal of Heredity 86:485–486. Hansen, M. M., D. E. Ruzzante, E. Nielsen, D. Bekkevold, and K. D. Mensberg. 2002. Long-term effective population sizes, temporal stability of genetic composition, and potential for local adaptation in anadromous brown trout (Salmo trutta) populations. Molecular Ecology 11:2523–2535. Heath, D. D., C. Busch, J. Kelly, and D. Y. Atagi. 2002. Temporal change in genetic structure and effective population size in steelhead trout (Oncorhynchus mykiss). Molecular Ecology 11:197–214. Hedgecock, D., V. Chow, and R. S. Waples. 1992. Effective population numbers of shellfish broodstocks estimated from temporal variance in allelic frequencies. Aquaculture 108:215–232. Hedrick, P. W. 1983. Genetics of populations. Science Books International, Boston. Jensen, L. F., M. M. Hansen, and J. Carlsson et al. 2005. Spatial and temporal genetic differentiation and effective population size of brown trout (Salmo trutta, L.) in small Danish rivers. Conservation Genetics 6:615–621. Kauer, M., D. Dieringer, and C. Schlo¨tterer. 2003. Nonneutral admixture of immigrant genotypes in African Drosophila melanogaster populations from Zimbabwe. Molecular Biology and Evolution 20:1329–1337. Lessios, H. A., J. R. Weinberg, and V. R. Starczak. 1994. Temporal variation in populations of the marine isopod Excirolana: how stable are gene frequencies and morphology? Evolution 48:549–563. Lynch, M., and M. O’Hely. 2001. Captive breeding and the genetic fitness of natural populations. Conservation Genetics 2:363–378. Mathworks, 2006. MATLAB. Natick, Massascusetts. McLean, J. E., P. Bentzen, and T. P. Quinn. 2003. Differential

1064

BLANKENSHIP ET AL.

reproductive success of sympatric, naturally spawning hatchery and wild steelhead trout (Oncorhynchus mykiss) through the adult stage. Canadian Journal of Fisheries and Aquatic Sciences 60:433–440. Morris, D. B., K. R. Richard, and J. M. Wright. 1996. Microsatellites from rainbow trout (Oncorhynchus mykiss) and their use for genetic study of salmonids. Canadian Journal of Fisheries and Aquatic Sciences 53:120–126. Moyer, G. R., M. S. Blouin, and M. A. Banks. 2007. The influence of family-correlated survival on Nb /N for progeny from integrated multi- and single-generation hatchery stocks of coho salmon (Oncorhynchus kisutch). Canadian Journal of Fisheries and Aquatic Sciences 64:1258–1265. Narum, S. R., C. Contor, A. Talbot, and M. S. Powell. 2004. Genetic divergence of sympatric resident and anadromous forms of Oncorhynchus mykiss in the Walla Walla River, U.S.A. Journal of Fish Biology 65(2):471–488. Nei, M. 1987. Molecular evolutionary genetics. Columbia University Press, New York. Nelson, R. J., and T. D. Beacham. 1999. Isolation and crossspecies amplification of microsatellite loci useful for study of Pacific salmon. Animal Genetics 30:228–229. Olsen, J. B., S. L. Wilson, E. J. Kretschmer, K. C. Jones, and J. E. Seeb. 2000. Characterization of 14 tetranucleotide microsatellite loci derived from sockeye salmon. Molecular Ecology 9:2185–2187. Østergaard, S., M. M. Hansen, V. Loeschcke, and E. E. Neilsen. 2003. Long-term temporal changes of genetic composition in brown trout (Salmo trutta L.) populations inhabiting an unstable environment. Molecular Ecology 12:3123–3135. Palm, S., L. Laikre, P. E. Jorde, and N. Ryman. 2003. Effective population size and temporal genetic change in stream-resident brown trout (Salmo trutta L.). Conservation Genetics 4:249–264. Pollak, E. 1983. A new method for estimating the effective population size from allele frequency changes. Genetics 104:531–548. Raymond, M., and F. Rousset. 1995. GENEPOP (version 1.2): population genetics software for exact tests and ecumenicism. Journal of Heredity 86:248–249. Rexroad, C. E., R. L. Coleman, A. M. Martin, W. K. Hershberger, and J. Killefer. 2001. Thirty-five polymorphic microsatellite markers for rainbow trout (Oncorhynchus mykiss). Animal Genetics 32:317–319. Rice, W. R. 1989. Analysing tables of statistical tests. Evolution 43:223–225. Ryman, N., and L. Laikre. 1991. Effects of supportive breeding on the genetically effective population size. Conservation Biology 5:325–329. Small, M. P., T. D. Beacham, R. E. Withler, and R. J. Nelson. 1998. Discriminating coho salmon (Oncorhynchus kisutch) populations within the Fraser River, British Columbia. Molecular Ecology 7:141–155.

Spies, I. B., D. J. Brasier, P. T. O’Reilly, T. R. Seamons, and P. Bentzen. 2005. Development and characterization of novel tetra-, tri-, and dinucleotide microsatellite markers in rainbow trout (Oncorhynchus mykiss). Molecular Ecology 5:278–281. Tajima, F. 1992. Statistical method for estimating the effective population size in Pacific salmon. Journal of Heredity 83:309–311. Tessier, N., and L. Bernatchez. 1999. Stability of population structure and genetic diversity across generations assessed by microsatellites among sympatric populations of landlocked Atlantic salmon (Salmo salar L.). Molecular Ecology 8:169–179. USACE (U.S. Army Corps of Engineers). 1975. Special report: lower Snake River fish and wildlife compensation plan. USACE, Walla Walla, Washington. Van Oosterhout, C., W. F. Hutchinson, D. P. M. Wills, and P. Shipley. 2004. MICRO-CHECKER: software for identifying and correcting genotyping errors in microsatellite data. Molecular Ecology Notes 4:535–538. Wang, J., and M. C. Whitlock. 2003. Estimating effective population size and migration rates from genetic samples over space and time. Genetics 163:429–446. Waples, R. S. 1989. A generalized approach for estimating effective population size from temporal changes in allele frequency. Genetics 121:379–391. Waples, R. S. 1990a. Conservation genetics of Pacific salmon, III. Estimating effective population size. Journal of Heredity 81:277–289. Waples, R. S. 1990b. Conservation genetics of Pacific salmon, II. Effective population size and the rate of loss of genetic variability. Journal of Heredity 81:267–276. Waples, R. S. 1998. Separating the wheat from the chaff: patterns of genetic differentiation in high-gene-flow species. Heredity 89:438–450. Waples, R. S., O. W. Johnson, P. B. Aebersold, C. K. Shiflett, D. M. Van Doornik, D. J. Teel, and A. E. Cook. 1993. A genetic monitoring and evaluation program for supplemented populations of salmon and steelhead in the Snake River basin. Bonneville Power Administration, DEA179-89BP0091, Portland, Oregon. Waples, R. S., M. Masuda, and J. Pella. 2007. SALMONNb: a program for computing cohort-specific effective population sizes (Nb) in Pacific salmon and other semelparous species using the temporal method. Molecular Ecology Notes 7:21–24. WDFW (Washington Department of Fish and Wildlife). 2005. Snake River summer steelhead: Lyons Ferry stock hatchery and genetic management plan. WDFW, Olympia. Weir, B. S. 1979. Inferences about linkage disequilibrium. Biometrics 35:235–254. Weir, B. S., and C. C. Cockerham. 1984. Estimating Fstatistics for the analysis of population structure. Evolution 38:1358–1370.