The Journal of Heredity 2001:92(2)
© 2001 The American Genetic Association 92:127-136
Kinship Analysis of Pacific Salmon: Insights Into Mating, Homing, and Timing of Reproduction
From the Marine Molecular Biotechnology Laboratory (Bentzen, McLean, and Seamons), School of Aquatic and Fishery Sciences (Quinn), University of Washington, Box 355020 Boat St., Seattle, WA 98105, and Gene Conservation Laboratory, Alaska Department of Fish and Game, Anchorage, Alaska (Olsen).
Address correspondence to Paul Bentzen at the address above or e-mail: pbentzen{at}u.washington.edu.
 |
Abstract |
|---|
 Top Abstract IntroductionMaterials and MethodsResultsDiscussionReferences |
|---|
Multilocus microsatellite genotypes were used to infer kinship and relatedness in two species of Pacific salmon from three populations in Washington State. Even in the absence of direct genetic data from parents, clustering of individuals according to allele sharing and reconstruction of parental genotypes allowed resolution of full- and half-sib relationships among 135 chinook salmon (Oncorhynchus tshawytscha) sampled as preemergent juveniles from 14 redds in the Dungeness River. Inferred reproductive behaviors included single-pair matings, polyandry in which females mated with two to three males at a single redd, polygyny in which males mated with two females at different redds, use of two redds by a single female, and use of one redd site by two females. Greater average relatedness (rxy) in the upper reach of the Dungeness River implied within-reach homing of returning adults. In steelhead trout (O. mykiss), the frequency of related pairs (dyads) of mature individuals that migrated up Snow Creek less than a week apart was greater than expected for randomly chosen dyads, as was the frequency of steelhead dyads that were spawned on the same day in the Forks Creek hatchery. These results imply a heritable basis for upstream migration date and maturation date in steelhead trout.
|
Introduction |
|---|
|
TopAbstractIntroductionMaterials and MethodsResultsDiscussionReferences |
|---|
Salmon, trout, and char (Salmonidae) are among the most intensively studied fishes because of their great importance in commercial, recreational, and subsistence fisheries, the diversity of habitats they occupy, and their ecological roles (National Research Council 1996). Reproductive behavior is easily observed in most salmonids and has been studied in great detail, revealing some interspecific variation but broadly similar patterns among all species (Fleming 1998; Quinn 1999 and references therein). Females select and prepare nest sites, competing for the best locations. Males provide no parental care but compete for access to ripe females. At the moment of egg deposition, subdominant males may attempt to fertilize some of the female's eggs. She then buries the eggs, excavates another egg pocket, and spawns with the same or other males until all of her eggs are deposited. In semelparous species, the female defends the nest site from encroachment by other females until she dies, but in iteroparous species, the female generally abandons the redd shortly after spawning. These discrete sex roles, alternative male reproductive tactics, and high levels of reproductive competition make salmonids ideal subjects for research into breeding success.
Anadromous salmon are also unique among vertebrates in that substantial proportions of most species are now bred in hatcheries, then released as juveniles to mature and intermix in the ocean with naturally reproduced fish. Hatchery reproduction inevitably entails mating systems and early life histories that differ from those in the wild, leading to questions about the reproductive and fitness outcomes of interactions between wild and hatchery-bred fish on the spawning grounds (see reviews by Utter 1998; Waples 1991). Given these factors, as well as the economic importance of salmonids, it is hardly surprising that salmonids have been the subject of many studies of genetic parentage and kinship.
Most such studies of salmonid fishes have examined the outcome of alternative reproductive tactics, especially the success of subdominant males, or males adopting alternative life-history pathways such as nonanadromy or precocious maturation (e.g., Chebanov et al. 1983; Garant et al. 2001; Garcia-Vazquez et al. 2001; Hutchings and Myers 1988; Maekawa and Onozato 1986; Moran et al. 1996; Martinez et al. 2000; Schroder 1981; Thomaz et al. 1997). Others have evaluated hybridization (Garcia-Vazquez et al. 2001 and references therein), the relative reproductive success or fitness of hatchery-produced and wild fish (e.g., Leider et al. 1990; Moran et al. 1994; Thompson et al. 1998), the performance of fish in artificial culture (Herbinger et al. 1995, 1999), the extent of outbreeding depression (Gharrett and Smoker 1991, 1993; Gharrett et al. 1999), or genetic influences on the timing of spawning migration (Gharrett and Smoker 1993; Quinn et al. 2000).
Some of the above studies have traced parentage to individuals, whereas others have assigned parentage to alternative groups of individuals, but all have entailed some knowledge of the parental genotypes. However, in many cases it would be useful to identify levels of kinship among progeny in the absence of genotypic information from the parental generation. Such studies would permit insights into many aspects of reproduction and behavior in cases where it is impractical to sample adults. For example, kinship analyses could be used to examine the relative incidence of monogamous, polygynous, and polyandrous mating, variance in family size and reproductive success, homing behavior in adults, dispersal behavior in juveniles, and the heritability of behavioral or phenotypic traits (Mousseau et al. 1998).
Here we report three case studies in which we assessed kinship or relatedness in salmonid fishes in the absence of genotypic data from parents. We had two broad objectives. The first objective was to evaluate our ability to resolve three levels of kinship—full-sib, half-sib, and "unrelated" (non-first- or second-order relatives) in a natural population, using microsatellite data. The second objective was to use microsatellite-based estimates of relatedness to test whether variation in particular behavioral and phenotypic traits is under some degree of genetic control.
To test our ability to resolve different levels of kinship, we studied chinook salmon (Oncorhynchus tshawytscha) from the Dungeness River in northwestern Washington State. These fish were originally sampled from redds (nests) as preemergent juveniles to form a broodstock for a captive breeding effort aimed at rebuilding the depleted population (Smith and Wampler 1995). In the fish culture facility, fish were individually marked according to the redd where they had been captured. Knowledge of the source redd for each genotyped fish enabled us to ground-truth our ability to resolve kinship in naturally produced families. We also used the kinship and relatedness estimates to make inferences about fine-scale homing behavior and population substructure. To test whether microsatellite-based estimates of relatedness (rxy) were correlated with two traits associated with reproduction—timing of entry into the natal river during spawning migration, and date of maturation—we studied steelhead trout [the anadromous form of rainbow trout (O. mykiss)] from two populations in Washington State.
We used three approaches based on multilocus microsatellite genotypes to estimate kinship or relatedness: individuals were grouped according to their allele sharing coefficient (DPS; Bowcock et al. 1994) using UPGMA clustering (Blouin et al. 1996); parental genotypes were manually reconstructed based on the genotypes of putative full or half sibs (Banks et al. 2000); and relatedness coefficients (rxy) and simulations of rxy for different levels of kinship were calculated using the program Kinship 1.2 (Goodnight 2000). In addition to our primary objectives, we consider the relative merits of these approaches.
|
Materials and Methods |
|---|
|
TopAbstractIntroductionMaterials and MethodsResultsDiscussionReferences |
|---|
Fin clips were taken from 147 chinook salmon in the Hurd Creek Hatchery where they were being held for captive breeding. These fish had been captured as preemergent fry from 14 redds in the Dungeness River in February 1993 (Smith and Wampler 1995), and were the progeny of fish that spawned in August or September 1992 (Figure 1). The source redds are named according to their position in river miles from the mouth of the river; redds less than 60 m apart are distinguished alphabetically. For analysis purposes, the 14 redds were split into three groups (lower, middle, and upper river reaches; Smith and Wampler 1995).
|
Fin clips from 53 Snow Creek steelhead were collected over a 10-week period (February–April) in 1998 at a weir on Snow Creek during the upstream spawning migration (Figure 1). Steelhead trout were sampled at four weekly intervals in December 1999 and January 2000 at the Forks Creek hatchery (Figure 1). Fin clips were taken from 52 fish (28 females and 24 males) on the day that they were spawned at the hatchery.
Fin clips were preserved in ethanol at ambient temperatures. DNA was isolated from chinook fin clips using a modified Gentra Systems (Minneapolis, MN) Puregene DNA isolation kit, and from steelhead fin clips using a CTAB protocol (Fields et al. 1989).
Fourteen microsatellite loci were scored in chinook salmon (Table 1). Polymerase chain reactions (PCRs) were carried out in 10 µl volumes [10 mM Tris-HCl (pH 8.3), 50 mM KCl, 1.5 mM MgCl2, 0.2 mM each dNTP, 0.5 U Taq DNA polymerase (Promega, Madison, WI), 0.05–0.35 µM each primer, and 100 ng DNA template] using a Perkin Elmer model 9600 thermocycler. DNA was amplified using the following conditions: (1) a five cycle "tuchdown" profile (described below); (2) seven cycles of 94°C (1 min) + x°C (30 s) + 72°C (15 s); (3) 17 cycles of 94°C (30 s) + x°C (30 s) + 72°C (15 s); (4) one cycle 72°C (30 min), where x was an annealing temperature that varied among primer pairs (Table 1). The touchdown phase of the PCR began with an annealing temperature of x + 5°C and decreased in 1°C increments; temperatures and times were otherwise as indicated for the second phase of the PCR. Loci were amplified in four multiplex groups as indicated in Table 1 and following methods described by Olsen et al. (1996).
|
Twelve microsatellite loci were scored in steelhead (Table 2). PCRs were carried out in 10 µl volumes [10 mM Tris-HCl (pH 8.3), 50 mM KCl, 1–2 mM MgCl2 (see Table 2), 0.25 mM each dNTP, 1 U Taq DNA polymerase (Promega), 0.5 µM each primer, and 100 ng DNA template] using an MJ Research PTC-200 thermocycler. Microsatellites were amplified using the following conditions: (1) three cycles of 95°C (1 min) ;pl x°C (30 s) + 70°C (1 min); (2) 22 cycles of 95°C (10 s) + x°C (30 s) + 70°C (1 min); (3) one cycle 70°C (30 min), where x was an annealing temperature that varied among primer pairs (Table 2).
|
Microsatellites were size fractionated using an Applied Biosystems Inc. (ABI) 373A automated DNA sequencer. Electrophoretic data were analyzed using GeneScan 672 version 1.1 and Genotyper software version 2.0 (ABI). Allele sharing coefficients (DPS) were calculated with the aid of the Microsat program (Minch et al. 1995). An UPGMA tree based on -ln DPS for the 147 chinook was prepared using the Neighbor program in PHYLIP version 3.5c (Felsenstein 1993), and TREEVIEW (Page 1996). Relatedness coefficients (rxy; Queller and Goodnight 1989) were calculated using the program Kinship 1.2 (Goodnight 2000). For chinook salmon, allele frequencies needed to calculate rxy were estimated from the reconstructed genotypes of the parents of the fish sampled from redds (see below). For steelhead trout, rxy was calculated using allele frequencies estimated from the fish genotyped for this study. Simulations of rxy for 5000 dyads for each of three levels of kinship (full sib, half sib, and unrelated) were also carried out using Kinship 1.2.
Estimates of misclassification rates for assigning individuals to different levels of kinship on the basis of rxy were made as described in Blouin et al. (1996). A cutoff value of rxy for assigning individuals to either of two levels of kinship was set at the average of the distributions of simulated rxy for the two levels of kinship. For example, the cutoff value of rxy for discriminating between full sibs and half sibs was set at the mean of the average simulated values of rxy for full sibs and half sibs.
Parental genotypes were reconstructed for each chinook salmon redd as follows: For full-sib families, two parental genotypes that were compatible with all progeny genotypes were constructed for each locus. In a few cases, either of two alleles was possible for the "second" allele of a single-locus parent genotype, in which case one of the two alleles was chosen arbitrarily for the parental genotype. In a few other instances, offspring genotypes in a putative sib group were compatible with the possibility that the parental genotype was either homozygous, or heterozygous with an allele not seen in the progeny. In most such cases, however, the number of offspring was large enough to make it unlikely that a parental allele went undetected (see Discussion); therefore it was assumed that the parental genotype was homozygous.
Genotypes were assigned on an arbitrary basis to either the "male" or "female" parent, except in cases where half- sib relationships provided additional information. For redds harboring representatives of two (or in one case, three) full-sib families related at the half-sib level, it was assumed that the genotype of the shared parent was that of the female. The analysis was begun by assuming that each terminal cluster on the allele sharing dendrogram represented a single full-sib family, and additional paternal genotypes were only proposed when at least one offspring in the cluster was incompatible with the first paternal genotype at one or more loci. In constructing parental genotypes, the possibility that progeny from different redds might be related at the half-sib level was also considered. In cases where position on the dendrogram or relatedness coefficients suggested the possibility that two redds might share a common parent, an attempt was made to create a single multilocus parental genotype that was compatible with both redds. With one exception, it was assumed that the parent shared between redds was male. The exception involved redds 4.2a and 4.2b, where evidence of a complex half-sib relationship with redd 6.2 forced the conclusion that one female deposited eggs in two redds.
|
Results |
|---|
|
TopAbstractIntroductionMaterials and MethodsResultsDiscussionReferences |
|---|
Dungeness River Chinook
The 14 microsatellites exhibited 5–19 alleles per locus (mean 9.9) in 147 chinook salmon originating from 14 redds (Table 1). Expected single-locus heterozygosities estimated for the parents of these fish were 0.54–0.95 (mean 0.79). An UPGMA tree based on –ln DPS for 147 chinook salmon resolved a number of clusters that with few exceptions corresponded to the redds from which the fish were captured (Figure 2). One exception involved individuals from redds 17.6a and 17.6b; these fish formed a single intermingled cluster rather than two discrete groups. Other exceptions were 13 fish that did not cluster according to their putative redds. Only one of these individuals (10.4F395) had a multilocus genotype consistent with either a full- or half-sib relationship with other members of its putative redd. Although this fish clustered with those from redd 15.2, its multilocus genotype was consistent with either a full-sib or half-sib relationship to all other individuals from its designated source (redd 10.4), and was inconsistent with a full- or half-sib relationship to other fish from redd 15.2.
|
The remaining 12 "aberrant" individuals displayed genotypes that were not consistent with either full- or half-sib relationships with other members from their putative redds. Repeated DNA extractions and genotyping analyses produced the same results, ruling out laboratory error. Ten of these fish clustered inside other redd groups (Figure 2) and had multilocus genotypes that were consistent with full- sib relationships with other members of their respective redds. The two remaining individuals, 15.7F285 and 10.9M256, clustered separately from all the redd groupings (Figure 2). Overall we conclude that the 12 "aberrant" fish were probably misidentified in the fish culture facility, and that 10 of them came from families represented in our collection. As a conservative measure, we eliminated these 12 individuals from further analyses.
Manual reconstruction of the parental genotypes for the 14 redds allowed us to infer the pattern of mating that produced the various families (Table 3). Ten redds were represented in our samples by single full-sib families, indicating a single mother and father per redd. Two redds (9 and 17.6b) harbored two full-sib families related at the half-sib level, and a third redd (10.4) contained representatives of three full-sib families related at the half-sib level. The progeny from each of these redds presumably resulted from a single female who had mated with two or three males. One redd (15.2) included two full-sib families that were not half sibs, indicating that two females used this redd and that each mated with a different male.
|
Five pairs of redds contained progeny related at the half-sib level. Redds 4.2a and 4.2b were half sibs, as were redds 4.2b and 6.2; however, redds 4.2a and 6.2 were unrelated. Hence redds 4.2a and 4.2b shared a parent of one sex, and 4.2b and 6.2 shared a parent of the opposite sex. On the assumption that a female is more likely to construct two redds close together rather than 3 km apart, but that a male might move that far to seek mates, we assigned redds 4.2a and 4.2b a common mother, and assigned a common father to redds 4.2b and 6.2. However, it could have been the other way around. Three other pairs of redds (9.4/10.9, 15.7/15.9, and 17.6a/17.6b) each shared a common (presumably male) parent. Redds that shared common parents were approximately 0.1– 3.2 km apart, indicating that some parents moved substantial distances between redds; however, we detected no evidence of parents shared among redds in different river reaches.
Most of the half-sib relationships within and among redds were evident in the topology of the UPGMA tree (Figure 2). For instance, the clusters corresponding to redds 9 and 10.4 each included two distinct subclusters corresponding to hypothesized full-sib families related at the half-sib level. Likewise, the half-sib relationships of redds 4.2b/6.2, 9.4/10.9, and15.7/15.9 were evident in the close clustering of these redds. However, as noted previously, redds 17.6a and 17.6b formed a single tight cluster despite the fact that these two redds evidently contained a total of three full-sib families with two half-sib relationships (Table 3). Redd 10.4 contained a single male offspring, 10.4M65, that was related to two other full- sib families in the redd at the half-sib level, but that branched within one of the two families. Finally, the position of the cluster for redd 4.2a in the tree presented no hint of its half-sib relationship to redd 4.2b.
Two features of the UPGMA tree suggested that fish from the upper reach of the river were more related to each other, on average, than were those from the middle and lower reaches (Figure 2): upper river redds tended to cluster together, and average branch lengths separating individuals within families appeared shorter in the upper reach than for the other two reaches. To test the hypothesis that levels of relatedness varied among river reaches for given levels of kinship, we calculated relatedness coefficients (rxy) for all pairs of individuals within each reach. We then compared the distributions of rxy for full sibs, half sibs, and unrelated individuals (as determined from our mating reconstruction) among river reaches (Figure 3). For each category of kinship, upper-reach fish had higher mean and median values of rxy than fish from the middle and lower reaches (Table 4; Mann–Whitney U test, P < .001). Lower-reach fish had mean and median values of rxy that were intermediate between those of upper- and middle-reach fish at every level of kinship. However, differences between the rxy distributions of lower- and middle-reach fish were only significant for comparisons among the half- and full-sib distributions (Mann–Whitney U test, P = .016 and .035, respectively).
|
|
The trend in the relatedness of offspring was mirrored to some extent by a reverse trend in the genetic variability of the parents. Mean HE estimated from reconstructed parental genotypes for upper-reach parents was lower (0.72) than it was for middle- and lower-reach parents (0.81 and 0.84, respectively).
We evaluated our ability to classify Dungeness River chinook according to kinship level on the basis of rxy coefficients. We calculated rxy values for all pairs of full sibs, half sibs, and unrelateds identified in the analyses above. The distributions of rxy for full sibs and nonrelatives showed relatively little overlap, but the distribution of rxy for half sibs overlapped extensively with the distributions for full sibs and nonrelatives (Figure 4). We estimated the error rate in classifying individuals between two alternative levels of kinship (e.g., full sib and half sib) using both the empirically derived and simulated distributions of rxy for each level of kinship (see Methods). Estimated misclassification rates averaged 12% higher using the empirically derived rxy distributions than the simulated distributions, but followed similar trends for classifications between given levels of kinship (Table 5). Misclassification rates were lowest for classification between full sib and unrelated (mean 5% simulated, 6.5% empirical) and substantially higher between full and half sib (mean 19.5% simulated, 26.5% empirical) and between half sib and unrelated (mean 20.5% simulated, 17.5% empirical).
|
|
Scrutiny of the multilocus genotypes from all 135 progeny revealed a few presumptive genotyping errors or mutations. One individual from redd 9.4 was scored as a 220/220 homozygote at locus Ogo2, but would only have been compatible with the hypothesized parental genotypes if it were a 220/262 heterozygote, suggesting the presence of a larger allele that went undetected. Two individuals, one each from redds 9 and 15.7, had scored genotypes at one dinucleotide microsatellite locus (Ots1 and Ots3, respectively), suggestive of a genotyping miscall or mutation involving one microsatellite repeat unit. Two individuals from redd 17.6b were scored as 201/201 homozygotes at the tetranucleotide locus Ots104, whereas 197/ 201 or 201/217 genotypes would have been more plausible, suggesting the possibility of a mutation involving a gain of one repeat unit (197 to 201), or a failure to detect the 217 allele. These results suggest a combined genotyping error/mutation rate of 0.27% per single-locus genotype (five genotypes among a total of 1873 single-locus genotypes scored). This is a conservative estimate, since some potential genotyping errors or mutations could have resulted in genotypes that remained compatible with their hypothetical parents, and would have gone undetected.
Steelhead
We genotyped 53 fish comprising the entire 1998 spawning population of Snow Creek and all 52 fish that were spawned in the Forks Creek hatchery in winter 1999–2000 at 11 and 12 loci, respectively. The microsatellite loci were highly variable (Table 2), with 10–41 alleles per locus (mean 20.8 and 19.2 for Snow and Forks Creeks, respectively), and expected heterozygosities of 0.8–0.96 (mean 0.89 in both populations).
The upstream spawning migration of Snow Creek steelhead occurred over 10 weeks in 1998, although 21 fish (40%) were intercepted at the weir during the week ending 23 March, and 11 fish (21%) were caught during the preceding week. To test whether related fish were more likely to be intercepted at the weir during the same week, we computed rxy for all pairs of fish (dyads) from the spawning migration. We compared the distribution of run timing differences (binned into 1-week intervals) for dyads for which rxy > 0.248 (see Methods for justification) to the "null" distribution expected for randomly chosen dyads based on the numbers of steelhead that migrated upstream each week. We also compared the distribution of run timing differences for dyads for which rxy < 0.124 to the null distribution. The proportion of related dyads intercepted at the weir less than 1 week apart (37%) was significantly greater than the expectation for randomly chosen dyads (23%) (
2 = 4.72, df = 1, P = .030). Surprisingly the unrelated (low rxy) steelhead group also included a greater proportion of pairs returning 1 week or less apart (28%) than expected for randomly chosen dyads (2 = 14.3, df = 1, P < .001).
We used the same approach to test the hypothesis that related steelhead were more likely than randomly chosen pairs of fish to be spawned on the same day in the Forks Creek hatchery. Steelhead were intercepted at a fence during their upstream spawning migration and were held in a raceway prior to being spawned. Once a week over a period of 4 weeks in December 1999 and January 2000, hatchery staff tested steelhead in the raceway for sexual maturity; those that were fully mature were killed and spawned artificially. The numbers of fish spawned per day (in order from first to last date) were 12, 10, 20, and 10.
We calculated rxy for all pairs of fish spawned in the hatchery and compared the distributions of spawning date differences for the upper and lower tails of the rxy distribution (>0.249 and <0.126) to the null distribution for randomly chosen dyads. The overall distribution of spawning date differences for the related fish (rxy > 0.249) differed significantly from the expectation for randomly chosen dyads (2 = 23.5, df = 3, P < .001; Figure 5). In particular, the proportion of related dyads that were spawned on the same day in the hatchery (53.5%) was greater than expected for randomly chosen pairs of fish (30%). In contrast, the distribution of spawning date differences for unrelated dyads (rxy < 0.1249) did not differ significantly from the random expectation (2 = 2.96, df = 3, P = .50).
|
We conducted simulations to estimate expected error rates for classifying steelhead according to kinship. We based our simulations on the allele frequencies for the 10 most variable loci common to the Snow Creek and Forks Creek datasets. We also evaluated the gain in power of classification expected from doubling or tripling the number of loci; for these simulations we assumed additional sets of 10 or 20 loci with allele frequencies identical to the original 10 loci. These simulations revealed a substantial decrease in expected misclassification rates with the use of increasing numbers of loci (Table 6). Expected misclassification rates between full sibs and half sibs averaged 17%, 9%, and 5% for 10, 20, and 30 loci, respectively. Similarly, expected misclassification rates between half sibs and unrelated fish averaged 15%, 7%, and 4% for the three numbers of loci, respectively. Expected misclassification rates between full sibs and unrelated fish averaged less than 3% for 10 loci, and less than 1% for greater numbers of loci.
|
|
Discussion |
|---|
|
TopAbstractIntroductionMaterials and MethodsResultsDiscussionReferences |
|---|
Using a combination of graphical analysis and manual reconstruction of parental genotypes we were able to identify likely full- and half-sib families of chinook salmon from the Dungeness River, even in the absence of any genotypic information from the parents. The two analytical approaches were complementary and together yielded results that would not have been possible with either method alone. An UPGMA tree based on allele sharing coefficients revealed clusters of individuals that (apart from 12 fish that were apparently misidentified in the fish culture facility) were consistent with the recorded redd origins of all but one fish (Figure 2). The topology of the tree also suggested potential half-sib relationships within and among redds. However, the graphical approach by itself failed to separate two adjacent redds (17.6a and 17.6b) comprised of closely related families, and more important, it provided no objective criteria for distinguishing between full- and half- sib levels of relationship, or even between full sibs and unrelated individuals. In contrast, manual reconstruction of potential parental genotypes allowed explicit tests of the feasibility of hypothesized relationships. However, the manual approach would not have been practical without the prior information about likely family groupings provided by the graphical analysis.
The success of the combined graphical and manual approach at discriminating the three levels of kinship (full sib, half sib, and unrelated) was likely due primarily to two factors. First, the relatively high variability (5–19 alleles per locus, mean HE = 0.79) of the chinook microsatellites increased the likelihood that the difference between full-sib and half-sib families would be signaled by an increase from four alleles to five or six alleles within families at at least some loci. Second, the number of offspring attributed to most hypothesized parents was large enough that the probability of detecting both parental alleles at a heterozygous locus, or correctly inferring a homozygous parental genotype, was very high for most parents. For 23 hypothesized parents, the number of sampled offspring per parent was six or more, corresponding to a probability of 
93% that both alleles from any given parent were seen in its progeny array. For the remaining six parents, five were represented by three or more offspring, corresponding to a probability of 75% that both parental alleles were detected.
The fact that the chinook salmon were originally sampled as preemergent fry from redds provides partial corroboration for the mating patterns that we inferred. However, we have no way of assessing the absolute veracity of our proposed relationships, other than probabilistic arguments. For example, any of our proposed full- or half-sib families could have included individuals sired by fathers in addition to those we inferred. However, the small number of fish that were likely present in the parental population, and the low probability of identical or closely similar multilocus genotypes for parents involved with the same redds suggest that the likelihood of such errors is minimal. We consider these two points in turn below.
There are no direct census data for the number of mature chinook present in the Dungeness River in 1992, but state biologists estimated the population to be 153, based on the assumption that there were 2.5 fish for each observed redd (Smith and Wampler 1995). We inferred an average of only 2.1 parents per redd, suggesting a lower estimate of 128 fish for all 61 redds observed in the river that year. Assuming that males represented half the population, there may have been 49–59 mature males that were not accounted for among our proposed parents. The presence of precocious males ("jacks") might have skewed the sex ratio in favor of males, but the extent of this effect is unknown. Given evidence from this study that the different river reaches may have harbored distinct breeding groups, it seems likely that the number of males present in any reach did not greatly exceed one-third of the total present in the entire river. Further, since spawning occurred over a 7- to 8-week interval (Smith and Wampler 1995), even fewer males were probably present in any reach at any particular time because fully mature salmon do not live that long on the spawning grounds.
The average probability that two unrelated chinook shared any given 14 locus genotype was on the order of 10-19. Thus the likelihood that additional unrelated males sired some of the fish that we genotyped seems remote. However, we cannot rule out the possibility that some of these fish were sired by a sib or other close relative of the inferred father (Olsen 1999). The same arguments apply to cases where we inferred that an individual parented offspring from more than one redd.
Due to the limited number of progeny we assayed (7–12 per redd), the number of parents we inferred per redd should be regarded as minimum estimates (see DeWoody et al. 2000). Such undetected parents would likely have been satellite males or jacks that fathered a small proportion of the offspring in a given redd. The two females that produced offspring in one redd (Table 3) presumably resulted from redd site reuse.
Notwithstanding such statistical caveats, we were able to infer several reproductive behaviors among the unseen parents of the offspring we genotyped. These included single-pair matings in 10 redds and polyandry in 3 others. The fact that one male sired most or all of the offspring in most redds indicated a greater degree of reproductive dominance than has been inferred in other species from behavioral observations (Keenleyside and Dupuis 1988) or from examination of the genotypes of offspring (Thompson et al. 1998). Among the redds with multiple sires, two had two fathers and the third had three (Table 3). It is not clear whether the comparatively low contribution from secondary males resulted from the low density of salmon, which might have facilitated dominance by the alpha male, or from species- specific patterns. Few observations have been made of spawning behavior in chinook salmon (but see Chebanov and Riddell 1998), because they are large-bodied fish that spawn in large rivers where observations are difficult. Female chinook salmon, like other salmonids, deposit their eggs in a series of discrete pockets within the redd (e.g., Hawke 1978), and presumably a single male was able to dominate access to the female during all or most of this process.
Two other aspects of salmon behavior inferred from the parentage analysis are noteworthy. First, one female apparently used two redds. Construction of multiple redds has been documented in iteroparous species but is not characteristic of semelparous salmonids (Barlaup et al. 1994), and parentage analysis might be a useful technique to determine the frequency of such behavior. Second, two unrelated females successfully spawned in one redd, suggesting that reuse of the site did not destroy the progeny of the first female. Disturbance of redds by later spawning females is a major cause of density-dependent mortality in salmon (Essington et al. 2000) and it has sometimes been assumed that the progeny of one female are destroyed if the site is subsequently used by another.
We also found evidence of nonrandom mating indicative of population substructure within the Dungeness River population (Table 4). These results imply localized homing of chinook salmon, particularly to the upper reach of the river. The general pattern of relatedness among the salmon (highest in the upper reach, lowest in the middle reach) suggests that the middle reach has been used for spawning by salmon from the other reaches.
This mixing of salmon among reaches could have an anthropogenic component: from 1951 to 1981, mature chinook salmon were intercepted at a weir at the upper end of the middle reach and spawned at a hatchery in an effort to enhance production. During this time, an average of 372,081 juvenile salmon per year were planted in the river (Smith and Wampler 1995). These management efforts may have distorted the population substructure in two ways. First, by intercepting salmon bound for the upper river, the weir presumably greatly reduced the number of fish spawning in the upper reach. This could account for the lower HE of the upper-reach parents. Second, the matings in the hatchery presumably involved a mixture of "upper"- and "middle"-reach fish; the release of their progeny below the hatchery could be expected to erode population substructure in the middle and lower parts of the river. In light of the magnitude of the hatchery releases, the fact that we detected evidence of substructure in the Dungeness River population within 10 years (two to three generations) after the removal of the weir and cessation of hatchery supplementation is testimony to the resilience of the salmon population (and possibly to the ineffectiveness of the enhancement effort).
There is relatively little information on the fine-scale homing behavior of salmonids. Experimental releases of smolts in the lower, middle, and upper sections of rivers generally result in some degree of assortative returns (Cramer 1981; Slaney et al. 1993; Wagner 1969). Genetic studies have documented varying degrees of microgeographic structure in salmon populations (Garant et al. 2000; Utter 2000), but analyses of relatedness seem a particularly promising way to study fine-scale homing using genetic markers.
Salmonid fishes exhibit extensive behavioral and phenotypic variation within and among populations. Much of this variation is associated with reproduction and has been interpreted as adaptive and under genetic control (Taylor 1991). Heritability estimates based on conventional breeding studies are available for some traits (e.g., Crandall and Gall 1993; Gall et al. 1988), but other traits cannot readily be studied using captive breeding. One complementary approach might be to assess genetic influences on behavior and phenotype by genotyping a set of naturally produced individuals that display variation in the trait of interest.
In steelhead trout we tested whether timing of river entry during spawning migration and date of sexual maturation were correlated with microsatellite-based estimates of relatedness. For the Forks Creek hatchery, we found that related trout dyads (rxy > 
l0.25) were more likely to be spawned on the same day than expected by chance. Since fish were held in the hatchery until they were sexually mature and because hatchery spawning occurred at weekly intervals, this result implies that related fish were more likely than unrelated fish to mature within 1 week of each other after 3 years of life. Similarly in the Snow Creek population we found that related dyads were significantly more likely to enter the river less than 1 week apart. These results are consistent with experimental evidence that the timing of return migration or spawning date is heritable in rainbow trout (Siitonen and Gall 1989), chinook salmon (Quinn et al. 2000), and pink salmon (Smoker et al. 1998). Our results thus demonstrate the potential of microsatellite-based relatedness estimates for testing correlations between phenotype and relatedness in populations that are natural or at least not subject to controlled breeding.
Our results on salmonids also indicate much room for progress in resolving kinship with microsatellite data. Although we were able to resolve different levels of kinship in the Dungeness River chinook salmon with high apparent accuracy, the combination of graphical analysis and manual reconstruction of genotypes that we used to achieve this result was tedious and would not be practical in all circumstances. In contrast, the computation of rxy coefficients for all possible dyads within a population was straightforward, but much less effective for resolving kinship because of high rates (>20%) of expected misclassification between half sibs and full sibs, and between half sibs and unrelated fish (Table 5). However, our computer simulations also suggested that misclassification rates could be greatly reduced by genotyping as many as 30 loci. This prospect is not unreasonable, given the number of microsatellite loci currently available for salmonid fishes (more than 500) and the ongoing development of high throughput genotyping equipment. Efforts are under way in several laboratories (including ours) to produce microsatellite linkage maps for salmonids. Such developments will make it increasingly practical to resolve kinship and estimate the heritabilities of behavioral and phenotypic traits in salmonid populations.
|
Acknowledgments |
|---|
We thank J. Shaklee, S. Young, C. Marlowe, T. Johnson, R. Cooper, and other staff of the Washington Department of Fish and Wildlife for help in acquiring fish samples and other forms of assistance. This research was supported by Saltonstall-Kennedy award NA76FD0299, NSF DEB-9903914, the Weyerhaeuser Foundation, and the H. Mason Keeler Endowment.
|
Footnotes |
|---|
This paper was delivered at a symposium entitled "DNA-Based Profiling of Mating Systems and Reproductive Behaviors in Poikilothermic Vertebrates" sponsored by the American Genetic Association at Yale University, New Haven, CT, USA, June 17–20, 2000.
Corresponding Editor: John C. Avise
|
References |
|---|
|
TopAbstractIntroductionMaterials and MethodsResultsDiscussionReferences |
|---|
-
Banks MA, Blouin MS, Baldwin BA, Rashbrook VK, Fitzgerald HA, Blankenship SM, and Hedgecock D, 1999. Isolation and inheritance of novel microsatellites in chinook salmon (Oncorhynchus tshawytscha). J Hered 90:281–288.
Banks MA, Rashbrook VK, Calavetta MJ, Dean CA, and Hedgecock D, 2000. Analysis of microsatellite DNA resolves genetic structure and diversity of chinook salmon (Oncorhynchus tshawytscha) in California's Central Valley. Can J Fish Aquat Sci 57:915–927.
Barlaup BT, Lura H, Saegrov H, and Sundt RC, 1994. Inter- and intra-specific variability in female salmonid spawning behaviour. Can J Zool 72:636–642.
Blouin MS, Parsons M, Lacaille V, and Lotz S, 1996. Use of microsatellite loci to classify individuals by relatedness. Mol Ecol 5:393–402.[Medline]
Bowcock AM, Ruiz-Linares A, Tomfohrde J, Minch E, Kidd JR, and Cavalli-Sforza LL, 1994. High resolution of human evolutionary trees with polymorphic microsatellites. Nature 368:455–457.[Medline]
Chebanov NA and Riddell BE, 1998. The spawning behavior, selection of mates, and reproductive success of chinook salmon (Oncorhynchus tshawytscha) spawners of natural and hatchery origins under conditions of joint spawning. J Ichthyol 38:517–526.
Chebanov NA, Varnavskaya NV, and Varnavskiy VS, 1983. Effectiveness of spawning of male sockeye salmon, Oncorhynchus nerka (Salmonidae), of differing hierarchical rank by means of genetic-biochemical markers. J Ichthyol 23:51–55.
Condrey MJ and Bentzen P, 1998. Characterization of coastal cutthroat trout (Oncorhynchus clarki clarki) microsatellites, and their conservation in other salmonids. Mol Ecol 7:783–793.
Cramer DP, 1981. Effect of smolt release location and displacement of adults on distribution of summer steelhead trout. Prog Fish Cult 43:8–11.
Crandall PA and Gall GAE, 1993. The genetics of body weight and its effect on early maturity based on individually tagged rainbow trout (Oncorhynchus mykiss). Aquaculture 117:77–93.
DeWoody JA, Walker D, and Avise JC, 2000. Genetic parentage in large half-sib clutches: theoretical estimates and empirical appraisals. Genetics 154:1907–1912.
Essington TE, Quinn TP, and Ewert VE, 2000. Intra- and interspecific competition and the reproductive success of sympatric Pacific salmon. Can J Fish Aquat Sci 57:205–213.
Felsenstein J, 1993. PHYLIP (phylogeny inference package) version 3.5c. Seattle: University of Washington.
Fields RD, Johnson KR, and Thorgaard GH, 1989. DNA fingerprints in rainbow trout detected by hybridization with DNA of bacteriophage M13. Trans Am Fish Soc 118:78–81.
Fleming IA, 1998. Pattern and variability in the breeding system of Atlantic salmon (Salmo salar), with comparisons to other salmonids. Can J Fish Aquat Sci 55(suppl 1):59–76.
Gall GAE, Baltodano J, and Huang N, 1988. Heritability of age at spawning for rainbow trout. Aquaculture 68:93–102.
Garant D, Dodson JJ, and Bernatchez L, 2000. Ecological determinants and temporal stability of within-river population structure in Atlantic salmon (Salmo salar L.). Mol Ecol 9:615–628.[Medline]
Garant D, Dodson JJ, and Bernatchez L, 2001. A genetic evaluation of mating system and determinants of individual reproductive success in Atlantic salmon (Salmo salar L.). J Hered 92:137–145.
Garcia-Vazquez E, Moran P, Martinez JL, Perez J, de Gaudemar B, and Beall E, 2001. Alternative mating strategies in salmonid fishes. J Hered 92:146–149.
Gharrett AJ and Smoker WW, 1991. Two generations of hybrids between even- and odd-year pink salmon (Oncorhynchus gorbuscha): a test for outbreeding depression? Can J Fish Aquat Sci 48:1744–1749.
Gharrett AJ and Smoker WW, 1993. Genetic components in life history traits contribute to population structure. In: Genetic conservation of salmonid fishes (Cloud JG, ed). New York: Plenum; 197–202.
Gharrett AJ, Smoker WW, Reisenbichler RR, and Taylor SG, 1999. Outbreeding depression in hybrids between odd-and even-brood year pink salmon. Aquaculture 173:117–129.
Goodnight KF, 2000. Kinship (version 1.2) 
http://gsoft.smu.edu/GSoft.html
.
Hawke SP, 1978. Stranded redds of quinnat salmon in the Mathias River, South Island, New Zealand. N Z J Mar Freshwater Res 12:167–171.
Herbinger CM, Doyle RW, Pitman ER, Paquet D, Mesa KA, Morris DB, Wright JM, and Cook D, 1995. DNA fingerprint based analysis of paternal and maternal effects on offspring growth and survival in communally reared rainbow trout. Aquaculture 137:245–256.
Herbinger CM, O'Reilly PT, Doyle RW, Wright JM, and O'Flynn F, 1999. Early growth performance of Atlantic salmon full-sib families reared in single family tanks versus in mixed family tanks. Aquaculture 173:105–116.
Hutchings JA and Myers RA, 1988. Mating success of alternative maturation phenotypes in male Atlantic salmon, Salmo salar. Oecologia 75:169–174.
Keenleyside MHA and Dupuis HMC, 1988. Courting and spawning competition in pink salmon (Oncorhynchus gorbuscha). Can J Zool 66:262–265.
Leider SA, Hulett PL, Loch JJ, and Chilcote MW, 1990. Electrophoretic comparison of the reproductive success of naturally spawning transplanted and wild steelhead trout through the returning adult stage. Aquaculture 88:239–252.
Maekawa K and Onozato H, 1986. Reproductive tactics and fertilization success of mature male Miyabe charr, Salvelinus malma miyabei. Environ Biol Fish 15:119– 129.
Martinez JL, Moran P, Perez J, De Gaudemar B, Beall E, and Garcia-Vazquez E, 2000. Multiple paternity increases effective size of southern Atlantic salmon populations. Mol Ecol 9:293–298.[Medline]
Minch E, Ruiz-Linares A, Goldstein DB, Feldman MW, and Cavalli-Sforza LL, 1995. MICROSAT (version 1.5d) http://www.lotka.stanford.edu/microsat.html.
Moran P, Pendas E, Garcia-Vazquez JT, Izquierdo JT, and Rutherford DT, 1994. Electrophoretic assessment of the contribution of transplanted Scottish salmon (Salmo salar) to the Esva River (northern Spain). Can J Fish Aquat Sci 51:248–252.
Moran P, Pendas AM, Beall E, and Garcia-Vazquez E, 1996. Genetic assessment of the reproductive success of Atlantic salmon precocious parr by means of VNTR loci. Heredity 77:655–660.
Morris DB, Richard KR, and Wright JM, 1996. Microsatellites from rainbow trout (Oncorhynchus mykiss) and their use for genetic study of salmonids. Can J Fish Aquat Sci 53:120–126.
Mousseau TA, Ritland K, and Heath DD, 1998. A novel method for estimating heritability using molecular markers. Heredity 80:218–224.
National Research Council, 1996. Upstream: salmon and society in the Pacific Northwest. Washington, DC: National Academy Press.
Nelson RJ and Beacham TD, 1999. Isolation and cross species amplification of microsatellite loci useful for study of Pacific salmon. Anim Genet 30:225–244.[ISI][Medline]
O'Connell M, Danzmann RG, Cornuet JM, Wright JM, and Ferguson MM, 1997. Differentiation of rainbow trout (Oncorhynchus mykiss) populations in Lake Ontario and the evaluation of the stepwise mutation and infinite allele mutation models using microsatellite variability. Can J Fish Aquat Sci 54:1391–1399.
Olsen JB, 1999. Genetic interpretation of microsatellite polymorphism in Pacific salmon: case studies in population genetics and kinship analysis (PhD dissertation). Seattle: University of Washington.
Olsen JB, Bentzen P, and Seeb JE, 1998. Characterization of eight microsatellite loci derived from pink salmon. Mol Ecol 7:1087–1089.[Medline]
Olsen JB, Wenburg JK, and Bentzen P, 1996. Semiautomated multilocus genotypic of Pacific salmon (Oncorhynchus spp.) using microsatellites. Mol Mar Biol Biotechnol 5:259–272.[ISI][Medline]
O'Reilly P, Hamilton LC, McConnell SK, and Wright JM, 1996. Rapid analysis of genetic variation in Atlantic salmon (Salmo salar) by PCR multiplexing of dinucleotide and tetranucleotide microsatellites. Can J Fish Aquat Sci 53:2292–2298.
Page RDM, 1996. TREEVIEW: an application to display phylogenetic trees on personal computers. Comput Appl Biosci 12:357–358.
Queller DC and Goodnight KF, 1989. Estimating relatedness using genetic markers. Evolution 4:258–275.
Quinn TP, 1999. Variation in Pacific salmon reproductive behaviour associated with species, sex and levels of competition. Behaviour 136:179–204.
Quinn TP, Unwin MJ, and Kinnison MT, 2000. Evolution of temporal isolation in the wild: genetic divergence in timing of migration and breeding by introduced chinook salmon populations. Evolution 54:1372–1385.[ISI][Medline]
Schroder SL, 1981. The role of sexual selection in determining the overall mating patterns and mate choice in chum salmon (PhD dissertation). Seattle: University of Washington.
Scribner KT, Gust JR, and Fields RL, 1996. Isolation and characterization of novel salmon microsatellite loci: cross-species amplification and population genetic applications. Can J Fish Aquat Sci 53:833–841.
Siitonen L and Gall GAE, 1989. Response to selection for early spawn date in rainbow trout, Salmo gairdneri. Aquaculture 78:153–161.
Slaney PA, Berg L, and Tautz AF, 1993. Returns of hatchery steelhead relative to site of release below an upper- river hatchery. N Am J Fish Manage 13:558–566.
Small MP, Beacham TD, Withler RE, and Nelson RJ, 1998. Discriminating coho salmon (Oncorhynchus kisutch) populations within the Fraser River, British Columbia using microsatellite DNA markers. Mol Ecol 7:141–155.
Smith CJ and Wampler P, eds, 1995. Dungeness River chinook salmon rebuilding project progress report 1992–1993. Northwest Fishery Bulletin. Project report series no. 3. Olympia, WA: Northwest Indian Fisheries Commission.
Smoker WW, Gharrett AJ, and Stekoll MS, 1998. Genetic variation of return date in a population of pink salmon: a consequence of fluctuating environment and dispersive selection? Alaska Fish Res Bull 5:46–54.
Taylor EB, 1991. A review of local adaptation in Salmonidae, with particular reference to Pacific and Atlantic salmon. Aquaculture 98:185–207.[ISI]
Thomaz D, Beall E, and Burke T, 1997. Alternative reproduction tactics in Atlantic salmon: factors affecting mature parr success. Proc R Soc Lond B 264:219–226.
Thompson CE, Poole WR, Matthews MA, and Ferguson A, 1998. Comparison, using minisatellite DNA profiling, of secondary male contribution in the fertilisation of wild and ranched Atlantic salmon (Salmo salar) ova. Can J Fish Aquat Sci 55:2011–2018.
Utter FM, 1998. Genetic problems of hatchery-reared progeny released into the wild, and how to deal with them. Bull Mar Sci 62:623–640.
Utter FM, 2000. Patterns of subspecific anthropogenic introgression in two salmonid genera. Rev Fish Biol Fish 10, in press.
Wagner HH, 1969. Effect of stocking location of juvenile steelhead trout, Salmo gairdnerii, on adult catch. Trans Am Fish Soc 98:27–34.
Waples RS, 1991. Genetic interactions between hatchery and wild salmonids: lessons from the Pacific Northwest. Can J Fish Aquat Sci 48(suppl 1):124–133.
CiteULike 
Connotea 
Del.icio.us What's this?
This article has been cited by other articles:
|
|
 |
|
  H. Yu and Q. Li Exploiting EST Databases for the Development and Characterization of EST-SSRs in the Pacific Oyster (Crassostrea gigas) J. Hered., March 1, 2008; 99(2): 208 - 214. [Abstract] [Full Text] [PDF]
|
|
| ||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||