1133 q 2003 The Society for the Study of Evolution. All rights reserved. Evolution, 57(5), 2003, pp. 1133���1141 DIFFERENTIAL REPRODUCTIVE SUCCESS AND HERITABILITY OF ALTERNATIVE REPRODUCTIVE TACTICS IN WILD ATLANTIC SALMON (SALMO SALAR L.) DANY GARANT, JULIAN J. DODSON, AND LOUIS BERNATCHEZ1 Universite �� Laval, De ��partement de Biologie, Sainte-Foy, Que ��bec G1K 7P4 Canada 1E-mail: louis.bernatchez@bio.ulaval.ca Abstract. A critical step in understanding the evolution and maintenance of alternative reproductive tactics is to obtain accurate comparisons of their fitness and to determine factors influencing individual status. In this study, we first used individual multilocus genotypic information to compare reproductive success between two alternative re- productive tactics of anadromous Atlantic salmon (Salmo salar L.) in their natural environments. We also documented the effects of the quality of the rearing environment and of paternal reproductive tactics on heritability of juvenile growth, which is an important component of individual status. Results showed that large dominant salmon (multisea winter) had higher reproductive success than smaller satellite individuals (grilse). Also, there was a status difference associated with both habitat and male tactic. Overall, offspring produced in streams were bigger than those produced in the main river stretch. Grilse also produced bigger offspring than those fathered by multisea winter males. Heritability of juvenile growth was significant but varied according to quality of habitat: higher heritability estimates were observed in higher quality habitats (streams) than in lower quality habitats (main river stretch). Heritability estimates for juvenile growth varied as well, depending on male tactic, with progeny fathered by multisea winter males having higher values than those fathered by grilse. Together, these results indicate that a combination of additive genetic effects, parental life history and habitat quality will ultimately shape juvenile growth rate, which is the main determinant of status and of subsequent choice of life-history tactics. Key words. Alternative strategy, environmental effects, fitness, juvenile growth, Salmonidae. Received September 27, 2002. Accepted December 6, 2002. A major objective in evolutionary biology is to understand processes by which polymorphisms associated with alter- native reproductive phenotypes are maintained within spe- cies. Theoretical frameworks, based on game theory (May- nard Smith 1982), have been elaborated to study alternative reproductive phenotypes in relation to their costs and benefits to evolutionary fitness (reviewed in Gross 1996). The first and rarest means by which phenotypic diversity within a sex may arise is through a genetic polymorphism for two strat- egies (Ryan et al. 1992 Lank et al. 1995). Two additional theoretical frameworks to explain the coexistence of alter- native reproductive morphs in a same population involve the ������mixed������ and the ������conditional������ strategy (Maynard Smith 1982 Gross 1996). Both of these strategies propose the ex- istence of genetically monomorphic alternative tactics within a strategy. A strategy refers to the underlying genetic basis of alternative life histories, whereas a tactic is the phenotype that results from the strategy (Gross 1996). Therefore, the mixed strategy proposes that frequency-dependent selection could result in equal fitness between alternative phenotypes, but little empirical evidence exists to support this model (Gross 1996 but see Sato 1998). The majority of known alternative reproductive phenotypes are thus interpreted as alternative tactics within a conditional strategy (Crespi 1988 Radwan 1995 reviewed in Gross 1996 see also Brockmann 2001). This theory proposes that individuals are required to make a ������decision������ regarding the tactic they will adopt based on their relative status (com- petitive ability) in the population that depends upon their individual condition (Gross 1996). This results in unequal average lifetime fitness of alternative reproductive tactics within the population, but it is hypothesized that the chosen tactic will provide the individual with the highest fitness pos- sible between alternatives, given its relative status at a given time. Furthermore, theoretical studies have shown that two alternative tactics with unequal fitness could be maintained in equilibrium, given a heritable component of the status (Hazel et al. 1990 Gross and Repka 1998a), and that such an equilibrium might be stable (Hazel and Smock 1993 Gross and Repka 1998b). In many species, individual status is related to juvenile growth or size attained at a given age (Bohlin et al. 1990 Wiegmann et al. 1997 Hofmann et al. 1999 Moczek and Emlen 1999 Garant et al. 2002). For natural populations, it is generally believed that local environmental conditions and environmental heterogeneity are the major determinants of juvenile growth and hence of status (Hutchings and Myers 1994). For example, in smallmouth bass (Micropterus dolom- ieui), it has been shown that environmental changes could alter the proportion of males that would adopt a given re- productive tactic through its influence on size attained at a given age (Wiegmann et al. 1997). Additional effects may be potentially important in deter- mining juvenile growth and status. Nonadditive parental ef- fects such as maternal (reviews in Mousseau and Fox 1998) and paternal (see Hunt and Simmons 2000 Pakkasmaa et al. 2001) effects have been proposed as major factors controlling offspring characteristics (reviewed in Qvarnstro ��m and Price 2001). Also, the heritable additive genetic component and its interaction with factors that act on offspring characteristics is clearly an important part of offspring growth and status that should be evaluated (Roff 1996). However, there are still relatively few estimates of heritability of phenotypic traits in the wild, largely due to logistical constraints (but see Si- nervo and Zamudio 2001), and consequently, most estimates available are derived from controlled experiments (e.g. Mousseau and Roff 1987). Heritability estimates in nature come mainly from bird studies using cross-fostering exper-

1134 DANY GARANT ET AL. iments for species in which parental identity can be assigned with a certain degree of confidence by field observation (Dhondt 1982 Merila �� 1997 Merila �� and Fry 1998). Such techniques, however, are almost impossible to use in species with high juvenile mortality and dispersal (Ritland 2000). Furthermore, heritability estimates have been shown to vary with the quality of rearing habitat, and are hypothesized to be greater in high- than in low-quality habitat (Qvarnstro ��m 1999). This underlines the importance of measuring herita- bility under natural conditions. Several studies have revealed that fitness pay-offs asso- ciated with alternative tactics could be unequal (Brockmann et al. 1994 Foote et al. 1997 Coltman et al. 1999). None of them, however, have fulfilled the complete set of criteria to demonstrate a conditional strategy convincingly (but see Hunt and Simmons 2001). Namely, fitness estimates in these studies have been equated to reproductive success quantified under controlled conditions or indirectly by means of be- havioral observations (Clutton-Brock 1988), which could be inappropriate in situations in which extra-pair fertilizations occur (Gibbs et al. 1990). Recent advances in the use of individual multilocus genotypic information have made it possible to accurately reconstruct pedigree and thus parental relationships in natural conditions. This has allowed precise measurements of reproductive success without manipulations in the wild (e.g., Weatherhead and Boag 1997 Zamudio and Sinervo 2000 Garant et al. 2001). DNA genotyping, in com- bination with new statistical approaches, has also improved the accuracy and availability of data for quantitative genetic analysis and thus permitted estimates of heritability in the field. Specifically, this has been performed using restricted maximum likelihood (REML) and animal model procedures and applied to population data for birds (Merila �� et al. 2001) and mammals (Re ��ale et al. 1999 Kruuk et al. 2000 Milner et al. 2000). Despite its obvious particular interest for the study of poikilotherms with high juvenile mortality and dis- persal capability, the use of individual multilocus genotyping to infer reproductive success has been limited thus far (Avise 2001), and no study has assessed the genetic basis of fitness- related traits in fishes. Atlantic salmon (Salmo salar L.) exhibit among the greatest within-population variability in size and age at maturity of all vertebrates (reviewed in Fleming 1998). In this species, males generally exhibit three alternative reproductive tactics associated with marked size differences. First, there is the small precocious male phenotype that consists of salmon who reach maturity in fresh water as early as their second summer of life and that attempt to sneak fertilization. There is also the larger anadromous male phenotype consisting of fish who require a minimum of three to four years, including an oce- anic phase, to mature and who compete among themselves for the opportunity to guard females (Belding 1934 Jones and Ball 1954 reviewed in Fleming 1998). There are two distinct phenotypes within anadromous males that are also associated with two distinct reproductive tactics. First, there is the multisea winter tactic (MSW), which consists of salmon that have stayed two or more years at sea before returning to freshwater to reproduce. Fish adopting this tactic attain large sizes and are typically dominant individuals on spawn- ing grounds. The second anadromous tactic involves fish, known as grilse, who remain only one year at sea before returning to fresh water to spawn. These males are much smaller than MSW fish and accordingly, they are known to behave like subordinate males on the spawning grounds (see Fleming 1998). Reproductive success in natural populations of Atlantic salmon has been recently assessed (Garant et al. 2001 Tag- gart et al. 2001). However, no estimates of reproductive suc- cess are yet available to specifically compare the alternative anadromous tactics found in this species. Furthermore, evi- dence that age and size at sexual maturity are both under genetic and environmental control in this species comes from experimental and controlled experiments (Naevdal et al. 1976 Thorpe et al. 1983 Glebe and Saunders 1986 Berglund 1992). Although there is mounting evidence that high growth rate of juvenile stages is one of the most important factors contributing to the development of precocious maturity in Atlantic salmon (Lundqvist 1980 Myers et al. 1986 Hutch- ings and Jones 1998 Whalen and Parrish 1999 Garant et al. 2002), very little is known about the influence of juvenile characteristics on the choice of the two anadromous tactics (MSW vs. grilse). Furthermore, as there are no heritability estimates of juvenile growth in the wild, it is unclear to what degree genetic and environmental components determine these life-history tactics in Atlantic salmon. At the habitat level, the amount of organic drift is more abundant in small streams than in bigger rivers (Naiman et al. 1987 Erkinaro and Niemela �� 1995), and invertebrates on which juveniles feed mainly in early life stages are bigger in streams (Keeley and Grant 2001). This influence of habitat might be so important that it could trigger individuals with typically lower growth rates to develop tactics different from that under restricted conditions, thus increasing the proportion of individuals be- yond the threshold value for early sexual maturation (see Hutchings and Myers 1994). The objective of this study was to use individual multilocus genotyping to compare reproductive success between the two alternative reproductive tactics associated with distinct phe- notypes in anadromous male Atlantic salmon. We thus tested the null hypothesis that there is no difference in reproductive success between the MSW and the grilse anadromous tactic. Second, we documented the effects of both the quality of the rearing environment and of the paternal reproductive tactic on juvenile growth and its heritability. We tested the null hypothesis that growth and its heritability value will be the same in both high- and low-quality habitats and for both paternal reproductive phenotypes. MATERIALS AND METHODS Sample Collection and Characteristics This study was conducted on the Sainte-Marguerite river (488209N, 708009W), Que ��bec, Canada, which sustains a nat- ural population of Atlantic salmon. Anadromous salmon mi- grate into this river in the middle of summer (July and Au- gust) and spawn during the fall (October and November). Salmon fry (young of the year) emerge from the spawning gravel in late June the following year and move to nursery grounds adjacent to spawning grounds. In July 1995, we caught 76 adult fish, specifically 41 males (31 grilse and 10