1 INTRODUCTION

Migratory species often exhibit coordinated movements underpinned by individual-specific timing of migration, whereby individuals arrive consistently early or late, relative to other individuals in a population, at specific habitats or life-history stages (Bell et al., 2009; Biro & Adriaenssens, 2013; Tamario et al., 2019). Timing of migration has proven to be crucial in defining fitness consequences related to reproductive success and offspring survival (Brodersen et al., 2012; Tibblin et al., 2016). Intra-individual consistency in animal movements and resultant repeatability in migration timing have received increased scientific attention as empirical evidence, based on advanced tracking technologies, accumulates for many migratory species (Herbert-Read, 2016; Westley et al., 2018), particularly birds (Gilsenan et al., 2020; Kentie et al., 2017) and fishes (Brodersen et al., 2012; Forsythe et al., 2012; van Wijk et al., 2016).
Migration timing has been shown to be partially determined by an endogenous circannual rhythm that synchronizes behaviors with photoperiod in birds and fishes (Eriksson & Lundqvist, 1982; Pulido et al., 2001; Styrsky et al., 2004). This rhythm arises from genetic evolution in animals, and it is therefore important to recognize that migratory behaviors are heritable and best understood in the context of evolutionary adaptation (Brönmark et al., 2013; Quinn et al., 2000; Thompson et al., 2019). Although these rhythms are driven by endogenous hormone cycles cued by photoperiod, migration timing in individuals can be affected by environmental factors (Franklin et al., 2022; Harrison et al., 2017). Because of consistent individual differences in animal behavior (Biro & Adriaenssens, 2013; Nilsson et al., 2014), the behavioral traits of individuals responding to environmental controls with pronounced seasonality tend to be repeatable across years (Dahl et al., 2004; Gilsenan et al., 2020; Quinn et al., 1997). However, identifying proximate environmental drivers and quantifying their effects on migratory behaviors have been technically difficult, especially in a field setting, because researchers generally can only recognize migration at its end point (e.g. a breeding site), and not at the time and location at which the migration was initiated (Porlier et al., 2012; Winkler et al., 2014). An additional complication is that animals typically migrate when environmental variables monotonically increase (or decrease), inevitably leading to spurious correlations with day-of-year, even though no causal mechanisms exist (Dahl et al., 2004; Sinnatamby et al., 2018).
Alternatively, repeatability in migration timing could be manifested when animals form cohesive groups (e.g. bird flocks and fish schools) of socially interacting individuals with strong, long-lasting group fidelity (Fraser et al., 2005; Hay & McKinnell, 2002; Klimley & Holloway, 1999). Group migration, or collective navigation, is known to facilitate rapid transfer of beneficial behavioral traits (Gil et al., 2018; Thorsteinsson et al., 2012), help avoid predation, cancel errors made by less experienced members (Berdahl et al., 2016), and improve homing precision to natal locations (Bett & Hinch, 2015). This mechanism, however, may not be entirely independent of the environmental control mechanism described above because at least some group members would still rely on external cues for decision making (Couzin, 2009; Herbert-Read, 2016, Herbert-Read et al., 2017). Even though the social aspect of animal migration has been investigated in laboratory and theoretical studies, it remains largely unexplored in field-based ecological studies because of logistical and methodological constraints (Brönmark et al., 2013; Gil et al., 2018).
Sakhalin taimen Parahucho perryi is a critically endangered, long-lived (>25 years), iteroparous salmonid inhabiting far eastern Russia and Hokkaido, Japan (Rand, 2006). During the spring spawning season, mature Sakhalin taimen migrate upstream to headwater streams in small groups from their non-breeding habitats in the estuary (Rand & Fukushima, 2014). A recent tagging study revealed that up to 87% of Sakhalin taimen spawners return to the same tributaries across consecutive years, one of the highest rates of site fidelity ever documented for iteroparous salmonids (Fukushima & Rand, 2021). The group migration and strong site fidelity of the species are consistent with the hypothesis that social interactions influence their movement patterns during the spawning migration. However, the rate of upstream migration of this species is significantly influenced by water temperature and stream discharge (Rand & Fukushima, 2014), suggesting the importance of environmental controls on migration timing as well.
In this study, we investigated the migratory behavior of adult Sakhalin taimen by tracking spawners and post-spawners in a river system in Hokkaido, Japan, with multiple spawning streams. We first characterized migration timing at specific waypoints and examined the degree to which the migration timing was repeatable among individuals across years. We then explored whether the repeatability in migration timing resulted from individual-specific responsiveness to seasonality in the environmental controls, or from social interactions among comigrating individuals. We discuss our findings in relation to the concepts of migratory connectivity (Webster et al., 2002) and differential migration (Briedis & Bauer, 2018; Brodersen et al., 2012).