4 DISCUSSION

Male Sakhalin taimen returned to the spawning grounds of the Karibetsu River several days earlier than their female counterparts. Conversely, the males departed from these sites later than the females after spawning, resulting in a longer stream residence for males. Earlier arrival at breeding sites by males (protandry) has been documented for some migratory fishes (e.g. Sinnatamby et al., 2018; Šmejkal et al., 2017; Tibblin et al., 2016), and is generally explained by the mate opportunity hypothesis that predicts that earlier arriving males of polygynous species at breeding areas increase their opportunity to mate with females (Morbey & Ydenberg, 2001). The earlier arrival and longer residence at the spawning grounds by the male Sakhalin taimen likely resulted in their lower rate of spawning across consecutive years (Fukushima & Rand, 2021). Therefore, even though the operational sex ratio can be temporarily skewed toward males earlier and again later in the season, the overall sex ratio of active spawners was female biased, according to our fish sampling and videotaped records of individual fish (Rand & Fukushima, 2014).
The number of tributaries Sakhalin taimen entered for spawning significantly influenced departure timings from tributary sites, such that individuals, especially males, spawning in two tributaries were on average later to depart from the last tributary than those spawning in a single tributary. However, subsequent arrival downstream at Site L was not influenced by the number of tributaries they entered, indicating that fish spawning in multiple tributaries tended to catch up with individuals with which they previously comigrated before spawning and restore cohesive groups during the downstream migration.
Sakhalin taimen migration timing was also different at the diel scale, between sexes and between ascent and descent in females. Male Sakhalin taimen were largely diurnal during both ascent and descent of the mainstem of the Karibetsu River, whereas females were diurnal during ascent but more likely to be nocturnal during descent. This may be explained by the propensity of male salmonids to be less risk averse than females during the spawning season (Fleming, 1996). Our model predicted that both sexes tended to migrate during night-time when departing from tributaries later in the season. Decreased water levels, especially in small streams, would increase the visibility of the post-spawners to predators during daytime (Finlay et al., 2020), which is a likely explanation for our observations.
Female Sakhalin taimen displayed remarkable intra-individual repeatability in seasonal migration timing between 2018 and 2019 at all migration stages. Their departure timing was repeatable even at the diel scale during MS3. Furthermore, the females descended the mainstem in a chronological order very similar to that observed during their ascent, despite most of them spawning in different tributaries in the interim. Therefore, Sakhalin taimen, at least females, can be characterized by highly synchronized movements during the spawning migration, and strong fidelity to arrival and departure timings at specific waypoints along their migration pathways. Meta-analysis on studies of behavioral repeatability in various animal taxa have revealed that, except for mate preference, females are generally more consistent in their behavior than males (Bell et al., 2009).
Individual-level consistency in migratory behavior has been rarely quantified for salmonids despite the long history of scientific investigation into their migration at the species or population level. This is in part due to the fact that the literature on salmonid migrations has largely focused on Pacific salmon (Oncorhynchusspp.), which nearly all die after the first spawning (Groot & Margolis, 1991). Even well-studied iteroparous salmonids have rates of repeat spawning so low (e.g. less than 10% in steelhead O. mykiss[Keefer et al., 2008; Leider et al., 1986], 11% in Atlantic salmonSalmo salar [Fleming, 1998]) that they also are considered functionally semelparous (Penny & Moffitt, 2013), making it impractical to investigate their long-term behavioral consistency in spawning at the individual level.
There was a significant correlation in both water temperature and water level measured in the Karibetsu mainstem between 2018 and 2019 in specific time windows relative to the arrival timing of female Sakhalin taimen in the spawning tributaries. Apart from the statistical significance of the correlations, the facts that the two critical time windows overlapped and particularly that their time lags (8.3 and 9.3 days) were very similar indicate that individual females became responsive to the two environmental signals during similar time periods in locations where the fish might have lingered before initiating migration upstream. The estimated mean travel time (approx. 12 days) by the females from the river mouth to the spawning ground exceeded the time lags of the critical time windows, leading to a counterintuitive conclusion that fish made the decision to initiate upstream migration and left the river mouth “before” the environmental conditions were met. However, it should be noted that this travel time was most likely over-estimated because it was derived by extrapolating the observed swimming speed (1.98 km day–1) between waypoints in the headwater area. This speed could be biased low given that fish are expected to swim slower through upper river reaches (Gauld et al., 2016) due to steeper elevation gradients, faster currents, and more numerous natural and unnatural obstacles, like debris jams and culverts, than they encounter in the lower reaches of the Karibetsu mainstem. If a more realistic (i.e. faster) estimate of swimming speed was available for the lower reaches, their travel time would likely have been much shorter and coincided more closely with the critical time windows than what is displayed in Fig. 5. Our presumption that the Sakhalin taimen spawners started upstream migration from the river mouth is justifiable according to a provenance study based on stable isotope analysis of otoliths (Fukushima et al., 2019). Strontium isotope ratios in the otolith samples of juvenile Sakhalin taimen collected throughout the Sarufutsu River drainage indicated that their mothers were inhabiting brackish waters prior to their spawning migration.
The time scale that Sakhalin taimen responded to temperature and water level appears to be different. The time window during which the fish responded to water level was much longer (6 days) than that for temperature (1.7 days), suggesting that river flow not only cued the start of migration but also continually influenced fish movements en route to the spawning ground, whereas water temperature cued migration with little influence thereafter. Simmons et al. (2021) observed that out-migrating Atlantic salmon smolts were significantly influenced by changes in water temperature early in the run with much less influence afterwards, but that changes in river discharge exerted stronger influences on the migration towards the end of the run.
Although water temperature and stream discharge have long been recognized as the key drivers of salmonid migrations (Dahl et al., 2004; Quinn et al., 1997; Rand et al., 2006), few studies have investigated individual variability in response to these environmental drivers. To our knowledge no studies have examined how individuals might respond uniquely to water temperature, but given the influence of river flow on migration speed and successful passage, it is plausible that they respond differently among individuals to river flows that cue migration. Body size likely plays an important role, in which migration of larger individuals is cued at higher river flows. Under a scenario of a falling hydrograph, during which salmonids commonly ascend rivers for spawning (Jonsson et al., 2018; Rand et al., 2006), the migration of larger fish upriver could be cued earlier in the season. Rand and Fukushima (2014) noted a trend of decreasing size of Sakhalin taimen migrating upstream on the basis of sonar observations. However, the earlier arrival of larger fish may be simply due to the greater capability of larger fish to migrate faster (Jonsson et al., 1991).
We found no evidence that coordinated movements and repeatable behaviors of Sakhalin taimen were rooted in social interactions. Although there were comigrating, same-sex pairs both ascending and descending the Karibetsu mainstem, we found no evidence that these pairs were cohesive during tributary entry or departure. However, the slightly smaller number of female pairs spawning in the same tributaries than expected by chance (albeit non-significant) may be indicative of some degree of conspecific repulsion and may help minimize competition for spawning space and thus redd superimposition (Fukushima et al., 1998). It should be emphasized that although the splitting of comigrating individuals into different tributaries could be viewed as straying, between-year consecutive return rates to the same tributaries were high (>80%) and likely a result of fine-tuned homing capabilities (Fukushima & Rand, 2021). The relatively small sample size of tagged individuals as well as the limited spatial extent over which fish movements were examined, relative to the species’ entire migration range, may have made it difficult to identify social interactions in the migratory behavior of this species.
Female and male Sakhalin taimen consistently differed in migration behaviors including seasonal and diel timings of arrival and departure, the degree to which the migration timings were coordinated and repeatable, reliance on environmental cues to initiate migration, the frequency of comigration, and swimming speed. Furthermore, female Sakhalin taimen have a higher rate of consecutive spawning than males (Fukushima & Rand, 2021). All these behavioral differences between sexes point to the hypothesis that female and male Sakhalin taimen may have different degrees of migratory connectivity (Webster et al., 2002). Although our study focused on the spawning migration period, we suspect that migration connectivity would also be an important process during other life-history stages. Migratory connectivity—the extent to which individuals of a migratory population behave in unison (Torniainen et al., 2014)—has been suggested to operate over the entire life cycle in some salmonid populations. For example, individual coded-wire-tagged steelhead released as juveniles at similar times and locations were recovered in the same fishing operation in distant locations in the North Pacific Ocean (McKinnell et al., 1997). Schools of migratory brook trout S. fontinalis contained significantly more kin than expected by chance for periods of up to 4 years (Fraser et al., 2005).
This migratory connectivity can be reinforced as individuals age and become more experienced (Tibblin et al., 2016), or it can be passed on to succeeding generations through learning by recruits or younger individuals from older and more experienced ones (MacCall et al., 2019). Thus, the connectivity can differ between different age-groups and developmental or life-history stages, a phenomenon known as differential migration (Briedis & Bauer, 2018). Differential migration between sexes in salmonids is exemplified by the migration protandry discussed above. Furthermore, a higher likelihood of adopting an anadromous life history in females than in males, a pattern often observed in salmonids with facultative migratory strategies, is an extreme example of sex-specific differential migration (Ferguson et al., 2019; Jonsson & Jonsson, 1993). Although anadromy is not obligatory, and resident populations of Sakhalin taimen exist, depending on river systems (Zimmerman et al., 2012), facultative or partial migration is yet to be reported from any rivers with the species, including the Karibetsu River. Nonetheless, our observations in this and previous studies suggest that even though Sakhalin taimen are broadly categorized as anadromous in this river system, there seems to be sex-specific differential migration in which females likely have stronger migratory connectivity than their male counterparts, enabling the females to better minimize variability in migration timing and form cohesive groups of migrants composed of the same members across years. The causes and consequences of the observed coordinated movements, migratory consistency, and the associated between-sex differences in migratory behaviors in Sakhalin taimen would be better elucidated by further research focusing on migratory connectivity, not just during the breeding stage but also during non-breeding stages of their life.