Justification for the use of a migratory route as a proxy for the past colonisation pathway
Our data on the migratory route of the Brown Shrike across the ECS is far too limited for population-level inference, thus the universality of the route for the entire L. c. superciliosus remains unclear. Careful interpretation of our tracking data is required because individual variation in migratory routes is recognized (Stanley, MacPherson, Fraser, McKinnon, & Stutchbury, 2012). Migratory routes inferred from the analysis also included large uncertainties (Figure 4) because of the methodological limitations of light-level geolocators (Lisovski et al., 2019). However, Lefranc & Worfolk (1997) stated that, based on direct observation during migration, many L. c. superciliosus fly at least 700 km across the sea from Japan to the coast of China’s Jiangsu Province or the Zhoushan Islands. Our geolocator data concur with their statement. Although further deployment of geolocators is needed, their statement guarantees that the focal sea crossing route is at least one of the major routes for the archipelagic population.
We have taken the recent view that idiosyncratic routes of obligate long-distance migratory species are highly persistent traits throughout their evolutionary histories (Winger et al., 2019) in order to interpret part of our tracking data as the route of past colonisation. However, the dominant view is to consider migratory routes as highly labile traits that are easily subjected to natural selection (Pulido, 2007; Winger et al., 2019). Thus, one could claim that the observed migratory route across the ECS is a result of optimization to the present environmental conditions and does not reflect past colonisation (Alerstam, 2011; Berthold, Helbig, Mohr, & Querner, 1992; Sutherland, 1998; Zink & Gardner, 2017).
For the following two reasons, we believe Winger et al.’s (2019) and our interpretation to be more plausible than the counterview. First, our system does not include any region that experienced severe glaciation. Previous studies have used species that inhabit regions previously covered by ice sheets during the LGM (Haché et al., 2017; Milá, Smith, & Wayne, 2006; Ruegg et al., 2006; Sokolovskis et al., 2018). Reverting from being migratory to becoming sedentary residents was probably a major response to glaciation, which was inferred by SDM analyses (Zink & Gardner, 2017). This implied that migration needed to be labile throughout a species’ evolutionary history and migratory routes should trace the post-LGM expansion to the present breeding range. By contrast, we have shown that suitable breeding habitat for the archipelagic population remained in the Japanese archipelago even during the LGM (Figure 3b-d). This implies that the focal migratory route traces a distribution change before the LGM, supporting our view that idiosyncratic routes over a large barrier have been conserved. Second, similarity between autumn and spring migration across the ECS (Figure 4) may support the conservatism of the focal route. If a migratory route is optimized with the present environment, different selection pressures upon passages in different seasons should independently shape the route (Stanley et al., 2012; Tøttrup et al., 2012). Generally, loop migration, in which spring and autumn migration routes are divergent, is thought to have evolved by retaining one ancestral route retracing colonisation while deriving the other to adjust to current environmental conditions (Newton, 2008). Wind is an important determinant (Alerstam 2001). This loop migration phenomenon has been shown for a population of the Oriental Honey Buzzard Pernis ptilorhynchus breeding in Japan. Its autumn migration involves crossing the ECS whereas its spring migration route detours around the Korean Peninsula to cross a narrow sea channel to enter the Japanese archipelago (Nourani, Yamaguchi, Manda, & Higuchi, 2016). It has been shown that a route over the Tsushima Strait was specifically suitable for spring wind conditions, when wind directions and strengths over the ECS were unstable (Nourani et al., 2016; Yamaguchi, Arisawa, Shimada, & Higuchi, 2012). These conditions would be unfavourable also for passerines such as shrikes (Alerstam 2001; Tøttrup, Pederson, Onrubia, Klassen, & Thorup, 2017).