Discussion

Detection probabilities

The probability of detecting a colour-ringed Whitethroat at our study site, when it was present, was 33%. Our results are consistent with theSylviidae family having relatively lower detection rates than other passerine birds (Johnston et al., 2014; Zwarts & Bijlsma, 2015) and are similar to detection probabilities at their breeding sites in the UK (30%; Johnston et al., 2014). This rate is relatively low when compared to detectability at the non-breeding grounds of other Afro-Palearctic migrants such as Whinchats Saxicola rubetra (63% detection probability; Blackburn & Cresswell, 2016b) and ChiffchaffsPhylloscopus collybita (recapture probability 66%; Catry et al., 2003), though there are few studies that have addressed and calculated detection probabilities during this period (e .g . Zwarts & Bijlsma, 2015). Nevertheless, despite Whitethroats having relatively low detection probabilities, we consider that our high sampling effort (sites were visited at least once a week, for over 20 weeks each year) was sufficient to compensate for this.

Site persistence

Site persistence varied significantly amongst individuals, ranging greatly between one and 165 days. Because of the high sampling effort as well as similar return rates between individuals of different residency categories, our evidence is fairly compelling that shorter stays truly reflect shorter residencies and not detectability issues or mortality. Overall, the mean persistence duration was similar throughout years, so residency dynamics at a species level may not be changing strongly with time. In the Gambia, Whitethroats were also observed to have different degrees of site persistence with 45% of captured individuals remaining in the area between two and 84 days, though most individuals were caught less than a month after ringing (King & Hutchinson, 2001). In Senegal, however, most individuals were on passage (King & Hutchinson, 2001).
Different wintering strategies of individuals at the same site have also been recorded for other long-distance migrants: 27% of BlackcapsSylvia atricapilla in Spain (Belda et al., 2007) and 8% of Chiffchaffs in Portugal were winter residents (Catry et al., 2003), whilst the rest were categorised as transients. This could have several explanations. First, individuals could have genetic differences due to parallel evolution of morphological and behavioural adaptations, making some individuals more inclined to lead either a nomadic or a resident lifestyle (Senar & Borras, 2004). However, we cannot be certain whether individuals that were categorised as “passage” or “unknown” at our study site remained itinerant throughout the season, if they were in fact en route to a stationary non-breeding site elsewhere, or if they died during the period. These genetic differences could also reflect individuals from different breeding populations although this seems highly unlikely because individuals switched strategies across years, and Whitethroats seem to have a somewhat low migratory connectivity (Tapia-Harris et al., n.d.). A second explanation could be due to habitat quality changes throughout the season. The broad residency spectrum, from continual and variable movement to winter residency, appears to reflect a gradient in predictability in food supplies (Newton, 2008). As time passes, habitats dry and resources change (Moreau, 1972), so competition could increase. Some individuals may decide to leave the area to find other more suitable habitats elsewhere, while others may risk staying (for example those that are better competitors or are already in the best territories).
Short-term residencies, during both autumn and spring, may indicate the use of multiple important non-breeding sites. Nigerian Whitethroats deployed with geolocators remained at a first stationary non-breeding site in the Sahel before arriving at our study site in November (Tapia-Harris et al., n.d.). If some of these birds were then to be short-term residents, individuals could have more than two important stationary non-breeding sites. Multiple site use by individuals has been well described for several migrants in the Palearctic systems (McKinnon, Stanley, et al., 2013).
Adult birds’ site persistence was longer than first-year birds likely due to their previous experience and potential dominance. Many first-year birds probably arrive stochastically at non-breeding sites looking for suitable habitats. Many will need to explore the terrain and scout for resources, and whilst some will remain at it until migration to a second site later in the non-breeding season, or back to Europe in the spring, others may continue their search elsewhere, making a relatively immediate migratory scale movement.

Between-years site fidelity

Many long-distance migrants return to the same non-breeding sites year after year both in the Nearctic-Neotropical and Afro-Palearctic systems (Blackburn & Cresswell, 2016; Moreau, 1969; Salewski et al., 2000). Here we found that a minimum of 19% of individuals returned from one year to the next, an intermediate return rate in comparison to other Palearctic migrants in Africa (Kelsey, 1989; Salewski et al., 2000; Thorup et al., 2019) and Whitethroats at their breeding grounds (0–64%; da Prato & da Prato, 1983, 14.5%; Boddy, 1992). Not only did a significant proportion of individuals return the following year, but individuals moved, on average, only 300 m. These results suggest that many individuals have a high degree of between-years site fidelity at a very small spatial scale, though less than Whinchats at the same study site, which moved < 30 m (Blackburn & Cresswell, 2016). Fidelity across years confers the same advantages as longer residency patterns, especially regarding knowledge of local and fluctuating food sources, competitor densities, and location of refuges, and this seems to secure and increase an individual’s survival.
Return rates were different amongst individuals from different age categories: first-years had lower return rates than adults, at least from year2 to year3. First-year birds lack knowledge of small and medium scale locations of where to arrive, thus their first non-breeding period is full of uncertainty and stochasticity (Cresswell, 2014). Some individuals will find a suitable site immediately and remain at it until spring migration, while others will continue their search elsewhere, many of them arriving at less suitable sites or even discovering new unknown suitable habitats. Older birds will tend to reuse non-breeding sites, so becoming more site faithful over time. Some studies have argued that higher return rates in adults could be due to greater survival from the previous year. Though this may be true for a few individuals, we think that because individuals had already undertaken a first migration, when high mortality rates occur, then overall, age-related survival differences are unlikely at this stage.
Even though most of the long-term winter residents remained as such the following year, some remained at the site for less time the following year. We expected individuals with a successful previous experience to repeat the same behaviour in the following years to guarantee their winter survival. When we further investigated these individuals, we found that most of them were seen in spring of the following season. Individuals may have remained longer at their previous wintering site due to better yearly habitat conditions that year and delayed the return to our site. If this were correct, we would expect that many individuals do not return to our study site when conditions are better further north and only return when conditions are suboptimal. On the other hand, passage birds increased the duration of their stay the following year. Given many of them were first-year birds, the lack of experience and dominance would have become a disadvantage, arriving at non-breeding grounds after adults, by which time fewer sites will have been available. By the following year, returning earlier in the season and as adults, they could occupy empty territories that allowed them to remain for longer periods.

Departure dates

Individuals did not seem to change their departure timing from year1 to year2, although further information over a span of many years is needed to draw stronger conclusions. Timing of migration is of critical importance in migratory species and is key for securing fitness (Drent et al., 2003; Kokko, 1999). Departure from the non-breeding grounds has been seen to correlate with arrival at breeding sites (Kristensen et al., 2013; Ouwehand & Both, 2017), though later departing individuals can migrate faster to compensate for lost time (Yohannes et al., 2009). As our study site is located at the southern part of the distribution and individuals have different breeding sites, the first individuals to depart are not necessarily the first to arrive at their respective breeding grounds (Tapia-Harris et al., n.d.).
Individuals at our study site showed lower intra-class correlation departure coefficients (r = 0.15) than most other Afro-Palearctic migrants (range 0.06 – 0.82, Both et al., 2016). Low repeatability fits well with the idea that individuals do not always remain for similar periods across years, and therefore it is perhaps not surprising that some individuals left the area on different dates, but there are a few things to consider with these results. Firstly, repeatability indicates how consistently individuals differ from each other and is not necessarily a measure of individual repeatability across years (Conklin et al., 2013). Secondly, departure from our study site does not necessarily imply that individuals commenced spring departure, they could have moved to another non-breeding site.

Conclusion

Our results are consistent with the serial residency hypothesis (Cresswell, 2014) but also suggest small modifications. Instead of individuals repeatedly settling at the same wintering sites during the same periods year after year, we hypothesise that individuals may vary their timing at the sites depending on yearly conditions. We predict that spatial site fidelity and repeatability are high with little variation across individuals, but temporal fidelity varies individually and according to yearly conditions. Individuals stay stationary at a suitable site and remain there for as long as possible. Some proportion of the population will stay there all the northern winter, most likely those that are better competitors and are at overall higher quality sites and better environmental conditions, whilst others will move to a second site. This might never happen to an individual, and for some individuals this might only happen in their second or third autumn non-breeding season, depending on the specific year’s conditions. The following year the same first site will be revisited; if conditions are stable and the site adequate to secure its survival throughout the period then there would be no need to migrate elsewhere. If not, then it would migrate to the same second site, and so on. When individuals reach the limits of their potential distribution, such as our study site, then they will return only in occasional years of widespread food shortage or suboptimal conditions in the previous sites (Newton, 2008). We would therefore predict that apparent survival at a sub-Saharan non-breeding site will correlate positively with latitude. Some individuals may get unlucky and will need to change sites several times, which makes them appear itinerant, whilst others will remain as long-term winter residents at few sites. Regardless of any particular year’s conditions, the number of migration steps and distance of migration is minimised and matched to environmental conditions.
In summary, results suggest that spatial fidelity is high and constant through years, but temporal use or temporal fidelity and site persistence may vary and a possible explanation for this might be variation in yearly and seasonal conditions. In other words, timing is important: individuals revisit locations at very precise scales but do not necessarily repeat them at the same time. There is temporal flexibility but not spatial flexibility, except in the sense there is always an option to make a potentially dangerous further migration to an unknown area if conditions became untenable. But these hypotheses can only be tested fully when small birds such as Whitethroats can be tracked with non-archival tags so that wintering locations regardless of site fidelity and long-term survival can be seen.