For the nine male nightingales in our study we found the timing of only one migration event namely arrival at the breeding site to be significantly linked to NDVI phenology there. In contrast, departure from non-breeding sites was only very weakly linked to local NDVI phenology. All other migration events such as arrival or departure from stopover sites were not related to any of the local phenology measures we tested. These results stem from only nine individuals and future studies with larger sample sizes are required to confirm our findings.
Our results are in contrast to the expectations and suggest that male nightingales used other cues for the departure decisions from the non-breeding sites. For the departure from the non-breeding sites, photoperiod might be the most important trigger. Indeed, photoperiod has been identified as a universal cue triggering (preparations for) migration in the majority of migratory species across all taxa, and even marginal changes of photoperiod can be sufficient to trigger preparation for migration in migrants wintering at low-latitudes. The fact that departures from non-breeding sites in our study showed the lowest variability also suggests a prominent role of photoperiod for the onset of migration. This also points out to a potential threat: if climatic changes differ in magnitude and direction between the various places migratory birds use, birds relying on the invariant photoperiod might mistime various activities in their annual cycle. For instance, departing too late from the non-breeding areas can have cascading effects, with mistimed arrival in the breeding grounds, a large mismatch between peak food availability and peak food requirement of offspring, and thus, negative fitness consequences. Hence, relying only on photoperiod for timing of migration might become an ecological trap and has been discussed as one major factor for the declines of many species within the Palaearctic-African migration system. It has sometimes been argued that migrants could catch up in the course of migration even if departure from the non-breeding site was non-optimal (usually: too late), e.g. by changing migration route or speed, but often migration strategy remained unaffected by climate trends.
Surprisingly, we also found no behavioural response to environmental conditions en route. Arrival at and departure from stopovers were unrelated to local phenology suggesting that birds used stop-over sites regardless of the progress of vegetation phenology at the focal site. This may be explained by a) differences in quality of sites and resulting differences in fuelling rates, b) carry-over effects of site-quality/fuelling rates in earlier sites, c) methodological issues inherent to geolocation and our assumed relation between NDVI and insect abundance (see below).
Accordingly, we infer, that fuel deposition rate diverged widely among the different individuals or stopover sites they visited. Fuel deposition rate depends on many factors, e.g. food abundance and accessibility, weather, predation risk as well as competition and individual state. Furthermore, we found that arrival at the breeding sites was related to stopover duration, i.e. individuals that stayed shorter at stopover sites also arrived earlier at their breeding sites, regardless of total migration duration. Although it remains elusive at this stage why staging times at stopover sites differ between individual nightingales, there are several non-exclusive explanations: First, the quality of stopover sites in terms of re-fuelling could differ substantially and individuals in “high-quality” sites may rapidly replenish their body reserves in preparation for the next migratory step. Second, individual condition on arrival at a stopover site might be co-determined by the quality of preceding sites, which affects their requirements at the present site (carry-over effects, see). If a spatial autocorrelation in quality between successive sites exists, migrants might adjust the timing of migratory progression according to conditions at sites ahead. Unfortunately, our results are inconclusive here – we found no significant relation between absolute NDVI data and individual differences in four measures of timing (e.g. the date of arrival at the breeding site, the time difference between arrival at the breeding site and local spring green-up, the time difference between the arrival at the breeding site and the date when the insects’ base development threshold was reached, and the period between the end of offspring’s peak food requirement and the end of larvae availability) suggesting the latter not being explained by differences in productivity between stopover sites.
Moreover, we found that arrival at breeding sites was related to proxies of food availability for adults and their offspring. Birds arriving temporally close to the onset of insect availability also enjoyed a long time period of high larvae availability after the offspring’s peak food requirement, which implies that covering the food requirements of arriving males and offspring is not a mutually exclusive task. The timing of arrival of adult birds in the breeding grounds can importantly influence the fate of their offspring and thus, their reproductive success.
For all individuals of our study populations, the periods of high availability of larvae covered the periods of high energy requirements, suggesting no apparent mismatch between food supply and demand. Possibly, this is facilitated by wide foraging niche and a seasonally flexible foraging strategy in nightingales. However, in dietary specialist species, like the pied flycatchers, significant mismatches between food supply and demand have been found recently.
Another important finding in our study is the link between the departure decision at the non-breeding sites and the matching of the spring green-up at the respective breeding sites about 3800 km away. Individuals, who left their non-breeding sites early, matched spring green-up at the breeding site more accurately. Even if departure from Africa is mainly driven by photoperiod, local conditions potentially fine-tune the decision, but we failed to verify this link. However, this would enhance the relevance of phenological trends at the non-breeding sites and their potential implications for demographic rates (survival, reproductive success) at later times at sites far away.
There are several methodological issues that potentially have confounded our findings – the accuracy of light-based positioning and the relation between NDVI data and higher-level productivity that we have implicitly assumed: Light-based positions naturally have a relatively large inaccuracy, e.g. compared to GPS positions, especially in woodland species. Therefore, the sites identified often comprise large areas, possibly including unsuitable habitats. Although we have explicitly excluded unsuitable habitats (desert, bare or sparsely vegetated ground) and thus, reduced this source of error considerably, our method is necessarily relatively coarse and might still confound finer-scale patterns. Many studies investigating the reliance of migrants on environmental conditions used NDVI data as a potential cue for timing (e.g.) or as a proxy for food abundance in non-herbivorous species, i.e. species foraging on higher trophic levels (e.g.). Although NDVI and primary production have been related explicitly[28, 54], its relation to productivity at higher trophic levels requires further specification (e.g.). By using temperature data stemming from an atmospheric model for modelling insect phenology and by pooling the thermal requirements of several insect taxa, we cannot predict small-scale variations in resource phenology. However, the approach allows for modelling the general insect abundance as for food for adults and offspring in insectivorous birds across large areas where field data are often impossible to achieve.