Our study provides the first direct-tracking evidence that swallows using a fly-and-forage strategy can also incorporate night flights into their migrations, particularly when crossing large, open-water barriers. The use of night flights in species usually characterized as diurnal, fly-and-forage migrants was predicted to occur at areas with reduced foraging opportunities, such as at barrier crossing . This has recently been shown in falcons , but had not been demonstrated for diurnal songbirds such as swallows, that commonly cross ecological barriers [13,14] where some night flight was also predicted to be required to cross such great distances. Our direct tracking of adult purple martins using high precision GPS units support hypotheses for the use of nocturnal flights at barrier crossing by otherwise diurnal migrants  and builds upon other recent direct-tracking evidence demonstrating the combination of day and night flights [e.g.28]. We also demonstrate how these flights are accomplished, wind conditions during night flights over water, and the rates travelled in comparison to daytime and overland flights.
Ten of the eleven purple martins that we tracked used night flight when crossing over water. In some ways this would be expected as birds migrating over open water would not have anywhere to stop for rest , and with a crossing of open water > 1000 km and flight speeds of 19–36 km/h, birds could not complete the cross during 12 hours of daylight. Barrier crossings may necessitate chiefly nocturnal or diurnal migrants to include flights during hours or light regimes where they do not typically migrate. This is because the distance required to cross these barriers, where landing and refueling is usually not possible, requires more than the available light or dark hours to complete. Despite a high prevalence of nocturnal flight in our study, the average migration distance over water was greater during the day as compared to the night. In part, this may be due to night flight being used primarily to finish off the crossing of the ecological barriers (13/19 open-water flights at night were to complete crossings). Similar to our findings for flights extending to cross day-night boundaries, several species of songbird incorporate some daytime flight into their otherwise nocturnal migrations in order to finish the crossing of the expansive migration barrier of the Sahara Desert [3,5,30].
However, we also found that 60% of individuals that used night flight for open water crossings also initiated these crossings at night. Night flights were therefore not only used to complete crossings initiated during daytime as could be predicted based on behavior of diurnal migrants crossing ecological barriers [9,16]. Our results support optimal timing hypotheses for fly-and-forage migrants, where it was predicted that daily travel schedules may shift toward night flights in areas that do not offer good foraging opportunities . Ocean crossing during dark hours may support an otherwise diurnal, fly-and-forage migratory strategy in songbirds. In fly-and-forage, migrants can both migrate and refuel during daylight hours. Aerial insectivores such as purple martins and other swallows may be able to catch insect prey while heading in a general migratory direction or make short stops or detours to accomplish this task while maintaining their migration , or migratory flights may be undertaken in the morning daylight hours, leaving the afternoon for foraging and refueling . Indeed, such a strategy may be evident in Bank swallows (Riparia riparia), where during fall migration travel speeds were slower suggesting they were actively refueling while migrating .
During open-ocean or other barrier crossing, the advantages of being able to forage while migrating may be reduced or eliminated, as aerial insect availability may be limited or absent over open ocean or desert . In this case, selection may favor night flights for barrier crossing in fly-and-forage migrants in the same way that it does in other species that are diurnal foragers, but that migrate at night. This may serve as an advantage particularly over ecological barriers such as the Sahara Desert where birds are at higher risk of evaporative water loss during the day [32,33]. Purple martins may also select barrier crossing flights at night to confer other advantages, such as cooler temperatures, lighter winds, or other environmental factors such as reduced predation that may facilitate open-water crossing at night . The migration speeds we measured for purple martins support this, where rates were highest when they were making open water crossings over the Caribbean Sea (e.g. up to 80 km/h).
Birds that engage in ocean crossing and incorporate night flights may reflect a time minimization strategy , where barrier crossing at the Caribbean Sea or Gulf of Mexico may greatly reduce overall distances travelled, and thus minimize the overall time and energy required to complete spring migration [35,36]. Indeed, the ocean crossings we documented reflect significant ‘short cuts’, as compared to an overland route throughout the same regions (i.e. Tag ID 48042: Gulf of Mexico crossing of 871 km versus around the Gulf route of 937 km). In some species, constraints of day-night light regimes for migration may preclude barrier crossing, as recently demonstrated in a nocturnally migrating nightjar that took the longer-route around the Gulf of Mexico, rather than crossing, which would have saved time and energy but required some daytime flight . Our results show, at least for purple martins, that nocturnal flight does not pose a restriction to large barrier crossing, as it may do for nightjars. It would be valuable for future work to compare and contrast how the constraints of diurnal, fly-and-forage migrants may differ from nocturnal-restricted species that must migrate and forage within dark hours.
In addition to large, open-water flights at night, we observed shorter night over land. However, like night flights over water where we anticipate foraging opportunities to be low, these flights too may also be over areas that offer fewer foraging opportunities . We found that night flights over land were much shorter than over water, suggesting night flights over land are a less used strategy in purple martins. There is also the potential that some of the flights we documented as ‘night’ flights, occurred around sunrise or sunset and therefore could have been completed in the twilight hours. Indeed, this was observed through automated radio tracking  of four swallow species that utilize a fly-and-forage strategy . Our GPS tags were programmed to collect positions around 4am or 6am and 4 pm or 6 pm local time, which across spring migration effectively divides flights between daylight and dark hours. However, flights occurring close to these times could be miscategorized. A positive relationship between ‘nighttime’ flight distances and amount of daylight available during each 12-hour track segment (Supplemental Figure S1b) indicates that some of the flight attributed to nighttime–especially in the shortest overland tracks–likely took place during daylight hours. However, the relatively small amounts of daylight available in each track could not reasonably account for long migratory flights. Further fine-scale tracking with higher sampling frequencies would enable confirmation of whether overland flights during strictly dark hours are used by martins or other swallows.
Incorporating water crossings into migrations between winter sites and breeding sites does not necessarily mean that purple martins were choosing the shortest routes (Table 2). If birds were to select the shortest routes, then those returning to breeding sites north or east of the Gulf of Mexico would benefit from taking shortcuts over the Caribbean Sea, Gulf of Honduras, and Gulf of Mexico, but Texas-breeding birds may actually add distance to their routes by navigating across those water bodies. Other factors may explain the broad-scale tendency to use coastal areas, hence putting the birds in positions that do make water crossings viable (Table 2). Overland routes through Central America contain a considerable amount of high-elevation terrain, so they may choose to avoid flight over mountain passes. For fly-and-forage migrants, low-lying coastal regions may also contain more resources for foraging. If purple martin migratory routes at a broad spatial scale generally favour coastal regions, then water crossings may be used as a tactic to save time and distance (Table 2), although may be avoided when those benefits are limited (Supplemental Fig. S2). When the birds are at the coast, facing a decision to cross the water or circumnavigate, supportive northward winds, combined with anticipated distance savings, may then play an important role in the departure decision. On the other hand, if atmospheric conditions over large water bodies are favourable for long migratory flights, birds may select routes that include water crossings, even if the total migration distance is longer as a result.
Our data indicate that the probability that purple martins will initiate water crossings during northbound spring migration increases with stronger winds flowing in a northward direction (Fig. 3). To a lesser degree, strong southward winds were likely to inhibit water crossings. While we did not have sufficient data to include a day/night distinction in the model, we do not suspect that wind has a varying effect on decision to initiate water crossings during day or night (see annotations of Fig. 3). However, regardless of wind factors, almost all instances of a bird reorienting to circumnavigate occurred during nighttime segments, and most water crossings were initiated during the day, suggesting that initiating water crossing at night is not generally favoured. Recent investigation has revealed that seascapes may provide energetic advantages through a combination of uplift and/or wind conditions (“energy seascape”, [34,37]). While these advantages would apply the most to birds using soaring flight, our results suggest that birds may make crossing decisions based upon the energetics of the seascape. Further investigation of conditions at-sea during day and night would be a valuable area of future investigation.
We found a statistically significant effect of increased moonlight on the probability that martins will initiate water crossings at night (Fig. 4), but we interpret these results with caution, due to a small sample size. Additionally, although the difference in MFI effect between flights over land and flights over water was not statistically significant, we also had a comparatively small sample of water crossings (Fig. 4). Moonlight illuminating the night sky may have contrasting effects when birds fly over land or over water. Over land, a brightly lit sky may put a migrating songbird at risk of being more visible to predators, and a species that normally migrates during daylight hours may be at a distinct disadvantage. Conversely, predation risk may be lower over open water, thus minimizing the risk that bright moonlight would impose. We suggest that further investigation into the influence of moonlight on night flights in typically diurnally-migrating songbirds could clarify a possible relationship.
If purple martins migrate near the coast because of foraging opportunities, then they are in a position to take local shortcuts via water crossings. However, there is a possibility that water crossings play their own role in full, broad-scale route selection. For example, 1) supportive northward winds over water may save birds energy, and make longer routes worthwhile, 2) they can incorporate night flight into the long water crossings where they would not have the opportunity to forage, and 3) water crossings may help to avoid predation, which may be suggested by our preliminary results in relation to moonlight.