Bird migration is defined as these animals’ regular and seasonal movements (Bernis, 1966; Alerstam, 1990; Dingle, 2014). These voyages are undertaken by several species to face climate and resource seasonality (Alerstam, 1990) and are legally protected through several European initiatives, such as the Bird’s Directive (79/409/EU) and the Bonn’s Convention (82/461/EU).

Due to their flight ability and efficiency, migrations are widespread among birds (Newton, 2008; Dingle, 2014). Approximately half of the world’s bird species migrate annually (Newton, 2008; Cox, 2010), and specifically 50 to 100% of the ones living at mid-latitudes and polar environments do so (Somveille et al., 2013). These migrations typically follow north-south axis routes along temperature and primary production gradients, among other factors (Alerstam, 1990; Newton, 2008). There are two annual migrations: prenuptial — generally from S to N, during spring — and post-breeding — from N to S, during late summer and autumn (Alerstam, 1990).

Birds fly by either gliding or flapping their wings. This has produced a selection for structures and body systems (Alerstam, 1990; Berthold, 2001; Newton, 2008; Pennycuick, 1975, 2008). Gliding birds usually have a large wing surface and depend on airflows formed on the earth’s surface to move with a small energetic expenditure (Onrubia, 2015). Several birds use upward flow to rise at a great speed in characteristic circular flights (Kerlinger, 1989; Newton, 2008). Because gliders use these airflows, these species concentrate their migration during the central hours of the day, when said airflows are available (Bildstein, 2006; Newton, 2008; Onrubia, 2015).

Migration between Africa and Europe corresponds with the Palearctic-African Migratory System, with 2,100-5,000 million birds of 215 species migrating annually (Moreau, 1972; Newton, 2008; Hahn et al., 2009). Important geographic barriers such as the Alps and the Mediterranean Sea largely condition these migrations, which has led to the formation of the eastern and western migration corridors. The later crosses the Iberian Peninsula and other western Mediterranean passes (Moreau, 1972; Mead, 1983; Berthold, 2001; Elphick, 2007; Newton; 2008; Filippi-Codaccioni et al., 2010), where several monitoring projects are being carried out (SEO/Birdlife 2000; Pérez-Tris & Santos, 2004; Filippi-Codaccioni et al., 2010; Isturiz et al., 2022).

The migratory phenology in the Iberian Peninsula has been previously studied (Filippi-Codaccioni et al., 2010; Onrubia, 2015; Isturiz et al., 2022), but what remains largely unexplored is the potential influence of meteorology on this phenomenon (Isturiz et al., 2022).

In this context, the aim of our work is to describe the post-breeding visible migration behaviour (below 400 m of height) of two gliding species (European Honey Buzzard Pernis apivorus and Black Kite Milvus migrans) and two flapping species (Common House Martin Delichon urbicum and Barn Swallow Hirundo rustica), and to study the effect of wind direction and speed on their passing through the Pyrenees. Non-visible migration was not considered (over 800 m of height). These species were selected for their abundance considering their taxonomic group allowing statistical analysis, and because they represent different flying strategies.

The Pyrenees are a mountain range located along the northern Iberian Peninsula. They represent the border between France and Spain and integrate the totality of Andorra, reaching elevations above 3000 m. The field work was carried out at two mountain passes of the Western Pyrenees: Somport and Portalet. Somport mountain pass (42º 47’ 44,79” N, 0º 31’ 33,08” W), is oriented NE-SO and has an elevation of 1,632 m. According to the Köppen (1900) classification, the climate is “Dfc, without a dry season and with a fresh summer”. During field work, weather was stable during summer, except for some storms, and rainy, cloudy, and windy in autumn. Fog was frequent due to the narrowness of the pass. The habitat is composed of Scots Pine Pinus sylvestris, Mountain Pine Pinus uncinata and Beech Fagus sylvatica in the lower areas and open subalpine grasslands in the higher ones. The Portalet mountain pass (42º 48’ 21, 72” N, 0º 25’ 6,35” W) has as N-S orientation and an elevation of 1,795 m. The climate is the same as in Somport, and so was the meteorology during field work, except for less fog.

Counts were performed simultaneously at both mountain passes from 7:00 to 17:00 h, between July 15^th and November 15^th of 2016 and 2017. At each mountain pass, five experienced ornithologists including one of us (Pablo Monroy) took individual turns to identify and count the passing birds using binoculars and spotting scopes. The following information was recorded: date, time, species, genus or taxonomic group, behaviour (migratory or non-migratory), number of birds per flock, age, and sex when possible.

The following meteorological parameters were recorded at the end of every census hour: wind direction (N, NE, E, SE, S, SE, W, NW), speed (km h-1), and temperature (ºC), the latter two with an anemometer and a thermometer, respectively.

To study the effect of wind speed and direction on bird migration we used Somport 2017 data (AEMET https://www.aemet.es/es/portada) with 120 days and 1,023 h of field work. For the analysis of each species’ behaviour, we only considered the ten days with a crossing over 5 % of total annual data. Wind effect was analyzed for the two mountain passes exclusively, not for Western or total Pyrenees.

Wind effect was studied using a Generalized Linear Model (GLM), that allows the dependent variable to have a non-normal distribution. Specifically, we used a log-lineal model, which is adequate when the dependent variable is count data, as is the case. When data showed overdispersion (variance larger than mean), we used the Negative Binomial regression (Doménech & Navarro, 2005; Karlsson, 2014; Alcaide, 2015). We used SPSS statistical package.

The dependent variable was the number of birds of a given species per hour. The independent variables were wind speed (0, 0-10 km h^-1; 1, 11-20 km h^-1; 2, 21-30 km h^-1 and 3 >30 km h^-1), wind direction (N-NE, E-SE, S-SO, O-NO) and T in ºC.

A total of 186,338 migrating birds were counted, 105,496 (56.6 %) at Somport and 80,842 (43.4 %) at Portalet.

Twenty-two percent of the detected birds belonged to gliding species and 78 % to flapping ones. Among the gliders, 15.9 % were European honey buzzards and 20.1 % were black kites. Among the flapping species, 23.7 % were common house martins and 16.2 % were barn swallows (Table 1). Migration was not constant throughout the day and increased progressively from 9 to 11 h and then decreased until practically zero after 18 h. (Figure 1).

The results of this work reveal the importance of the Portalet and Somport mountain passes during post-nuptial migration. Despite the fact that the number of birds to have crossed these passes is notably less than that recorded in other Pyrenean passes further west such as the Organbidexka (Filippi-Codaccioni et al., 2010) and Lindus (Isturiz et al., 2022) passes, more than 90,000 specimens per year constitutes a very high number, for which it seems necessary to establish an annual monitoring at the Portalet and Somport passes.

It has been possible to confirm that the start and end dates of migration are similar to those recorded in the westernmost Pyrenean mountain passes (Filippi-Codaccioni et al., 2010; Isturiz et al., 2022), while the migration in the Strait of Gibraltar occurs some weeks later (Onrubia, 2015).

The apparently longer seasonal migration period detected in our work, compared to that of the Strait of Gibraltar, could possibly be related to a broader use of thermal flows, given their relevance in mountain environments (Agostini & Logozzo, 1997; Panuccio et al., 2013).

The four species show different migration time patterns, as it also happens in the Alps (Agostini & Logozzo, 1997) or Gibraltar (Onrubia et al., 2009; Panuccio et al., 2013; Onrubia, 2015).

Gliding and flapping birds do not show a common time migration pattern, except for the number of passing days. Gliders pass in fewer days, maybe because they do it faster, even though there are species differences.

The Black Kite shows a similar time pattern as in Gibraltar (Panuccio et al., 2013; Onrubia, 2015). The European Honey Buzzard has a similar time pattern as in the Alps (Agostini & Logozzo, 1997) and Gibraltar (Onrubia, 2015). This may be because both are large gliders with high flight speed.

Both flapping birds show different time patterns. The Common House Martin has a different time pattern compared with Gibraltar (Onrubia et al., 2009), which happens a month later. This species is small, has a less efficient glide. The Barn Swallow shows a similar pattern as the one observed in Gibraltar, with few days of delay to cross the Iberian Peninsula (Onrubia et al., 2009).

Gliders cross the Pyrenees during fewer days than flapping birds, as it also happens in the Alps and Gibraltar (Agostini & Logozzo, 1997; Onrubia et al., 2009; Panuccio et al., 2013; Onrubia, 2015). This may be due to their larger size and the use of thermal flows, which facilitate their migration by improving speed and energy saving (Kerlinger, 1989; Newton, 2008; Onrubia, 2015).

The concentration of sightings of both glider species during midday is possibly due to their use of thermal flows, which are produced more easily at that time of the day (Bildstein, 2006; Newton, 2008; Onrubia, 2015). However, the European honey buzzard does not show this pattern in Gibraltar, where the peak is in the early morning (Onrubia, 2015). Again, this could be facilitated by the effect of thermal flows, which help them cross the Pyrenees, as happens in Gibraltar to cross the sea (Onrubia, 2015).

The flapping birds weren’t found to concentrate their crossing in the midday, as previously described (Alerstam, 1990; Berthold, 2001; Newton, 2008; Pennycuick, 1975, 2008). They seem not to depend on thermal flows to cross the Pyrenees.

Meteorological factors do not seem to make a clear difference for gliders and flapping birds. The factor that affected the European honey buzzard most was wind direction, as this species seems to avoid N-NE winds (Agostini, 1992), chooses wind directions different to its flight direction and can fly with lateral wind (Agostini, 1992). The black kite choses strong winds, skilfully flying with the worst winds, probably using thermal flows (Agostini & Logozzo, 1997). This choice could be the reason to select lateral E-SE wind rather than a NE tailwind. The meteorology associated with E-SE winds are Atlantic storms coming from the N and anticyclones from the S. Both can produce thermal flows and southward winds that may aid migration (García De Pedraza, 1972) and establish a positive association with T, again related with the use of thermal flows.

Regarding the flapping birds, both studied species show different patterns. The common house martin’s avoidance of strong winds is probably due to its small size and inability to use these winds. This species also shows a positive association with temperature, as it uses a wide daily timeframe. On the other hand, the barn swallow selects moderate winds, which could be interpreted as a better flight ability with stronger winds due to its larger size. This species chooses N-NE tailwind and avoids high temperatures, which is probably why it crosses during the first hours of the day.

Finally, analysing wind effect on bird migration exclusively in these two mountain passes could have some limitations regarding this effect in Western Pyrenees. A simultaneous monitoring in this whole area would allow to have more consistent data and consequently sound interpretation.