Data interpretation

The GPS-transmitters directly provide information on locations and speed. If instantaneous speed = 0 m/s birds were not flying. Other low speeds, which indicate e.g. walking, did not occur in the dataset. The number of logged GPS locations is strongly influenced by the battery load of the device. In cloudy circumstances or if feathers cover the solar panel the battery is insufficiently charged, resulting in gaps in information. For ARGOS-transmitters, birds were assumed resting if two subsequent positions were received within 10 km within at least a one-hour interval. If the transmitter did not move from its exact position for more than two days birds were assumed dead or the transmitter lost. The bird was also assumed to be dead if the behaviour before the last position pointed to a bad condition of the bird such as slow flight (< 10 km/h), regular stops in hostile environment (desert), and in some of our cases the presence of sandstorms just before the last positions (based on the Royal Netherlands Meteorological Institute weather reports and Google Earthcloud coverage).

We defined the ‘start of migration’ as the moment when birds actually took flight and started southward movements, so excluding the pre-migratory fattening period. If the exact onset of migration could not be recorded, the departure date was estimated based on average travel speed assuming an evening departure around sunset, a period with a distinct peak of migration onset (van den Bosch & Vlek 2006; van der Winden et al., unpubl.). All of the studied herons moved to the region south of the Sahara and as this is the nearest regular wintering habitat (Cramp & Simmons 1977), we defined the ‘end of migration’ as the moment the herons reached this area and stopped flight movements for more than one day.

Instantaneous speed (km/h) was directly measured for birds with a GPS-transmitter. In addition, for these herons as well as herons with a ARGOS-transmitter, hourly travel speed (km/h) was derived by calculating distances between fixes during flight according to the shortest path between points (great circle distance). All fixes with high quality 1, 2 and 3 were used for this, as well as low quality fixes A and B if they coincided with the previous migration direction and speed based on high quality fixes. Since most of the distances were relatively short and predominantly north-south in orientation the differences with loxodrome distances (distances with a constant course on a globe) were minor. Migration speed (expressed as average n km/day) is defined as the speed from start to end of migration, including stops.

Figure 1.

Migration routes of seven Purple Herons with satellite transmitters migrating from The Netherlands to Africa after breeding in 2007. The symbols distinguish between active flight, diurnal stopover or unknown (ARGOS-fixes, 1,2,3, A and B). ID 33995, 33998 and 34001 died in dust storms in the desert. Dotted line indicates dispersal movement of juveniles (A and B). Map in Equidistant Conic Projection with Africa as central median to prevent north-south distortion.

Fast travellers

Seven herons succeeded in covering at least 3500 km (Fig. 1, Table 1). All but one bird showed a fast flight covering the route to the Sahel in 5–7 days (with an average migration speed of 699 km/day). The average instantaneous ground speed of three migrants was 45.9 km/h with a maximum of 80.0 km/h (n = 35) (Table 2). Instantaneous speed did not differ from travel speed (Table 2). Also the travel speed did not differ between day and night except for heron ID 33999 (Welch's t-test, t = 5.537, df = 6.3, P = 0.0013), which faced a strong headwind in Mauritania ( http://www.wetterzentrale.de). Two adults died in the Sahara during sandstorms. Both birds slowed down flight speed significantly in a desert storm (confirmed by weather satellite pictures of the Royal Netherlands Meteorological Institute) and stopped moving in an extremely hostile environment (sand dunes). One other adult also slowed down flight speed substantially in the desert and stopped in a hostile environment for unclear reasons. One juvenile (ID 34002) flew in a westerly direction and ended up in the Mid-Atlantic Ocean, c. 2000 km west of Senegal and c. 1800 km northeast from the Brazilian coast, after a 5600 km non-stop flight. Based on the high average traject speed (48.3 km/h) and because of its continuous route over sea, stops were presumed impossible for this individual. Finally, we assume that it drifted for one day on the water surface (travel speed less than 2.4 km/h for at least 7 hours) and after this, the signals stopped. Lastly, two adults and one juvenile successfully reached the African wintering areas directly south of the Sahara.

Table 2.

Diurnal and nocturnal flight speeds of three Purple Herons flying over Africa in 2007. Presented is instantaneous speed measured by the GPS satellite transmitter and calculated travel speed for the same transects for which GPS fixes were obtained. The diurnal and nocturnal flight speeds can be compared with the travel speed of bird 34002 (ARGOS-transmitter, so only travel speed available), which was flying continuously over water and could not stop. (n.d.) no data were available (no GPS).

Short diurnal stops

Due to problems with the batteries, GPS data were unfortunately very scarce, especially in Europe, resulting in little information on stopover time and habitat use. However, there were indications of short diurnal stops on a daily basis in Europe and North Africa (Fig. 1). For example, for one bird (ID 34001) the transmitter provided regular signals during a part of the journey from North Africa southwards (Fig. 1). This bird was flying over the Mediterranean north of Morocco around 10:00 on 1 Sept 2007. At 12:00 noon it landed in a river valley where it stayed until 16:00. This proved a short stop of about 4 hrs. At 18:00 it was flying 64 km west of this stopover point. At 10:00 on 2 Sept the bird landed again for a diurnal stop at a river valley for about 11 hrs and in the evening it took off for the flight over the Sahara.

Few other indications of stops in North Africa were obtained (Fig. 1); these stops were also diurnal. Only for one bird one stop was recorded in the Sahara (ID 34004), all other information pointed to non-stop crossings of the Sahara, based on the time of passage south of the Atlas Mountains, the time at the end position of the route in the Sahara or Sahel, and the average recorded GPS and travel speeds of around 45 and 43 km/h, respectively (Table 2). Because GPS fixes were lacking during parts of the journeys, especially in Europe, total travelling time (Table 1) could be used to reconstruct a daily stopover time based on average travel speeds and migration speed (Table 2). For four birds the non-stop crossings of the Sahara took on average 1.4 days, meaning that on average 4.3 days was spent on migration in Europe and North Africa. Assuming the same average travel speed of around 45 km/h during this part of the journey, the herons must have stopped for about 47% of the time. Although GPS and ARGOS signals were scarce in Europe, the timing of passage of the birds indicates that this stopover time was evenly divided over the travel days, meaning mainly nocturnal migration alternated with diurnal stops taking less than the entire daylight period.

One adult heron (ID 33999) behaved differently to the other six herons in respect to its suspended migration halfway in Morocco. This bird encountered the same severe weather conditions in the Atlas region in which two adult birds died. However, this individual avoided the bad weather by returning north to the coast of Morocco where it suspended migration for 33 days. After this stopover it continued migration over the Atlas and Sahara non-stop like the other herons.

A fast travel to the Sahel

Our results show a yet undescribed and unexpectedly fast travel for Purple Herons, without a long stopover en route, from the breeding areas to the Sahel. In Europe and North Africa the herons flew throughout the night and part of the day, whereas the Sahara was crossed non-stop. Moreover, they travelled in continuous flapping flight and only short stops were made. Excluding the pre-migratory fattening period, the Purple Herons completed their journey of 3500–4000 km in just 5–7 days. This makes Purple Herons the fastest heavy migrants known so far covering such distance. Other large species like geese, swans, storks or raptors need 35–50 days to cover similar distances. Geese and swans need at least one long stopover en route (comparison with exclusion of the pre-migratory fattening period), while storks and raptors need thermals for soaring and can only cover about 450 km/day and therefore their migration speed is relatively slow (Meyburg et al. 2004, Shamoun-Baranes et al. 2003, Trierweiler et al. 2007, Newton 2008, Klaassen et al. 2008). Comparing Purple Herons to the Osprey Pandion haliaetus, another piscivorous species with a similar migration route, highlights the difference in travel strategy. Ospreys are slower than Purple Herons with an average travel speed of about 260 km/day, with a range of ground speeds of 17–47 km/h (Kjellén et al. 2001). The travel speed of Purple Herons, however, is lower than that of geese, who also use flapping flight and, in the case of Brent Geese Branta bernicla spp, can fly up to 1300–1900 km within 24 hrs (Dau 1992 in Newton 2008) compared to the close to 700 km per 24 hrs found for the herons in our study. This difference is mainly caused by the short diurnal stops made by the herons, and also by a lower flight speed. But as the Purple Herons flight:stopover ratio is 1:1 (excluding the pre-migratory fattening period) over a distance of 3500–4000 km, it results in a quicker overall journey than for Brent Geese covering a similar distance. Light-bellied Brent Geese B. b. hrota migrating from Greenland to the UK stop for at least two weeks in western Iceland in spring and autumn (Madsen et al. 1999) and Dark-bellied Brent Geese B. b. bernicla, which migrate 5000 km from the Wadden Sea towards northern Russia, stop also halfway, resulting in a flight:stopover ratio of 1:13 (without the pre-migratory fattening period) (Green et al. 2002). If pre-migratory fattening is included, these ratios are about 1:11 for Purple Herons and 1:25 for Brent Geese.

In this sense, the Purple Heron migration strategy resembles those of smaller long-distance migrant birds like Knot Calidris canutus or Bar-tailed Godwit Limosa lapponica, with few or no stopovers at travel distances of 4000 km or more, resulting in a fast travel (e.g. Piersma 1987, Gill et al. 2008). Such long-distance flights can only be accomplished by fattening substantially prior to migration. Based on the empirical data for Purple Heron, and the flight distance model of Pennycuick (1989, 1998), a fat mass of 0.217 kg (24% body mass) is estimated to be necessary to reach the winter quarters at a distance of 5000 km (Table 3). The assumed 24% is high for a large bird, so it is likely that the flight distance is more constrained by this, probably in the range of 3500–4000 km as was observed in six of the seven tracked birds. Further investigations should be undertaken to get more information on the pre-migratory strategy of the Purple Herons. But it is unlikely that the 3–4 short diurnal stops en route during migration might be sufficient to store enough energy for such a flight. Moreover, indications for feeding during the (mostly) diurnal, relatively short stops are lacking.

Table 3.

Flight characteristics and predicted fat fraction required for an assumed non-refuelling flight of 5000 km for Purple Heron (Pennycuick 2007, Flight model 1.18). Fat fraction is fat mass/body mass at departure.

Figure 2.

The relation between body mass and wing load in long-distance migrants bird species subdivided in birds with known strategies: active flapping flight with pre-migratory fuelling, (partly) soaring flights or active flapping flight with long en route refuelling stops. Data from Alerstam et al. (2007).

Although many long-distance records of migrating birds using flapping flight are known, the ability of Purple Herons to use this migration strategy despite their large size may be a result of a relative low wing load compared to other large birds. So far, bird species with a maximum wing load of 4–5 kg/m² are known to be able to perform long-distance flights without long stopovers (Fig. 2). Covering long distances in flapping flight without prolonged stopovers is probably only just feasible for Purple Herons, given their size and wing loading (Fig. 2). In this regard, the even heavier Grey Heron Ardea cinerea and Great Blue Heron A. herodias may constitute the very limit of long-distance migration without the need for long stopovers. Accordingly, herons are potentially good colonisers, as they are capable of passing areas without refuelling opportunities and thermals, such as oceans. Our finding of the juvenile bird flying 5600 km non-stop supports this idea. The colonisation of the Americas by the Cattle Egret Bubulcus ibis from Europe further illustrates this point (Del Hoyo et al. 1992, Kushlan & Hafner 2000).

The question can be raised why herons make this long-distance migration without making long stopovers to restore reserves under way like other large birds? Perhaps migrant herons were forced into this strategy as they face insufficient and unpredictable habitat availability along their migration route in the Mediterranean and North Africa; such as during periods of summer droughts that result in wetlands of decreased size and thus increased competition with resident heron populations. This might have selected for a long pre-fattening period and a high migration speed rather than for a stopover strategy en route. The necessity of serious refuelling of at least a week or more in these regions might be risky.

Figure 3.

(A) The number of Purple Herons starting migration at a bird observatory at Ridderkerk in 2007 registered by visual counts ( http://home.tiscali.nl/elzerman/vwgpureiger2.htm) in relation to the wind direction in the lower air layer. Triangles indicate days when also the herons with satellite transmitters started migration. (B) Wind directions (%) in the period 15 August to 5 October 2007 corresponding with Purple Heron departure period. Surface wind direction based on daily average surface wind direction for De Bilt, The Netherlands ( www.knmi.nl).

Migration not possible without tailwinds?

Favourable tailwinds are essential for large long-distance migrants using active flight, especially at take-off when fat load is high (Alerstam 1979, Liechti 2006, Gill et al. 2008). Indeed, all Purple Herons, with or without transmitters started southward migration in evenings with favourable northerly tailwinds. At such evenings peak departures of hundreds of Purple Herons were registered at a bird observatory near Ridderkerk in The Netherlands (van den Bosch & Vlek 2006) (Fig. 3). This suggests that at least a convenient start for the herons might be important from an energetic perspective.

Conservation implications

This unexpected migration strategy shows the fragile balance between the choice for a high travel speed and the need for high quality premigratory fattening sites in the breeding range, and disturbance and hunting free sites in Spain and Morocco.

From a conservation perspective our findings imply that the quality of habitat (abundant food and rest) both in the departure areas and in the Sahel might be of crucial importance for replenishing reserves. In the nineteen-seventies, a severe dry period in the Sahel was clearly linked with a decrease in the Dutch Purple Heron population as a result of increased adult mortality (Den Held 1981). So far, the decline in surface area of foraging habitat in dry periods was assumed to have been the critical factor in adult mortality, although this was conflicting with the presence of abundant suitable wintering habitat further south in the tropical West African regions, where Purple Herons also commonly winter (Den Held 1981, Zwarts et al. 2009). But why is mortality high in the Sahel if abundant feeding habitat is present more to the south? We propose that the observed vicissitudes of mortality are better explained by a change in suitable habitat for arriving and (presumably) exhausted Purple Herons at the southern edge of the Sahara. Especially if, for this large species, the covered distances reported here are close to their physiological limit and particularly if wind conditions are not favourable throughout. If this is true, most fatalities of migrants might be expected at the far end of the route, which was the case for three of the studied herons. Although the reason for the fatalities is not known, severe weather occurred in two cases. It cannot be excluded that the transmitters hindered additionally, although other herons migrated successfully over the Sahara. At this stage no information is available if our relatively high loss of 3 adults in the Sahara was accidental or not. The study results might also explain the paradox of the correlation of winter mortality with Sahel droughts while substantial numbers of Dutch Purple Herons winter further south in areas without droughts. It will be very interesting to test this hypothesis in more detail and for more species.