Introduction

The Purple Heron Ardea purpurea is a colonial migratory species, which is highly sensitive to disturbance and breeds predominantly in reed marshes. Because reed marshes are scarce habitats, the breeding distribution of Purple Herons is very patchy and the species is considered one of the most vulnerable herons in Europe and the Mediterranean region (Hafner 2000, Marion et al. 2000). In France, most individuals breed on the Mediterranean coast where the population size was about 800 pairs in 2004 (B. Poulin & Y. Kayser, unpubl. data). Aerial surveys revealed large annual fluctuations in numbers breeding in the Camargue, with a steep decline from 1982 to 1994 and a further decline between 1996 and 2000 (Hafner et al. 2004). Anthropogenic changes of the breeding sites, such as reed harvesting and water management practices, have led to low water levels in spring and probably explain part of the observed inter-annual variation (Barbraud et al. 2002).

Because the Purple Heron is migratory, its population size may also be affected by conditions encountered in wintering areas or at stopover places during migration. Several studies showed that the breeding populations of Purple Herons in Europe are influenced by the amount of rainfall in sub-Saharan Africa (Den Held 1980, Cavé 1983, Barbraud & Hafner 2001), but considerations of exactly how wintering conditions affect survival is hampered by our scant knowledge of their distribution and habitat on wintering grounds. Thanks to ring recoveries, the French populations are believed to use three different migratory routes (i.e. south-west towards Spain, south-east to Italy, and eastwards along the Po River towards the Adriatic Sea), and to predominantly winter south of 10°N in West Africa (Voisin 1996). However, further research is needed on the distribution, habitat use, winter mortality, and environmental limiting factors in the wintering range (Kushlan & Hancock 2005).

To gather complementary information on the stopover places and wintering quarters of Purple Herons from southern France, we equipped six juvenile birds with Argos transmitters during the spring 2004. We were also aiming to determine whether juvenile birds migrate back to Europe in spring or spend their first summer in Africa, as has been suspected (Voisin 1996).

Methods

Satellite transmitters (numbered 942 to 947) were fitted on 2 July 2004 on six juvenile Purple Herons from a single breeding colony in the Camargue area (43°30′N, 4°30′E). In order to minimize disturbance at the breeding site, the colony size (42 breeding pairs) and the approximate laying date had been estimated beforehand using aerial photographs, as described previously (Kayser et al. 1994, Moser 1986). The birds were captured on the nest and released within 10 minutes, so that the time spent in the heronry did not exceed one hour.

We used 30-g platform transmitter terminals or PTTs (Microwave Telemetry Inc., PTT 100) that we attached as backpacks using flexible Teflon ribbon harnesses. Because we did not plan to recapture the birds, a breakaway link made of biodegradable surgical wire had previously been sewed on each harness, so that the birds would lose the transmitter and the harness after a few months. We scheduled each PTT to transmit for 8h every 1–8 days, with more frequent sampling during times of the year when we expected most long-range movements.

The PTTs were tracked by CLS/Service Argos in Toulouse, France. Locations provided by the Argos system are divided into different classes (labeled A, B, Z and 0–3) depending on validation, number of messages received and location accuracy. The accuracy is supposed to be within 1 km for three classes of locations (1–3), while the accuracy for 0, A, B and Z is unspecified. The high-quality locations represented 13% of our data set, which is close to what was found in other studies (Hake et al. 2001). Unless specified, only high-quality locations were used.

Results

Of the six juvenile birds equipped, one faced transmitter failure or died within the first week (943), and a second stopped transmitting in early September (944). The four remaining birds successfully started migration (Fig. 1). One of them (946) seemed to have left the Camargue in late August and spent some time in south-west France near the Spanish border before flying to Algeria in mid-September. The corresponding location classes were low (A and B), which means that less than four signals were received by the satellites, but it is very likely that the locations registered were accurate (Hays et al. 2001). Unlike PTTs 942 and 946, location accuracy was high for PTTs 945 and 947 and allowed to assess migration departure dates: PTT 945 and 947 were last detected in the Camargue on 12 and 14 September, and were reported in the Mediterranean Sea north of Oran and Alger on 14 and 16 September. All four birds travelled south-south-west across the Mediterranean Sea, flying over the Balearic Islands (942 and 947) before reaching the Algerian coast near Cherchell, where PTT 945 was registered near the Damous river in Algeria. Then, they pursued their migration south-west towards Morocco or Mauritania (Fig. 1). In Morocco, one bird followed the Moulouya valley, south-east of Melilla: PTT 945 was located north of the river on 18 September and further south on 20, 24 and 26 September. The other bird detected in Morocco (942) was located near the Atlantic coast of Western Sahara on 22 September, south-east of Cape Bojador, about 2 000 km from the place where it had been contacted six days earlier. In contrast, the bird carrying PTT 946 seemed to have chosen another route, over the Sahara desert. On 18 September it was located in southern Algeria, near the Mauritanian border and 16 hours later about 660 km south-west in Ouarane Sands, which means that it was flying with an average speed of about 40 km h-1. All four signals stopped moving between 20 and 24 September (Fig. 2).

Figure 1.

Main dates of detection and corresponding locations of the four Purple Herons equipped with PTTs on 2 July 2004 in the Camargue area (black star). All indicated locations have been obtained with high accuracy (class 1, 2 or 3) except for PTT 942 on 16 September (class A). Ring recovery data (adapted from Voisin 1996) from birds ringed in the Camargue between 1930 and 1982 are indicated (grey square: autumn recovery, black square: winter recovery).

Figure 2.

Recording of transmitter locations indicating when at least one highly accurate (black dot) or unspecified (white dot) signal was received within 5-days periods. Black arrows point out when transmitters were first detected with a high-accuracy signal outside of the Camargue and grey arrows when signals stopped moving.

Discussion

All four juvenile birds tracked in this study followed the same migration route, going south-west towards Spain, Algeria, Morocco and/or Mauritania. The results obtained are highly consistent with ring recovery data from birds ringed in western France between 1930 and 1982 (Voisin 1996). Interestingly, transmitters were detected in areas where ringed birds had previously been recovered in spring or autumn before the 1990s (e.g. Balearic Islands, Cherchell area in Algeria, Moulouya valley in Morocco), which suggests that migratory routes have not changed. Because PTTs 942 and 947 were located on the same day south of the Balearic Islands, we may hypothesize that these birds were travelling together, which supports previous observations (Voisin 1996, Kushlan & Hancock 2005).

Unfortunately, migratory movements could not be followed as long as expected. All four signals stopped moving before the birds reached their wintering quarters, indicating either that the birds died or had dropped their transmitter. The latter hypothesis is the most likely as each harness had a breakaway link and all PTTs stopped moving almost simultaneously. This mishap is illustrative of the trade-offs that have to be made when dealing with both research and conservation issues. The data obtained on the first part of migration however allowed to gain information about departure dates and flight speed of juvenile herons and confirmed the western migratory route previously suggested by ring recovery data (Voisin 1996).