This study reports the effects of age and culling of breeding adults on movements and dispersal rates of Yellow-legged Gulls (Larus michahellis) from a western Mediterranean colony (the Medes Islands) based on recaptures and resightings of birds banded as chicks. Juveniles (1-year-old birds) were most frequently located in the French coast of Biscay and in the Western North Mediterranean. Older gulls became concentrated near the colony and its surroundings (core area), with 3rd year sub-adults and adults accounting for more than 70% and 90% of resightings in this area, respectively. Culling of breeding adults increased the dispersal rate of juveniles (pre-cull: 62.6% ± 6.0 SD; post-cull: 71.2% ± 6.4 SD) and mean distance of resightings of juveniles (pre-cull: 250.5 km ± 267.6 SD; post-cull: 367.6 km ± 300.1 SD), and favored displacement of juveniles to the French Atlantic coast. Culling also increased the dispersal rate of adults (pre-cull: 31.7% ± 16.6 SD; post-cull: 38.8% ± 16.3 SD) and mean distance of adult resightings (pre-cull: 27.4 km ± 108.7 SD; post-cull: 35.7 km ± 113.0 SD). This could be attributed to an increase in the number of gulls recruited to other colonies near the Medes Islands after culls. Culling performed in the Medes Islands colony seems to have effects at metapopulation level, conditioning the dynamics and management of other colonies.
Culling, because of its apparent effectiveness in the short term (Thomas 1972; Blokpoel and Spaans 1990; however see Bosch 1996; Oro and Martínez-Abraín 2007) has been used for decades to reduce possible problems linked to large populations of large gulls. The number of gulls breeding in culled colonies decreases drastically due to the direct removal of individuals, as well as due to the changes it causes in philopatry, dispersal and recruitment rates (Duncan 1978; Coulson et al. 1982; Coulson 1991). Consequently, culls can influence the number of gulls beyond the boundaries of the colonies treated. In this way, culling a single gullery can have unpredictable effects at the metapopulation level because it may affect dispersal rates among colonies (Coulson 1991; Defous du Rau 1995; Oro 2003). For this reason, the evaluation of the effects of a culling programme in a given colony should include an analysis of the possible changes in the movement and dispersal rate of individuals.
The Yellow-legged Gull (Larus michahellis) is a large gull that has shown a marked increase in its number of colonies, breeding pairs and distribution area over the last four decades (Yésou and Beaubrun 1995; Thibault et al. 1996; Bosch and Carrera 2004). This increase has been linked to the exploitation of anthropogenic resources, which have improved the breeding performance of pairs (Bosch et al. 1994; Oro et al. 1995; Real et al. 2017). From the second half of the last century, several banding programs have been implemented to study the movements of individuals of this species. They have shown that the movements of Yellow-legged Gulls vary largely depending on the area in which their colonies of origin are located. While gulls from the Atlantic north Iberian coast colonies are sedentary (Munilla 1997; Arizaga et al. 2010), those from Mediterranean basin colonies move long distances, especially when they are young, and follow very different routes. Thus, Yellow-legged Gulls from the Adriatic area move to the Baltic and North Sea coasts (Kralj et al. 2014), while those from Eastern Iberian and Southern French coasts (or nearby islands), as well as the Balearic Islands, mainly move to Atlantic coasts from the Bay of Biscay to Portugal (Carrera et al. 1993; Martinez-Abrain et al. 2002; Rodríguez and Muntaner 2004; Galarza et al. 2012). Individuals from southern Iberia disperse along the southern and eastern Iberian coasts (Cuenca and Delgado 2014), and those from the Algerian coast follow two distinct routes, one to the Balearic Sea and Bay of Biscay and the other from southern Iberian coasts to the Atlantic coast of northwestern Iberia (Baaloudj et al. 2012). The large increase in the population of this species has led to the performance of culls by management agencies on several of its colonies, some of them included in banding programs (e.g., Medes Islands, Ebro Delta and Balearic Islands). However, data reflecting the effects of culling on the movements and dispersal of Yellow-legged Gulls are lacking. On the other hand, juveniles of some colonies of this species have been resighted at large distances from the natal colonies, in contrast to older individuals (Carrera et al 1993). This could be due to a reduction of the dispersal rates as the age increases, although studies about this are lacking.
The aim of our study is to analyze the possible effects of age and culling on the movements and dispersal rates of Yellow-legged Gulls from a western Mediterranean colony based on the recapture or resighting of birds banded as chicks.
Methods
Study Area
The Medes Islands archipelago (42° 03′ 00″ N, 3° 13′ 15″ E; NE Spain) holds a large breeding colony of Yellow-legged Gulls that was subjected to annual culls from 1992 to 1996 (for a detailed description of the colony see Bosch and Sol 1998). During those years, 25,000 breeding gulls were killed using baits of bread with butter mixed with α-chloralose and secobarbital placed in their nests (Sargatal et al. 1992). Culling greatly reduced the size of the colony within a short time, i.e., from 14,000 pairs in 1991 to 5,400 in 1997 (Bosch et al. 2000). The colony slowly increased to 7,700 pairs by 2008, but since then, numbers have declined to 4,673 pairs currently (M. Bosch, pers. obs., unpubl. data). Preliminary data on the movements of gulls from the colony banded with metal bands in the 1970s and 1980s have been analyzed in previous studies (Carrera et al. 1981; Carrera 1987; Carrera et al. 1993).
We studied the movements and dispersal rates of Yellow-legged Gulls from the colony using recoveries and resightings (hereafter, all referred to as resightings) of individuals marked as chicks with metal or engraved darvic bands in the colony from 1976 to 2003. During this period, 15,595 gull chicks were banded: 11,082 with a hard metal band before the culls (prior to 1992) and 4,513 afterwards (1997 onwards; 4,274 of them with both metal and engraved darvic bands). Four types of resightings were excluded: those reported as “not freshly dead”, those with “unknown find conditions”, those of fledglings found dead at the colony, and duplicate findings of individual in the same geographical zone and same age group (Coulson and Nève de Mévergnies 1992; Oro and Martinez 1994). After discarding these, the number of resightings was 1,618 referring to 1,259 individuals.
Resightings were grouped by three factors: age of gulls at resighting, geographical region of resighting, and pre- or post-culling period. Four age classes were distinguished: juveniles (in their first year of life), two-year-old sub-adults, three-year-old sub-adults, and adults. Gulls sighted after 1 May were considered to belong to the next age class (Martinez-Abrain et al. 2002). Eight large geographical areas (hereafter referred to as regions) were considered when locating resightings: R1, North-Western Mediterranean (between Cap de la Nao, Alicante, and Toulon, Golfe du Lion); R2, French coast of the Bay of Biscay (between Hendaye and Ploudalmézeau, Bretagne); R3, English Channel (between Ploudalmézeau and Oostende, Belgium, including the English coast on the other side of the channel); R4, North Atlantic Iberian (between Hendaye and Cape Fisterra, Galicia); R5, North Sea (between Oostende, Belgium, and Niebüll, German–Danish border); R6, Western South Mediterranean (between the Cap de la Nao and the Cape of San Vicente, Portugal; a resighting from Morocco was included); R7, Central Mediterranean (from Toulon to Terracina, East Coast of Central Italy); R8, Central Europe (comprising the eastern part of Central France, Switzerland, Austria and the South of Germany). A “core area” consisting of the Medes Islands colony and the surrounding area within 40 km was also identified. Two culling periods were distinguished: pre-culling (until the onset of culls initiated in 1992) and post-culling (from after the last cull of 1996 to 2004). Resightings recorded during the culling years were excluded, thus excluding possible resightings of individuals (mainly subadults) that were out of the colony during the first culls. As in other studies, it was not possible to quantify any possible bias linked to differences in observation effort among regions (Kralj et al. 2014).
Effects of Age and Culling Period on Geographical Distribution of Gulls
The effects of age and culling period on the distribution of resightings among different geographical areas were first tested by multifactorial Chi-square test. This test rejected the independence between these two factors (
= 693.2, P < 0.001), so the age effect was analyzed independently for each culling period, and the culling effect was analyzed separately in each age group. The effects were tested using the Chi-square or Fisher exact tests (significant if P < 0.05). Resightings in regions 3 to 8 were not included in the analysis because they had very low expected frequencies (less than 3%). Moreover, the resighting frequencies of juveniles and two-year-old sub-adults were grouped in multifactorial Chi-square test.
Effects of Age and Culling Period on Dispersal of Gulls
Dispersal rates were calculated by grouping resightings according to successive concentric 100 km circular zones extending outwards from the colony. The percentage of gulls resighted within each zone, grouped by age and culling period, was plotted on a log scale against distance to show whether a constant proportion (dispersal rate, r) of the gulls entering a zone remained within it (i.e., r = percentage of gulls moving beyond consecutive 100 km zones from the colony; see Coulson and Brazendale 1968; Parsons and Duncan 1978; Coulson and Nève de Mévergnies 1992; Oro and Martínez 1994). Resightings within the breeding colony were not included in this analysis. Seasonal periods were not distinguished to avoid some dispersal rates being calculated from small samples. Variations of dispersal rates among age groups and between culling periods were tested using the Kruskal-Wallis and Mann Whitney U tests (significant if P < 0.05). As in previous analyses, the age effect was analyzed independently for each culling period, while the culling effect was analyzed separately in each age group.
Effects of Age and Culling Period on Mean Distance Moved by Gulls
The distance between the colony and the location of resightings was calculated, and mean distances were calculated, distinguishing between ages and culling periods. Resightings performed within the colony (distance zero) were included in this analysis since mainly adult Yellow-legged Gulls tend to stay closer to the breeding colonies (Sol et al. 1995; Kralj et al. 2014). Mean distances were compared using the Kruskal-Wallis and Mann Whitney U tests (significant if P < 0.05). Again, the age effect was analyzed independently for each culling period, while the culling effect was analyzed separately in each age group.
Results
Effects of Age and Culling Period on Geographical Distribution of Gulls
Gulls were unequally distributed among the eight defined regions: three of the regions (0, 1, and 2) contained more than 95% of the resightings (Fig. 1; Table 1). The geographical distribution of resightings varied with the age of the gulls (pre-culling period excluding two- and three-year-old birds:
= 63.1, P < 0.0001; post-culling period: 
= 545,2, P < 0.0001). Consequently, the distribution of resightings was analyzed separately for each age group. Juvenile distribution did not vary among the three main regions (Regions 0, 1, and 2) before culling ( = 1.1, P = 0.575), but did vary after culling ( = 18.2, P = 0.0001), with a larger resighting frequency of gulls along the French coast of the Bay of Biscay (Region 2) (a posteriori tests) (Table 1). The distribution of two-year-old gulls did not vary significantly among the three main regions ( = 5.3, P = 0.072). Three-year-old gulls showed significant differences in their distribution among the three main regions after culling ( = 473.8, P < 0.0001), with a larger resighting frequency in the colony and nearby (a posteriori tests).
Adult gull distribution varied significantly among the three main regions, both before and after culling (before: = 148.1, P < 0.0001; after = 1289.4, P < 0.0001) due to a high resighting frequency of individuals in the core area (a posteriori tests). To ascertain whether this simply reflected the need to stay in the colony for breeding, the analysis was repeated only considering data from the non-breeding season (from July to February) during the post-culling period (data from the pre-culling period were not included because of the small sample size). Again, significant differences were detected due to a higher frequency of adults in the core area (post-culling period: = 570.5, P < 0.0001).
The geographical distribution of resightings varied significantly between the pre-culling and post-culling periods in the case of adult gulls ( = 13.5, P = 0.001) due to a higher frequency of individuals near the colony after culls (a posteriori tests). In the other age groups, there were no significant differences between periods (juveniles: = 3.2, P = 0.201; 2nd year sub-adults: Fisher Exact test grouping areas 0 and 1, P = 0.629; 3rd year sub-adults: Fisher Exact test grouping 0 and 1 areas, P = 0.547) although in the case of three-year-old gulls, it may be due to the small sample size in the pre-culling period.
Effects of Age and Culling Period on Dispersal Rate of Gulls
The dispersal rates of gulls (percentage of gulls moving beyond consecutive 100 km ranges from the colony) varied significantly among the age groups, both before and after culling (before: H = 9.88, P = 0.0197; after: H = 19.94, P < 0.001) (Table 2). A posteriori tests showed that adult gulls had a lower dispersal rate than the other age groups, both before and after culling (Table 2; Fig. 2).
On the other hand, dispersal rates also varied between the culling periods. Thus, juvenile and adult gulls showed a significantly higher dispersal rate after the culls (juvenile: Z = -2.6, P = 0.010; adults: Z = 2.5, P < 0.014) (Table 2; Fig. 2); two-year-old gulls showed the same tendency, although it was not significant (Z = -1.5, P = 0.142), possibly due to the small sample of resightings for this age class before culling. Three-year-old gulls showed the opposite tendency, although the differences in the rates of dispersal between periods were not significant (Z = 1.0, P = 0.336) (Table 2; Fig. 2).
Effects of Age and Culling Period on Mean Distance Moved by Gulls
The mean recovery distance decreased significantly with age, both before and after the culls (before: H = 126.9, P < 0.0001; after: H = 1427, P < 0.0001) (Table 3; Fig. 3). Mean distances increased significantly after culling both in juvenile gulls (Z = -2.71, P = 0.007) and adult gulls (Z = 5.38, P < 0.0001) (Table 3; Fig. 3). In contrast, for the three-year-old gulls, the mean recovery distance decreased significantly after the culls (Z = 3.24, P = 0.002); no significant differences were detected in two-year-old gulls (Table 3; Fig. 3).
Table 1.
Percentage of resightings of Yellow-legged Gulls (Larus michahellis) from the Medes Islands colony (in the western Mediterranean) in each geographical region for each age class. Data are distinguished by culling period (n = number of resightings). Core area = Medes Islands colony and the surrounding area within 40 km; R1 = NorthWestern Mediterranean; R2 = French coast of the Bay of Biscay; R3 = English Channel; R4 = North Atlantic Iberian; R5 = North Sea; R6 = Western South Mediterranean; R7 = Central Mediterranean; R8 = Central Europe (see Fig. 1).
DISCUSSION
The results of this study show that both gull age and massive culls have a significant effect on the movements and dispersal of Yellow-legged Gulls. With regard to age, juveniles were more frequently located along the French coast of Biscay, Region 2, (as previously described by Carrera 1987; Carrera et al. 1993), especially in the post-culling period, and also in the northwestern Mediterranean (Region 1). In contrast, older gulls were concentrated in the colony and its surroundings (core area). In this way, more than 70% of resightings of three-year-old sub-adults were in this area; when considering adult resightings, the percentage exceeded 90%. Data from this study show that juveniles arrive at the Biscay coast not only through the Loire and Garonne valleys (as previously described by Carrera 1987; Carrera et al. 1993) but also through the Ebro Valley. The presence of a large proportion of juvenile Yellow-legged Gulls along the Biscay coast has been reported since 1970s (Is-enmann 1973), and they come from a large number of colonies along the Mediterranean coast (or nearby islands) of Spain and France (Yésou 1985; Carrera et al. 1993; Martinez-Abrain et al. 2002; Galarza et al. 2012) and the Balearic Islands (Carrera et al. 1993; Rodriguez and Muntaner 2004; Galarza et al. 2012), as well as some colonies in Sardinia (Carrera et al. 1993; Galarza et al. 2012) and along the Atlantic North Spanish coast (Arizaga et al. 2010). Such a concentration of juveniles has been linked to the high availability of food in the region (linked to its high fishing productivity) relative to the Mediterranean Sea during summer (Le Mao and Yésou 1993). Moreover, juvenile Yellow-legged Gull movements begin at the onset of the post-breeding period (Carrera 1987), when food competition is expected to be larger. However, the main food resource of the gulls of the Medes Islands colony (i.e., garbage from dumps) is largely available throughout the year, and the dependence on this resource has been very high for more than 50 years (Carrera and Vilagrasa 1984; Bosch et al. 1994, 2000; Bosch 2010). This could support the hypothesis that the patterns of the Yellow-legged Gull movement are not a recent development nor linked to circumstantial factors, but rather result from its evolutionary history (Kralj et al. 2014), although more data are needed to test this.
Table 2.
Dispersal rates (r = % moving beyond each 100 km zone) of Yellow-legged Gulls banded as chicks in the Medes Islands colony and resighted out of the colony, distinguished by age and culling period. Data do not include resightings performed in the colony.
Figure 2.
Percentage of Yellow-legged Gulls (Larus michahellis) moving beyond consecutive 100 km ranges from the Medes Islands in the western Mediterranean, plotted on a logarithmic scale against distance. Pre-culling and post-culling.
Table 3.
Dispersal distances moved by Yellow-legged Gulls (Larus michahellis) from the Medes Islands colony in the western Mediterranean according to age group and culling period.
As gull age increases, individuals become concentrated in the colony and surroundings, showing decreasing dispersal rates and mean distances of resightings, and thus a return to this area. In fact, among the 116 gulls sighted in more than one geographical area, 79% were last sighted in the colony or its surroundings. Moreover, the high proportion of adults in the core area were sighted not only in the breeding season, but also outside it (as observed by Sol et al. 1995), so it was not due only to breeding requirements.
Figure 3.
Mean distance (± SE, vertical bar) moved by Yellow-legged Gulls (Larus michahellis) from the Medes Islands colony in the western Mediterranean according to age group and culling period.
Culling affected the geographical distribution of resightings, as well as dispersal rate and mean distance of resightings. The dispersal rate and mean distance of resightings of juvenile gulls increased after culls. Furthermore, in this period the proportion of juveniles dispersing throughout the north-western Mediterranean decreased significantly in favor of the French Atlantic Coast. This is supported by the significant differences in resighting frequencies in these two areas between culling periods. On the other hand, culling also increased both the dispersal rate and the mean distance of adult resightings. These results could be explained by an increase in the numbers of gulls recruited to other colonies near the Medes Islands colony after the culls (such as Cap de Creus, Aiguamolls de l'Empordà or Sant Feliu de Guíxols, all of which are located less than 40 km away). This hypothesis is supported by the fact that 263 out of 725 adults (i.e., 36%) sighted after the culls in the core area were never sighted in the colony; furthermore, 19 adults (2.6%) sighted in the colony, were sighted in the core area in the two years after their last resighting in the colony. In fact, since culls of the colony were performed, the number of new colonies as well as the size of the colonies that already existed has greatly increased (Bosch et al. 2000; Bosch and Carrera 2004). The most noted case is the Ebro Delta colony, which increased from 1,100 to 10,500 pairs within 9 years, since the start of culls in the Medes Islands colony (Parc Natural del Delta de l'Ebre, unpubl. data). Consistent with this, previous studies on other colonies of gulls subjected to culling showed an increase in the frequency of recovery of individuals that were breeding in a colony other than that in which they had hatched (Chabrzyk and Coulson 1976; Coulson 1991). In turn, gull movements can condition the management of other colonies (Duncan and Monaghan 1977; Gabrey 1996). In this way, the large increase in size of the Yellow-legged Gull colony of the Ebro Delta has resulted in massive culls since 2016 (Parc Natural del Delta de l'Ebre, unpubl. data). Thus, the culling performed in the Medes Islands ended up necessitating more culling in other colonies, expanding its effects beyond what was expected. In conclusion, culling in the Medes islands colony may have led to uncontrollable effects at the metapopulation level, consistent with Brooks and Lebreton (2001), and this should be taken into account when planning new gull management measures.
Acknowledgments
We thank ALL the people who have reported resightings/recoveries of Yellow-legged Gulls banded in the Medes Islands colony (especially Carlos Alvarez, Michel Fouquet, Andreas Guyot, Patrick Mayet and Alain Kim); without them, this study would not have been possible. We thank the Port Autònom de Barcelona and Servei de Vigilància de les Illes Medes (Parc Natural del Montgrí, les illes Medes i el Baix Ter) for their logistical support. Daniel Oro, Andrew Kasner and anonymous referees provided very helpful comments on the manuscript. Toffa Evans (Servei Lingüístic UB) and Esther Gil improved the English text, and Vittorio Pedrocchi improved the map of resightings. Toni Curcó (Parc Natural del Delta de l'Ebre) provided the census data for the Ebro Delta. We thank Anna Galdeano, Mikel Zabala, Jofre Carnicer, Santi Mañosa, Daniel Oro and Vittorio Pedrocchi for their support. Banding was conducted in accordance with permits issued by the regional wildlife agency (Departament de Territori i Sostenibililitat, Generalitat de Catalunya) and Spanish government (Ministerio de Medio Ambiente) and was performed with all the ethical guidelines for handling wild birds. This study was partially financed by the Generalitat de Catalunya. We wish to state that we did not participate in the culls performed in the Medes Islands, nor were we responsible for these culls, nor did we advise on these measures.
