Lesser Black-backed Gulls (Larus fuscus fuscus) were studied over 10 breeding seasons (2005-2014) to assess factors influencing nesting distribution among 7 subcolonies with two distinct habitats (open-rocky or vegetated) in an archipelago on the Norwegian Coast. The study was divided into an early (2005-2008) and late period (2009-2014), depending on a predatory event in 2008, where one subcolony suffered a complete reproductive failure. In the early period, three open-rocky subcolonies in the periphery appeared to contain birds of higher quality (assessed by stability in number of pairs among years, reproductive timing, clutch size, and chick production) compared to a large vegetated subcolony. In the late period, the proportion of the population nesting in the large vegetated subcolony increased, as did the quality of individuals, a result of birds from the depredated subcolony settling there. In subcolonies not subject to complete reproductive failure, philopatry to natal subcolonies was high among juveniles (∼80%), and the rate of among-year change between subcolonies by adult breeders was as low as 0-3%, although the rate of change increased up to ∼15% following years of poor reproductive success. However, there was no evidence that either habitat consistently offered better nest protection and reproductive success than the other.
Nests provide protection from predators and environmental exposure for parent birds and their offspring, and selection of sites that minimize such risks is important to optimize fitness (Cody 1985). For ground-nesting species, it may be advantageous to nest at protected sites, such as those concealed in vegetation (Kim and Monaghan 2005; Oro 2008). In contrast, nesting openly may be advantageous by facilitating predator detection and defense, and trade-offs regarding reproduction may exist between open and concealed nesting success (Ewald et al. 1980; Götmark et al. 1995; Wiebe and Martin 1998; Kim and Monaghan 2005).
Gulls are mostly ground-nesting and colonial species that breed in diverse habitats, both in vegetated and open locations, but to establish what is a preferred nesting habitat for gulls is not always straightforward. For example, vegetated habitats may seem to offer better nest protection than open habitats (Parsons et al. 1976; Calladine 1997; Kim and Monaghan 2005; Oro 2008), but gulls may still show preference for the latter (Pierotti 1982; Rodway and Regehr 1999; Borboroglu and Yorio 2007; Suárez et al. 2010). Several bird studies have demonstrated that central nesting places within colonies offer better protection from predators than the periphery, and birds of higher quality would therefore be expected to prefer central nesting sites at the expense of inferior individuals (Aebischer and Coulson 1990; Vergara and Aguirre 2006; Yorio and Quintana 1997; Gaston and Elliott 1996; Indykiewicz et al. 2019).
This study focuses on the Lesser Black-backed Gull Larus fuscus fuscus in northern Norway. The subspecies is threatened (Juvaste et al. 2017), and the Norwegian population has declined strongly since the early 1970s (Bustnes et al. 2010). Our study area is an archipelago with 7 separate subcolonies; 4 rocky inlets in the periphery where birds nested openly, and 3 in the center in which ∼70-90% of the nests were concealed in vegetation. Hence, in this study there are two dimensions to nest site selection: within and among subcolonies, and here we focused on the distribution of breeders among the 7 subcolonies with different nesting habitats.
Based on previous studies, we hypothesized that the central vegetated subcolonies would be preferred by dominant birds of high quality (i.e. birds that are likely to successfully reproduce) due to the protection offered (Calladine 1997; Bosch and Sol 1998; Kim and Monaghan 2005; Oro 2008), and that subordinate individuals were excluded from these areas. We also predicted that the high-quality individuals would be more prone to breed, resulting in the number of birds nesting being more stable among years in preferred habitats compared to the non-preferred ones. In addition, we expected that additional birds would try to establish in the preferred habitat once they gained breeding experience and status (Serrano and Tella 2006; Oro 2008). An alternative hypothesis is that habitat selection has a genetic component (Rodway and Regher 1999) and suggests that recruits have an inherited preference for certain habitats or areas. This would be facilitated and maintained if recruits show consistent natal philopatry to the colony where they hatched. Yet another alternative hypothesis is that none of the habitat types provides any fitness advantage (Bosch and Sol 1998), and that nest selection is flexible and governed by available habitats and previous experience (Pajero et al. 2006; Coulson and Coulson 2009). Under this hypothesis, selection of nesting habitats is a dynamic process in which gulls move between areas depending on the individuals' or subcolonies' reproductive success in the past. This hypothesis also predicts changes in the number of birds nesting in different habitats and shifts in habitat preferences, and low return-rates to colonies with poor reproductive success. Testing of the different hypotheses was facilitated by a ‘natural experiment’ where a predatory event led to a complete reproductive failure in one of the open-rocky subcolonies. Reproductive parameters, recruitment, and intra-colony movements by adults and recruits were monitored in the different subcolonies.
The study location was Horsvær (65°19' N 11 °37' E), an archipelago at Helgeland (Nordland County) in northern Norway (Fig. 1), where close to 400 pairs of Lesser Black-backed Gulls might breed in good years, but very few in poor years (383 in 2006 and only 15 in 2013). At Horsvær there were 7 subcolonies (denoted by letters A-H in Fig. 1). Two of the subcolonies were on the same island (Fig. 1, A and B) whereas all other subcolonies were on different islands. Four of the subcolonies (B, E, G and H) were dominated by open-rocky habitats and had no vegetation taller than a few cm and nearly all nests were openly exposed (>90%) in all years (Fig. 2). The other subcolonies (A, C and D) were dominated by dense vegetation (∼70 cm tall) consisting of meadowsweet Filipendula ulmaria: in the A- and D-subcolonies, >70% of the nests were in the vegetation, whereas in the C-subcolony >90% of the nests were in the vegetation (Fig. 2).
The study started in 2005, and the first season fieldwork lasted from mid-June to late July. In subsequent years, the fieldwork was divided into two periods: 1) a period of 6-10 days in the latter half of June; and 2) 1-3 days in late July when large chicks were individually marked with tarsus bands. This set-up was chosen to reduce the human disturbance of this threatened subspecies. At arrival, all colonies were searched, all nests were marked with a numbered wooden stick, and eggs were marked with a waterproof pen. After 5-7 days, a new search was conducted where all nests were revisited, and unrecorded nests were marked. During both visits, clutch sizes were recorded. Egg laying started in early June and was finalized in the latter half of June when we were present. We are thus confident that our estimate of the number of nests in the colony is accurate.
Birds were trapped on their nests during the incubation period using a nest trap (Bustnes et al. 2008). All birds caught on their nests and juveniles were equipped with steel bands and colored bands with alphanumeric codes visible from distance with binoculars and telescopes. The sex of 83 individuals was determined using molecular methods (Erikstad et al. 2009). Among these, males (mean = 774 g; range = 700-920 g) were 20% heavier than females (mean = 643 g, range = 585-695 g), with no overlap. For the remaining birds, we thus assumed that individuals weighing < 700 g were females and those >700 g were males. In addition, some individuals were sexed by comparing birds in pairs when at their nest site.
We assessed timing of nesting differently among years. Fieldwork in 2005 and 2006 covered the hatching period, and the proportion of nests hatching at the date of the last nest-checks was compared between subcolonies. In the remaining years with sufficient sample sizes (2008-2011 and 2014), the timing of reproduction was assessed by comparing egg weight/volume ratio between subcolonies, since egg mass in birds is reduced by ∼15% over the incubation period (Ar and Rahn 1980; Zicus et al. 2004); i.e., lighter eggs controlled for volume indicate earlier egg-laying. At the second visit (the same day for all subcolonies in a given year), the eggs from 10 randomly selected nests in all subcolonies were measured with sliding calipers (± 0.1 mm) and weighed (± 1 g) with a digital scale. Preferably, 3-egg clutches were selected, but due to the low number of nests in some subcolonies and years, clutches with two eggs were sometimes included. The variables used in the analyses were the mean weight/volume ratio of each clutch. Egg volume (mm3) was estimated using the formula: Volume = 4.76×10-4×Length×Width2 (Hoyt 1979).
The rate of nest-loss (all eggs disappearing from the nests) was assessed by comparing the proportion of nests in which eggs disappeared between the two nest checks in the different subcolonies. Nest-loss was mostly caused by predators such as crows, ravens and other gulls present at Horsvær. Production was assessed as the number of juveniles banded in late July, in addition to number of unmarked chicks observed (e.g., those which took to the sea), in relation to the number of nests in each colony.
We checked all subcolonies for color-banded birds at least once, but often two or three times per day each year, to maximize the number of re-sightings. Checking the whole colony took about 4 hours depending on weather conditions. Occasionally, bad weather precluded us from checking for color-banded birds. We used binoculars, spotting scopes, or took photographs of birds with a zoom lens. Individual IDs were noted at every check to track observation histories for all known individuals, including birds marked both as adults and as juveniles. A bird was classified as “changing subcolony” if it was observed more than once consecutively in a new subcolony (either within the same years or in subsequent years), and not seen returning to the former subcolony. Birds observed only once in a new subcolony were excluded from the analyses. Returning juveniles were classified as established in a subcolony when they reached 3 years of age, when they might become breeders (Bosman et al. 2013), and were repeatedly observed in the same colony.
Individual quality is commonly assessed using the definition proposed by Wilson and Nussey (2010) as an axis of among-individual heterogeneity that is positively correlated to fitness. In this study, direct measurements of individual fitness were impossible since we were rarely able to assign banded individuals to specific nests. This was because there were few nests that could be directly observed from distance to establish the identity of the breeders, and the gulls were very difficult to catch more than once. Thus to reduce disturbance and risk of nest-loss, we used fitness proxies such as breeding propensity of individuals (stability of number of nests in different subcolonies among years), timing of egg-laying (hatching date), clutch size, nest-loss, and chick production (Indykiewicz et al. 2019). Hence, we compared the means of these traits between subcolonies, assuming that subcolonies with high scores were inhabited by individuals of high quality relative to subcolonies with lower scores.
In 2008, the predation event (“natural experiment”) occurred in the G-subcolony (Fig. 1). There were 55 nests in the subcolony when we left Horsvær on 20 June. When we returned on 24 July there were no chicks, whereas in all other subcolonies chick production was relatively high. Based on more recent observations, the only likely predators were ravens (Corvus corax) with fledglings. At Horsvær, ravens with five fledglings have been observed to clear entire subcolonies of eggs in a matter of days (J.O. Bustnes, pers. obs.). Raven occurrence and detection in this area varies with human presence, and the G-colony is relatively concealed so raven activities could easily have gone undetected, which was less likely for the other colonies.
Statistical tests were carried out in SAS (SAS 2011). Low sample size precluded a detailed statistical analysis on the year-subcolony level in some cases. Hence, we merged different years into an early and a late period (2005-2008 and 2009-2014, respectively) distinguished by the timing of the natural experiment in which the predation event led to complete reproductive failure in the G-subcolony (Fig. 1). Moreover, low sample sizes also precluded analyses of all subcolonies; i.e., some subcolonies were merged for comparisons, such as the open-rocky subcolonies: E-, I- and G-subcolonies (before 2009) were compared to the large vegetated C-subcolony (Fig 1.). Student's t-tests and ANOVAs were used to compare the different subcolonies with regard to reproductive timing (after 2006), egg, and clutch size. Tukey-Kramer tests were used for individual post-hoc comparisons. Logistic regression was used for reproductive timing in 2005 and 2006, and for nest-loss. For clutch size, we subtracted the yearly mean from each clutch to control for differences between years.
Number of Nests
The total number of nests at Horsvær varied from 15 (2013) to 383 (2006; Table 1). In the good breeding seasons of 2005 and 2006, the vegetated C-subcolony (Fig. 1) was the largest with 108 nests, whereas the other subcolonies did not exceed 60-70 nests (Table 1). In contrast, 2007 was a poor breeding season (few birds nesting), and only 5 nests were found in the C-subcolony, whereas the open-rocky subcolonies still had 35-65% of the numbers in the previous years (Table 1). As the breeding situation improved in 2008, the number of nests was equal to the numbers for 2005 and 2006, except the C-subcolony in which the number was still low (Table 1). However, as the stable and productive G-subcolony (open-rocky habitat) was abandoned after reproductive failure caused by predators (“the natural experiment”) in 2008 (Table 1), the gulls moved predominantly to the vegetated and more central subcolonies (A and C). As a result, over the entire study, the proportion of the nests at Horsvær in the C-subcolony increased from 4-30% annually in the early period to 41-48% in the late period (Table 1). In this analysis, 2012 and 2013 were excluded, as the number of nests in the archipelago was too small for any meaningful comparisons (Table 1).
Number of nests in 7 subcolonies of Lesser Black-backed Gull (Larus fuscus fuscus) at Horsvær, northern Norway, 2005-2014.
Reproductive Timing and Clutch Size
In 2005 and 2006, birds in the vegetated C-subcolony laid eggs later than in most of the other subcolonies (χ25, 688 = 26.04, P < 0.0001; controlling for year χ21, 688 = 20.94, P < 0.0001: logistic regression); i.e., in 2005 no eggs were hatching in the C-subcolony at the last nest check (30 June and 1 July), whereas 10-12% of nests were hatching in the other subcolonies ( Appendix 1 (2-19229 p163-173 A4 COLOR Apendix ON-LINE ONLY WBdj indd.pdf)). The corresponding values for 2006 (26 and 27 June) were 0.9% in the C-subcolony and 33-35% in the other subcolonies ( Appendix 1 (2-19229 p163-173 A4 COLOR Apendix ON-LINE ONLY WBdj indd.pdf)). Moreover, in 2008 the egg weight/volume ratio was higher in the vegetated C-subcolony than in the open-rocky subcolonies (F5, 73 = 6.36, P < 0.0001), suggesting later egg-laying ( Appendix 1 (2-19229 p163-173 A4 COLOR Apendix ON-LINE ONLY WBdj indd.pdf)); significantly different from the E- and the H-subcolony (P < 0.05), but not the G-subcolony (P > 0.3). The A-subcolony appeared to be somewhat intermediate ( Appendix 1 (2-19229 p163-173 A4 COLOR Apendix ON-LINE ONLY WBdj indd.pdf)). After 2008, however, the situation changed, and the C-subcolony had lower egg weight/volume ratio than most other colonies (F4, 170 = 15.85, P < 0.0001), when controlling for year (2009, 2010 and 2014: F4, 170 = 226.21, P < 0.0001; Appendix 1 (2-19229 p163-173 A4 COLOR Apendix ON-LINE ONLY WBdj indd.pdf)), significantly lower than the A, B and H subcolonies (P ≤ 0.025), but not the E-subcolony (P = 0.24). This suggests that after 2008 the C-subcolony switched from late to early egg-laying relative to the other colonies, indicating that individuals of higher quality had established in the subcolony.
Clutch size varied from one to four eggs (mostly 2-3 eggs, Appendix 1 (2-19229 p163-173 A4 COLOR Apendix ON-LINE ONLY WBdj indd.pdf)). A significant subcolony×period interaction (P < 0.01) suggested that subcolonies with large clutches in the early period (2005-2008) were not to the same degree associated with large clutches in the late period (2009-2014). In the early period, the vegetated C-subcolony tended to have small clutches relative to the other subcolonies, but shifted towards having large clutches in the late period ( Appendix 1 (2-19229 p163-173 A4 COLOR Apendix ON-LINE ONLY WBdj indd.pdf)). In the early period, the open-rocky E- and H-subcolonies had larger clutches than the C- and B-subcolonies (P ≤ 0.05), but when controlling for multiple comparisons, only the differences to the B-subcolony remained significant (P = 0.04). In the late period, none of the differences between the colonies were significant (P > 0.1), when controlling for multiple comparisons.
In 2005 and 2006, hardly any nests were lost between nest checks: 1.7% in 2005 and 0% in 2006. Afterward, the overall rate of nest-loss in the subcolonies increased to 8-10% (2007, 2008 and 2010), and then to 33% (2009 and 2011, Table 2). In 2012 and 2013, nearly all nests were lost, whereas in 2014 the nest-loss rate varied between 12 and 17%, but no nests disappeared in A- and B-subcolonies. Statistical analyses (logistic regression) could be conducted for the years 2009-2011 and for 2014, and there were no significant differences in the rate of nest disappearance among the subcolonies (0.46 < P < 0.96). The exception was 2011, when 59% of the nests disappeared in the vegetated C-subcolony, which was significantly higher (P = 0.0014) than in the open H-subcolony (10%). This suggests that structure of the nesting habitat has very little influence on nest protection at Horsvær.
Percentage of nests lost between checks in different subcolonies of Lesser Black-backed Gull (Larus fuscus fuscus) at Horsvær, northern Norway, 2005-2014 (n = number of nests checked twice in each colony; colony D was not included due to low number of nests).
The production of chicks varied greatly between years: 0 (2012 and 2013) to 365 (2005, Table 3). The number of chicks produced per nest also varied, being highest in 2005, 2006 and 2010 (Table 3). The vegetated C-subcolony had a low number of chicks per nest in 2005 and 2006 compared to the other subcolonies, whereas after 2008 this number tended to be higher. Notably, in 2009-2010 this subcolony had the highest number of chicks produced per nest compared to all other subcolonies (0.61-1.3 chicks per nest, Table 3). After 2010, the production was very low, and a limited sample size precluded any further analysis.
Site Fidelity and Movement Patterns Among Subcolonies
A total of 157 birds (86 females and 71 males) were caught on their nest between 2005 and 2011 and observed in subsequent years (total of 1009 observations including repeated sightings in the same year). Twenty-eight individuals (16 females and 12 males) were from the “natural experiment” subcolony (G-) and subsequently excluded from this analysis. None of these 28 birds, however, changed subcolony before the “experimental manipulation”. Of the remaining 129 birds, 25 (19.4%) changed subcolony over the observation period, 17 females (20%) and 8 males (11%). Of these, only 2 females were found to return to their original subcolony after being observed in a different subcolony in one and three years, respectively.
The rate of subcolony change varied among years; i.e., after good breeding seasons in 2006 (n = 74), 2008 (n = 56) and 2010 (n = 39), none of the birds changed subcolony in the subsequent year. Year 2005 was also a good breeding season, but the marginal D-subcolony was abandoned by all marked birds (n = 3), increasing the total rate of change to 8% (n = 62). Hence, when excluding the D-subcolony, only 2 birds (3%) changed subcolony following the 2005 season. After the relatively poor 2007 season, 6% (n = 69) changed subcolony, and after the very poor 2009 year, nearly 15% (n = 48) moved to another subcolony. After 2011, no birds changed subcolony, but very few birds were nesting, and the number of birds observed in two consecutive years was low (n = 17, 8 and 11 in 2012, 2013 and 2014, respectively). In general, our data suggests that the proportion of adult gulls changing subcolonies between years was low but increased after years with poor breeding success. Although birds in the vegetated C-subcolony seemed of inferior quality in the early period, only 2 out of 25 birds (8%) from this subcolony have established themselves in other colonies over the whole study period. Similarly, the B-subcolony seemed of relatively low quality, but no birds (n = 13) were observed to establish permanently in other colonies.
Number of chicks produced and number of chicks per nest in different subcolonies of Lesser Black-backed Gull (Larus fuscus fuscus) at Horsvær, northern Norway, 2005-2014 (colony D was not included due to low number of nests).
A total of 144 juveniles returned after two years or more (a total of 461 observations), of which 18 were from the natural experiment G-subcolony (Fig. 1) and subsequently excluded from this analysis. Of the remaining 126 recruits, 91 (72.2%) were first observed in their natal subcolony. There appeared, however, to be great differences between the subcolonies. Notably, the homing (first observation) was strong for the open-rocky H- and vegetated C-subcolony recruits (86% and 83%, respectively), whereas on the other end of the scale, the A- and B-subcolony were low (31% and 20%), although the sample size was low for the latter subcolonies.
Of the 126 recruits, 75 were classified as established (repeatedly observed); 80% in their natal subcolony. There were, however, some differences between years; i.e., 67% (n = 32) of the recruits from 2005 established in their natal subcolony, whereas 83% (n = 30) of the 2006 recruits did, however not significant (P = 0.15; Fischer exact test). All recruits produced after 2006, and recorded as established (n = 13), settled in their natal subcolony.
The “Natural Experiment”
After reproduction failed in the G-subcolony in 2008 (“natural experiment”), nearly all breeders abandoned the site. Fifteen birds were recorded in other subcolonies in subsequent years, and only one was observed in or near its natal G-subcolony in 2009 and 2010. Of the 15 birds, 11 were observed to establish in other subcolonies: 8 (73%) in the two vegetated A- and C-subcolonies in the center of the archipelago; 2 birds in the open-rocky H-subcolony; and one nested solitarily. Hence, birds that abandoned the open nesting G-subcolony mostly moved to the nearby vegetated subcolonies (Fig. 1), reflected in the tripling of the breeding numbers in the C-subcolony in 2010 compared to 2008 (Table 1).
There were 18 recruits from the G-subcolony observed returning, of which 5 (28%) were seen in their natal subcolony. However, three of these were only observed before 2009. After 2008, there were thus 13 recruits observed: 7 (47%) in the vegetated C-subcolony; 1 (8%) and 5 (33%) in the open H- and E-subcolonies, respectively; and one was observed solitarily. After 2008, 8 G-subcolony recruits were classified as established: 4 in the open-rocky H-subcolony; 3 in the vegetated C-subcolony; and one bird established briefly in its natal G-subcolony in 2010 and 2011, but it was not observed in the subsequent years.
This study unraveled a complex pattern of nesting habitat use by Lesser Black-backed Gulls, but our findings do not suggest that the physical habitat (vegetated or open-rocky) is the key to understanding the nesting distribution of these Lesser Black-backed Gulls. Our first hypothesis was that the vegetated subcolonies would be preferred by birds of high quality because of the protection offered (Calladine 1997; Bosch and Sol 1998; Kim and Monaghan 2005; Oro 2008), and that subordinate individuals were excluded from these areas. However, in the early period (2005-2008), high quality individuals seemed to prefer the three open-rocky subcolonies (E-, G- and H-) in the periphery of the archipelago, and these subcolonies seemed to have an upper limit to the number of nests (∼ 60), especially subcolonies E and G. Even in the poor 2007 season, these subcolonies had relatively high numbers of nests (35-65% of the maximum number), whereas in the large, vegetated C-subcolony hardly any birds nested. In fact, 9 birds were found to have starved to death in the C-subcolony in 2007, while no dead gulls were found in the other subcolonies that year. However, in the late period (2009-2014) the situation was reversed as the vegetated C-subcolony became the most productive one, suggesting that the quality of birds in this subcolony increased after 2008. Hence, despite apparent preference for the open-rocky subcolonies in the early period, there was no consistent difference in nest-loss or chick-production between the two habitat types over the whole 10-year study period. Additionally, if one of the habitats was preferred, we would expect to see birds trying to establish themselves in the preferred subcolonies once they gained breeding experience and status (Oro 2008). This was not supported in our data, since we found no evidence that recruits tried to move towards any specific habitat or subcolony as they gained experience (Serrano and Tella 2007), as found in Yellow-legged Gulls (Larus michahellis, Oro 2008).
It has been suggested that gulls may have an innate preference for one type of habitat when they are recruited into the breeding population, and afterwards show little flexibility (Rodway and Regher 1999), e.g., through natal philopatry. Clearly, natal philopatry was high among recruits, but birds changed both subcolonies and nesting habitats. Hence, the alternative hypothesis that habitat preference has a genetic component (Rodway and Regher 1999) gained no support in our data. We are aware of no studies of gulls that have followed individuals over time with regard to habitat features, so it is not known whether our findings are general among gulls. However, other studies have found that gulls may move to areas with other habitats because of reproductive devastation, such as culling (Oro 2008; Coulson and Coulson 2009; Coulson 2015)
The last hypothesis was that selection of nesting site is a dynamic process in which gulls move between different locations and habitats depending on available space and their own, or the subcolony's, previous reproductive success (Pajero et al. 2006). Under this hypothesis, whether the habitat is open or vegetated does not matter to the individuals. Instead, they select breeding habitats based on their perception of the safety or other factors, both intrinsic and extrinsic, affecting their reproductive success. This hypothesis predicts changes in the number of birds nesting in different colonies, shifts in habitat preferences, and low return rate to subcolonies with poor reproductive success. Our data suggests that this is the best model for understanding the spatiotemporal distribution of breeding Lesser Black-backed Gulls. First, individuals used both types of nesting habitats and, although most of them were observed to stay in their subcolony independent of former breeding success, a higher proportion of the birds changed subcolony after experiencing poor seasons. In addition, the ‘natural experiment’ in the G-subcolony showed that all birds might depart in the case of complete reproductive failure, and most of the individuals moved from one type of nesting habitat to another (in this case from open-rocky to vegetated habitats). This is similar to previous observations by Coulson and Coulson (2009, and references therein), where synchronous removal of nests from colonies of Lesser Black-backed Gulls resulted in complete abandonment, whereas removal of a smaller portion of colonies did not reduce the number of breeders.
The question then is what causes and maintains the distribution of seemingly high- and low-quality subcolonies. A central feature was the high natal philopatry among the recruits (80%), and interestingly the highest philopatry rates were found in a peripheral open-rocky subcolony (H: 86%), and in a vegetated subcolony (C: 83%). This suggests that recruits were able to settle in both types of habitats if they were raised there. This might be as simple as birds allowing their offspring to settle near them, or that high-quality birds produce high-quality offspring capable of securing a place for themselves in a good colony in competition for space with other high-quality breeders. Hence, this study suggest that such situations may be stable until some changes disrupt the breeding structure.
One concern with this study is that in several of the years, only a fraction of the birds actually nested (2007, 2009, 2012 and 2013) due to poor feeding conditions, and it may thus not be appropriate to define all movements as changes of nesting subcolonies/habitats. This is clearly a weakness, but we only included observations where birds were repeatedly seen in a new subcolony. In addition, of 25 birds that changed subcolony, only two females were observed to return to their original subcolony, suggesting that the observed movements were permanent rather than just by chance. Another issue is that no birds were recorded to change subcolony after 2011, although breeding success was poor. A probable reason for this is that breeding success was poor in all subcolonies, and prospecting birds would have little to gain from changing subcolony. Finally, a question is whether high production in one type of habitat compared to the other is attributable to the habitat features or the quality of individuals that breed there. This is difficult to resolve, but this study indicated that differences in reproduction were a result of individual quality, since both types of habitats turned out to be the most productive at different times.
In conclusion, this study suggests that nesting habitat use by Lesser Black-backed Gulls is flexible regarding habitat characteristics, but that high natal philopatry and low rate of exchange between subcolonies is essential for maintaining a structure of high- and low-quality subcolonies. However, such structures may collapse if a reproductively devastating event occurs in a subcolony with high quality individuals, and the breeders subsequently establish in colonies with available space, where the resident breeders may be of inferior quality. Many species of seabirds are in decline, notably surface-feeding species such as the Lesser Black-backed Gull, mainly due to consistent food shortage (Cury et al. 2011; Paleczny et al. 2015). In this perspective, nesting habitat selection clearly plays a minor role. However, breeding motivation and thus nest attendance is strongly dependent on the feeding situation, and for poorly motivated birds, the probability of successful nesting may depend on features in the habitat. This might be both the social environment (subcolony size etc.) and the ability to conceal the nest and chicks. Reproductive success may thus be dependent on nesting habitat. In this perspective, information about nesting habitat selection is important when trying to develop good strategies for protecting colonies of vulnerable gulls.
We are grateful to a number of people assisting us during fieldwork, notably Trond Johnsen, Nils H. Lorentsen, Klaus Torland, Geir A. Bustnes and Jorg Welcker. We are grateful to Harald Bustnes, Sigfred Jørgensen and Runar Omnø for logistical support during our stay at Horsvær. We also thank two reviewers for very valuable input to the manuscript. The study is a part of the Norwegian seabird program SEAPOP and was funded by SEAPOP and the Norwegian Environment Agency.
Please find the supplemental content for this article online by clicking on the Supplemental Content tab at: https://doi.org/10.1675/063.043.0204