Study colony

The main study was conducted in one of the largest breeding colonies on the island Texel (Kelderhuispolder, Geulduinen; 53°00.5′N, 04°43.1′E), The Netherlands, at the crossroads of the western Wadden Sea and the southern North Sea (Camphuysen & Gronert 2010, Camphuysen 2013). The Lesser Black-backed Gull colonised this area in 1970 (Dijksen & Dijksen 1977, Dijksen 1996) and the colony has not been studied before. The study colony is part of a much larger colony, on the south tip of the island where, around 2006, when the study started, over 11,000 pairs of Lesser Black-backed Gulls bred in a mixed colony together with c. 4000 pairs of Herring Gulls. It is within Natura 2000 area NL2003060 (‘Duinen en Lage land Texel’) and the Lesser Black-backed Gull is listed by conservationists with the specific objective to maintain a breeding population of c. 14,000 pairs (Aarts et al. 2008). The study colony is an undisturbed and ungrazed nature reserve, inaccessible to public in summer. There are few mammalian predators: Feral Cats Felis catus, Stoats Mustela erminea, European Hedgehogs Erinaceus europaeus and Brown Rats Rattus norvegicus. Colony data were collected from April to August, 2006–2021. Two prime study plots for Lesser Black-backed Gulls were chosen initially, but in later years, following gradual changes in the distribution of nests in the core colonies (areas gradually being abandoned and new sites colonised), four further study plots were selected to ensure a sufficiently large sample of nests available for research.

Annual cycle and non-breeding distribution: colour-ringing and GPS tracking

The annual cycle was reconstructed using field observations in and around the colony, plus re-sightings of colour-ringed individuals (adults and juveniles), captured as breeding birds or fledglings, throughout their (annual) range. The colour-ring data will be instrumental to assess apparent annual mortality in a follow-up study, and have provided information on changes with increasing age in timing, range, and the use of stop-over or wintering sites age during annual migrations for this study. Breeding birds were captured at Texel while incubating and marked with a permanent stainless-steel ring on one leg and a four-letter coded, green PMMA colour ring on the other, to allow for individual recognition (first letter F for females, M for males). Walk-in traps were deployed over well-incubated clutches (eggs c. 10–22 d old). During ringing, standard biometrics were taken always by the same observer. Head length (head+bill) was used as a noninvasive method to indicate the sex of adults or subadults (males are on average larger than females, cutoff point 113 mm, accuracy 95.0 ± 2.0%, ±SD), following Coulson et al. (1983). Over the first years it appeared that some of the smaller males had been wrongly sexed, indicating that Dutch birds were on average larger than birds from the United Kingdom. After sexing birds with molecular techniques (blood samples) and during necropsies (gonadal inspection), a new cut-off point between the sexes of 116 mm was chosen, which fitted the bi-modal pattern in head length well (Camphuysen 2013). Breeding gulls were aged using plumage characteristics (Adriaens et al. 2021). Age was expressed in (annual) cycles, i.e. from the start of one primary moult to the next (1^st–4^th cycle, versus adults; following Adriaens et al. 2021).

Chicks were colour-ringed just before fledging, using the same rings as for adults but with different codes (first letter K or P), when the tarsus had reached a length of c. 60 mm, or when the intertarsal joint was thin enough (young chicks have a thick intertarsal joint, but when around 30–35 d of age, the leg becomes much thinner, just as in later life). To enlarge numbers of colour-ringed fledglings in anyone season, to allow the study of apparent survival and return rates with meaningful samples, an additional 40–80 free-running near-fledglings were captured annually in July. To boost the total number of Lesser Black-backed Gulls colour-ringed in the same general breeding area even further, c. 30 adults and c. 40 near-fledglings were captured annually in a large neighbouring colony at IJmuiden (52°27.5′N, 04°34.3′E; using the Y as first engraved character). Total numbers colour-ringed arrived at 428 adults and 757 fledglings on Texel (2006–2021) and 394 breeding adults plus 551 fledglings at IJmuiden (2008–2021). The result was a population of ‘known’ individuals, used to assess return rates, site- and mate-fidelity in either colony, but also to track foraging trips and to reveal timing and migratory pathways across ages throughout the annual cycle.

Colour-ring reading was conducted within and outside the colony during the entire prospecting phase (with increasing frequency, Jan–Apr), and during all field work associated with nest marking and nest controls (May–August, with declining frequency in Sep–Oct), to assess individual colony returns, pair bonds, (potential) recruits, prospectors and (known) nest owners (32,658 sightings in all; 2006–2021: 1226 colony visits, 4561 field hours of observation). Within and around the colony at IJmuiden, 31,932 sightings were obtained from prospecting through breeding (courtesy Fred Cottaar and co-workers; no field time recorded; 2008–2021: 716 field days with over 25 sightings), while another 4344 sightings were obtained from elsewhere along the flyway. In addition, a ‘worldwide’ network of colour-ring readers provided re-sightings of colour-ringed individuals all over the known flyway of this species. The 2130 ringed individuals yielded 73,700 re-sightings in 17 European and NW African countries. To track movements from colour-ring re-sightings for all age categories (1^st–4^th cycle and adults) as discrete groups, for each individual, each month (each year) the mean latitude and longitude positions were calculated and plotted against time throughout the annual cycle (Figure 1, Table S1).

A GPS tracking system was used since 2008, to study foraging whereabouts and flyways continuously and over the entire annual cycle. The system was developed at the University of Amsterdam and consists of c. 17-g solar-powered GPS tags, with a wireless ZigBee transceiver that communicates with a base station in the centre of the breeding colony (Bouten et al. 2013). The GPS tags were attached to the back of adult breeding birds using a Teflon three-strap permanent backpack harness configuration with one strap fitting across each wing and one strap below the crop. Birds were immediately released following standard biometric measurements, ringing, and secure placement of the tags. Data logged on the tags were downloaded to a base station and new tag settings could be uploaded to the instruments if needed. Speed, direction and time duration were calculated between consecutive GPS fixes. The GPS data were collected for a whole range of detailed studies, including flight energetics (e.g. McLaren et al. 2016, Shamoun-Baranes et al. 2016). Tracking individuals by using colour rings or by using GPS tags has its own specific pros and cons and depending on the research question the suitability of the techniques may differ (see e.g. Thorup et al. 2014 for a more detailed review). Here specifically, we used the GPS data to complement the migration data collected with colour rings (in which all age categories could be followed simultaneously, while GPS trackers were deployed on breeding adults only), as well as the use of foraging habitats, especially those away from land (no observers) or in areas where colour-ringed birds cannot be detected easily.

Time intervals between fixes longer than 60 minutes indicated data gaps, generally due to low battery power, and were excluded from analysis. The time interval between consecutive GPS positions was attributed to the first of two consecutive locations. This interval and the first GPS position was then used to calculate the presence of birds (hours) in 2′N × 3′E rectangles within 150 km around the colony. For this paper, we summarised seven years' worth of data (51 adult Lesser Black-backed Gulls, 24 males and 27 females), trapped and fitted with GPS tags between 2008 and 2014 on Texel. Individual differences in foraging whereabouts, just exemplifying individual site fidelity, were mapped for ten adult nesting birds, tagged at Texel in 2020 and 2021, now using plots of all fixes, but only for the first 30 d worth of data after tag deployment (i.e. late egg phase, well into chick phase).

Timing of egg-laying, monitoring incubation (egg phase)

From late April to early June, nest searches were conducted along a pre-set trail through fixed study plots, to mark 80–100 nests covering the main laying period. Exceptionally late nests (outside the normal distribution; mean ± 2 SD and beyond) were ignored in our studies (maximally 1–2 nests within the study area per annum). Nests containing at least a single egg were marked (unique numbers), photographed, described and a GPS position was recorded (latitude, longitude). Nest descriptions involved position in a dune landscape (at the top, along a slope (steep or gradual) of a dune, or in flat valley areas) and the amount of vegetation covering the nest bowl, assessed using an 8-point scale, with categories ranging from open ground nest (just a scrape in the soil) via more or less half-covered (25–75% cover) to completely covered in tall grass or bushes (invisible from above). For the analysis, nest cover categories were grouped as ‘open’ (<10% nest cover: category 1–2), ‘half covered’ (10–75% hidden in vegetation: cat. 3–5) or largely ‘hidden from view’ (>75% concealed by vegetation: cat. 6–8).

Over the years, 1356 nests were monitored until hatching (84.8 ± 12.5 nests/yr; ±SD), in these nests 3658 eggs were laid (228.6 ± 33.8 eggs/yr). During each nest visit, the presence and incubation stage of the eggs were checked and recorded into a logbook. For each egg, a prediction was made of the expected day of hatching (assuming an incubation time of 27 d), but the exact duration of incubation (d) was assessed and used for analysis if laying and hatching dates were both exactly known. Marked nests were visited every third day until all eggs had hatched, were lost, or were abandoned.

During incubation, laying date, clutch size (n/nest), laying sequence (A, B, or C-eggs), egg size (max. length [L], max. width [B], mm) to calculate egg volume (ml), egg depredation, egg mortality and hatching success were assessed. Egg measurements were all taken by the same observer or after calibration by a frequent coworker. Egg volume was calculated from linear egg dimensions, using Volume = kLB², where k is a constant (Barth 1968, Hoyt 1979). The constant (k) used was derived from Spaans & Spaans (1975), to allow for direct comparisons with earlier Dutch data: k = 0.5035. Differences in egg shape were ignored, even though some eggs were considerably rounder or more pear-shaped than others (Barth 1968, Makatsch 1974). Truly unusual eggs, such as dwarf eggs or bright blue eggs were recorded and described. The timing of egg laying was based on all eggs laid in marked nests within a season (expressed as mean or median laying date), whereas in some earlier studies the onset of laying (first eggs) for each nest has been used. If mean or median dates of clutch initiation would be preferred, the difference with data presented in this paper would mean a 1.5 ± 0.7d shift forward (1314 nests).

Trends in most vital rates over time were tested using linear least squares regression. Nesting attempts were grouped as ‘early’, ‘peak’ and ‘late’ relative to median egg laying for each year. The interquartile range (IQR) is presented as the laying peak (d), used to illustrate the level of synchronisation in a given season, where ‘early’ nesting attempts occurred prior to the 1^st quartile, ‘late’ nests after the 3^rd quartile, and ‘peak laying’ in between. Vital statistics over early, peak or late breeding attempts (timing relative to annual colony laying median) were compared by means of one-way ANOVA. To test whether hatching success (number of eggs/nest) was related to relative laying date (nest versus annual colony mean), a generalized linear mixed effect model with a Poisson distribution was used (R-package ‘lme4’). Model fit, and whether such a relationship was linear or quadratic, was compared with a log-likelihood test. To account for annual variation, year was included as a random effect.

Lesser Black-backed Gull chicks, Kelderhuispolder, Texel, The Netherlands (photo CJC, 27 June 2011).

From hatching to fledging (chick phase)

The colony is situated in a well-vegetated area and hatchlings quickly disappear in the vegetation if not somehow contained. To be able to study the development of chicks, a randomly selected subset of marked nests was fenced in with 50 cm high chicken wire to form an enclosure (minimum size for isolated nests: 2 × 2 = 4 m², ensuring inclusion of the chick provisioning patch near each nest and sufficient vegetation for chick cover), where chicks could be retrieved during colony visits every third day until fledging or death. The use of enclosures was a choice, even though unwanted side-effects may occur (Griffiths 1992, Gyimesi et al. 2016), because it would provide detailed data on growth, causes of mortality and fledging rates. Because it was our aim to obtain productivity data representative for the colony at large, enclosed plots should not be more nor less susceptible to (chick) depredation than unenclosed sections of the colony. Fortunately, the study colony was not (or known to be; pers. obs.) under particular stress from rats or other mammalian predators, in which case enclosure data might have resulted in biased data. Of the 1356 nests that were initially marked during laying, 37% were randomly selected to be enclosed (497 nests, or 31.1 ± 8.0 nests/yr, ±SD). In these enclosures, 989 hatchlings were monitored (61.8 ± 23.7 chicks/yr), of which 251 (25.6%) individuals fledged (15.7 ± 8.9 fledglings/yr).

Levels of chick survival, depredation, or mortality were based on regular enclosure controls: every third day until fledging. Hatchlings were allocated to the appropriate egg number when possible and marked with numbered aluminium tarsus rings, so that they could be recognised during handling. Aluminium rings were replaced by stainless steel and PMMA colour-rings (see above) later in life. Data gathered from chicks in enclosures to assess growth included structural size measurements (see below; mm/d), mass increments (g/d) and primary moult scores. Measurements included total head+bill length (mm), bill length (exposed culmen, 0.1 mm), wing (flattened, mm), tarsus (diagonal, mm, not in 2006; methods following Balwin et al. 1931) and body mass (g), all taken by the same observer or after thorough calibration by a frequent co-worker. The primary moult score was a description of the stages of flight feather growth using a 5-point scale, from 0 (down), 1 (feather shafts emerging), 2 (feather shafts opening), 3 (feathers between 1/3 and 2/3 final length), 4 (>2/3, not fully grown) to 5 (fully grown), following Camphuysen et al. (2009). With few exceptions (some specific projects), chicks were not sexed. Chicks were recorded as ‘predated’ when they had entirely vanished from the (intact) enclosure while very young (<10 days of age, incapable of escaping), or when partly scavenged remains were found at later age in or outside the enclosure on the first nest visited after an earlier control. Chicks were recorded as ‘died’ when intact remains were found on such a subsequent control (even if scavenged later). Chicks were recorded as ‘fledged’ when they reached 40 d of age; any individuals that died within the colony when older were logged under ‘first year mortality’.

Reproductive output (fledglings/nest) was calculated over all randomly selected, enclosed nests combined, separately for each year of study or for any other grouping with biological relevance, such as for nests with different clutch size or laying date (timing). The World Seabird Union and the Circumpolar Seabird Group prefer fledglings per nest over mean clutch size (F/CS×100) as productivity indices (PI) to facilitate comparisons between colonies and seabird species over larger geographical scales (CSG 2010). From that perspective, good productivity for ‘seabirds’ is ≥50%, poor productivity ≤ 10% and moderate productivity between >10% to <50% of eggs producing fledglings. As in the egg phase, trends over the years were tested using linear least squares regression. Mean values over (grouped) early, peak or late breeding attempts are presented and compared by means of one-way ANOVA. To test whether fledging rate (chicks/nest) was related to relative laying date (nest versus annual colony mean), a generalized linear mixed effect model with a Poisson distribution was used (R-package ‘lme4’). Model fit, and whether such a relationship was linear or quadratic, was compared with a log-likelihood test. To account for annual variation, year was included as a random effect.

Censuses of the colony, breeding densities and non-breeders

Colony counts were conducted late May (incubation) in 2009, 2010, 2011, 2015 and 2016, in one day, timed such that both sympatric breeding species had completed clutches and could be included. Active territories with at least one well-constructed nest (so-called ‘play nests’ were excluded using expert judgement) were counted within a 10 m wide strip-transect, walking band-transects through the colonies, using three observers tied together along an 8 m long rope, preceded by a navigator/recorder to lead the way. Nesting species were identified on the basis of observations ahead (birds flushed while approaching), but especially based on prey remains found on the territories (prey remains dominated by crab or shellfish in Herring Gulls, clean sites with fish remains or grass pellets characterizing Lesser Black-backed Gulls). During nest marking, the same method was used to identify the nesting species, but with the possibility of checking results during later nest controls. This showed that an error occurred of around 1.3% misidentified Lesser Black-backed Gull nests (n = 1316). During the colony census, clutch size or occasional brood size (hatchlings) were recorded. Densities were expressed as number of nests per unit area (per m² or ha) for c. 25–35-m long stretches of transect. Extrapolations for non-surveyed parts of the colony (between transects) were used to assess relative abundance and breeding numbers within study plots and within the study colony as a whole. The transect sampling technique provided insight in spatial patterns in species composition and breeding densities throughout the study area.

Recruitment, age at first breeding, colony exchanges

Prior to egg laying and throughout the breeding period, the colony was rigorously checked for colour-ringed individuals. Indications of immigration were based on ‘foreign’ rings observed (prospecting, likely breeding or definitely breeding) and similar scans in other colonies were used to obtain evidence for breeding dispersal. Age at first breeding was based on birds trapped on nests to be ringed, with age being based on plumage characteristics.

Resource base (foraging habitat and diet)

Diets were studied from spontaneously regurgitated matter (pellets, large chunks of regurgitated matter, partly eaten food remains, food boluses produced during handling of adults and chicks) and from chick-feed sub-sampled within the territories, each considered as individual samples. During colony visits, marked territories were always inspected for the presence of discarded prey items, and these were bagged, thereby keeping individual pellets and boluses apart as discrete samples, numbered, and kept frozen for later analysis. Additional prey samples (pellets mostly) were collected from study plots, to enhance the overall sample in each of the phases of breeding. With pellets, boluses and regurgitated matter, some easily and fully digested prey are overlooked, but the microscopical inspection of prey samples ensured that even very small remains (such as earthworm setae and minute otoliths) were detected. This paper is not intended as a diet paper, and therefore the results were simply summarised and expressed as ‘frequency of occurrence’ (FO, %, see Barrett et al. 2007) for individual prey types over all analysed prey samples, per phase in the breeding cycle.

In the following results section, we present our findings based on the above methods. To provide a comprehensive overview, wherever necessary, we also provide findings already published elsewhere.

Annual cycle and non-breeding distribution

Colour-ring data showed that most adults left the breeding grounds after late August (Jul–Oct), and that successful breeding birds were preceded by failed breeders and prospecting immature (sub-adult) non-breeders (Figure 1, Table S1). Winter (Dec–Jan) sightings of colour-ringed adult birds (580 individuals) were recorded in Spain (49.8% of the individuals), France (21.7%), Morocco (10.9%), Portugal (10.2%) and the UK (5.3%), with far fewer birds in Nigeria, Gambia, Senegal, Mauritania, Algeria, Tunisia, Italy or Belgium. The GPS tracking data largely confirmed these results, but indicated that the UK is of greater significance than colour-ring data had suggested (a likely effect of differences in ring-reading observation effort). Virtually all birds migrated in a south-westerly direction (c. 200°) and all age groups reached their southernmost destinations around January. Colour-ringed juveniles abandoned the breeding grounds late (Sep–Nov) and travelled on average 37% in kms further away than adults (Figure S1, Table S1). Later in life, as immatures (2^nd–4^th cycle), they still tended to travel further to the south than adults (up to 10–25% in kms). Many birds used stop-over sites in northern France (Aug–Nov), roosting mostly in industrial areas, for up to a few months before travelling towards their southernmost winter destinations. Immatures leaving the breeding grounds used these same stopovers earlier on (Jun–Sep), juveniles later than adults (Oct–Nov). Adults showed strong winter site fidelity, over a smaller wintering range than most immature birds: almost a quarter of individual adult re-sightings were reported in at least three years (up to 13 consecutive non-breeding seasons) in the same country, usually at the exact same location. Strong site-fidelity throughout the annual cycle (nests, stop-overs and winter destinations) was confirmed with GPS tracking data (Brown et al. 2021).

Most adults returned in March and April (range: Jan–May), followed by older immatures (3^rd and 4^th cycle, mostly April–May), that were more likely to travel towards their breeding grounds in summer than younger birds (1^st and 2^nd cycle; Figure S1). Prospecting immatures arrived later than most adults in the breeding colonies, but returned earlier every season when growing older. Juveniles that survived their first winter sometimes travelled to the northeast in May and June of their second year, but normally without reaching their natal breeding grounds (mean latitude ± SD: 48.8 ± 4.9°N; Table S3). Adults that returned early (Feb–Mar) on the breeding grounds assembled for weeks on roosts on nearby beaches, before the colony itself was visited (field observations and GPS tracking results). Adults that spent the winter in Africa arrived last (later than birds wintering within the temperate zone), sometimes not before May, to join colony life (Shamoun-Baranes et al. 2016, Kentie et al. 2023). Pair bonds were seen to be established or renewed, at least partially, in assemblies or roosts on beaches around the colony, often well before territories within the colony were re-occupied. The longest-distance migrants largely missed these earliest periods of pair-bond (re-) establishments. Overnight stays, territorial defence and nest-building within the colony area itself were observed no earlier than mid-April (GPS data and visual observations during regular colony visits), three weeks to a month before the first eggs were laid. The breeding grounds were in use from February (early arrivals) to October (late departures), but colonies were only occupied in April–August (earliest prospectors in March; Figure 1).

Figure 1.

Schematic representation of the annual cycle of Lesser Black-backed Gulls Larus fuscus based on systematic colony observations at Texel (2006–2021) and on migratory movements of adult birds that were colour-ringed at Texel or in IJmuiden (Netherlands). For details on mean values (bold, ± SD) and range (in parentheses) of timings and annual fluctuations therein, see Results. Inset: violin-plot of average latitudinal distribution of sightings of colour-ringed adult birds ringed at Texel and IJmuiden, throughout Europe and NW Africa. Median values are indicated with a dot. Main colony attendance: April–August, main wintering: period September–February.

Nesting habitat, pair-bond renewals, site and mate fidelity (pre-laying phase)

Lesser Black-backed Gulls established breeding territories mainly on grassy, flat surface areas (44.8%) or on gradual hill slopes (52.3%), largely avoiding steep slopes (0.7%) and dune tops (1.3%; 1356 nests), thereby differing from the sympatric Herring Gull that had a larger preference for steep slopes and dune tops. In all, rather few (11.1%) of the 1356 marked nests were open (poorly covered), another small fraction (8.9%) were largely or completely hidden from view.

Of 42 males with known pair-bonds lasting more than one season (range spanning 2–14 years per individual), 29 birds (69%) were with the same female partner in subsequent seasons (range: 2–6 years), 13 changed a partner once (2–8 years) and two birds had three partners (5–14 breeding years). Nine partner changes were the result of a divorce (47.4%), ten followed the death of the initial partner (52.6%). Similar statistics were found for 42 females with known pair-bonds lasting more than one season (range: 2–10 years): 27 (62.3%) with the same (male) partner (2–5 years), 13 with one partner change (31.0%, 2–10 years), one with three partners (over 6 years) and one with four partners (6 years): 12 changes (66.7%) occurred following a divorce, six (33.3%) because the earlier partner died. These results suggest that the likelihood of encountering the same pair was 87% from one year to the next and 79% over a three-year period, eventually declining to 45% over five years, and to below 20% over longer time intervals.

Lesser Black-backed Gulls displayed a high degree of nest site-fidelity, generally choosing breeding sites but also club sites within or very near the study plot where they were found as breeding birds when they were ringed. Based on colour-ring data, only five re-nesting males (8%) abandoned their original study plot, and these were always small displacements (20–40 m) in immediately adjacent nesting areas. Area displacements between nesting attempts were commoner in females (8 cases, 16%), but again, most changes were minor and towards neighbouring study plots. With 30 study plots used for ring-reading in the main study colony and with 1223 individuals ringed, re-sightings were typically within the same or perhaps in one other, neighbouring plot. Only 159 (13%) were seen in three or more study plots and only 52 (4%) in four or more sites. Multi-year GPS tracking data confirmed that re-nesting attempts by adult breeders far away from initial (ringing) sites in later seasons were rare. Of 110 subsequent breeding attempts or confirmed territoriality (one year to the next), 90% re-nested within only 20 m of the original nest site, 96% within 40 m (median distance between subsequent nests: 3 m, IQR: 1.4–8.6 m). Longer distances between two subsequent nesting attempts of GPS carrying individuals were rare but included 87 and 118 m in females and 154, 194, and 210 m in males. One exceptional bird, a female, GPS-tagged in 2012 and re-nesting on the exact same location in 2013 and 2014, was rediscovered in 2017 as a breeding bird in Havergate NNR, Orfordness, Suffolk (UK), 238 km to the WSW of Texel, and returned there annually until at least 2022.

Laying date, clutch size, egg volumes, incubation and hatching success (egg phase)

Laying was highly synchronised (peak duration: 8.7 ± 2.0 d, ±SD), but the median laying date, initially around 12 May became progressively later after 2008 and stabilised on c. 20 May since 2013, widening the gap in laying with the consistently earlier nesting Herring Gulls from 2.9 ± 0.8 d (2.3–3.8 d) in 2006–2008 to 9.2 ± 3.2 d in 2010–2021 (7.3–12.7, but 4.3 d in 2019 as a result of a particularly ‘late’ season in Herring Gulls). The delay in egg-laying was highly significant, and while the duration of the (short) laying peak was variable, there was also a significant tendency for the peak to become wider over the years (Table 1, annual details in Table S2).

Clutches contained on average 2.7 ± 0.1 eggs (excluding disturbed nests followed by relaying: 7.4% single egg clutch (C/1), 18.9% C/2, 73.7% C/3, 1324 nests). Clutches with four or even five eggs were only encountered during the colony census (0.2% of nests checked, see below). Clutch size varied slightly, with no consistent change over the years (Table 1).

The difference in volume between A-eggs (first egg in a clutch; 74.3 ± 6.2 ml, n = 878) and B-eggs (76.8 ± 6.5 ml; n = 617) was small and not significant, whereas C-eggs (third egg in a clutch; 70.2 ± 6.4 ml, n = 707) were significantly smaller (t₁₃₂₂ = 10.4, P < 0.001). Dwarf eggs (<50 ml) were rare (<0.1%, n = 3979), bright blue eggs occurred 20 times (0.5%), of which 12 eggs (four clutches) were produced by the same female. By volume, A-eggs in two-egg (1.0%) and three-egg (4.3%) clutches were larger than in single egg clutches and this difference was considerably larger in late-laying pairs (5.6 and 10.5% larger, respectively; Table S5). Highly consistent was a long-term, significant decline in mean total volume of three-egg clutches from 225 ml to 215 ml (–4.4%) between 2006 and 2021 (Table 1).

Incubation typically commenced when at least two eggs were laid (egg temperature when touched, logbook notes). GPS tracking data had revealed that males remained longer at the nest than females during incubation bouts in the egg phase, thereby taking a greater share in the incubation process (7.3 ± 6.2 h versus 5.4 ± 4.7 h per bout in males (n = 387) and females (n = 514 bouts), respectively; Camphuysen et al. 2015). Eggs hatched on average after 26.2 ± 0.4 days of incubation, with the first hatchlings being found 30 May ± 3.6 d, median hatching around 11 June ± 3.7 d. Hatching success was fairly stable throughout the study period (71.0 ± 8.4% of eggs laid). There were no apparent trends in either of these egg parameters over the years of study (Table 1). Hatching success was similar for all eggs in which the sequence of laying could be established without doubt (A: 69.1%, B: 75.0%, C: 72.5%). Relaying frequencies varied (14.8 ± 13.7% of predated nests, range: 0–46.2%), but the likelihood more than halved when the laying window of the colony as a whole became substantially delayed (2006–2013: 20.3 ± 17.0% failed pairs relayed, 2014–2021: 9.3 ± 7.5% relaying).

Table 1.

Statistics describing the egg laying and incubation phase based on all marked nests in Lesser Black-backed Gulls breeding in the Kelderhuispolder, Texel, 2006–2021.

Table 2.

Statistics describing chick development in the chick care phase including fledging rates (n/pair) in Lesser Black-backed Gulls breeding in the Kelderhuispolder based on randomly selected nests in enclosures, Texel, 2006–2021.

Figure 2.

Reproductive success (chicks fledged/nest, grey bars) and factors reducing the output, portraying losses from the initial clutch size (i.e. entire bar length) as negative values.

Egg predation amounted to 21.0 ± 8.1% per annum (Table 1), varied between 15 and 30% in most years, and one third of the eggs that were predated disappeared within five days after being laid. As expected, nest cover reduced the risk of egg predation (27.4% of the eggs in open nests, 21.0% in half covered nests, versus 11.7% in clutches that were almost completely hidden from view; G_adj = 12.6, df = 2, P < 0.01, 3642 eggs). There was some anecdotal evidence for egg depredation by Carrion Crows Corvus corone roaming the colonies, but by far the most eggs were taken by conspecifics or Herring Gulls (pers. obs.). A-eggs (30.7%) suffered higher losses from predation than B (20.3%) or C-eggs (16.9%), and late clutches (33.8%) suffered more than eggs laid early (20.3%) or in the peak period (21.7%; all Table S5). Predation levels per egg were much higher in single-egg clutches (78.8% of all eggs laid, n = 99) than in two-egg clutches (33.8%, n = 494) and three-egg clutches (15.9%, n = 2927). The relatively high hatching success in C-eggs is the result of the overall much higher success rates of three-egg clutches (76.7% of all eggs hatched) over two- (67.1% hatched) and one-egg clutches (23.1% hatched).

Less than a tenth of all eggs produced were addled or died during incubation and did not hatch (7.9 ± 1.9%). However, of all eggs recorded as ‘predated’ of which the laying date is precisely known, 12.1% were taken after the expected date of hatching (>27 days after being laid; n = 711). Adding these eggs to the category that died, would lower the fraction that was predated to 18.7 ± 8.2% and increase the proportion of eggs that failed to hatch to 10.3 ± 2.3%. After correction for addled eggs predated after 27 days (see above), 53% of the eggs were predated within the first 9 d after nest initiation, 24% during the next and 23% during the nine final days prior to hatching. Egg predation was particularly high in 2013: 41.4% (n = 215, or 37.7% after correction for eggs taken >27 d after being laid), which led to an all-time low in hatching success of 51.6%.

Overall, 1.92 ± 0.29 eggs/pair hatched to produce chicks of the original clutch of on average 2.70 ± 0.14 eggs (interannual variability in Figure 2).

Hatchling mass, chick growth, chick mortality, fledging success and productivity index (chick phase)

Overall, 0.49 ± 0.23 eggs/pair fledged chicks of the original clutch of the on average 1.92 ± 0.29 eggs/pair that had hatched (c. 25% of all hatchlings; interannual variability in Figure 2). The productivity index (PI), using fledglings over clutch size (CSG 2010), showed that reproductive success was moderate to poor in all years (16.3 ± 8.2, range: 3–40%), even in the ‘best’ season (2020). Single egg clutches fledged not a single chick, two-egg clutches fledged 0.25 chicks/nest, whereas three-egg clutches were the most productive (0.6 chicks/nest; Table S5).

Hatchlings weighed 56.2 ± 1.9 g, but their body mass declined significantly over the years (Table 2, annual details in Table S4), likely as a result of the progressively smaller eggs that were produced (see above). Duties were shared between nest partners and GPS tracking data revealed that males spent only marginally longer at territories than females in the chick phase (3.8 ± 3.4 h versus 3.0 ± 2.8 h per nest bout in males (n = 698) and females (n = 1008 bouts), respectively; Camphuysen et al. 2015). While hatching success was similar for all properly sequenced eggs (A, B, C), rather fewer hatchlings fledged from C-eggs (22.7%, n = 172) than from either A- (31.6%, n = 190) or B-eggs (32.0% n = 153), but the difference was not significant (G_adj = 4.65, df = 2, P > 0.05; Table S5). Two thirds of the breeding pairs lost all chicks or even failed to hatch any eggs (Table 3). The body mass of chicks at around 40 d of age (while fledging) was highly similar, between pairs fledging one, two or the maximum of three chicks (Table 3).

Table 3.

Brood size fledged in Lesser Black-backed Gulls, Texel, 2006–2021 based on randomly selected nests in enclosures, and the mean mass (± SD and range) of fledglings against brood size.

Table 4.

Chick size with age (left) and growth increments to reach that size (right) in 5-day intervals for Lesser Black-backed Gull chicks raised and successfully fledged in the Kelderhuispolder, Texel, 2006–2021 based on randomly selected nests in enclosures. Growth increments are based on individual assessments every third day from first handling (usually 0–3 d of age) until fledging (40 d). In the second half of the table, biometrics of adult breeding individuals are provided, to be compared with the size of various body components (mm) and the body mass (g) of (unsexed) fledglings around 40 days of age.

About one fifth of the chicks died from starvation, disease or for other (unknown) reasons (21.7 ± 9.2%). Chick body mass gain was least variable, and usually uninterrupted during the first 20 d of life (Figure 3). Logistic growth models, based on chicks that fledged successfully, typically showed a point of inflection at c. 15 d of age (period of linear growth: c. 10–20 d of age). Between day 10 and day 20 (linear phase of growth) body mass increased by on average 24 g/d (Table 4). Of 3686 measurements over the years of young chicks (<21 d of age), only 5.5% demonstrated mass loss between subsequent assessments. In later stages of chick care, mass losses between subsequent checks became increasingly more common: age 20–25 d: 20.3% (n = 644), 26–30 d 24.7% (n = 543), 31–35 d 27.6% (n = 431), 36–40 d 43.2% (n = 315) of the measurements indicating mass loss. Raising a full brood (three chicks) from hatching to fledging was rare: 2.3% of all cases (n = 497 nests).

Figure 3.

Box plot of body mass increments with age of chicks that fledged successfully (n = 296), compared with median body mass of (sexed) breeding adults (n = 538) ringed at Texel, 2006–2021 (inset). The body mass of 234 fledglings weighed between 36 and 45 days of age averaged 658.7 ± 132.4, range: 300–971 g. The body mass of unsexed adult breeding birds averaged 826.5 ± 105.1, range: 630–1065 g.

With 2.70 ± 0.14 eggs as mean clutch size over all seasons, 37.1 ± 9.1% of all losses between laying and fledging could be attributed to chick depredation, making this the prime factor influencing fledging rates. Chick predation, largely by conspecifics and by Herring Gulls, but occasionally by raptors such as Marsh Harriers Circus aeruginosus, Goshawks Accipiter gentilis and perhaps crows (Corvidae), varied around 52.8 ± 13.2% over all hatchlings, with a significantly declining trend over the years. Chick depredation could reach very high levels such as 67% (2007, 2009) or even 78% (2017) of all hatchlings in some seasons. In such years, chick depredation occurred mostly in sudden, particularly violent, often initially localised but gradually spreading cannibalistic events during which, when observed, numerous agitated, fighting adults were involved, pecking at and eventually killing free roaming chicks. During such outbreaks, chicks were attacked and killed or torn apart by adult gulls rather than eaten, although carcasses were later scavenged (gulls typically aim at the intestines and fleshy thighs). Major outbreaks tended to spread like oil-slicks, with some parts of the colony severely impacted, and other sectors relatively spared. The advantage of nest cover was absent during chick care. More or less open ground nests did not suffer from higher chick predation rates (57.0% of the chicks) than the half-covered nests or nests that were largely hidden from view (51.9 and 53.5%, respectively; G_adj = 0.22, df = 2, P > 0.05, 1019 chicks). Overall, 49.3% of all 515 chicks with known hatching dates were predated before they had reached an age of only 10 d, a further 29% were taken before 20 d of age, another 18% before 30 d and only 4% during the last 10 days prior to fledging. The median age of chicks when depredated was relatively high (16–18 d) in 2006, 2008 and 2021, very low (3–4 d) in 2019 and 2020, and varied between 6–13 d in the other years.

At c. 40 d of age, most chicks were capable of flight and could leave the enclosures at will. Around that age, while flying around, the offspring were still considerably smaller than the adults. Head and bill were around 80–90% of adult size, the wings were c. 70% and the body mass was c. 80% of an adult bird. Only the feet (tarsus) were fully developed (98%; Table 4). While head length in adults differed between males and females such that a barely overlapping bimodal frequency distribution occurred (see Coulson 1983), the size-distribution in free running near-fledglings (exact age unknown), captured for colour-ringing, was normally distributed with a considerable spread: mean: 105.8 ± 5.8 mm, range: 88–121 (n = 954). The same was true for properly aged chicks in enclosures at c. 40 d of age (mean: 103.5 ± 5.9 mm, range: 88–120, n = 179). Structural size (head, wing) and mass around fledging declined over the years, but this decline was only significant when considering chick body mass (Table 2). A similar (non-significant) decline was found when a body mass index was calculated (body mass corrected for size, i.e. using head length as proxy; ANOVA χ²₁ = 3.53, P = 0.06).

Breeding densities

Breeding numbers of Lesser Black-backed Gulls within the two adjacent valleys of the Kelderhuispolder colony (15 ha), ranged from 2336–4817 (mean: 3640 ± 1025) ‘active’ territories in each of the colony censuses (2009–2016). Breeding densities in 2015 and 2016 (hence, breeding numbers), were distinctly lower than in the earlier counts (2009, 2010 and 2011): 0.015 and 0.019 nests/m² (156–193 nests/ha) in 2015–2016 versus 0.025–0.032 nests/m² (250–320 nests/ha) in 2009–2011. Of 2810 inspected nests during area censuses, 5.7% were single egg clutches, 14.7% 2-eggs, 54.2% 3-eggs, 0.2% 4–5-egg clutches and 23.5% were territories with empty nests. Empty nests were more frequent in 2011 (33.1%, n = 483) and 2016 (32.8%, n = 808) than in 2010 (18.6%, n = 802) and 2015 (11.9%, n = 707), so non-breeding varied irrespective of fluctuations in breeding densities. Peak densities of up to 0.14 nests/m occurred in sections of counting strips averaging 298 ± 131 m² in surface area, and such high densities occurred only in the flat ground valley areas, or on the lower gradual dune slopes.

Consequences of nesting late

Nest cover (ranked along an 8-point scale) was similar, irrespective of laying date, but clutches produced in late nests were significantly smaller than in earlier breeding attempts (Table 5). Egg size, expressed as three-egg clutch volumes, did not differ between early, late or peak-produced clutches. Both the number of hatchlings and the number of fledglings were significantly reduced in late nests. but the penalty for being late was most obvious in the chick phase. Early nests had the lowest incidence of failed broods (54.2%), while 87% of the late nests did not fledge a single chick (Table S7). From these late nests, just under 10% of the hatchlings fledged, 21% died and nearly 70% was predated. Of the early nesting birds, over all seasons, 7% fledged full broods (three chicks), against 2% of nests from the peak laying period and none of the late-nesting birds. Hatching and fledging rates and egg volumes for properly sequenced A-, B-, and C-eggs for gull pairs incubating 1-, 2- or 3-egg clutches, all relative to (grouped) laying date, are tabulated in Table S5.

Table 5.

Vital statistics describing nest cover, egg production, hatching and fledging rates, failures and brood size (fledglings) relative to the timing of nesting attempts in Lesser Black-backed Gulls, Texel, 2006–2021.

The number of eggs hatched per nest had a significant, linear, negative relationship with laying date (χ²₁ = 16.63, P < 0.001; Figure 4A, Table S4). The number of chicks produced per nest, showed a significant quadratic relationship with median laying date (χ²₁ = 12.98, P < 0.001; Figure 4B, Table S4). The earliest nests produced slightly fewer chicks per nest than those that were initiated later, but still before the median laying date. Note that the confidence intervals for early nests are wide, indicating variation in fledging success for early nests. Nests that were initiated after the median laying date hardly produced any chicks, with low confidence intervals around the prediction.

Recruitment, age at first breeding, colony exchanges

Of the breeding birds trapped for colour-ringing (with a breeding history otherwise unknown), 0.4% were in late 3^rd cycle, 1.0% in 4^th cycle and 98.6% were in adult plumage when captured (n = 507). Immatures were even uncommon as prospecting ‘visitors’. During ring-reading, 78 colour-ringed immature individuals were seen, including seven for the first time as 2^nd cycle (9.0%), 38 as 3^rd cycle (48.7%) and 33 as 4^th cycle birds (42.3%; all true age from rings). One of these was ringed as chick in Zeebrugge (Belgium; 1.4%), one at Helgoland-Düne, Schleswig-Holstein (Germany; 1.4%), two in IJmuiden (2.9%) and the rest originated from the study colony itself (94.3%). At the nearby colony of IJmuiden, where breeding success is not measured, but where visual observations of assemblies of fledglings in August 2008–2021 indicated that the reproductive success is much higher than on Texel (Fred Cottaar pers. comm.), a higher proportion of breeding subadults was found during ringing: 1.4% in 3^rd cycle, 5.6% in 4^th cycle and 93.0% in adult plumage (484 incubating individuals). Subsequent ring-reading activities in the colony IJmuiden yielded nearly 1634 sightings of immature prospecting birds, representing 162 individuals: two (1.2%) for the first time during their first cycle, 59 (36.4%) as 2^nd cycle, 83 (51.2%) as 3^rd cycle and 18 (11.1%) as 4^th cycle individuals.

Figure 4.

(A) Number of hatched eggs and (B) fledged chicks in relation to relative laying date (annual colony median laying date). The light grey area represents the confidence interval. Early nests, compared to the median laying date, hatched significantly more eggs than late nests, while nests produced just after the first nests, but before the laying peak, fledged most chicks.

Of fledglings ringed on Texel prior to 2018, only 91 individuals were confirmed to have returned as a prospector or to breed (9.2%, n = 994). For 17 chicks raised at Texel (1.7%), breeding evidence was obtained from other colonies: two on Helgoland (FRG, 247 km distance), four in IJmuiden (61 km), five individuals elsewhere in the Dutch Wadden Sea (4–129 km), and six in urban and industrial areas, deeper inland in The Netherlands (66–147 km). Three birds ringed as breeding adults were found breeding in later years in the UK (238 km), in IJmuiden (61 km) and near Petten in Noord-Holland (30 km). Confirmed breeding at Texel of birds fledged in other colonies included 11 individuals that had fledged in industrial areas at Europoort/Maasvlakte and Moerdijk (125 km distance), one from Zeebrugge harbour (Belgium, 212 km), one from Helgoland (FRG, 247 km) and one from Havergate in the UK (238 km). Seven other ringed individuals from other colonies were seen as prospectors on Texel and four juveniles fledged from Texel were seen prospecting other breeding sites.

Resource base (foraging habitat and diet)

Using 13,624 hours of GPS tracking data of 51 adult Lesser Black-backed Gulls (24 males, 27 females) within 150 km from the colony (Apr–Jul, 2008–2014), it appeared that 50% of the time spent away from the colony was recorded within a radius of action of 28 km (85% <41 km, 95% <72 km). Most foraging activities, sexes alike, were within a crescent shaped area to the west, southwest and south of the colony (Figure 5). As a result, the overlap in feeding whereabouts with Lesser Black-backed Gulls breeding at Vlieland and Terschelling (53°14′N, 04°55′E and 53°25′N, 5°27′E) or IJmuiden was minimal (Sage 2022). The foraging habitat of birds breeding in Texel ranged from pelagic to inshore, coastal and almost entirely terrestrial. ‘Time spent’ includes commuting (flight) and roosting. Terrestrial foraging activities focussed on ‘old’ polders on the mainland (often circumnavigating ‘modern’ polders with larger farms and monocultures that were constructed over the past 100 years), and waste or wastewater processing plants.

GPS tracking data also revealed considerable individual variation in foraging locations. Strong site-fidelity plus a tendency to behave similarly across years occurred primarily in individuals foraging in terrestrial habitats, where foraging flights were usually highly directed flights to and from frequently visited foraging opportunities (Figure 6). Amsterdam was used as a distant urban foraging area by some specialised individuals, or opportunistically by other birds. Individual foraging site-fidelity in urban areas was often quite extreme, visiting the same bridge, square, building or street over and over. In more recent years (2017–2021), much longer inland foraging trips were recorded, that often went beyond the Amsterdam region and well into Friesland (especially in 2019), than in the earlier years of GPS tracking (2008–2016). This material still awaits further analysis. The North Sea was used for ‘wide-range searching’ (pelagic or free-swimming surface prey) and trawler visits (discards), while the Wadden Sea was almost exclusively used to forage behind shrimpers (discards). Intertidal mudflats, important foraging grounds for numerous other waterbird species in the Wadden Sea, were utilised only by one specialist.

Figure 5.

Prime foraging distribution based on GPS tracking data (time spent (min) per 2 × 3′ rectangle) for adult male (24 individuals, 29,106 h) and female (27 individuals, 46,503 h of tracking) Lesser Black-backed Gulls breeding at Texel (Kelderhuispolder), 2008–2014, Apr–Jul data combined.

Figure 6.

Examples of individual foraging trips using GPS tracking data of 10 individuals during the first month after capture (late egg phase and half-way into chick care phase). Four-letter codes represent colour-ring codes and colours match the colours used on the maps. Left two maps: 3 June – 3 July 2020, right map 5 June – 5 July 2021.

Prey items collected around nests confirmed the use of habitats as indicated by the tracking data (Table 6). The dietary spectrum was enormous (307 prey species or types) and varied in response to availability (seasonal, diurnal, tidal, lunar cycles, otherwise), but probably also by (individual) preference. Of all prey samples collected (8203 samples), at least 82.2% (FO) contained marine prey, 31.0% contained terrestrial prey. The third most important category was waste material with an anthropogenic origin (FO: 7.3%). Prey types normally found along the tideline or on the intertidal zone were very scarce (FO: 3.0%, while abundantly represented around nests of the sympatric Herring Gulls). Another rare category was freshwater prey (FO: 1.4%), in which 75% of remains found around marked nests (105 samples) originated from one specialised individual bird or pair in two seasons.

The most frequently encountered marine prey were roundfish (FO: O 67.7%, n = 6739) and flatfish (FO: 52.5%), which were mostly Whiting Merlangius merlangus, Horse Mackerel Trachurus trachurus, Grey Gurnard Eutrigla gurnardus, Plaice Pleuronectes platessa, Dab Limanda limanda and Dover Sole Solea solea, most of which are demersal fish species, only available as discards from large beam trawlers. Marine crustaceans were represented in 27.2% of all prey samples holding marine prey, and these were typically swimming crabs (primarily Swimming Crab Liocarcinus holsatus), while marine polychaetes (FO: 14.7%) formed a final important group, mostly represented by Eunereis longissimi.

Table 6.

The origin and frequency of occurrence (FO, %) of remains of principal prey of Lesser Black-backed Gulls breeding in the Kelderhuispolder, sorted by relative abundance per breeding phase, Apr–Aug 2006–2021.

From pre-laying through the egg phase and into the chick phase, the frequency of occurrence of particularly lean (Gadidae, Syngnathidae, Trachinidae), difficult to handle (Belonidae, Bothidae, Pleuronectidae) or bony fish (Callionynidae, Cottidae, Triglidae) species declined, whereas fatty, small pelagics (Ammodytidae, Clupeidae, Osmeridae, Scombridae) doubled during the chick phase (Table 6). Swimming crabs and marine polychaetes typically occurred in pulses, pointing at shifts in availability for foraging gulls at sea.

Species or types of most frequently recorded terrestrial prey were various insects and spiders (49.7%, n = 2541), plant materials (mostly livestock food from farms, seasonally also Empetrum nigrum berries), Oligochaetes (mostly earthworms Lumbricus terrestris, 21.6%), plus non-passerine birds (mostly gull eggs) among some 95 other, less commonly encountered terrestrial prey types. A remarkable peak in small rodents (mostly Common Vole Microtus arvalis) was recorded in 2019 (12× usual levels given long-term results, FO in 2019: 14.3%, versus 1.9 ± 1.8% over all other years).

Among the anthropogenic waste materials (597 samples) were mostly plastic fragments, bread seeds, plastic line, thread, plastic packaging, bread, chicken, plastic pellets and paper, plus 75 other, less frequent types, which points at a mixture of items from a marine origin (free floating plastics), rubbish bin contents (most likely from urban environments) and human interventions (e.g. feeding birds bread). The likely origin of food identified as garbage were cities and pickings from holiday resorts and one waste processing plant on the mainland (Opperdoes, 36 km to the SE). Open landfills do not exist within the foraging range around Texel.

The frequency of occurrence of terrestrial prey in analysed samples (not considering species composition) was similar throughout the study, with no apparent trend (r² = –0.067, P = 0.68, P > 0.05.). The occurrence of marine prey, however, declined substantially and significantly, from a representation between 80 and 95% in all food samples in earlier years to around two-thirds or sometimes even less in recent years (r² = 0.399, P = 0.009). Prey items with an anthropogenic origin, even though consistently at low levels, increased in frequency (r² = 0.269, P = 0.033).

Annual cycle

Lesser Black-backed Gulls are long-distance migrants that leave their breeding grounds in early autumn and return in spring. Colour-ring data and GPS tracking results revealed a south-westerly flyway of this subspecies up to NW Africa (Morocco to Gambia and Senegal), but with the majority of the birds wintering in Europe (mainly Portugal and Spain; Brown et al. 2021). This is in accordance with even the earliest published work (Landsborough Thomson 1924, Barnes 1953, Harris 1962). Shifts in wintering areas have been reported some decades ago, with an increasing proportion remaining in the UK or in northern France in winter (Lack 1986, Alerstam 1990). That perceived shift northwards, estimated at 1.6 latitudinal degrees or 150–200 km per year between 1965 and 1975, has not continued (as predicted in Alerstam 1990, p. 213). In fact, the entire original flyway still exists (Brown 2022), with more northerly breeding birds leapfrogging populations breeding in The Netherlands and in the UK (Hallgrimsson et al. 2021). Only birds wintering in NW Africa are almost entirely marine in their orientation (coastal fish factories, fisheries and offshore resources; own unpublished GPS tracking data). Birds wintering on the Iberian Peninsula have a large variety of foraging areas including open sea and the seashore (Oro & Ruiz 1997), but including numerous terrestrial options such as large, open landfill areas, rice fields, and olive plantations (Martin-Velez et al. 2020, Martin-Velez 2021). Birds wintering in France and in the UK are largely terrestrial in winter, where they mostly feed on landfills, pig farms, and agricultural fields (Lack 1986, Shamoun-Baranes et al. 2016, Baert et al. 2018).

The gradual, stepwise change in wintering destinations, range, and timing for birds with increasing age (1^st–4^th cycle versus adults), now only available from colour-ring re-sightings, supports the exploratory migration model (Baker 1978, 1980), in which, by a process of area exploration, young birds become familiar with a larger area than that they will eventually use as adults (Figure S1, Table S1). Juvenile birds had a different timetable and alternative routes compared to adults (see Kilpi & Saurola 1984, Helberg et al. 2009, Marques et al. 2009). Colour-ring data have provided evidence for strong site-fidelity as adults, with regard to stopover use and wintering destinations (as in Helberg et al. 2009), which would mean that individuals develop intimate knowledge of foraging opportunities at only a few preferred locations. GPS tracking data have confirmed that strong site-fidelity in stopover sites and wintering quarters in adults, with considerable individual variation in timing (Shamoun-Baranes et al. 2016, Brown 2022). Only a handful of tracked individuals changed their winter destinations completely from one year to the next, thereby suggesting that gulls can be flexible if needed (Brown et al. 2021). Unexpectedly, the annual cumulative distance travelled by long distance migrants did not differ significantly from individuals wintering further to the north (Shamoun-Baranes et al. 2016), indicating that the travel required to feed particularly in wintering areas was negatively correlated with distance travelled along the entire flyway.

Adults that returned in late winter and early spring assembled at nearby beaches weeks before they visited the colony itself, and initially only briefly on days with calm weather and overnight stays within the colony occurred no earlier than mid-April. As a result, the breeding grounds on Texel are in use from February (early arrivals) to October (late departures), but residency at the colony effectively spans only April–August. As long-distance migrants returned so much later than most birds that spent their winter within the temperate zone (Shamoun-Baranes et al. 2016, Kentie et al. 2023), this meant that the available time to (re-)establish pair bonds, to (re-)occupy and successfully defend territories, or to boost energy reserves required to produce eggs, were all compromised for the birds wintering furthest away.

The absence of gulls in the colony from September to March will discourage potential predators from staying in the area between seasons. Frequent rains in winter will likely wash away rotting prey remains that have accumulated around nests and on roosts and leftovers of dead chicks from the previous breeding season (in part, a hygienic reset). The complete absence of birds within the colony during the non-breeding season is typical for this site, but perhaps not a general rule for all colonies (see however Paludan 1951).

Breeding strategy

Historical published work on the breeding biology of the Lesser Black-backed Gull was conducted during that expanding phase of Lesser Black-backed Gull with colonisations around Europe (Lockley 1932, Darling 1938, Paludan 1951, Goethe 1957, Harris 1964, Brown 1967b, Hausmann 1969, Davis & Dunn 1976). Clutch size (c. 2.7), hatching success (c. 79%) and fledging rates (c. 1.2–1.8 fledglings/nest; Darling 1938, Davis & Dunn 1976, Wanders 1980) were revealed and assuming an adult annual survival of 85 to 90% (i.e. typical for long-lived large gulls; Schreiber & Burger 2002, Ross Smith et al. 2013, Dierschke et al. 2021), such life history parameters would certainly stimulate population growth. More recent studies were often short term (student) projects (de Wijs 1982, 1984, Bekhuis 1994, Calladine 1997), conducted as a sideline of experimental work such as food supplementation (Spaans et al. 1994, Bukacinski et al. 1998), cross-fostering (Harris 1970), courtship behaviour (Brown 1967a), GPS tracking and diet studies (Garthe et al. 1999b, Gyimesi et al. 2016) or colonies suffering from culling (Duncan 1981, Wanless et al. 1996), floodings (Davis et al. 2018), botulism (Kim & Monaghan 2006), fox depredation (Wanders 1985, Davis et al. 2018) or other additional mortality factors enhancing variability in the measured reproductive success. Annual variations in reproductive output are difficult to appreciate from such short-term studies.

At the onset of the studies at Texel (2006–2010), while still investigating factors underlying the apparently ongoing increase of the Lesser Black-backed Gull against consistent declines of the sympatric Herring Gull, good reproductive output was expected, but the opposite was found (Camphuysen & Gronert 2010). Reproductive success at Texel remained low, certainly in comparison with historical work (cited above), in virtually all seasons that followed. At the time at Texel (2006–2011), with additive year effects rather than sex providing highest model support, apparent annual survival for both sexes combined averaged 91% (range: 81–100%; Camphuysen & Gronert 2012). Low reproductive success in combination with a ‘high’ (or normal) apparent annual survival of the adults breeding on Texel would lead to an ‘ageing’ population, but it remains to be seen how survival probabilities have changed in recent years.

Breeding densities and population trends

For colonial seabirds, one of the often-highlighted advantages of nesting synchronised and in high densities is: safety in numbers (collectively defending eggs and chicks when disturbed by predators; Danchin & Wagner 1997). In large gulls, however, most eggs and chicks are taken by conspecifics, so next-door neighbours may suddenly change from co-defenders to predators. Wilkens & Exo (1998) compared the reproductive success for Herring Gulls at two plots with different population densities at the island Mellum (German Wadden Sea island). With similar clutch sizes, the lower density area produced more offspring than the high-density area, largely as a result of higher (conspecific) chick depredation rates in the more densely populated parts. A three-fold increase in nesting density coincided with a significant reduction of reproductive success in Herring Gulls breeding on Terschelling (Dutch Wadden Sea), again, largely as a result of strongly increased levels of chick depredation by conspecifics (Spaans et al. 1987). The breeding density of the two (sympatric) species at Texel (2009–2016) was at least two times higher than that in the mixed colony at Terschelling at the time, but the reproductive output at Texel was even marginally higher than that measured at Terschelling (Spaans et al. 1987). There is no evidence for a shortage of nesting space or suitable habitat at the Wadden Sea islands in most of the larger colonies studied in dune areas along the North Sea in recent decades. Breeding density itself is probably not the issue. That populations in which the nesting density has increased suffer more from cannibalism than before, has probably more to do with increased levels of intra-specific competition for food: with a resource base now shared with more competitors.

Resource base and foraging opportunities

Central place foragers often segregate in space, even without signs of direct agonistic interactions: a common avoidance strategy among and within species from neighbouring breeding colonies (Wakefield et al. 2013, Sage 2022). Individual-level memory of resource availability can be sufficient to cause spatial segregation in the foraging range of neighbouring colonial animals (Aarts et al. 2021), but that seabirds from neighbouring colonies forage in largely mutually exclusive areas is often assumed to result from density-dependent competition (Wakefield et al. 2013). Lesser Black-backed Gulls equipped with GPS data loggers from six study colonies along the German North Sea coast did indeed forage in barely overlapping home ranges, and that range increased with increasing colony size (Corman et al. 2016). Colony-specific foraging patterns, habitats, and resources could reduce intra-specific competition. Similarly, with most foraging activities of Lesser Black-backed Gulls breeding on Texel directed to the west, southwest and south of the colony (Figure 5), an overlap in feeding whereabouts with large neighbouring colonies at Vlieland, Terschelling and IJmuiden was indeed almost non-existent (Ens et al. 2009, Sage 2022). As a result of the spatial segregation with birds breeding on Vlieland and Terschelling, the Frisian Front area (c. 53°45′N, 04°30′E), an important feeding ground for Lesser Black-backed Gulls (Camphuysen 1995, Baptist et al. 2019), was for Texel birds within range (c. 52 km to the NNW), but effectively ‘out of reach’, and thus rarely visited (Figure 5).

Individual foraging site fidelity, another widespread phenomenon in nature (Bormann et al. 2019), probably benefits individuals through higher foraging efficiency. It may arise as a consequence of dietary specialisation, and the resulting resource partitioning might guarantee semi-exclusive foraging opportunities for individual birds (Ceia & Ramos 2015). From the diet studies at Texel, it appeared that only few birds specialised on mammals (mostly rodents) or freshwater fish. Birds that otherwise shared a common resource with many others, often specialised on specific locations. So, while numerous birds foraged on farmland on Texel, or in old polders on the mainland, GPS tracking data typically showed that individuals used only two or three different ‘preferred’ feeding areas out of multiple options (given the population at large) for most of their time during breeding (Figure 5). Flexibility was demonstrated when several GPS tracked individuals suddenly ventured into Friesland grasslands in 2019, while the frequency of occurrence of common voles in prey samples throughout the colony peaked at 12× long-term average levels, during a major outbreak of common voles in agriculture grasslands (Wymenga et al. 2021). Flexibility was also shown in short-lived, but distinct peaks in the frequency of occurrence of prey types such as swimming crabs, marine polychaetes (only available during spawning bouts), ripening berries or insects.

Marine prey (primarily obtained in the North Sea and in deeper parts of the Wadden Sea) supplemented by food found in agricultural areas (Texel and mainland) and small amounts of junk food (garbage from urban areas mostly) currently forms the basis for Lesser Black-backed Gulls breeding at Texel (Table 6). In an earlier analysis of tracking data, it appeared that males travelled on average further from the colony than females, and spent more foraging time at sea, offshore (Camphuysen et al. 2015). Males fed mostly on fisheries discards at offshore trawlers while females foraged predominantly on land or near-shore and in the Wadden Sea around shrimpers (Tyson et al. 2014). Discards from large offshore beam trawlers (to a lesser extent nearshore shrimpers), formed the most important resource base, given prey remains found in c. 8200 food samples. For Lesser Black-backed Gulls at Terschelling in the 1990s, the presence or absence of small pelagics (clupeids) was at the time seen to influence chick growth and reproductive success most (Spaans et al. 1994), but even though small pelagics were frequent prey for birds breeding on Texel, they were too uncommon (in mass) to likely ‘drive’ success or failure for the breeding population as a whole. Beam trawlers typically return to the harbours for the weekend (Rijnsdorp et al. 2011) and discard-scavenging specialists, tracked with GPS, responded by switching to foraging on terrestrial habitats (Tyson et al. 2014).

The demonstrated dependence on discards, of this colony, but in fact of most colonies in this part of the North Sea (Flore 1999, Schwemmer & Garthe 2005, Sotillo et al. 2014), has shed a different light on the rather sudden increase in breeding numbers that occurred in the 1970s and 1980s. This increase followed the modernisation of the Dutch fishing fleet: from otter trawl to beam trawl that commenced in the 1960s (Rijnsdorp et al. 2008). Beam trawling is a discards-intense form of fishing (Van Beek 1998), that seemingly formed a sound foundation, at the time, for the colonising species all over the Southern and German Bights (see also Camphuysen 1995). A successful utilisation of discards from large (offshore) beam trawlers can explain the timing and the scale of the population increase throughout the 1970s–1990s. So, rather than outcompeting other gulls for existing foraging opportunities (Noordhuis 1987, Noordhuis & Spaans 1992), the increase was more likely driven by the successful exploitation of an emerging (marine) resource, associated with offshore beam trawlers (Camphuysen 1995, Garthe et al. 1999a). The decline in fishing effort, especially a shrinking beam trawl fleet, and the EU discards ban (Bicknell et al. 2013, Borges 2013, Anon. 2014, Kuhlman & Van Oostenbrugge 2014) is an ongoing and gradual process that commenced earlier this century, and that will reduce this anthropogenic resource base more and more. A consistent decline of prey samples containing marine prey over the study period, in line with various other signals in the breeding colony pointing at food stress, are clear indications that the birds breeding on Texel find it difficult to adapt within the constraints of their colony-specific foraging home range. An increasing tendency to utilise inland rather than marine foraging opportunities in Lesser Black-backed Gulls was reported from many areas in Europe, even though factual evidence was sometimes sparse, but indeed, some recently established colonies are even almost entirely terrestrial, and often with great success (Gyimesi et al. 2016, Sotillo et al. 2019ab). The birds breeding at Texel, despite their longer inland foraging trips in recent years, are seemingly unable to boost their current reproductive success with existing (gradually changing) foraging opportunities, as if still ‘hopeful’ for returning fishing fleets offshore.