Introduction
Gulls in the genus Larus are found in a wide variety of habitats, from urban environments and agricultural land to remote, uninhabited islands. They are omnivorous and feed on anthropogenic waste, naturally hunted fish, fish and offal scavenged from the fishing industry, intertidal organisms, and other seabirds and small mammals (Harris Reference Harris1980, Buckley Reference Buckley1990, Steenweg et al. Reference Steenweg, Ronconi and Leonard2011, Westerberg et al. Reference Westerberg, Brown, Eagle and Votier2019). This flexibility has made them highly resourceful and adaptable, and a common feature of many ecosystems. As observed by Anderson et al. (Reference Anderson, Shlepr, Bond and Ronconi2016), this adaptability has perhaps led ecologists and conservationists to a feeling of complacency regarding the conservation status and marked ecological knowledge gaps of Larus gulls. However, accurate and contemporary information about the ecology and status of populations is key to shape conservation policy.
Limited attention to the population trends of Larus species has been compounded by their status as “nuisance species” resulting in persistent negative perceptions towards them. They may negatively impact agriculture and fisheries or create human-wildlife conflict by stealing food, defending their offspring, or defecating on vehicles and street furniture. Gulls have been “managed” and culled in great numbers in the early 20th century. For example, over 800,000 Herring Larus argentatus and Great Black-backed Gull L. marinus eggs were destroyed in New England, United States, between 1934 and 1950 due to concerns over the potential impacts on crops and fishing weirs (Anderson et al. Reference Anderson, Shlepr, Bond and Ronconi2016). Similarly, non-lethal and lethal gull control is a common tool used at local scales in the event of wildlife conflicts, typically when gull populations negatively impact other bird populations (Anker-Nilssen and Tatarinka Reference Anker-Nilssen, Tatarinkova, Anker-Nilssen, Strøm, Golovkin, Bianki and Tatarinkova2000, Bosch et al. Reference Bosch, Oro, Cantos and Zabala2000, Guillemette and Brousseau Reference Guillemette and Brousseau2001, Finney et al. Reference Finney, Harris, Keller, Elston, Monaghan and Wanless2003). Oro and Martínez-Abraín (Reference Oro and Martínez-Abraín2007) examined the use of lethal control in gull management, the most prevalent approach, finding that it is often ineffective in the long-term and that its implementation is generally driven by the negative perception of gulls rather than scientific evidence. Large knowledge gaps regarding the foraging ecology of some Larus species can result in faulty assumptions about gulls’ impact on anthropogenic resources, further challenging the justification for these management measures.
Additionally, it has been argued that recent declines in Larus populations should be of no concern after increases in abundance observed in the 20th century were linked to anthropogenic activities, and populations may at present be returning to “baseline” levels of abundance. However, although the distribution of gull species was relatively well known prior to the 20th century, quantitative records of population size are scarce during that time (Holloway Reference Holloway2010). Therefore, it is extremely challenging to identify a “natural” baseline abundance of Larus populations, particularly for species that have long been associated with human settlements. Although there is evidence that anthropogenic resources have led to a significant increase of some populations, others do not rely on these resources and have experienced declines due to environmental change and anthropogenic impacts in a similar fashion to other seabird species (Regehr and Rodway Reference Regehr and Rodway1999, Stenhouse and Montevecchi Reference Stenhouse and Montevecchi1999, Davoren and Montevecchi Reference Davoren and Montevecchi2003). Accurate and objective classifications of the status and population trends of the species are therefore needed to inform management and conservation of their populations.
Accordingly, concern has grown regarding the declines seen in several, generalist Larus species across their range (Anderson et al. Reference Anderson, Shlepr, Bond and Ronconi2016). One such species is the Great Black-backed Gull, Larus marinus, which inhabits temperate and Arctic coastal areas on either side of the Atlantic Ocean, their core breeding area being the United States, Canada, Iceland, Scandinavia, and the British Isles. They are primarily diurnal foragers and exploit a wide variety of coastal and terrestrial habitats and food resources (Coulson Reference Coulson2019). The number of breeding Great Black-backed Gulls has fluctuated greatly throughout the 20th and 21st centuries. During most of the former, populations increased in both numbers and geographical range across the North Atlantic basin (Nisbet et al. Reference Nisbet, Auer, Veit and White2013). A remarkable example of this increase was reported by Drury (Reference Drury1974) in New England, USA, where the breeding population of Great Black-backed Gulls increased from 30 to 12,400 breeding pairs between 1930 and 1972. This trend observed in Great Black-backed Gulls was mirrored in other Larus species and was associated with a decrease in persecution and an increase in the availability of anthropogenic food resources, particularly landfill waste and fishery discards (Wilhelm et al. Reference Wilhelm, Rail, Regular, Gjerdrum and Robertson2016). More recently, however, steep declines have been recorded in the last 20–40 years. For example, between 2008 and 2013, the number of Great Black-backed Gulls in Maine, United States, declined by 30% from approximately 10,000 to 6,000 breeding pairs (Mittelhauser et al. Reference Mittelhauser, Allen, Chalfant, Schauffler and Welch2016). In eastern Canada, the estimated number of breeding pairs declined from approximately 41,000 to 12,000 between 1990 and 2014 (Wilhelm et al. Reference Wilhelm, Rail, Regular, Gjerdrum and Robertson2016). In Europe, similar declines have been observed in some countries such as Iceland, where the population declined from 22,500 to 7,000 pairs between 1980 and 2016 (BirdLife International 2015). Although these local and regional estimates suggest that the species is in decline, the species’ extensive breeding range and difficulty in censusing all populations has meant that accurate estimates of global breeding population trends have not been undertaken. Accurate and contemporary information about the status of Great Black-backed Gull populations is key to inform conservation and research priorities, as well as environmental impact assessments required in the regulation of some marine activities (e.g., renewable energy generation). Accordingly, in this study we sourced the most recent breeding population estimates from throughout the species’ range and analysed the change in abundance at a global, continental, and national scale over the most recent three-generation period, a time frame used to evaluate species against the IUCN Red List criteria.
Methods
Population counts
Population count data were sourced from 17 countries, spanning the entirety of the Great Black-backed Gull’s breeding range (Figure 1 and Table 1). For a small number of countries, states, or regions, where total population counts were not available, a sample from selected breeding colonies was used (see “Data analysis” for how missing populations were incorporated in the analysis). All population counts represented the number of breeding pairs only, excluding data from the non-breeding season and non-mature birds, in line with IUCN criteria.
Global breeding range of Great Black-backed Gulls Larus marinus (areas coloured black) and countries where population counts were obtained (dark grey). Breeding range data were obtained from BirdLife International and Handbook of the Birds of the World (Del Hoyo et al. Reference Del Hoyo2019).
Parameters used in the analysis. These include: two population counts per region (Count 1 and Count 2), the years these counts were obtained (Year 1 and Year 2), the error associated with each count (Error 1 and Error 2; see Table 2 for more information about error categories), and the locations each population count represents.
1 BirdLife International (2000)
2 BirdLife International (2015)
3 BirdLife International/European Bird Census Coucil (2020)
4 Bond et al. (Reference Bond, Wilhelm, Robertson and Avery-Gomm2016)
5 Cotter and Rail (Reference Cotter and Rail2007)
6 Lock (Reference Lock1979)
7 Fauchald et al. (Reference Fauchald, Anker-Nilssen, Barrett, Bustnes, Bårdsen, Christensen-Dalsgaard, Descamps, Engen, Erikstad, Hanssen and Lorentsen2015)
8 New York State Department of Environmental Conservation; unpubl. data
10 Guzzwell et al. (Reference Guzzwell, Wilhelm and Gjerdrum2015)
11 Hario and Rintala (Reference Hario and Rintala2016)
12 Massachusetts Division of Fisheries and Wildlife; unpubl. data
13 Mittelhauser et al. (Reference Mittelhauser, Allen, Chalfant, Schauffler and Welch2016)
14 Nager and O’Hanlon (Reference Nager and O’Hanlon2016)
15 North Carolina Wildlife Resources Commission; unpubl. data
16 Ronconi et al. (Reference Ronconi, Stephens, Crysler, Pollet, Fife, Horn and Taylor2016)
17 Watts et al. (Reference Watts, Paxton, Boettcher and Wilke2019)
18 Watts et al. (Reference Watts, Byrd, Watts and Byrd1998)
19 Rail (Reference Rail2018)
20 Washburn et al. (Reference Washburn, Elbin and Davis2016)
21 Chapdelaine and Brousseau (Reference Chapdelaine and Brousseau1991)
22 These population estimates were calculated as 10% of the global population in 1985. See “Data analysis” section.
23 CWS unpubl. data (Wilhelm)
24 Cotter et al. (Reference Cotter, Rail, Boyne, Robertson, Weseloh and Chaulk2012)
25 Lock (Reference Lock2003)
26 Regular et al. (Reference Regular, Wilhelm, Gjerdrum, Boyne and Robertson2015)
27 Boyne et al. (Reference Boyne, Toms and McKnight2006).
The geographical resolution of population counts varied across the species’ range. Country-level counts were available for Europe, whereas for the United States and Canada counts were generally available at a higher resolution (state or region; Table 1). Population estimates were collected using a variety of methods of different accuracy. Methods of higher accuracy included individual counts carried out at ground level, as well as boat and aerial surveys of breeding individuals. Methods of lower accuracy were typically linked to some European countries, where the data had already been classed by BirdLife International as “based on extrapolation from a subset of colonies” or “based on expert opinion” (see references from Table 1 for sources).
Selection of counts and uncertainty
Generation length, defined as “the average age of parents of the current cohort”, is used in conservation for assessing extinction risk. The IUCN Red List criteria contrast changes in population size over the most recent three-generation period of a species to quantitatively assign each species to an extinction risk category (IUCN 2012). Generation length in Great Black-backed Gulls has been estimated as 12 years (Bird et al. Reference Bird, Martin, Akçakaya, Gilroy, Burfield, Garnett, Symes, Taylor, Şekercioğlu and Butchart2020, IUCN 2022), so we estimated changes in Great Black-backed Gull abundance from 1985 to 2021 (36 years).
Because population counts were available at different geographical scales (country, state, regions), the term “population” is used hereafter to represent any geographical scale for which there was an individual set of counts (i.e. each input to the analysis presented in Table 1). We chose to use two counts per population, as this was the common denominator for all populations. As we were interested in presenting the overall change in abundance over the three-generation period rather than changes in abundance at a higher temporal resolution, the two counts closest to the beginning and end of the study period (1985 and 2021, respectively) were chosen for populations where more than two counts were available. This ensured consistency and the highest degree of accuracy in the analysis (see Data Analysis, below).
Each population count was allocated an estimate of uncertainty based on the method used to produce the count. This was done according to a five-point scale (Croxall and Kirkwood Reference Croxall and Kirkwood1979, Lynch et al. Reference Lynch, Ratcliffe, Passmore, Foster and Trathan2013) ranging from low uncertainty (N1) to high uncertainty (N5). We redefined the N5 category from Lynch et al. (Reference Lynch, Ratcliffe, Passmore, Foster and Trathan2013) to account for a potential margin of error of up to 100% in some instances as we were confident no population estimate would be erroneous by more than twice its size (Table 2). For data from the USA, Canada, and some European countries, where raw population counts were available from count coordinators or published literature, error categories were either assigned by the count coordinators themselves or selected based on the methodology detailed in the literature. Counts from the remaining 12 European countries were obtained from BirdLife International European Red List species factsheets (2021). These data had already been allocated an accuracy/quality classification by BirdLife International according to their own criteria. We therefore translated these classifications into our study’s error categories using a conservative approach whereby the highest possible degree of accuracy issued by BirdLife International (“Good - Based on reliable and complete or representative quantitative data”) was allocated to the third error category in this study (“N3- Estimate of nests with an accuracy of 10–25%”) (see Table 2 for conversion criteria). Such a conservative approach was taken to limit the potential to overestimate the accuracy of population trend estimates calculated from BirdLife International data.
Error categories, BirdLife International criteria conversion, and the descriptions used to classify population counts in terms of uncertainty.
Data analysis
All analyses were carried out in R version 4.0.4 (R Core Team 2020). A stochastic simulation was conducted for each population, which allowed us to account for the error associated with each population count to present confidence intervals around the estimated changes in population size. Random draws from the simulation provided two “estimated” counts per population, which were then used to calculate the finite rate of change of the population.
Simulation parameters were adapted from Lynch et al. (Reference Lynch, Ratcliffe, Passmore, Foster and Trathan2013) to fit our data in the following way: A fractional error (representing the uncertainty of the count) was randomly sampled from a uniform distribution defined by the error category associated with each population count.
$$ {\displaystyle \begin{array}{l}\left(\mathrm{N}1\right)E\sim Unif\;\left(0.00,0.05\right)\\ {}\left(\mathrm{N}2\right)E\sim Unif\;\left(0.05,0.10\right)\\ {}\left(\mathrm{N}3\right)E\sim Unif\;\left(0.10,0.25\right)\\ {}\left(\mathrm{N}4\right)E\sim Unif\;\left(0.25,0.50\right)\\ {}\left(\mathrm{N}5\right)E\sim Unif\;\left(0.50,1.00\right)\end{array}} $$
-
1. An estimated population count
$ (\hat{n} $
) was randomly sampled from a truncated normal distribution limited at 0 in the lower tail to avoid sampling negative population counts. This truncated normal distribution was defined by the original population count
$ (n) $
and the randomly sampled error (E) from the previous step.(Equation 2)
$$ \hat{n}\sim N\left(n,{\left[\frac{E}{2}n\right]}^2\right) $$
-
2. Equations 1 and 2 were bootstrapped 50,000 times for the two counts from each population. Using the estimated population counts in each year, we calculated the finite rate of population change (λ):
$$ \lambda ={\left(\frac{{\hat{n}}_2}{{\hat{n}}_1}\right)}^{\frac{1}{Year2- Year1}} $$
where Year1 was the year of the first population estimate (n1), and Year2 was the year of the most recent estimate
$ (n $
2).
-
3. Lastly, the estimated population sizes in 1985 and 2021 were calculated as follows:
-
a. The time gaps between
$ n $
1 and 1985, and
$ n $
2 and 2021 were calculated:(Equation 4)
$$ {Gap}_1= Year1-1985 $$
(Equation 5)
$$ {Gap}_2\hskip0.35em =\hskip0.35em 2021- Year2 $$
-
b. Once the time gaps had been obtained, the fractional change (Change1 and Change2) in population size between
$ {\hat{n}}_1 $
and 1985 and
$ {\hat{n}}_2 $
and 2021 was then calculated as follows:(Equation 6)
$$ {Change}_1\hskip0.35em =\hskip0.35em {\lambda}^{Gap_1} $$
(Equation 7)
$$ {Change}_2\hskip0.35em =\hskip0.35em {\lambda}^{Gap_2} $$
-
c. Lastly, the 1985 start (StartPopulation1985) and the 2021 end populations (EndPopulation2021) were calculated, as well as the fractional change in population size over the three-generation period (3GenPeriod):
(Equation 8)
$$ StartPopulation1985\hskip0.35em =\hskip0.35em \frac{{\hat{n}}_1}{Change1} $$
(Equation 9)
$$ EndPopulation2021\hskip0.35em =\hskip0.35em {\hat{n}}_2\times Change2 $$
(Equation 10)
$$ 3 GenPeriod\hskip0.35em =\hskip0.35em \frac{\left( End\ Population\hskip0.35em 2021- Start\ Population\hskip0.35em 1985\right)}{StartPopulation1985} $$
-
-
4. Equations 3 to 10 were repeated for each of the 50,000 bootstrapped population estimates to obtain distributions of estimates for each population for 1985 and 2021.
Due to the large uncertainty associated with the counts of some European countries and the need to use a normal distribution with a truncated lower limit of 0, the resulting distributions for some regions were marginally non-normal. To stop extreme, positive outliers from heavily influencing the mean statistics, counts outside the 2.5% and 97.5% quantiles were excluded in figures involving European data, which resulted in normal distributions.
There were populations for which no trend data were available (unsurveyed populations). These were: 27% of the Quebec population (Canada), approximately 40% of the New York population (USA), and the entirety of Delaware (USA). These figures accounted for <10% of our 1985 global population estimate. Two scenarios were used to include these figures in the analysis. The first approach (Scenario A) was conservative and assumed these populations had remained stable over the three-generation period. The second approach (Scenario B), likely more realistic, assumed unsurveyed populations declined at the same rate as the global average of the surveyed populations (Table 1). Mean statistics for global and continental estimates were calculated from the sum of every iteration from all populations that fell within that geographical area. Unsurveyed populations were only included in global estimates and were excluded from continental, national, and regional estimates. Finally, we compared overall population changes over three generations to IUCN Red List criterion A2, defined as “Population reduction observed, estimated, inferred, or suspected in the past where the causes of reduction may not have ceased OR may not be understood OR may not be reversible” (IUCN 2012). We did not apply criterion A1 because the causes of decline were not known, did not appear to have ceased, and were not reversible (see “Discussion” section). We also did not apply criterion A3 because we did not project populations into the future. Lastly, we did not apply criterion A4 because we used two count estimates per population and therefore a “moving window” scenario would not yield any additional decline estimates. Because we had population trend data for virtually the entire range (90% of the global population), we assumed that population trends appropriately reflected the global population.
Results
a. Global population trend
Under “Scenario A” (unsurveyed populations assumed to have remained stable between 1985 and 2021), the number of Great Black-backed Gull pairs assessed in this study declined from approximately 291,000 (95% CI: 261,000; 324,000) to 165,000 (95% CI: 146,000; 187,000), a three-generation change of -42.8% (95% CI: -31.4%; –52.9%), and an annual decline of 1.2% (95% CI: 0.8%; 1.5%) (Table 3). Under “Scenario B” (unsurveyed populations assumed to have declined at the same rate as the global average), the number of Great Black-backed Gull pairs declined from approximately 291,000 (95% CI: 261,000; 324,000) to 152,000 (95% CI: 132,000; 173,000), a three-generation decline of 47.5% (95% CI: -36.7%; -57.3%), and an annual decline of 1.3% (95% CI: 1.0%; 1.6%). According to IUCN Red List criterion A2 (applicable when the population decline has not ceased), the results from both scenarios correspond to a listing of Vulnerable (a decline of ≥30% over 10 years or three generations, whichever is longer), and Scenario B approaches the threshold of an Endangered listing (decline of ≥50%). Indeed, the upper 95% CIs from both scenarios overlapped with an Endangered listing and the lower 95% CIs did not drop below a Vulnerable listing.
Summary of changes in the abundance of Great Black-backed Gull pairs globally, in North America (Canada and USA), and in Europe (rest of countries from Table 1). Change is represented as the estimated total number of breeding pairs in 1985 and 2021, and the corresponding percentage change in population size and finite rate of change (λ) over that time. For global estimates, “Scenario A” assumes unsurveyed populations have remained stable between 1985-2021, whereas “Scenario B” assumes unsurveyed populations have declined at the same rate as the global average.
b. North American and European population trends
The three-generation decline was greater in North America than Europe (two-sample t-test: df =50,070, t = –772.37, P < 0.001), with a mean three-generation change of -68.7% (95% CI: –63.9%; –72.8%) compared to -28.1% (95% CI; –49.7%; –0.4%), respectively (Table 3).
c. Country and sub-country population trends
While there were populations that showed growth between 1985 and 2021, these tended to be smaller and/or were present within a larger state or country where most populations declined, such as the Bay of Fundy (75% increase) in Canada or New Jersey (1686% increase) in the United States. Generally, increases were recorded in populations at the edges of the species’ range, or in areas that have been recently colonised such as Spain, Germany, Netherlands, North Carolina (United States) or Virginia (USA) (Figure 2). Overall, however, declines were recorded across most of the species’ strongholds: Atlantic Canada, United States, United Kingdom, and Iceland. The exception was Norway, which apparently remained relatively stable, but large uncertainty surrounded the estimated change in abundance (Table 4; Figure 2).
Changes in Great Black-backed Gull abundance over the three-generation period 1985-2021 per population. Change is represented by the estimated number of breeding pairs at the beginning and end of the period, and the corresponding percentage change in the number of breeding pairs and finite rate of change (λ).
Bubble map showing the size and trends of Great Black-backed Gull populations per country (except Svalbard, which is shown separate to Norway) between 1985 and 2021. Circles reflect the size of populations and symbols reflect the population trend. Stable populations are depicted as those where the three-generation change has been less than 5% between 1985 and 2021.
Discussion
Globally, Great Black-backed Gull populations were estimated to have declined between 42.8% and 47.5% from 1985 to 2021, depending on the assumption regarding unsurveyed populations. This overall decline is consistent with trends observed in local and regional analyses from recent years (Mittelhauser et al. Reference Mittelhauser, Allen, Chalfant, Schauffler and Welch2016, Nager and O’Hanlon Reference Nager and O’Hanlon2016, Washburn et al. Reference Washburn, Elbin and Davis2016, Wilhelm et al. Reference Wilhelm, Rail, Regular, Gjerdrum and Robertson2016), though North American populations declined more steeply than European ones (-68.7% and -28.1% per annum, respectively).
Most Great Black-backed Gull populations increased for most of the 20th century until peaking sometime in its later decades. In areas where large-scale surveys were undertaken with some regularity, this peak was reached between the 1970s and 1990s (Cotter et al. Reference Cotter, Rail, Boyne, Robertson, Weseloh and Chaulk2012, Anderson et al. Reference Anderson, Shlepr, Bond and Ronconi2016, Mittelhauser et al. Reference Mittelhauser, Allen, Chalfant, Schauffler and Welch2016). Following this peak, our results showed declines across most populations that historically held a large percentage of the global population: Atlantic Canada and the United States, Iceland, Sweden, United Kingdom, and Ireland. Norway, which supports almost half of the breeding European Great Black-backed Gulls, remained relatively stable over that period, though confidence intervals around estimates of λ dip into negative growth (95% CI: 0.9817; 1.0222), and most of the colonies that are monitored annually are in decline (Anker-Nilssen et al. Reference Anker-Nilssen, Barrett, Christensen-Dalsgaard, Dehnhard, Descamps, Systad, Moe, Reiertsen, Bustnes, Erikstad, Follestad, Hanssen, Langset, Lorentsen, Lorentzen and Strøm2021). Increasing trends were typically observed in countries and states with smaller populations, generally at the southern or northern edges of the species’ range such as Germany, the Netherlands, Spain, Svalbard, North Carolina (United States), and Virginia (United States). France showed a remarkable increase from approximately 2,000 to 10,000 pairs. Increases seen at these locations could perhaps be due to birds emigrating from larger populations; Great Black-backed Gulls from Canadian colonies have replenished US populations in the past (Anderson et al. Reference Anderson, Shlepr, Bond and Ronconi2016).
Due to the vast distribution and inaccessibility of some Great Black-backed Gull populations, absolute counts of breeding pairs are rarely undertaken. Therefore, the number of available counts differed between countries, as did methodology and count accuracy. Using two population counts and standardising the uncertainty associated with each count following the five-point scale detailed in Lynch et al. (Reference Lynch, Ratcliffe, Passmore, Foster and Trathan2013; Table 2) was the approach we used to factor in the limited count data and varying levels of accuracy. The main caveat of this approach was the potential to over- or underestimate 1985 and 2021 population estimates. Depending on how close the two population estimates per region were in time to the initial (1985) and final (2021) years of the most recent three-generation period, estimates could have been over- or underestimated. We minimized this bias by choosing the two counts closest to 1985 and 2021. Additionally, our estimated global population size for 1985 (291,000) was within the range estimated by BirdLife International in previous species assessments (283,000–403,000; BirdLife International 2015, Wetlands International 2015). Our approach also meant we were not able to provide finer-scale changes in abundance over time periods shorter than three generations. For example, Mittelhauser et al. (Reference Mittelhauser, Allen, Chalfant, Schauffler and Welch2016), reported an annual rate of change of -6.3% between 2008 and 2013 in Maine (USA, a figure much higher than the longer-term three-generation decline of -1% reported in this study. However, our focus was on linear change in abundance over a three-generation period in line with the IUCN Red List criteria which aimed to provide an overall change in abundance over a standardised time period.
The causes of decline remain largely unexplored. The foraging plasticity and overall adaptability of Great Black-backed Gulls is presumed to be a buffering mechanism against variable availability of resources (Coulson Reference Coulson2019). Identifying the individual causes of population change for the Great Black-backed Gull would be challenging due to existing knowledge gaps regarding their foraging ecology and population dynamics (lack of demographic rate data) (Bicknell et al. Reference Bicknell, Oro, Camphuysen and Votier2013). However, significant contributors to the above-mentioned declines of populations have been suggested. Reductions in the availability of fish discards may be a primary reason that some Larus populations have declined. In the North Sea, Great Black-backed Gulls have been identified as one of the species at risk of being impacted by an ongoing, gradual reduction in discards since 1990 and a newly implemented discard ban in European Union waters (Bicknell et al. Reference Bicknell, Oro, Camphuysen and Votier2013, Sherley et al. Reference Sherley, Ladd-Jones, Garthe, Stevenson and Votier2020). However, the impacts of these changes on North Sea Great Black-backed Gull populations remain largely unknown. In eastern Canada, a reduction in landed groundfish (which provided offal and discards) due to a moratorium in 1992 was correlated with a decline in the number of Great Black-backed Gull pairs in the following years (Cotter et al. Reference Cotter, Rail, Boyne, Robertson, Weseloh and Chaulk2012). Conversely, some populations forage mostly on naturally occurring habitats and food such as seabirds, mammals, intertidal organisms, and forage fish (Harris Reference Harris1980, Borrmann et al. Reference Borrmann, Phillips, Clay and Garthe2019, Westerberg et al. Reference Westerberg, Brown, Eagle and Votier2019) and their declines are likely unrelated to changes in anthropogenic fish and waste management. Between the 1970s and 2000, delayed spawning of capelin Mallotus villosus was correlated with lower breeding success in several seabird species in Canada, including Great Black-backed Gulls (Stenhouse and Montevecchi Reference Stenhouse and Montevecchi1999, Davoren and Montevecchi Reference Davoren and Montevecchi2003, Regehr and Rodway Reference Regehr and Rodway1999). In Finland, where a three-generation decline of 50% occurred, the eutrophication of water bodies and culling could be contributory factors (Hario and Rintala Reference Hario and Rintala2016). Lastly, predation can directly or indirectly impact populations. In North America, Bald Eagles Haliaeetus leucocephalus, whose populations are currently recovering, depredate Herring and Great Black-backed Gulls and exclude them from areas near eagle nest sites (White et al. Reference White, Heath and Gisborne2006, Hipfner et al. Reference Hipfner, Blight, Lowe, Wilhelm, Robertson, Barrett, Anker-Nilssen and Good2012). Overall, the drivers of change remain largely unknown across most of the Great Black-backed Gull populations reported in this study and the causes of decline appear not have ceased; see annually monitored Norwegian (Anker-Nilssen et al. Reference Anker-Nilssen, Barrett, Christensen-Dalsgaard, Dehnhard, Descamps, Systad, Moe, Reiertsen, Bustnes, Erikstad, Follestad, Hanssen, Langset, Lorentsen, Lorentzen and Strøm2021) and United Kingdom populations (Seabird Monitoring Program 2022). It is likely that changes in abundance are driven by several factors acting simultaneously upon populations due to the widely acting nature of human threats.
We estimated that Great Black-backed Gulls declined globally in the range of 43% to 48% between 1985 and 2021. We suggest that uplisting Great Black-backed Gulls to ‘Vulnerable’ on the IUCN Red List is appropriate and further assessments should be carried out regularly since it is approaching an ‘Endangered’ listing. Causes of decline remain largely unknown and unexplored for most populations. Further research should focus on reducing knowledge gaps to support a better understanding of factors influencing population trends and status across its range. A key knowledge gap is the lack of estimates of demographic parameters. These data, in particular survival and productivity, are key to understanding the drivers and mechanisms of population change and can be used to model population trajectories and inform targeted conservation action. Additionally, improved and consistent monitoring of Great Black-backed Gull populations (including standardisation of survey methodologies, more frequent population counts, and coordination between countries) would allow for more accurate assessments of population size, status, and trends.
Acknowledgements
We are very grateful for all workers and volunteers who were involved in the data collection and estimation of population counts. We also thank C. M. Johnson and B. E. Washburn for providing population count data. The findings and conclusions in this article are those of the authors and do not necessarily represent the views of the U.S. Fish and Wildlife Service. This studentship has been funded under the NERC Scottish Universities Partnership for environmental Research (SUPER) Doctoral Training Partnership (DTP) (Grant reference number NE/S007342/1 and website https://superdtp.st[1]andrews.ac.uk/). Additional funding has been provided by the University of the Highlands and Islands, and the Scottish Government.
Open access
