Peri-urban systems alter trophic niche size and overlap in sympatric coastal bird species

Urban habitats can create empty trophic niche space by providing abundant alternative dietary resources, allowing some generalist species to either shift or expand their trophic niches. Resulting changes to trophic niche size may consequently affect interspecific interactions and competition, although trophic impacts are less clear in peri-urban systems, where both natural and urban resources are readily accessible to highly mobile species. We combined stable isotope ( δ 13 C, δ 15 N, δ 34 S) data with GPS tracking of two sympatric coastal bird species ( Larus argentatus , L. marinus ) across four study sites on the east coast of the United States, ranging from a study site in New York City (the most popu-lous city in the United States) to less-urban sites off Massachusetts. We quantified the trophic niche size and trophic niche overlap in three-dimensional isotopic space and assessed spatial overlap in foraging habitat between species at both peri-urban and less-urban study sites. We found that for both species, birds at peri-urban study sites had significantly larger trophic niches


INTRODUCTION
The urbanization of landscapes is a fundamental component of life in the Anthropocene (Elmqvist et al., 2021), a time period characterized by human presence and alterations to natural systems (Crutzen, 2006;Waters et al., 2016). Urbanization is expected to continuously increase in the coming decades (Seto et al., 2012) and understanding the effects of urban landscapes on wildlife is critical for efforts to minimize human-wildlife conflicts and improve conservation outcomes (Ritzel & Gallo, 2020;Schell et al., 2021). Urbanization is often associated with reduced species richness (McKinney, 2008;Ordeñana et al., 2010) and functional and phylogenetic diversity Palacio et al., 2018). However, urban habitats also act as novel ecosystems with empty niche space (Corlett et al., 2020;Rebele, 1994), providing abundant alternative habitats and dietary resources for species that can adapt to these systems. By modifying ecological niches, urban habitats can have wide-ranging implications for interspecific interactions (Manlick & Pauli, 2020;Pagani-Núñez et al., 2019), food web dynamics (Alp & Cucherousset, 2022;Faeth et al., 2005;Olds et al., 2018), and species assemblages (Battles et al., 2018;Gamez et al., 2022;Henderson et al., 2020;Pagani-Nunez et al., 2022).
Many generalist species can shift or expand their trophic niches to make use of urban dietary resources (Devictor et al., 2016;Larson et al., 2020;Manlick & Pauli, 2020;Murray et al., 2015;Pagani-Núñez et al., 2019). Assessing how urbanization influences trophic ecology is key to understanding the impacts of urban landscapes on ecosystem stability as wider niches are often associated with an increase in weak and intermediate trophic relationships, improving ecosystem resilience (Mccann et al., 1998). Niche expansion may also affect interspecific interactions and competition by altering trophic niche overlap between species. For example, Larson et al. (2020) found that urban areas increased chances of interspecific competition between terrestrial mammalian carnivores by increasing trophic niche width and overlap between species. However, the relationship between urban areas, niche size, and interspecific overlap remains understudied for species in peri-urban environments, which are defined as areas contiguous and interfacial to both urban and natural landscapes (Morello et al., 2000). Teta et al. (2012) suggested that peri-urban environments may allow animals to have a wider trophic niche breadth compared with urban and non-urban populations by providing close access to a greater variety of natural and anthropogenic dietary resources, following the theory that species diversity peaks in peri-urban habitats (Andrade et al., 2018;Blair, 1996;Germaine & Wakeling, 2001;Pawlikoski & Pokorniecka, 1990;Racey & Euler, 1982). Therefore, urbanization may have different impacts on species occurring in peri-urban areas than those occurring in more isolated urban areas.
Coastal urbanization and subsequent decreases in natural habitat and prey availability have caused many coastal species to shift from foraging in marine or coastal areas to urban (e.g., Dorn et al., 2011;Evans & Gawlik, 2020) and more populous regions. For example, polar bears (Ursus maritimus), an ice-obligate marine apex predator in the Arctic, increased their presence in human-populated areas during recent years of low sea ice cover and natural prey abundance, likely searching for substitutive anthropogenic resources (Smith et al., 2022;Stirling & Parkinson, 2006;Wilder et al., 2017). Other species that have historically been important coastal predators, such as the red fox (Vulpes vulpes; Zabel & Taggart, 1989) and raccoon (Procyon lotor; Wilson, 1989;Witherington, 1992), have adapted to urban areas and are now abundant foragers in urban and peri-urban systems (Bateman & Fleming, 2012;Killengreen et al., 2011;Kimber et al., 2020;Prange et al., 2004;Rulison, 2022;Scholz et al., 2020;Schwemmer et al., 2021). The effects of peri-urban systems on animal foraging patterns have received little attention to date. Yet, understanding these effects is central to quantifying the ecological impacts of coastal urbanization as it accelerates and expands (Chen et al., 2020;Seto et al., 2012).
Coastal birds offer a unique vantage point for assessing changes in trophic ecology at the interface between urban and natural systems as their high mobility allows them to easily access both natural marine and urban foods rather than being restricted to one environment or the other. Gulls (Larus spp.), in particular, are ideal study species for addressing this gap in knowledge as they are generalist foragers, feeding on both natural marine prey and urban refuse (Maynard & Ronconi, 2018;Méndez et al., 2020;Spelt et al., 2019;Washburn et al., 2013), and breed at both remote (less-urban) colonies and colonies within peri-urban coastal regions (e.g., Fuirst et al., 2018;Maynard & Ronconi, 2018;Mendes et al., 2018;Shaffer et al., 2017). Furthermore, gulls often nest sympatrically (e.g., Jouanneau et al., 2022;Lato et al., 2021), allowing interspecific interactions to be assessed at the same study site.
We used stable isotope analysis to quantify the three-dimensional trophic niche size of herring gulls (L. argentatus) and great black-backed gulls (L. marinus) in coastal peri-urban and less-urban coastal environments, testing the hypothesis that urbanization results in niche expansion at peri-urban colonies. We expected that gulls at peri-urban colonies would show significantly larger trophic niches than in less-urban colonies. We further assessed the relationship between urbanization and interspecific interactions by using stable isotope analysis to quantify the trophic niche overlap and GPS tracking data to quantify the overlap in foraging space between herring gulls and great black-backed gulls within peri-urban and less-urban environments. Based on previous literature, we expected that peri-urban colonies would show greater overlap in trophic niche and foraging space than less-urban colonies.

Data collection
We collected data from four gull breeding colonies that differ in their proximity to urban habitats ( Figure 1a). We compared foraging patterns between colonies as opposed to larger geographic areas (i.e., New York vs. Massachusetts), as marine birds show colony-specific foraging patterns even among colonies that are in close proximity to one another (Masello et al., 2010;Paiva et al., 2010;Redfern & Bevan, 2014;Shaffer et al., 2017). Jamaica Bay (40.5948 N, 73.8392 W) and Young's Island (40.9216 N, 73.1522 W) were considered peri-urban sites due to their proximity (<5 km) to densely populated urban areas and coastal and marine foraging areas (all study sites were islands). Jamaica Bay is located within the borough of Queens in New York City, the most populous city in the United States (United States Census Bureau, 2022), but also provides access to more coastal and pelagic habitats in the Jamaica Bay Wildlife Refuge and the North Atlantic Ocean. Young's Island, located within Suffolk County (one of the most populous counties in the country; United States Census Bureau, 2021), is on the north shore of Long Island, NY, USA, and provides close access to both populous urban areas and salt marshes, bays, and pelagic habitats in the Long Island Sound. Muskeget Island (41.3370 N, 70.3014 W) and Tuckernuck Island (41.3035 N, 70.2551 W), located off the coast of Massachusetts, USA, were considered less urban (more remote) than Jamaica Bay and Young's Island. Muskeget Island is completely undeveloped while Tuckernuck Island is sparsely developed (a few residential houses). Both Muskeget and Tuckernuck Island are adjacent to pelagic habitats in the North Atlantic Ocean and much further from populous regions (>10 km to Nantucket, MA, and >25 km from mainland MA) than Jamaica Bay and Young's Island. Additionally, the human population density in the areas surrounding Muskeget and Tuckernuck Island is considerably lower than in the areas around Jamaica Bay and Young's Island (Figure 1a).
Data were collected from herring gulls and great black-backed gulls during egg incubation between the months of April and June in 2016-2022. For stable isotope analyses, 0.25-2 mL of whole blood was taken from the medial tarsal vein using a 25-gauge needle. Blood samples were collected from herring gulls from Jamaica Bay and Tuckernuck Island in 2016-2018, from Young's Island in 2017-2022, and from Muskeget Island in 2021-2022. Great black-backed gull samples were only collected from Young's Island in 2019-2022 and Muskeget Island in 2021-2022 due to limitations in field personnel (Appendix S1: Table S1). Samples were either placed in a 2 mL cryogenic vial or 4 mL sodium heparin-coated BD vacutainer (Franklin Lakes, NJ) and frozen at approximately −80 C until further processing. Regurgitant samples were collected opportunistically from both species during handling from the Young's Island and Muskeget Island colonies during the 2021 and 2022 field seasons. Gull regurgitants often include either undigested or partially digested items, and thus all regurgitants collected were able to be visually identified to either the species level (for nonurban diet items; e.g., fish) or the food-item level (for urban diet items; e.g., chicken finger).
To quantify foraging movements, archival GPS loggers (CatLog Generation 2, Perthold Engineering LLC, USA and igotU-GT120, Mobile Action Technologies, Taiwan) were attached to an individual's three to four central tail feathers using Tesa Tape (Beiersdorf AG, GmbH, Hamburg, Germany) and collected data at 2-min intervals for an average of 7.61 days (±4.46 days). Archival GPS loggers provide high-resolution data, allowing us to investigate fine-scale habitat use and foraging movements, which would not be possible with satellite or GSM/GPS devices. Given that recaptures of specific individuals are required for archival GPS logger retrieval and that gulls are difficult to recapture, studies of gull habitat use typically have relatively small sample sizes (Borrmann et al., 2019;Cimino et al., 2022;Maynard & Ronconi, 2018) compared with studies of other coastal bird species. Herring gulls were tracked at all four study colonies while great black-backed gulls were tracked at Muskeget Island and Young's Island (Appendix S1: Tables S1 and S2). Blood samples for birds tracked with GPS tags were obtained upon tag retrieval and subsequently used for stable isotope analysis in addition to the blood samples taken from individuals not fitted with a GPS device. Year did not have a significant effect on habitat use or isotope values, and thus data from all years were pooled together in analyses.

Stable isotope analysis
Blood samples were dried for 48 h using either a freeze dryer (FreeZone 6 Liter Benchtop Freeze Dry System Model number 553477) or a drying oven (Heratherm OGS400) at 60 C, homogenized using a mortar and pestle until they reached a fine powder consistency, and weighed to the nearest 0.01 mg prior to encapsulation in tin capsules (Elemental Microanalysis, Okehampton, UK). Samples were analyzed for δ 13 C and δ 15 N at the University of Hawaii SOESTT lab using a Costech ECS 4010 Elemental Combustion System using a Zero Blank F I G U R E 1 Location of study colonies across New York and Massachusetts, USA. (a) Study locations shown with associated population size by congressional district (Source: United States Census Bureau, 2021) used as a proxy for degree of urbanization. Jamaica Bay and Young's Island are peri-urban study sites (squares) while Tuckernuck and Muskeget Islands are less-urban study sites (circles). 95% (solid line) and 50% (dotted line for great black-backed gulls, dashed line for herring gulls) kernel utilization distribution of herring gulls (yellow) and great black-backed gulls (blue) are shown from (b) Muskeget Island, MA, and (c) Young's Island, NY. Overlap values are reported in Table 2. Autosampler, coupled to a ThermoFinnigan Delta Plus XP. Samples were analyzed for δ 34 S either at the University of California Davis Stable Isotope Facility using a PDZ Europa 20-20 continuous flow isotope ratio mass spectrometer coupled with a PDZ Europa ANCA-GSL elemental analyzer (Sercon Ltd., Cheshire, UK) or at the University of Utah SIRFER lab using an isotope ratio mass spectrometer (Finnigan Delta Plus XL) coupled to an elemental analyzer operated in continuous flow mode (Costech EA 4010 via Finnigan Conflo III). Stable isotope ratios in samples are reported using standard delta notation (δ) relative to an international standard (Vienna Pee Dee Belemnite, atmospheric air, and Vienna-Canyon Diablo Troilite for δ 13 C, δ 15 N, and δ 34 S, respectively) in units per mill (‰).
Quality assurance (QA) and quality control (QC) of stable isotope measurements were assessed using blood sample replicates (one replicate performed every tenth sample) in addition to in-house reference samples (glycine, acetanilide, and white tuna muscle for δ 13 C and δ 15 N; silver sulfide, zinc sulfide, ground feather, cysteine, hair, whale baleen, mahi-mahi muscle, and taurine for δ 34 S). For replicate blood samples, the SE between replicate samples was 0.1‰ for δ 13 C, δ 15 N, and δ 34 S. Using in-house reference samples, machine error did not exceed 0.2‰ for δ 13 C and δ 15 N and 0.3‰ for δ 34 S. To account for the effect of lipid concentration on δ 13 C values, δ 13 C values of whole blood were mathematically corrected following Post et al. (2007). To account for the effects of sodium heparin on δ 34 S values, we mathematically corrected δ 34 S values for samples stored in vacutainers (n = 35) based on the mean effect (−0.4‰) of sodium heparin on δ 34 S values in gull whole blood . We did not perform a mathematical correction to δ 13 C and δ 15 N values of samples stored in vacutainers as sodium heparin does not significantly affect these values in gull whole blood .

Estimation of isotopic niche size and overlap
Raw isotope values were not normally distributed and were thus compared within and between species across the two study sites using Kruskal-Wallis tests with a post hoc Dunn comparison and Benjamini and Hochberg correction for multiple comparisons. Parametric testing (ANOVA) following data transformation was also explored, but results did not differ from those of the Kruskal-Wallis tests, and thus Kruskal-Wallis tests were used in all final analyses. We estimated the 95% isotopic region size and 95% probabilistic isotopic niche overlap, respectively, using all three isotope values (δ 13 C, δ 15 N, δ 34 S) to assess isotopic niche size and trophic niche overlap in three dimensions using the "nicheROVER" package in R (Swanson et al., 2015). This method estimates the Bayesian standard elliptical area (SEA B ) while estimating the uncertainty in the sampling process and is a preferable and more robust method than estimates using the total ellipse area (Jackson et al., 2011;Syväranta et al., 2013). Additionally, the SEA B method accounts for the distribution of points in isotopic space, whereas other methods (such as the convex hull method) do not (Cucherousset & Villéger, 2015). Using a threedimensional assessment of trophic niche space (i.e., incorporating δ 13 C, δ 15 N, and δ 34 S values) provides a more powerful method for disentangling foraging and trophic patterns than using δ 13 C and δ 15 N values alone (Rossman et al., 2016;Swanson et al., 2015). Significant differences in the posterior estimations of niche size were assessed within species across colonies using a Kruskal-Wallis test with post hoc Dunn comparison and Benjamini and Hochberg correction.

Analysis of GPS tracks
During incubation, gulls regularly travel outside the colony multiple times per day to forage (Ceia et al., 2014;Fuirst et al., 2018;Maynard & Ronconi, 2018). As gulls also frequently exhibit movements adjacent to the nesting colony (presumably to drink, rest, etc.), foraging trips were delineated as consecutive movements at least 500 m from the nesting colony with a minimum of 30-min durations (15 GPS points).
Foraging behavior was identified as areas of restricted search (ARS) along each foraging trip track line using first passage time (FPT) analysis as in Fuirst et al. (2018) and Lato et al. (2021). FPT is the time it takes for an individual to travel through a spatial circle with a radius of a given size (Fauchald & Tveraa, 2003). The scale of ARS (i.e., the scale at which foraging took place) was identified as the radius size at which the maximum variance in the log of FPT values occurred (Fauchald & Tveraa, 2003). Spatial autocorrelation between GPS points was adjusted for according to Suryan et al. (2006), as described in Lato et al. (2021), and the upper quartile of FPT values of remaining points was classified as foraging (ARS) points. Foraging points were then overlaid onto satellite imagery in ArcGIS (version 10.6.1) and manually categorized as being within either urban, lake, coastal, or pelagic habitats. To account for differences in track length and deployment duration, we averaged the proportion of foraging points within each habitat over every trip and individual within each colony.
To determine whether and how individual foraging differed between colonies, we further assessed within-and between-individual differences in foraging patterns. We quantified the proportion of tracked individuals from each study site that utilized both urban and natural (pelagic and coastal) habitats and the proportion of foraging points occurring within each habitat type at both the individual and colony levels. We used the "lmerTest" package in R (Kuznetsova et al., 2017) to perform linear mixed effect models (with species and colony as random effects) to investigate the relationship between the proportion of foraging points in urban habitats and δ 13 C, δ 15 N, and δ 34 S values to draw a direct link between foraging area and isotope ecology.
Analyses of overlap in foraging space focused on colonies and years for which we could gather data from both species (Young's Island and Muskeget Island; 2019-2022). To quantify the spatial overlap in foraging areas between species, we quantified the utilization distribution overlap index (UDOI) and Bhattacharya's affinity (BA) using the "adehabitatHR" package in R (Calenge, 2015). UDOI values typically range from 0 (indicating no overlap) to 1 (indicating total overlap); however, it is possible that values can be more than 1 if spatial distributions are nonuniform and show a high degree of overlap (Fieberg & Kochanny, 2005). BA values, though similar to UDOI, measure the degree of similarity in utilization distributions between populations and are strictly bounded between 0 (indicating no similarity in utilization distributions) and 1 (indicating identical utilization distributions; Fieberg & Kochanny, 2005). When calculating utilization distributions, choosing a bandwidth, or smoothing parameter (h), is necessary prior to calculation (Worton, 1989). Though the least squares cross-validation method for identifying bandwidth is often a preferred method for bandwidth selection (Worton, 1989), it failed to converge. Thus, we chose bandwidth ad hoc in which an initial value of bandwidth is set high and incrementally reduced until bimodality in spatial distributions was detected (Reisinger et al., 2020;Schuler et al., 2014). Using this method, we determined h to be 1800 m for all Muskeget Island trips and 1600 m for all Young's Island trips. To account for individual differences and variability in the length of GPS deployments, we averaged the kernel utilization distribution estimation across individuals of each species within each colony. We used the 95% and 50% UDOI and BA to represent the spatial overlap between species over most of their foraging range and in their core foraging areas, respectively (Berlincourt & Arnould, 2015;Jordan et al., 2022;Reisinger et al., 2020). The approximate area (in square kilometers) of most of the foraging range and core foraging areas was also calculated and compared between species and colonies.

Stable isotope analysis
Herring gull δ 13 C, δ 15 N, and δ 34 S values differed significantly between breeding colonies. Herring gulls at Young's Island, a peri-urban colony, had significantly lower δ 13 C and δ 15 N values than herring gulls at the less-urban colonies, Muskeget and Tuckernuck Islands (Figure 2; p values in Table 1). However, herring gulls at the Jamaica Bay colony did not differ significantly from Muskeget or Tuckernuck Island herring gulls in their δ 13 C values (Figure 2; p values in Table 1). Additionally, δ 15 N values of herring gulls at Jamaica Bay were significantly lower compared with herring gulls from Muskeget Island and were not significantly different from those at Tuckernuck Island. Herring gulls at Jamaica Bay and Young's Island had significantly lower δ 34 S values than herring gulls at Muskeget and Tuckernuck Islands (Figure 2; p values in Table 1).
For great black-backed gulls, δ 15 N and δ 34 S values, but not δ 13 C values, were significantly different between study colonies ( p values in Table 1). Great black-backed gulls at Young's Island had significantly higher δ 15 N values and significantly lower δ 34 S values than those at Muskeget Island (Figure 2; p values in Table 1).
Colony had a significant effect on the trophic niche size for both herring gulls and great black-backed gulls. Herring gulls at Jamaica Bay and Young's Island had a significantly greater trophic niche size than the herring gulls at Tuckernuck and Muskeget Islands (p < 2.0 × 10 −16 for all comparisons; Figure 3). Similarly, great black-backed gulls at Young's Island had a significantly greater trophic niche size than the great black-backed gull population from Muskeget Island (p < 2.0 × 10 −16 ; Figure 3). Lastly, the 95% probabilistic overlap in trophic niche space between herring gull and great black-backed gulls was much greater at Muskeget Island (59.84% overlap) than at Young's Island (9.24% overlap; Figure 4).

GPS tracking
For both great black-backed and herring gulls, a larger proportion of foraging points occurred in pelagic areas at less-urban colonies (Muskeget and Tuckernuck Islands) than at peri-urban colonies (Young's Island and Jamaica Bay; Figure 5). Conversely, in peri-urban colonies, great black-backed and herring gulls showed a greater degree of urban foraging than in less-urban colonies ( Figure 5). However, great black-backed gulls primarily foraged in coastal habitats at the peri-urban site where they were studied (Young's Island). For both species, a larger proportion of individuals utilized both urban and natural habitats for foraging at the peri-urban colonies than at the less-urban colonies (Appendix S1: Figure S1). At Muskeget and Tuckernuck Islands (less-urban colonies), 41% and 67% of herring gulls, respectively, used both urban and natural habitats to some extent. At Young's Island and Jamaica Bay (peri-urban colonies), 100% and 96% of herring gulls, respectively, used both urban and natural habitats to some extent. For great black-backed gulls, 0% and 57% of gulls from Muskeget and Young's Islands, respectively, used both urban and natural habitats to some extent. Similarly, birds from the peri-urban colonies exhibited larger between-individual variability in their degree of urban habitat use (Appendix S1: Figure S1). At Muskeget Island and  Tuckernuck Island, the variance in urban habitat use between individual herring gulls was 1.6 × 10 −3 and 3.4 × 10 −2 , respectively. At Young's Island and Jamaica Bay, the variance in urban habitat use between individual herring gulls was 4.2 × 10 −2 and 6.0 × 10 −2 , respectively. For great black-backed gulls, the variance in urban habitat use between individuals at Muskeget Island and Young's Island was 1.15 × 10 −5 and 5.2 × 10 −2 . We found significant negative linear relationships between the proportion of foraging points in urban areas and δ 13 C, δ 15 N, and δ 34 S values, respectively ( p values in Appendix S1: Table S3). δ 34 S had the strongest linear relationship to the proportion of foraging points in urban areas (Appendix S1: Figure S2) while δ 13 C had the weakest (Appendix S1: Figure S2).
The spatial areas of most of the foraging and core foraging areas were greater for gulls at Muskeget Island than at Young's Island for both species (Table 2). The overlap over most of their foraging range and core foraging areas between herring gulls and great black-backed gulls was also greater at Muskeget Island than at Young's Island according to both UDOI and BA (

Regurgitant collection
We collected a total of 23 and 28 regurgitant samples from Young's and Muskeget Islands, respectively (Appendix S1: Table S4). At Young's Island, herring gull regurgitant samples primarily consisted of urban diet items (e.g., bread, fast-food meat, pizza, pasta) while great black-backed gulls regurgitant samples primarily consisted of Atlantic menhaden (Brevoortia tyrannus) with some urban diet items (e.g., fast-food meat). At Muskeget Island, herring gulls regurgitant samples primarily consisted of longfin squid (Doryteuthis pealeii) and fish (e.g., scup [Stenotomus chrysops]) while great black-backed gull regurgitant samples primarily consisted of scup and other fish species (e.g., Atlantic menhaden, hake [Merluccius spp.]). In addition to regurgitant sample collection, we observed multiple feedings of great black-backed gulls on Atlantic horseshoe crabs (Limulus polyphemus; n = 3) and scup (n = 2) and multiple feedings of herring gulls on hard clam (Mercenaria mercenaria; n = 3) at Young's Island. No feeding observations took place at the Muskeget Island colony.

DISCUSSION
Coastal birds have shown to utilize urban and less-urban habitats when both are easily accessible (Carmona et al., 2021;Evans & Gawlik, 2020;Yoda et al., 2012). Our results demonstrate consistent impacts of urbanization on the trophic ecology and trophic niche size of a highly mobile and abundant coastal predator within two of the most heavily urbanized areas in the United States. Additionally, we found that within peri-urban systems, sympatric species can show greater overlap in trophic niches and overlap in foraging space than their less-urban counterparts. Our results show that urban foraging significantly impacts the trophic ecology of gulls, as it does for other consumers (Moreno et al., 2010;Scholz et al., 2020;Sugden et al., 2021). We found a significant linear relationship between the degree of urban foraging and the δ 13 C, δ 15 N, and δ 34 S values of individual gulls. Our results suggest that δ 34 S values may be more indicative of urban feeding than δ 13 C and δ 15 N values alone; δ 34 S had the strongest relationship to the proportion of urban foraging and we found significant differences in δ 34 S values F I G U R E 4 Overlap in trophic niche space between great black-backed gulls (blue) and herring gulls (yellow) in three dimensions at the less-urban colony of Muskeget Island (left panels) and the peri-urban colony of Young's Island (right panels). Isotope values are shown in three-dimensional space in panels (a) and (b). Visualization of 95% probabilistic elliptical overlap in three-dimensional space is depicted conceptually in panels (c) and (d).
between all peri-urban and less-urban colonies across both species. Other studies have also found that δ 34 S is a strong indicator of terrestrial versus marine and nearshore versus offshore feeding in marine consumers (Barros et al., 2010;Elliott & Elliott, 2016;Whitney et al., 2018). Historical isotopic records of gulls further support the notion that the degree of urban foraging within a population is inversely related to δ 15 N and δ 13 C values (Blight et al., 2015;Farmer & Leonard, 2011;Osterback et al., 2015) given that urban dietary resources are often comprised of wheat and corn-based products that have lower trophic levels and different photosynthetic pathways than marine species (Jahren & Kraft, 2008;Lato et al., 2021;Moreno et al., 2010). At the colony level, Young's Island herring gulls also showed significantly lower δ 13 C and δ 15 N values from Tuckernuck and Muskeget Island herring gulls, while Jamaica Bay herring gulls only showed significantly lower δ 15 N values than those at Muskeget Island. This difference in nitrogen isotopes observed between Jamaica Bay and Young's Island herring gull populations may be a result of Jamaica Bay herring gulls foraging in coastal marine areas to a slightly greater extent than Young's Island herring gulls ( Figure 5; Fuirst et al., 2018;Thorne et al., 2021). For great black-backed gulls, significantly lower δ 15 N values were also observed at the less-urban colony of Muskeget Island when compared with the peri-urban colony of Young's Island. Regurgitant samples further demonstrated that gulls from the peri-urban colony feed on urban refuse to a greater extent than gulls from the less-urban colony.
Gulls from both peri-urban colonies (Jamaica Bay and Young's Island) showed significantly larger trophic niche sizes than gulls from less-urban colonies (Tuckernuck Island and Muskeget Island), consistent with our prediction and previous studies conducted on F I G U R E 5 Proportion of foraging points for great black-backed gulls and herring gulls in four different habitat types across four study colonies. Location abbreviations: JB, Jamaica Bay; MS, Muskeget Island; TN, Tuckernuck Island; YI, Young's Island. Peri-urban colonies are shown in a striped pattern and less-urban colonies are shown in solid colors.
T A B L E 2 Utilization distribution overlap indices (UDOI) and Bhattacharya's affinity (BA) and approximate area of 95% and 50% utilization areas (UA). terrestrial animals (Manlick & Pauli, 2020;Pagani-Núñez et al., 2019). This highlights the potential for generalist consumers to expand their trophic niches as a result of urban foraging (Larson et al., 2020;Manlick & Pauli, 2020;Murray et al., 2015;Pagani-Núñez et al., 2019). Using GPS tracking data, we found that the proportion of individuals using both urban and natural (pelagic and coastal) habitat types was higher at peri-urban colonies, demonstrating strong within-individual variability in foraging. Similarly, there was greater variability in urban habitat use between individuals at the peri-urban colonies than at the less-urban colonies. This, combined with our findings that urban foraging is linearly linked to isotope values, suggests that both within-and between-individual differences in urban foraging resulted in trophic niche expansion at the peri-urban colonies. Liang et al. (2020) reported similar findings in multiple passerine and frog species, where a combination of between-and within-individual differences drove changes in trophic niche size across urban and natural populations. Several studies have found that urban diet items show distinctly different isotope values than natural diet items (Lato et al., 2021;Moreno et al., 2010;Shlepr et al., 2021), and thus likely further drives differences in isotopic values and niche sizes in urban populations.

Colony
Our results provide evidence of decreased trophic niche overlap within peri-urban coastal habitats and highlight the complexity of urban impacts on trophic ecology and the importance of distinguishing between urban and peri-urban habitats in urban studies of wildlife. We found herring and great black-backed gulls at the less-urban colony showed lower trophic niche overlap than at the peri-urban colony, which was inconsistent with our initial prediction and findings from other studies (Larson et al., 2020;Manlick & Pauli, 2020;Pagani-Núñez et al., 2019). This increase in trophic niche overlap in the less-urban environment was further supported by collected regurgitant samples, which showed herring and great black-backed gulls consuming more similar prey items (e.g., scup, Atlantic menhaden) at the less-urban colony. This finding suggests that urban environments may reduce interspecific competition and drive resource partitioning in highly mobile sympatric species, which can access a variety of habitat types. Furthermore, the use of a broader range of diet items by highly mobile consumers in peri-urban systems may lead to greater resource partitioning among functionally similar species. Interspecific dietary and habitat use overlap in natural ecosystems often results in competitive pressure between species (Kozlowski et al., 2008;Larson et al., 2020), but peri-urban habitats may alter this competitive pressure by providing an opportunity for generalist species to shift and/or expand their trophic niches. We posit that in peri-urban systems, where both urban and less-urban food types may be available, changes to trophic niche overlap and interspecific interactions may be different than in isolated urban systems where consumers are limited to foraging in a single habitat type. These changes may be magnified in coastal peri-urban areas, where an abundance of both marine and urban dietary resources can be easily accessed within similar distances. Additionally, coastal peri-urban areas may offer higher marine diversity than less-urban coastal areas , which may further drive reduced interspecific competition in top coastal predators, such as the gull species considered in this study. We acknowledge our limited sample size for assessing overlap in trophic niche space and encourage other studies to further explore the relationship between urbanization and interspecific interactions in coastal and other peri-urban systems.
The mobility and diet specificity of focal species is a critical consideration when assessing the effects of urban habitats on trophic dynamics. Mobile generalist consumers may make use of both urban and natural habitats for foraging when available (Dorn et al., 2011;Evans & Gawlik, 2020;Hobson & Stirling, 1997;Stirling & Parkinson, 2006) while less mobile species may be isolated and limited to foraging in a single habitat type (e.g., Bateman & Fleming, 2012;Prange et al., 2004;von Merten et al., 2022). For example, priorly mentioned studies that found higher trophic niche overlap between species in urban areas (e.g., Manlick & Pauli, 2020;Pagani-Núñez et al., 2019) were focused on studying isolated urban populations of terrestrial mammalian carnivores and passerines, which are generally less mobile than the seabirds studied in our study.
Trophic niche expansion in urban areas could have important implications for alterations in food web dynamics and population structures as human-modified landscapes may select for generalist species with wider trophic niches, inherently reducing the abundance of specialist species (Clavel et al., 2011;Devictor et al., 2016;Henderson et al., 2020;Rocha & Fellowes, 2020). Furthermore, the increased use of urban areas by wildlife has led to increased human-wildlife conflicts (Belant, 1997;Lewis et al., 2015;Wilder et al., 2017). Gaining a better understanding of the degree to which animals residing in peri-urban areas forage on urban subsidies is necessary to mitigate human-wildlife conflict. Urban foraging has also been linked to altered body condition (Cypher & Frost, 2017;Otali & Gilchrist, 2004;Townsend et al., 2019), reproductive success (Pineda-Pampliega et al., 2021;Seress et al., 2012), and increased pathogen and parasite exposure in animal populations (Fuirst et al., 2018;Murray et al., 2016;Sugden et al., 2020). Thus, the utilization of urban food subsidies, such as that observed in this study, may also result in changes to individual and population health, population structure, or community dynamics.
We demonstrate that urbanization of coastal regions can have profound impacts on trophic dynamics, both within and between species. We found that gulls in peri-urban coastal areas foraged more in urban regions than gulls in less-urban marine areas, which subsequently impacted trophic signatures, induced trophic niche expansion, and increased interspecific overlap in trophic niche and foraging space in these species. Given the global increase in studies focusing on the effects of urban systems on wildlife over the last decade (e.g., Chejanovski et al., 2022;Ordeñana et al., 2010;Osterback et al., 2015;Rocha & Fellowes, 2020;Rodewald & Gehrt, 2014), we urge future studies to consider their results within the context of urban, peri-urban, and less-urban settings to gain a more holistic understanding of the impacts of global urbanization on animals and ecological phenomena. We highlight that the mobility and plasticity of a species is an important consideration when examining the effects of urban development on wildlife. Research assessing how and at what scale urban and peri-urban systems influence wildlife is key to understanding how further globalization will shape ecosystem structures, particularly along coastal regions.