1 INTRODUCTION

The Laridae family (gulls) is a recent cosmopolitan family of seabirds, in the Charadriiformes order, presenting an exceptional ecological and life-history diversity (Thomas et al., 2004). Laridae includes the gulls, terns, noddies and skimmers. The latter present powerful flight and hunting behaviours. The yellow-legged gull, Larus michahellis, is a typical surface-feeding seabird on the Mediterranean coast. It is known as the yellow-legged gull because its legs are usually yellow in colour. They show relatively long wings during flight and a broad bill with a curved tip. Adults present a bigger red patch on their bill, which usually reaches the upper mandible and a slightly darker dorsal side with a lighter blue shade. They are generalist seabirds as they can fly, swim, walk, dive etc. In addition, they are omnivorous and can live off different animals (fish, crustaceans, worms, eggs and other nestlings' species) and human waste. The yellow-legged gull is considered an invasive species because of its competitive behaviour and its abundance. It is an opportunistic predator and a non-migratory species. It usually nests in colonies, although it can also nest in solitary pairs, normally on cliffs, coastal islands and lake shores. Although it is a diurnal bird, it is partially active at night during its life cycle. Emond et al. (2006) and Vidal et al. (2018) described the retinal morphology of different gulls, and Victory et al. (2021) recently characterized the yellow-legged gull's foveae. Very little is known, however, of the colour vision and visual acuity of gulls and of seabirds generally.

Colour vision and visual acuity have been studied in passeriformes, psittaciformes, galliformes, anseriformes and birds of prey (Baumhardt et al., 2014; Bowmaker et al., 1997; Ödeen & Håstad, 2003). Despite an increasing interest in this field, vision in seabirds has not been studied in depth. Visual field and optic structures have been studied in the Humboldt penguin, Spheniscus humboldti (Martin & Young, 1984), and the manx shearwater, Puffinus puffinus (Martin, 1991; Martin & Osorio, 2008), and spatial resolution and optical sensitivity in the leach's storm petrel, Oceanodroma leucorhoa, and northern fulmar, Fulmarus glacialis (Mitkus et al., 2016).

Light microscopy and spectrometry studies have established that birds possess highly developed colour vision because of the presence of five different cone photoreceptors with maximum sensitivity to long (LWS), medium (MWS), short (SWS) and either violet (VS) or ultraviolet (UVS) wavelengths and a type of long-wavelength-sensitive (LWS) double cone (Bowmaker et al., 1997; Hart, 2001a, 2001b). Moreover, tetrachromatic vision in birds may be more complex due to the presence of coloured oil droplets positioned at the distal end of the inner cone segments. These droplets cover the entire photoreceptor's entire width, acting as spectral filters and limiting wavelengths that reach the external segments (Stavenga & Wilts, 2014; Toomey & Corbo, 2017; Toomey et al., 2016). Oil droplets are designated as R (red), Y (yellow), C (colourless) and T (transparent) in LWS, MWS, SWS and VS/UVS single cones and P (principal) and A (accessory) in the principal and accessory members of the LWS double cones respectively (Hart, 2001a, 2001b). Oil droplets are situated immediately prior to the light-sensitive outer segment in the light path, and their role is to filter light before it enters the outer segment, thereby sharpening the spectral sensitivity function of individual photoreceptors and contributing to colour discrimination (Wilby & Roberts, 2017). Many oil droplets contain mixtures of carotenoid pigment (Arteni et al., 2019; Johnston & Hudson, 1976; Toomey et al., 2015). They have mainly been studied for their spectral filtering properties (Hart, 2001a, 2001b; Partridge, 1989) and their influence on tuning the spectral sensitivity of colour vision, thereby improving colour discrimination and colour constancy (Vorobyev & Vorobyev, 2003).

The distribution of photoreceptors and oil droplets of cone cells have been described in many avian retinas, and it is known that the distribution of different cone types varies across the retina not only between species, but also between individuals (Budnik et al., 1984; Mariani & Leure-Dupree, 1978; Meyer & May, 1973; Rahman et al., 2007; Rahman et al., 2007; Tyrrell et al., 2019). These variations reflect differences in sensory capacities and in visual ecology (Coimbra et al., 2015; Hart, 2001a, 2001b, 2004; Hart et al., 2006; Martin, 2017; Mitkus et al., 2014; Rahman et al., 2006, 2007a, 2007b). On the other hand, the foveae cone composition has been described in raptors (Mitkus et al., 2017; Montoyo et al., 2018) and in the Larus michahellis (Victory et al., 2021). These studies show that the retinal structure presents a high degree of variation that may result in differences in colour vision and spectral sensitivity in different parts of a single eye's visual field (Martin, 2017). These differences highlight the presence of interspecific differences, especially regarding precise tasks, such as foraging, which are related to the species' ecology (Martin, 2017). The latter implies that vision variations between species are likely to occur within and between orders, families, genera and even individuals.

Although birds need to accurately detect, identify and track fast-moving objects, their visual resolution, that is the eye's upper limit of spatial resolving power, has been poorly characterized compared to other vertebrates. Some studies have shown the spatial visual acuity of raptors (Harmening et al., 2009; Reymond, 1985, 1987), two parrot species (Mitkus et al., 2014), several seabirds such as the penguins Aptenodytes patagonicus and Eudyptula minor (Coimbra et al., 2012), two procellariiforms, Leach's storm petrel, Oceanodroma leucorhoa, and the Northern fulmar, Fulmarus glacialis (Mitkus et al., 2016). Improved visual acuity in birds may be correlated with the presence within the cone-rich avian retina of well-developed areae and foveae (often one or two) of variable depth and size, generally observed within either a central or lateral area (Bringmann, 2019; Fite & Rosenfield-Wessels, 1975; Meyer, 1977; Pumphrey, 1948; Reymond, 1987; Ruggeri et al., 2010). In addition, a bandlike area or visual streak of high cell density has been found to extend horizontally across the retina (Coimbra et al., 2014, 2015; Hayes & Brooke, 1990; Lisney et al., 2012, 2013; Mitkus et al., 2014, 2016; Tyrrell et al., 2013), including in seabirds (Hayes & Brooke, 1990; Mitkus et al., 2016; Victory et al., 2021). According to ‘terrain theory’, this horizontal visual streak, with a greater density of cones combined with an increase in the number of bipolar and ganglion cells, is characteristic of animals that inhabit open environments and provides appropriate horizontal plane vision (Collin, 1999; Hughes, 1977). In this regard, retinal ganglion cells are the retina's only source of output. Therefore, the ganglion cell layer's topography has proved to be an unrivalled source of information on a species’ visual capability when combined with visual optics, and it had been associated with the ecology of the species (Hughes, 1977). Most studies on avian visual acuity use the retinal wholemount technique (Stone & Johnston, 1981; Ullmann et al., 2012) and retinal ganglion cell density counts. Coimbra et al. (2015), however, recommend using retinal ganglion cell density in foveate retinas, and photoreceptor density in foveate bird species. The topographic analysis of cone photoreceptor cells provides information on the importance a species places on sampling specific areas of its visual environment (Lisney & Hawryshyn, 2010). Moreover, the different types of retinal cone photoreceptors allow to generate models of how the different types of retinal cone photoreceptors in birds mediate colour discrimination (Endler & Mielke, 2005; Vorobyev & Osorio, 1998; Vorobyev & Vorobyev, 2003).

Although variations in the relative densities of different cones, according to visual ecology, have been reported between bird species (Bowmaker, 1977; Bowmaker & Knowles, 1977; Coimbra et al., 2015; Goldsmith et al., 1984; Hart, 2001a, 2001b, 2002; Hart et al., 1998, 2000; Kram et al., 2010; Meyer, 1977; Moore et al., 2012; Rahman et al., 2006), few studies have addressed the differences in seabirds (Hart, 2001a, 2001b; Muntz, 1972). Recent studies have reported the retinal morphology and foveae characterization of the yellow-legged gull (Victory et al., 2021; Vidal et al., 2018). In the present work, the six photoreceptors were identified in the yellow-legged gull retina, the cone distribution was analysed, and the spatial resolution was estimated from cone photoreceptors in central fovea. Although vision is the most important sense for birds, surprisingly, little is known of their anatomical spatial resolution. Our findings therefore provide significant data and make a substantial contribution to our general understanding of the field.

2 MATERIAL AND METHODS 2.1 Study specimens

The sample consisted of three adult male yellow-legged gulls collected from the Santa Faz Wildlife Recovery Centre (in Alicante, Spain). They were euthanized by means of an intravenous overdose of sodium pentobarbital (Eutanax®, Fatro Ibérica, SL) for reasons not connected with this work. The study was conducted in conformity with the European Union and the Spanish government regulations. In addition, it was approved by the Experimentation Ethics Committee of the University of Alicante (UA-2020–03–04). The eyes were collected within 30–60 min after the bird's death.

2.2 Retinal wholemount preparation

After being extracted, the left eyes of three individuals were dissected and the anterior portion, including the cornea, lens and iris, was cut away. After removing the vitreous humour, the retina was excised from the eyecup and placed in an 0.9% saline solution at 37°C until the pigment epithelium became detached from the retinas, in accordance with Kolb and Jones (1982). Each retina was then mounted (photoreceptor side up) on a glass slide. To allow it to lay down flat, radial cuts were made and the pecten oculi was cut out. Its position, however, was utilized to conserve its orientation. A few drops of 50% glycerol in buffer were placed onto the retina, and the slide was gently coverslipped, preventing bubbles from forming.

To avoid the degradation of the oil droplets or colour disappearance, the retinal samples were examined, microphotographed and analysed under the microscope immediately after they were mounted. Following Ullmann et al. (2012), a basic drawing was made of each retina and they were scanned under a low-power objective. The oil droplet distinction criteria consisted of the diameter and colour appearance under bright-field illumination (Table 1), based on the criteria established by Hart (2001a), with some modifications. Moreover, we viewed the slides under an epi-fluorescence microscope (Leica DMRB) using UV illumination in order to distinguish the P-, C- and T-type oil droplets. T droplet was no observable under UV illumination; C droplet was strongly fluorescent, and P droplet was medium fluorescent under UV illumination. The distribution of different types of oil droplets was determined.

TABLE 1. Criteria used to distinguish the different oil droplet types based on their diameter and colour appearance under bright-field illumination Oil droplet R-type Y-type C-type T-type P-type Cone type LWS MSW SWS UVS LWS double Diameter (µm) 2.6–4 2.4–3.6 2.3–3.4 1.5–2.5 2.6–3.5 Colour under bright-field illumination Red Golden yellow to orange Colourless to bluish green Colourless to pale blue Pale green or greenish yellow Stratification Intermediate Intermediate Vitread Vitread Sclerad Note Each oil droplet is also assigned to its respective spectral class of cone photoreceptor. 2.3 Counting and measuring oil droplets

Based on Ullmann et al. (2012), a basic drawing was made of each retina. The retinas were scanned under a low-power objective (10×), and relevant features were added. To analyse the oil droplet distribution across the retina, the retina was divided into 9 central regions and 11 peripherals, taking into account the temporal, nasal, dorsal and ventral regions as indicated in Figure 1a. The retina was then broken down into four quadrants, each separated into several regions: the dorsotemporal (DT) region, which includes the CDT1, CDT2, PDT1 and PDT2 regions; the dorsonasal (DN), with regions CDN1, CDN2, PD, PDN and PN2; the ventronasal (VN), including the CVN, PN1, PV2 and PVN regions; and the ventrotemporal (VT) quadrant, comprising the CVT, PT, PV1 and PVT regions (Figure 1a). The equatorial retina, which is immediately dorsal to the pecten oculi, was divided into three different regions: the CC, CT and CN. The latter were treated independently from the rest of the retina, since they potentially presented retinal specializations with a high density of cells. Such regions are representative and comparable between different species (Hart, 2001a, 2001b).

(a) Graphical representation of a right retina showing the 20 areas analysed in this study. P, peripheral, C, central, D, dorsal, V, ventral, T, temporal, N, nasal, PO, Pecten oculi. (b) Photomicrograph showing the topography of cone oil droplets on a fresh and whole-mounted retina under a bright-field microscope. R, Y, C, T, P and A: red, yellow, colourless, transparent, principal and accessory oil droplets respectively

A graticule of 1 × 1 mm was used on the whole mount coverslip to select retinal areas and to perform the counting. In each 1 × 1 mm square marked in the graticule, several microphotographs were taken with the 63X, a mosaic of an area of 0.75 × 0.75 mm was made and counted on this mosaic. All oil droplet types were counted and converted into cells/mm2. To prevent double counting, we only counted the cells lying within each square grid and ‘touching’ the top and right sides of the square grid. Since oil droplets are located on different planes and their size varies throughout the retina, microphotographs were taken using at least three different focal planes within the same visual field. An orthogonal projection was then performed from each visual field. The density and percentage of the different oil droplet types were determined in each retinal region, and the percentage and density average were then calculated in the central, peripheral, temporal, nasal, dorsal and ventral areas. Finally, the diameter (in µm) of each oil droplet type was measured in each region. All counting and measurements were conducted using the public domain ImageJ software (Schneider et al., 2012).

2.4 Processing for light and transmission electron microscopy

In order to associate each oil droplet with a specific type of photoreceptor, and to determine their morphological characteristics, samples of two right eyes were processed under a light and electron microscope. The criteria established by Hart (2001a) were applied.

In accordance with Segovia et al. (2016), the retinas were immersed in the fixative (1% paraformaldehyde, 1.6% glutaraldehyde, 0.15 mm CaCl2 in 0.1 M phosphate buffer, pH 7.4) for 2 h at room temperature and then at 4°C overnight. Each retina was divided into central and peripheral zones, and retinal fragments were post-fixed in 2% OsO4 in an 0.1 M phosphate buffer for 1 h, pH 7.4 and then dehydrated in ascending series of ethanol, cleared in propylene oxide and flat-embedded in Epon 812 epoxy resin. Blocks were oriented and sectioned vertically using a Leica LKB-III ultramicrotome (Leica, EZ4D). Semithin sections of 1.0 µm were mounted on gelatinized slides, stained with 0.5% toluidine blue and examined under a Leica DMRB light microscope. Photomicrographs were taken with a Lumenera Infinity microscope camera. Finally, ultrathin sections were cut with a diamond knife, double-contrasted with uranyl acetate and lead citrate and examined under a JEOL JEM-1400 Plus transmission electron microscope at 120 kv (Tokio, Japan), equipped with a Gatan Orius digital camera for taking images (Pleasanton, USA).

2.5 Estimation of visual acuity

The theoretical, spatial resolution was calculated using the average oil droplet diameter in the central fovea based on the cross-sectional retinas, and the posterior nodal diameter (PND) as a proxy for focal length (Hart, 2002). The PND was assumed to be 0.6 × of the eye axial length (Hart, 2002; Hughes, 1977; Lisney & Collin, 2008; Lisney et al., 2012, 2013; Martin, 1994). The axial length was measured based on ten eyes from five individuals after enucleation using digital callipers as described by Lisney et al. (2012, 2013).

In order to calculate the spatial resolution, Fnc, we used the cone oil droplet's diameter in accordance with Mitkus et al. (2016) and following the equation of Miller (1979): where PND is the posterior nodal distance, d is the oil droplet diameter and the resolution is expressed in cyc deg−1. 2.6 Statistical analysis

Differences in oil droplet densities, percentages and diameter among the different retinal regions were analysed by conducting a Student t or Mann–Whitney test and a one-way ANOVA test followed by Tukey's multiple comparisons test. A two-way ANOVA test was performed for each type of oil droplet in order to determine significant differences in the 20 regions studied among the three retinas analysed. All statistical analyses were performed using the GraphPad Prism Software version 8.00 for Windows.

3 RESULTS 3.1 Light microscopy of fresh samples

Five types of brightly coloured oil droplets were found in fresh flat preparations with the photoreceptor side up: red, yellow, colourless, transparent and green. Based on Hart (2001b), these oil droplets were designated as (R-type), (Y-type), (C-type), (T-type) and (P-type), respectively, and associated with the LWS, MWS, SWS and UVS single cones and the LWS double cone pair (Figure 1b). A fragmented greenish microdroplet was sometimes observed in the accessory member of the double cone pair. It was named (A-type), but it was not considered in the identification of the double cones (Figure 1b). Oil droplets differed in diameter, which was significantly lower in the central areas with the highest density than in the peripheral retina with lower density areas (Figure 2a,b). The R-type oil droplet had the largest diameter (3.59 ± 0.271 µm in the peripheral region and 3.04 ± 0.218 in central areas). In contrast, the T-type oil droplet was clearly the smallest (2.36 ± 0.159 µm in the peripheral region and 1.78 ± 0.212 µm in the central areas) (Figure 2b). Table 2 summarizes the diameters of each oil droplet type in the 20 regions analysed. Although the diameter of all oil droplets decreased as the retinal centricity increased, the mean diameter of each oil droplet type was constant in each quadrant (Figure 2c).

(a) Photomicrograph of CC and PT regions showing the topography of cone oil droplets on a fresh and whole-mounted retina under a bright-field microscope (a, b). Digitized versions for each photomicrograph field were elaborated. Coloured dots represent an oil droplet type: R-type (c,d), Y-type (e, f), C-type (g, h), T-type (i, j), P-type (k, l). Scale bar: 10 µm. Note the difference in densities and diameters between central and peripheral regions. (b) Graphical representation comparing means and standard deviations of the diameter of each oil droplet type between central (C) and peripheral (P) regions. p < 0.001(***). (c) Graphical representation of the oil droplet diameter found in the four quadrants of the retina: quadrant DT, DN, VN and VT. The average of each quadrant was obtained from the following regions: DT (CDT1, CDT2, PDT1 and PDT2); DN (CDN1, CDN2, PD, PDN and PN2); VN (CVN, PN1, PV2 and PVN) and VT (CVT, PT, PV1 and PVT). The difference between the means was not statistically significant for all variables with the same letter within the quadrants (p < 0.05)

TABLE 2. Mean and standard deviation of the oil droplet diameter in the 20 regions analysed from three different retinas Region Oil droplet diameter (µm) (Mean ±SD) R-type Y-type C-type T-type P-type CC 2.589 ± 0.153l 2.422 ± 0.127i 2.335 ± 0.222j 1.525 ± 0.103h 2.556 ± 0.073i CT 2.971 ± 0.094k 2.724 ± 0.132h 2.463 ± 0.11i 1.751 ± 0.093f 2.751 ± 0.107h CN 3.007 ± 0.146k 3 ± 0.111g 3.067 ± 0.129e 1.821 ± 0.099f 2.837 ± 0.141gh CDN1 3.158 ± 0.096ghi 3.229 ± 0.104ef 2.591 ± 0.108gh 1.621 ± 0.112gh 2.744 ± 0.145h CDN2 3.131 ± 0.105hij 3.036 ± 0.128g 2.5 ± 0.098hi 1.639 ± 0.154g 2.622 ± 0.115i CDT1 3.02 ± 0.134jk 3.152 ± 0.132f 2.677 ± 0.111g 1.728 ± 0.09fg 2.945 ± 0.118f CDT2 3.066 ± 0.153ijk 2.988 ± 0.103g 2.837 ± 0.115f 1.793 ± 0.122f 2.861 ± 0.163fg CVN 3.268 ± 0.09fg 3.248 ± 0.116ef 3.12 ± 0.124de 2.119 ± 0.143e 3.262 ± 0.116de CVT 3.16 ± 0.102ghi 3.24 ± 0.124ef 3.062 ± 0.157e 2.019 ± 0.107e 3.315 ± 0.106cd PT 4.021 ± 0.126a 3.557 ± 0.141ab 3.416 ± 0.093b 2.541 ± 0.213a 3.497 ± 0.098a PN1 3.434 ± 0.111de 3.259 ± 0.149ef 3.143 ± 0.107cde 2.309 ± 0.094bcd 3.299 ± 0.096cd PN2 3.373 ± 0.157ef 3.324 ± 0.129de 3.154 ± 0.097cde 2.368 ± 0.091bc 3.285 ± 0.105cd PD 3.92 ± 0.121ab 3.624 ± 0.125a 3.544 ± 0.172a 2.567 ± 0.116a 3.457 ± 0.111ab PDN 3.245 ± 0.197gh 3.622 ± 0.081a 3.249 ± 0.133c 2.288 ± 0.177bcd 3.161 ± 0.194e PDT1 3.63 ± 0.097c 3.458 ± 0.116bc 3.223 ± 0.105cd 2.253 ± 0.085d 3.377 ± 0.114bc PDT2 3.901 ± 0.095b 3.489 ± 0.131bc 3.204 ± 0.116cd 2.31 ± 0.128bcd 3.354 ± 0.093bcd PV1 3.508 ± 0.088d 3.317 ± 0.108de 3.141 ± 0.118cde 2.388 ± 0.108b 3.277 ± 0.107cd PV2 3.487 ± 0.148de 3.308 ± 0.108de 3.167 ± 0.15cde 2.345 ± 0.098bcd 3.277 ± 0.085cd PVN 3.473 ± 0.111de 3.391 ± 0.104cd 3.143 ± 0.117cde 2.37 ± 0.091bc 3.382 ± 0.099bc PVT 3.455 ± 0.111de 3.247 ± 0.093ef 3.143 ± 0.118cde 2.262 ± 0.11cd 3.35 ± 0.074bcd One-way ANOVA + Tukey's multiple comparison test. Difference between means is not statistically significant for all variables with the same letter within columns (p < 0.05).

Oil droplets were stratified according to colour. P-type oil droplets were seen at the most sclerad level of focus. Also visible were the R-type oil droplet and Y-type droplet, although at a more vitread level. Finally, at the most vitread level, C-type and T-type oil droplets were visible. Moreover, P-, C- and T-type were distinguishable by their relative brightness under UV illumination: the P-type showed medium fluorescence, the C-type strong fluorescence and the T-type no observable fluorescence.

3.2 Photoreceptor morphology

Six different photoreceptors were distinguished in the blue toluidine semithin sections, but oil droplet colour alterations were found. Photoreceptors were then differentiated on the basis of presence and oil droplet characteristics (diameter and stratification), nuclei position and axon morphology (Figure 3a). Moreover, transmission electron microscopy allowed characterizing the cone based on mitochondrial morphology, the presence of paraboloid and cytoplasm electron density.

(a) Schematic representation of rod and cones found in the L. michahellis retina. Each cone is related to the wavelength of their visual pigments, and they are associated with LWS, MWS, SWS and UVS single cones respectively. The P-type corresponds to the principal member of the LWS double cone pair. A: accessory member of the LWS double cone pair; OS, outer segment; IS, inner segment; OLM, outer limiting membrane; ONL, outer nuclear layer; OPL, outer plexiform layer; N, nucleus; white arrowhead: oil droplets; black arrow: mitochondria/ellipsoid; H, hyperboloid; P, paraboloid; black arrowheads: axon; ST, synaptic terminal. (b) Photomicrograph of ultrathin sections showing the different cones in a central region of L. michahellis. R, Y, C and T: R-, Y-, C- and T-type oil droplets and accessory oil droplet (A); OLM, outer limiting membrane; N, nucleus; E, ellipsoid and M, myoid

3.2.1 Rods

Rod photoreceptors had no oil droplets in their inner segments, and their outer segments were wider and longer than those of the cones. The ellipsoid was characterized by an accumulation of mitochondria and on the immediate vitread side to this was an aggregation of glycogen known as hyperboloid (Figure 3a). Numerous profiles of rough endoplasmic reticulum and cytoskeleton fibres constituted the myoid. The nucleus, which was in the most vitread layer of the outer nuclear layer (ONL), displayed a slightly condensed chromatin and a clearly differentiated nucleolus. The synaptic terminal was in the outermost stratum in the outer plexiform layer (OPL).

3.2.2 Single cones

The R-type oil droplet was assigned to LWS single cones and the Y-type oil droplet to MWS single cones. Under electron microscopy, the R-type oil droplet was recognizable due to its pale appearance, its larger size and its location above the Y-type oil droplet (Figure 3b). Both cones had an ellipsoid with numerous mitochondria that were arranged parallel to the longitudinal axis of the inner segment; the latter were more electron-dense in LWS cones than in MWS ones (Figure 3b). The nuclei of both cones were near the outer limiting membrane (OLM).

The C-type oil droplet belonged to SWS cones, and the T-type oil droplet was associated with UVS cones. Both oil droplets were localized at the most vitread level of focus. The T-type oil droplet was the most electron-lucent under electron microscope examination and considerably smaller than the C-type. UVS cones had the slenderest inner segments and, typically, had an oblique axon connecting its perikaryon to the synaptic terminal which were adjacent to the inner nuclear layer (INL) (Figure 3a).

3.2.3 Double cones

Double cones were composed of principal and accessory members. The principal member was distinguished by an oil droplet located in the most sclerad level that corresponded to the green oil droplet observed in fresh retinas. This oil droplet was named P-type and, according to Hart (2001b), belongs to the LWS double cone. It was always accompanied b