Sexual and individual foraging segregation in Gentoo penguins Pygoscelis papua from the Southern Ocean during an abnormal winter

Knowledge about sexual segregation and gender-specific, or indeed individual specialization, in marine organisms has improved considerably in the past decade. In this context, we tested the “Intersexual Competition Hypothesis” for penguins by investigating the feeding ecology of Gentoo penguins during their austral winter non-breeding season. We considered this during unusual environmental conditions (i.e. the year 2009 had observations of high sea surface and air temperatures) in comparison with the long term average at Bird Island, South Georgia. Through conventional (i.e. stomach contents) and stable isotopic values from red blood cells, plasma and feathers of both male and female Gentoo penguins, we showed that there were significant differences between sexes, with males feeding mainly on fish (54% by mass) followed by crustaceans (38%) whereas females fed mainly on crustaceans (89% by mass) followed by fish (4%). Themisto gaudichaudii was the most important crustacean prey for males (64% by mass; 82% by number; 53% by frequency of occurrence) and females (63% by mass; 77% by number; 89% by frequency of occurrence), contrasting with all previous studies that found Antarctic krill Euphausia superba were generally the main prey. Stable isotopic data showed that, in terms of habitat use (based on δ 13C), there were significant differences in short-term carbon signatures between males and females (based on plasma and red blood cells), suggesting that both sexes explored different habitats, with females exploring more offshore pelagic waters and males feeding more in coastal benthic waters. Based on δ 15N, males fed on significantly higher trophic level than females (based on plasma and red blood cells), in agreement with our diet results., Thus, Gentoo penguins behave in a similar manner to other non-breeding penguins species (e.g. king, macaroni and rockhopper penguins), albeit at a smaller spatial scale (as they do not disperse as these other penguins do), in that they have a wider habitat and trophic niche during the Antarctic Winter (in comparison to Summer). We also detected individual specialization in feeding/trophic levels for each gender, with certain males feeding mainly on fish and certain females mainly on crustaceans, which may be driven the prevailing environmental conditions that lead individuals to search for alternative prey, and cause sexual diet segregation. Our results provide further information to help improve understanding about sexual segregation and individual specialization of marine organisms, while contributing valuable information on the winter diet for Antarctic monitoring programs and for modelling Antarctic marine food webs.

Therefore, it is important to understand both diet segregation and individual specialization within a population. In our study, we looked at these mechanisms when animals have no reproductive obligation (to assess their diet flexibility) during the austral winter at Bird Island, South Georgia.
We focused on a period of unusual environmental conditions as such conditions are most likely to reveal individual differences, and because information on such conditions is important as they are likely to affect the ecology, management and conservation of Antarctic ecosystems [42].
Ocean warming has been recorded at South Georgia during the austral winter (i.e. August), with a mean increase of~2.3˚C since 1925 [43]. In 2009 (when our study took place), high sea surface and temperatures occurred across many consecutive months, and coincided with extremely low catches in local fisheries and poor breeding success in higher predators at South Georgia [44,45] (Fig 2). Prior to the fieldwork at Bird Island, a research cruise (March-April 2009) in the Scotia Sea (including around South Georgia), showed that Antarctic krill density was lower (2.2 g m -2 ) than in previous years (Antarctic krill density assessments made annually since 1996) [46]. These findings suggest that this region, during the austral winter of 2009, was atypical for prey availability/abundance to penguins and other predators in the region. Indeed, anomalous oceanographic, sea-ice and/or weather conditions can cause changes in prey availability [12,47,48] and are known to have negative impacts on populations of seabirds, including mass mortalities (also known as "wrecks" of seabirds) [49][50][51][52].
In the context of low Antarctic krill abundance, the behaviour of male and female penguins could provide valuable data on diet sexual segregation and individual specialization. The "Intersexual Competition Hypothesis" (see also [53] for intraspecific competition) predicts that, in dimorphic species, reduction in competition for food between sexes results from the capture of different prey types by each sex and that this is one of the main selective forces maintaining sexual dimorphism [54,55]. Here, we had the unique opportunity to investigate the feeding ecology of Gentoo penguins during an oceanographically abnormal non-breeding season. The objectives of the study were: • Characterize the winter diet of Gentoo penguins using conventional techniques (i.e. stomach contents), and stable isotope analyses of red blood cells, plasma and feathers to evaluate diet changes, under known environmental conditions; • Evaluate differences in diets according to sex and individual specialization; • Assess implications of these results on the conservation of Gentoo penguins, under extreme and low food availability.

Fieldwork and stomach content analyses
Fieldwork was carried out during the austral winter of 2009, between June and September, at Bird Island, South Georgia. The British Antarctic Survey (BAS) provided all the support related to the permits for the fieldwork (as fieldwork was carried out from a British Antarctic research base): "The animal procedures used in this this study were reviewed and approved by the Joint BAS-Cambridge University Animal Welfare and Ethical Review Committee. Permits to operate were issued by the Government of South Georgia and the South Sandwich Islands". Stomach samples were obtained each month (N = 13-15 penguins handled, always ensuring that at least 10 samples contained food) from non-breeding Gentoo penguins at one colony (known locally as Landing Beach).
For each month, all samples were collected within a 4-day period. Penguins were selected randomly when returning to the colony at dusk. Birds may, or may not have used the site for breeding, as Gentoo penguins move around the archipelago during the winter [25]. Each penguin was handled as follows: after putting a cover on the penguin's head (to reduce stress), the bill was measured (length and depth), followed by the height of the penguin and its weight (a harness was produced to specifically hold each penguin comfortably) using a 5 kg Pesola1 spring balance (Pesola AG, Barr, Switzerland). Post-molt feathers (6-8 chest feathers) and blood were collected, following Ceia et al. [56], adapted to penguins; the blood samples were collected using 1 ml syringes (25 G needles) and were separated into plasma and red blood cells (RBC) using a centrifuge (15 min at 3000 rpm), stored frozen (-20˚C), and later freezedried and homogenized prior to stable isotopic analyses. Finally, a food sample was obtained by stomach flushing, following Xavier et al. [57] adapted to penguins (stomach samples were successfully obtained with a maximum of 3 flushes), following the Convention for the Conservation of Antarctic Marine Living Resources (CCAMLR) Ecosystem Monitoring Program (CEMP) Standard Methods. If the first flush produced a green or yellow coloured sample, it was assumed that the penguins were without food and they were released. All penguins handled were then marked to ensure they were sampled only once, and released. The procedures lasted, on average, 15 minutes.
The analyses of the food samples were carried out at the Bird Island research station laboratory within 24 hours of collection. Each food sample was analysed following Xavier et al. [58]. The samples were weighed and the overall mass was recorded. To remove the liquid, each food sample was washed carefully through two sieves (1.00 and 3.35 mm). Only fresh (recently consumed) material was found (i.e. no old cephalopod beaks nor other indigestible material (e.g. stones or plastics) was found). All components were then sorted into categories (crustaceans, fish, cephalopods and others; the latter comprising other fresh prey). Crustaceans were identified when possible using the BAS and Royal Belgium Institute of Natural Sciences reference collections, and the key reference bibliography of Kane [59] and Boltovskoy [60]. The total length of crustaceans was measured when possible (from tip of the eyes to the end of uropods/ telson). The fish otoliths were identified following Hecht [61], Williams and McEldowney [62], Smale et al. [63], Reid [64] and fish size relationships used were given by Adams and Klages [65], Hecht [61], Williams and McEldowney [62], Smale et al. [63], Reid [64], Olsson and North [66], Berrow and Croxall [67] and Croxall et al. [68,69]. We putatively identified one group of very small otoliths as? Gymnoscopelus braueri (preliminary identification suggested by Marcella Libertelli) but subsequent genetic analyses using flesh did not confirm this identification. The number of fish was estimated from the number of intact crania containing both otoliths, and loose otoliths, in each sample. These loose otoliths were compared with each other (right otolith compared with left otolith by size and level of erosion) and paired if possible (e.g. if four loose otoliths were found with similar sizes and similar sizes of erosion, and two were left otoliths and the other two were right otoliths, it was assumed that there had been two fish). The cephalopod beaks were counted (both upper and lower), identified and measured. Only the lower cephalopod beaks were measured, using the lower rostral length (LRL) following Xavier and Cherel [70] and the BAS beak reference collection. Allometric equations, of mantle length (ML) and estimated mass (W) for cephalopods were taken from Xavier and Cherel [70]. The components and species were analysed by frequency of occurrence, number and by mass following Xavier et al. [58].

Stable isotopic analyses
Tissue δ 13 C and δ 15 N values provide useful information about seabird diet at different temporal scales [71]. The carbon stable isotope value (δ 13 C) mainly reflects consumers' foraging habitat, while the nitrogen stable isotope value (δ 15 N) is mainly used to define consumers' diet and trophic position. The isotopic niche of each adult was determined by using three complementary tissues (plasma, RBC and feathers) that record trophic information at different time scales [72]. Plasma and RBC retain information on diet (carbon source and trophic level) from a few hours/ days prior to sample collection to the previous 3-4 weeks, respectively [73,74]. Hence, the isotopic signature of plasma is representative of the food and feeding ecology of the penguins during their last few daily foraging trips. Breast feathers represent the diet during the previous premoulting stage, since feather keratin is metabolically inert after synthesis, which in Gentoo penguins from South Georgia occurs generally between March and April [75]. Therefore, based on stable isotopic analyses of different tissues from the same penguin we are able to test for shortterm (i.e. days/weeks-RBC versus plasma) and medium-term (i.e. weeks to months-RBC versus feathers, when these were grown) consistency in individual foraging niche [56].
Lipids are depleted in 13 C relative to whole tissues and were removed from plasma using successive rinses in a 2:1 chloroform: methanol solution [56,76]. The low lipid content of whole blood (or RBC) does not typically require lipid extraction [77]. Prior to stable isotopic analyses, feathers were cleaned of surface contaminants using successive rinses in a 2:1 chloroform: ether solution, air-dried and then ground to a fine powder in a freezer mill operating at liquid nitrogen temperature.
Nitrogen and carbon isotope ratios were determined by a continuous-flow isotope ratio mass spectrometer (Delta V Advantage, Thermo Scientific) coupled to an elemental analyser (Flash EA1112, Thermo Scientific) in the LIENSs, Université de La Rochelle, France. Approximately 0.3 mg of each sample was combusted in a tin cup for the simultaneous determination of nitrogen and carbon isotope ratios. Results are presented in the usual δ notation based on the Vienna PeeDee Belemnite (V-PDB) for carbon and atmospheric N 2 (AIR) for nitrogen. Replicate measurements of internal laboratory standards (acetanilide) indicate measurement errors <0.15 ‰ for both δ 13 C and δ 15 N.
To analyse stable isotope data in the context of isotopic niche between sexes, we adopted the recent metrics based in a Bayesian framework (Stable Isotope Bayesian Ellipses in R: SIBER [78]), which allows for robust statistical comparisons. The Bayesian approximation of the standard ellipse area (SEAb) is a metric used to test whether Group 1 (males) standard ellipse area (SEA) is smaller than Group 2 (females) and is calculated based on 1000 replications. The SEA corrected for small sample sizes (SEAc, an ellipse that contains 40% of the data regardless of sample size) was adopted to compare niche width between sexes (see Jackson et al. (2011) for more details). The SEAb and the layman metric of convex hull area (TA) [79] were also calculated as a measure of isotopic niche area. SEAb was used to test whether Group 1 is smaller than Group 2 (i.e. p, the proportion of ellipses in males that were lower than females), following Jackson et al. (2011). We used the computational code to calculate the metrics from SIBER implemented in the package SIAR [80] under R 3.2.1.

Molecular sexing
Blood samples from Gentoo penguins were collected, as mentioned above, from 55 individuals to identify their sex. DNA from blood was isolated using an adaptation of the Chelex extraction method [81]. All samples were centrifuged for 3 min and a small portion of blood was removed for extraction with a spatula. 50 μl of distilled H 2 O and 20 μl of InstaGene™ Matrix (BioRad) were added to each sample. The samples were then incubated at 50˚C for 30 min, followed by 8 min at 100˚C. One negative control (a tube with all the reagents but without a blood sample) was included for each set of 24 extractions to monitor for possible contamination with exogenous DNA. Primers P2/P8 [82] were used for PCR amplification. These primers have been commonly used for penguins [83,84] and provided sufficient separation of bands (~20 base pairs) to be differentiated on an agarose gel. All PCRs included two positive controls to test for the success of the amplification and two negative controls, prepared with distilled water, to test for possible contamination. Each male result was repeated at least three times and each female result was repeated at least twice. Amplifications were performed using a Multiplex kit, carried out in 10 μl reactions containing 1x of QIAGEN 1 Multiplex PCR Master Mix, 0.2 μM of each primer and 0.8 μl of DNA template (~1 ng/μl). The thermal conditions were 95˚C for 15 min, 35 cycles of 95˚C for 1 min, 47˚C (annealing temperature) for 1 min 30 s, 72˚C for 1 min 30 s, and a final extension at 72˚C for 10 min. All reactions were carried out using an Applied Biosystems Veriti 1 Thermal Cycler PCR machine. Samples were run for about 2h on 3% weight/volume agarose gels stained with ethidium bromide.

Dead counts
In September 2009 (1 st and 8 th Sept.), two surveys for dead Gentoo penguins (n = 111) were carried out at Bird Island (at Landing Beach, Iceberg Point, Freshwater Beach, Stinker Point and Everman Cove), due to the appearance of numerous dead Gentoo penguins on the beaches of Bird Island, South Georgia. After identifying a body of a Gentoo penguin, the bill length and bill depth were measured, as above. Sex was estimated using equations from Williams [85]. It was also reported if it was recently dead or an old carcass. To avoid duplicating the identification of dead animals in the second survey, stock marker was used to paint the bill of the penguins.
Data were statistically analysed using Minitab statistical software (Sowers Printing Company, PA, USA) and R [86]. The values are as mean ± standard deviation, unless stated (significance threshold: 0.05).

Results
The feeding ecology of 55 Gentoo penguins from Bird Island (South Georgia) was studied during the austral winter of 2009 (Table 1; S1 Dataset). Of these, 12 had empty stomachs (or only minor residues of food), which were removed from further analyses. Based on individuals whose stomach contents were analysed, our study showed that male Gentoo penguins (N = 17; body mass = 6666 ± 653 g) were significantly heavier than females (N = 26; body mass = 5334 ± 520 g; Mann-Witney U test, U = 409, P < 0.01). The mean solid proportion of Gentoo penguin food samples ranged from 4 to 414 g (141 ± 101 g), with females having significantly more solids than males (Mann-Witney U test, U = 114, P < 0.01).

Diet composition
Overall, when samples for both sexes were combined, Gentoo penguins fed primarily on crustaceans (68% by mass) followed by fish (25%; Table 2). Cephalopods and other prey/debris represented <1% and 7% by mass, respectively. By prey species, Themisto gaudichaudii was the most important crustacean by frequency of occurrence (74%), by number (77%) and by mass (63%; Table 2). The proportion of crustaceans and fish consumed differed significantly Oedicerotidae  The mean solid proportion of male Gentoo penguin food samples ranged from 4 to 298 g (93 ± 84 g). Males fed mainly on fish (54% by mass) followed by crustaceans (38%), other species (5%) and cephalopods (< 1%). T. gaudichaudii was the most important crustacean prey (64% by mass; 82% by number; 53% by frequency of occurrence) ( Table 1). Euphausia superba represented only 29% by mass and 14% by number, although it had 47% by frequency of occurrence (Table 1). Within the fish component, P. georgianus was the most important fish Foraging segregation in penguins   prey by mass (55% by mass; 27% by number; 12% by frequency of occurrence) and L. larseni by number and frequency of occurrence (10% by mass; 37% by number; 47% by frequency of occurrence; Table 1). The mean solid proportion of female Gentoo penguins food samples ranged from 25 to 414 g (172 ± 100 g), significantly heavier than males (see above). Females fed mainly on crustaceans (89% by mass) followed by fish (4%), others (8%) and cephalopods (< 1%). T. gaudichaudii was the most important crustacean prey (63% by mass; 77% by number; 89% by frequency of occurrence), followed by E. superba (19% by mass; 6% by number; 77% by frequency of occurrence). Of the fish, C. gunnari (67% by mass; 19% by number; 46% by frequency of occurrence) was the most important fish prey by mass (Table 1). P. georgianus (the most important fish species in male Gentoo penguins) was absent in female Gentoo penguins ( Table 1).

Stable isotopic and niche analyses
Stable isotopic and niche analyses, both δ 13 C and δ 15 N, were carried out on RBC, plasma and breast feathers of male and female Gentoo penguins (all individuals; n = 55) (Table 4) (Table 4).
There were also positive significant relationships between isotopic values of δ 13 C and δ 15 N, both in plasma and RBC, with relationships significantly higher in males (δ 13   However, no significant relationships between isotopic values of δ 13 C and δ 15 N in feathers were found in males (Pearson correlation 0.08, P = 0.18) or females (Pearson correlation 0.12, P = 0.06). The isotopic niche width (SEAb) was estimated and found to be significantly higher in males than in females using plasma (p = 0.01) and RBC (p = 0.04), and practically no overlap was detected in the isotopic data (i.e. the area of the standard ellipses; SEAc) of males and females, with males having higher levels of δ 15 N (Table 5; Fig 4). On the other hand, although SEAb was also higher in males than in females using feathers (p = 0.05), a relatively high overlap was detected in the isotopic niche between sexes, with overall larger areas than in blood tissues (Table 5).

Individual specialization
Both males and females exhibited individual specialization (Figs 4,5 and 6). A strong positive relationship relating δ 13 C in RBC versus plasma was found for both males (Pearson correlation 0.80, P< 0.01) and females (Pearson correlation 0.77, P< 0.01) (Fig 5). Similarly, when relating δ 15 N in RBC versus plasma, a strong positive relationship was found for both males (Pearson correlation 0.88, P< 0.01) and females (Pearson correlation 0.76, P< 0.01) (Fig 5). Furthermore, a group of female individuals (n = 4) segregated with lower δ 13 C and δ 15 N from most females whereas a group of male individuals (of variable number) clearly segregated with higher δ 13 C and δ 15 N from most males (Figs 4 and 6). Also, when assessing individuals with highly divergent diets, male individuals that fed on > 80% by mass on fish and compared with females that fed on > 80% by mass on Antarctic krill, the differences in δ 15 N is even more obvious (Fig 4).
Finally, when relating δ 13 C in feathers versus RBC, no relationships were found for either males (Pearson correlation 0.22, P = 0.29) or females (Pearson correlation 0.24, P = 0.24). Similar results were found when assessing δ 15 N in feathers versus RBC, with no relationships found for either males (Pearson correlation 0.26, P = 0.21) or females (Pearson correlation 0.01, P = 0.95).

Dead counts
A total of 111 individual dead Gentoo penguins (of which 34 individuals were classified as recently dead) were found along Bird Island beaches in the two surveys. The bill length Table 5. Metrics of isotopic niche width in plasma and red blood cells' tissues of Gentoo penguins obtained at Bird Island, South Georgia in winter 2009. The area of the standard ellipse (SEAc), the layman metric of convex hull area (TA) and the overlap between males and females for each metric were calculated from SIBER (see Jackson et al. 2011 for more details on these metrics).

Discussion
Our study showed that the feeding ecology of Gentoo penguins during their winter non-breeding season, under abnormal environmental conditions, presented significant differences in sexual and individual segregation between males and females, with males feeding more on fish and females more on crustaceans. Further evidence came from differences noted both in terms of foraging habitat (δ 13 C values; using plasma and feathers) and trophic level (δ 15 N; using plasma and red blood cells). Individual specialization was also detected with individual males and individual females segregating from the majority of the individuals. Such levels of sexual  Table 6). Within the crustacean component, our study showed that Gentoo penguins at South Georgia fed more on Themisto gaudichaudii (63% by mass,) rather than Antarctic krill Euphausia superba (20% by mass), contrasting with previous studies [26][27][28]. These previous studies have shown that Antarctic krill dominated the diet in winter 1987 and 1988 (> 87% by mass). In 1989, fish (not identified to species level) dominated the first sampling period, in early July (73-76% by mass) and Antarctic krill dominated the second sampling period, in late July (> 89% by mass) [27].
Williams et al. [26] also conducted their study in late July, with Antarctic krill also dominating the diet (63.3-100.0% by mass). The variation in diets of Gentoo penguins at South Georgia have been linked to food availability locally [26][27][28], as Gentoo penguins are inshore-feeding species. Under the context of low Antarctic krill availability in autumn 2009 [44,46], the diet shift of Gentoo penguins to T. gaudichaudii implies that high abundances of Antarctic krill did not occur during the following austral winter 2009 in inshore areas where Gentoo penguins forage. Moreover, the occurrence of Gentoo penguins found dead on the beaches (potentially a seabird wreck; see [47]) may support such a statement, although we need to be cautious as no counts of dead penguins from previous years are available, nor their cause of death. From other data available, it suggests that in years of good Antarctic krill availability, Gentoo penguins do feed mainly on Antarctic krill (Table 6). In years of low Antarctic krill availability, the importance of fish increases and the crustacean component is replaced by T. gaudichaudii or other (not Antarctic krill) crustaceans [20,45,[87][88][89][90] (Table 6). Foraging segregation in penguins

Sexual foraging segregation and individual specialization in Gentoo penguins
To our knowledge, this is the first study to assess sexual dietary differences in austral winter under known (abnormal) environmental conditions, when Gentoo penguins are not constrained by breeding duties. Female Gentoo penguins had significantly heavier meals (i.e. solids) than males (see Results) but as prey were inadequate (i.e. the availability of T. gaudichaudii was low), this fact was reflected in females struggling during austral winter 2009; with more dead female Gentoo penguins found than males on shores (see Results).
Although both sexes of Gentoo penguins are known to forage in inshore waters [25], males exhibited a different diet in comparison with females (Tables 1, 4 and 7): sexual differences in Gentoo penguins were found, with females taking more crustaceans (i.e. T. gaudichaudii) and Foraging segregation in penguins males more fish. Also, Bearhop et al. [91] found that Gentoo penguin males foraged at a slightly higher trophic level than females (although not statistically significant), during the austral summer. Therefore, the "Intersexual Competition Hypothesis" for dimorphic species, does apply to Gentoo penguins. This is potentially due to the ability of male Gentoo penguins, with slightly larger bills compared to females (and also heavier mass; see Results) being able to catch fish prey when confronted with low availability of Antarctic krill around South Georgia during their non-breeding period [26][27][28], as suggested for other penguins elsewhere [92,93]. Moreover, male Gentoo penguins are larger, heavier and are able to go deeper to forage than females [94,95], and are able to feed on more bentho-pelagic prey (i.e. Lepidonotothen larseni, Parachaenichthys georgianus [96,97]) than females, which tend to be limited to shallower pelagic waters. However, Croxall et al. [89] and Bost et al. [98] found no significant differences in the diving parameters of male and female Gentoo penguins. With such flexibility in their diet and in foraging habitat, male Gentoo penguins may be more resilient, especially in years when pelagic prey (i.e. Antarctic krill, T. gaudichaudii) are generally scarce (see below); this interpretation is also supported by our results in finding more females dead on the beaches than males (see results). Volkman et al. [99] also showed that males fed significantly more on fish than female Gentoo penguins, breeding at King George Island, despite the lack of differences in energy expenditure or foraging trip duration demonstrated for this species [75]. This supports the theory that sex differences in dietary preference changes of Gentoo penguins may largely reflect local changes in the availability of particular prey species within the inshore area exploited by this all-year-round predator, where intraspecific competition is intense (particularly during the winter when food availability is low). To avoid competition, Gentoo penguins switched their foraging behaviour by foraging at deeper depths to limit competition with Adélie penguins Pygoscelis adeliae at West Antarctic Peninsula [23].
In comparison with other penguin species, comparable diet and foraging data is mostly available for the summer breeding season (Table 7). Adélie penguins [99][100][101], southern rockhopper penguins Eudyptes chrysocome [93,102,103], Magellanic penguins Spheniscus magellanicus [92], Emperor penguins Aptenodytes forsteri [104] are also known to exhibit sex differences in trip duration, foraging areas and diets. With these data, it is not clear why these patterns vary across species (Tables 6 and 7), but differences during the breeding period in some species (e.g. Emperor and Humboldt penguins Spheniscus humboldti) may be related to body condition, and the role of each sex (e.g. females must produce the egg, whereas Emperor penguin males are solely responsible for incubating the egg) [104,105]. In terms of habitat use (based on δ 13 C), there were variable results in short-term carbon signatures between males and females of Gentoo penguins (based on plasma and red blood cells), emphasising a changeable habitat use through the austral winter. Such a result is possibly a reflection of male and female Gentoo penguins trying to avoid (when possible) competition for the same prey (due to the lack of prey availability locally), and thus broadening their habitat use.
In terms of the trophic level (based on δ 15 N), males fed significantly at higher trophic levels than females (based on plasma and red blood cells; Figs 4 and 5), confirmed by male Gentoo penguins feeding on bentho-pelagic/benthic high trophic level prey (i.e. fish) whereas females feeding on pelagic low-trophic level prey (i.e. crustaceans; Table 2); fish has a higher trophic level than crustaceans [106]. Moreover, within males, some individuals clearly fed on a higher trophic level than other males whereas within females, some individuals fed on a lower trophic Foraging segregation in penguins level than others. This suggests that, within sexes, there are clear feeding/habitat segregations not identified before, which may be a reflection of the abnormal environmental conditions causing individuals to search for alternative prey. Individual feeding preferences and adjustments to spatial, seasonal and inter-annual variations in resources by exploiting different foraging niches were found in several other seabird species [39]. Therefore, fluctuations in individual specialization within a population may be related to temporal changes in the availability and predictability of resources [107], and could be, to some extent, a consequence of dietary sexual specializations. Indeed, when assessing sex-related specializations, females feeding on Antarctic krill (i.e. individuals that fed more than > 80% on Antarctic krill) versus males feeding on fish (i.e. individuals that fed more than > 80% on fish), females clearly feed on lower trophic levels (on pelagic crustaceans) in comparison with males (on high trophic levels (on benthic/bentho-pelagic fish); Fig 4). Therefore, in this abnormal season of low food availability, males and females do exploit different prey in a consistent way (i.e. in various foraging trips, in various days, as confirmed by the red blood cells data). Despite the significant differences in feather carbon signatures for the end of summer (March-April; at the end of their breeding period, when their feathers were created), biologically both sexes may explore similar habitats, as these carbon stable isotopic values are similar (from Southern Ocean waters; see Results). However, at this time of the year (i.e. pre-moulting period), penguins may disperse more widely and feed in more profitable areas, despite their foraging ranges during the breeding season being still relatively short (< 50 km, and often much less) [37]. In terms of the trophic level (based on δ 15 N), no differences were found in feathers, suggesting that both males and female Gentoo penguins also fed on similar prey (likely to be T. gaudichaudii, that dominated the diet of C. gunnari [108]. Antarctic krill only represented 12% by mass in Gentoo penguins diets during their breeding period [44]).