Fine-scale dietary changes between the breeding and non-breeding diet of a resident seabird

Unlike migratory seabirds with wide foraging ranges, resident seabirds forage in a relatively small range year-round and are thus particularly vulnerable to local shifts in prey availability. In order to manage their populations effectively, it is necessary to identify their key prey across and within years. Here, stomach content and stable isotope analyses were used to reconstruct the diet and isotopic niche of the little penguin (Eudyptula minor). Across years, the diet of penguins was dominated by anchovy (Engraulis australis). Within years, during winter, penguins were consistently enriched in δ¹⁵N and δ¹³C levels relative to pre-moult penguins. This was probably due to their increased reliance on juvenile anchovies, which dominate prey biomass in winter months. Following winter and during breeding, the δ¹³C values of penguins declined. We suggest this subtle shift was in response to the increased consumption of prey that enter the bay from offshore regions to spawn. Our findings highlight that penguins have access to both juvenile fish communities and spawning migrants across the year, enabling these seabirds to remain in close proximity to their colony. However, annual fluctuations in penguin isotopic niche suggest that the recruitment success and abundance of fish communities fluctuate dramatically between years. As such, the continued monitoring of penguin diet will be central to their ongoing management.

2. Introduction

Unlike most seabird species that undertake annual migrations, resident seabirds remain in close proximity to their breeding areas throughout the year. They depend on local resources during both the breeding and non-breeding seasons [1]. The availability and abundance of resources during the breeding season can shape their breeding events (e.g. lay date) and determine breeding success [2–4]. By contrast, resources obtained during the non-breeding season are critical to the successful completion of moult [1,5,6] and surviving the environmental constraints imposed by winter [7], and can have carry-over effects that influence subsequent breeding performance [8–10]. Therefore, to ensure their survival and reproductive success, resident seabirds require access to relatively predictable and local prey resources year-round.

As adults and chicks are most accessible for study during the breeding season, most seabird dietary studies (including resident seabird species) are undertaken at this life-stage [11]. These studies have been crucial in identifying particular prey species or age classes of prey that influence breeding performance and require protection [12–14]. By contrast, few studies have identified important foraging locations or resources necessary for the survival of seabirds during their non-breeding season, with most of these having focused on seabird winter diets [15,16]. This is primarily due to the logistical difficulties of tracking the diet of migratory or widely dispersing seabirds. Surprisingly, despite the year-round presence of resident seabird species, few studies have assessed their diet in the non-breeding season. The few existing studies comparing life-stage dietary shifts have found varying results. For example, in the Isle of May, Scotland, fatty acid analysis demonstrated that the pre-breeding and breeding diet of common guillemots (Uria aalge) differ substantially [17]. But, due to insufficient dietary analysis in the non-breeding season, the non-breeding diet of guillemots remains unknown [17]. By contrast, the winter and breeding diet of yellow-legged gulls (Larus michahellis) in the Bay of Biscay differ slightly, in a consistent manner across years, and this is attributed to seasonal fluctuations in resource availability [18]. These studies highlight that the breeding and non-breeding diets of resident seabirds differ, and that in order to manage their populations effectively it is necessary to have an understanding of their trophic relationships at all stages of the annual cycle. This is especially important for resident species that have a small foraging range and that are particularly vulnerable to local shifts in prey availability.

Stable isotope analyses of seabird tissues, in combination with conventional assessments of diet, are powerful tools to investigate the year-round foraging ecology of seabirds [19–21]. Recent advancements in isotope ecology have provided the statistical frameworks to reconstruct the diet of individuals or groups at specific temporal scales [22,23]. Consequently, the stable isotope ratios of consumers and their prey can be used in stable isotope mixing models to estimate the proportion of each prey group in the diet of consumers. Additionally, stable isotopes can be used to calculate the ‘isotopic niche’ of seabirds and provide quantitative information on resource and habitat use, parameters that can be used as proxies to define the ecological niche of populations [24,25].

Little penguins (Eudyptula minor) are resident seabirds with one of the shortest foraging ranges among seabird species [26,27], and local fluctuations in prey availability strongly influence their foraging and reproductive ecology [28]. The St Kilda penguin colony are thought to forage exclusively within Port Phillip Bay [29,30] on a predominantly clupeoid-based diet [31]. Their short foraging range and narrow dietary breadth make them particularly vulnerable to changes in the distribution and abundance of their prey [30]. In this study, little penguins were used as a model species to assess how the diets of small home-range, resident seabirds fluctuate between life-stages and years. Specifically, we assessed if these residents display inter-annual dietary and isotopic niche variation and evaluated if they display shifts in diet and isotopic niche between the pre-moult, winter and breeding stages of the annual cycle.

3. Material and methods

3.1 Study area and species

Fieldwork was carried out at the St Kilda breakwater, within Port Phillip Bay, Victoria, Australia (37°51′ S, 144°57′ E) over four years (2007, 2008, 2011 and 2012). This temperate, semi-enclosed tidal embayment is joined to Bass Strait through a 3-km-wide, shallow, channel [32]. The embayment has an approximate 1930 km² area, with a mean depth of 13.6 m [33]. The St Kilda breakwater is located in the north of Port Phillip Bay and is occupied by approximately 1000 little penguins who reside on the breakwater year-round (Z. Hogg 2006, unpublished data). The annual cycle of little penguins comprises the non-breeding (moult and winter) and breeding seasons. During pre-moult (approx. February to March), adults accumulate sufficient reserves to sustain them during their annual moult. Little penguins fast ashore during moult and the moulting cycle lasts approximately 17 days [34]. After moult, adults return to sea and regain lost energy reserves. They return to the colony throughout the winter period (approx. March to September) and increase the time spent at the colony in preparation for the breeding season [35]. The commencement date of the breeding season is highly variable (May to September) both within and between colonies from year to year, but peak breeding occurs in the austral spring [36]. Typically, females lay one clutch of two eggs but have been recorded to lay up to three clutches in a season [37]. Males and females share the task of egg incubation, which spans approximately five weeks, and once chicks hatch they are brooded by at least one parent for two to three weeks, a period termed ‘guard’ [38]. After the ‘guard’ stage, both parents forage at sea, leaving the nest unguarded during the day and return to feed the chicks at night, a period termed ‘post-guard’. Chicks fledge at approximately eight weeks of age [28]. In this study, the incubation and guard stages of the breeding season are hereafter referred to as the ‘breeding life-stage’ as they comprise most of the breeding season.

3.2 Stomach content analysis

Ten penguins were caught when entering the St Kilda breakwater each month between February 2007 and October 2008. Individual penguins were identified via passive integrated transponders (Trovan, Ltd., Australia), and stomach contents from the sampled penguins were obtained using a modified water offloading technique [39] and frozen prior to analysis. Prey items were measured to the nearest 0.1 mm and identified from otoliths and squid beaks using methods outlined previously [31]. The length and weight of each fish and squid were calculated from published regression equations of otoliths and beaks, respectively [40–42]. Anchovy were categorized into broad age classes according to size, based on data from anchovies collected within Port Phillip Bay [43]. The less than 1-year cohort corresponded to standard lengths less than 63 mm, 1–2-year cohort 63–91 mm and 2–3-year cohort 92+ mm (none larger than 98 mm was recorded), recognizing that there is overlap in size between the age cohorts. No stomach analysis was conducted after October 2008 due to monthly consistencies in stomach content prey items.

Stomach content samples were quantified using a modified weighted relative occurrence method [44]. The percentage contribution of each item to the stomach sample based on mass (calculated by linear regression) was determined and averaged across all samples to provide a percentage contribution value for each prey item for each month [31].

3.3 Penguin tissue collection and preparation for stable isotopes

For animals of similar mass as little penguins (approx. 1 kg), the half-life of δ¹³C and δ¹⁵N stable isotopes in whole blood is 10–23 days [45]. Accordingly, in 2007 and 2008, individual penguins were identified via passive integrated transponders (Trovan, Ltd.) and a single blood sample was collected to represent the dietary intake of birds at either the pre-moult, winter or breeding (incubation and guard) stage of the annual cycle (table 1). Approximately 80 μl of blood was collected from the tarsal vein of adults using venipuncture and capillarity. Blood samples were stored in 70% ethanol at room temperature until analysis. Ethanol-based blood preservation does not appear to change stable carbon and nitrogen readings [46].

In 2011 and 2012, blood samples were collected to represent winter and breeding dietary intake (table 1). Approximately 150 μl of blood was collected from the tarsal vein using venipuncture and capillarity and was then transferred onto a microscope slide and dried at ambient air temperature [47]. To provide dietary information on the pre-moult diet of penguins in 2011 and 2012, a feather sample was collected from the lower back of post-moult adults. As little penguins replace their feathers during their three-week fast ashore and because feathers are metabolically inert after growth, feathers are thought to reflect food consumed during the pre-moult foraging bout [48,49].

In the laboratory, blood samples were freeze-dried; blood lipids were not extracted prior to analysis given that the lipid component of blood is less than 1% of the total wet mass of whole blood [50]. Blood samples were powdered, loaded into tin capsules (8×5 mm), weighed (0.4–0.6 mg) and sealed. Entire adult feathers were washed with distilled water before being freeze-dried and finely cut using stainless steel scissors. Surface lipids were not removed using a chloroform/methanol solution because this process has been shown to have negligible effects on isotope ratios [51]. Feathers were homogenized, and a subsample of the entire feather was loaded into a tin capsule (8×5 mm), weighed (0.4–0.6 mg) and sealed.

3.4 Prey collection and preparation for stable isotopes

In 2007 and 2008, muscle tissue from anchovy (Engraulis australis), southern garfish (Hyporhamphus melanochir) and bay squid (Loliolus noctiluca) was collected from St Kilda penguin stomach contents for stable isotope analysis (table 2). These species were selected for stable isotope analysis due to their dominant presence in penguin stomach contents. Additional anchovy (n=23) and southern garfish (n=10) samples were obtained from commercial fishing boats that operate within Port Phillip Bay in the winter of 2008 (table 2). Anchovy were categorized into three ontogenetic stages based on their morphometric measurements, as described above [43]. Ontogenetic categories included: (i) less than 1 year class (n=6), (ii) 1–2 year class (n=8) and (iii) 2–3 year class (n=9). Due to the small sample size of prey in 2007 and the similar isotope ratios within species across years, prey items collected in 2007 and 2008 were pooled to reconstruct the 2007 and 2008 diet of penguins, similar to procedures adopted previously [52]. Prey items collected in these years are hereafter referred to as 2007/2008 prey.

In 2011, anchovy and southern garfish (bay squid was not available) were obtained from commercial fishing boats that operate within Port Phillip Bay (table 2). In 2012, in addition to anchovy and southern garfish, sandy sprat (Hyperlophus vittatus), blue sprat (Spatelloides robustus) and pilchard (Sardinops sagax) were obtained from commercial fishing boats that operate within Port Phillip Bay (table 2). Clupeoids collected in 2011 and 2012 were collected for stable isotope analysis due to their increased abundance in Port Phillip Bay in 2011 [53], and were thus a potential prey source for little penguins as they have been found in the stomach contents of penguins at St Kilda and elsewhere [31,39,54]. Pilchards were categorized into juvenile and adult age classes in accordance with length-frequency data for pilchards obtained from commercial catches in Port Phillip Bay [55]. Size variations in other prey sources were not distinct, therefore these species were not separated into age classes. As all potential prey species were not collected in both 2011 and 2012, prey species were pooled across years so as to provide a wider range of dietary sources within mixing models. Prey obtained in 2011 and 2012 are hereafter referred to as 2011/2012 prey.

In 2007/2008, a section of prey caudal muscle was prepared for δ¹³C and δ¹⁵N stable isotope analysis. Lipids were not extracted prior to analysis. Samples were freeze-dried, ground, and 0.4–0.6 mg tissue samples were loaded into tin caps prior to stable isotope analysis. In 2011/2012, a section of the caudal muscle of prey samples was rinsed in deionized water and dried at 60°C in a glass vial until it reached a constant weight. Dried samples were ground and two samples were obtained from each vial; one was immediately prepared for stable isotope analysis (samples were freeze-dried, ground and loaded into tin caps) and the second underwent lipid extraction [56,57]. To remove lipids, samples were placed in glass centrifuge tubes and submerged in 2:1 chloroform:methanolsolution. Samples were stirred and centrifuged for 10 min at 1318g. The supernatant containing solvent and lipids was removed. This process was repeated until the supernatant solvent was clear and colourless after centrifugation. Samples were then dried at 60°C for 24 h. Treated samples were freeze-dried, ground and 0.4–0.6 mg tissue samples were loaded into silver caps prior to stable isotope analysis.

Lipid extraction can induce shifts in isotope ratios (particularly δ¹³C values) [57], and the effects of lipid extraction are greatest on tissues when their C:N ratios are more than 4.0 [58]. The C:N ratios in penguin prey in 2007/2008 were at times more than 4.0 and therefore were anticipated to have some influence on the SIAR reconstructed diet of penguins. To accommodate differences in lipid extraction protocols between years 2007/2008 and 2011/2012, 2007/2008 prey δ¹³C values were normalized. Values were normalized by detracting the mean difference between lipid extracted and non-extracted δ¹³C values from 2011/2012 anchovy and southern garfish samples. As no squid were collected in 2011 and 2012, the mean difference between lipid extracted and non-extracted δ¹³C values could not be determined. We therefore used published values for squid to normalize data [59]. Normalized values were used for all statistical analyses.

3.5 Stable isotope analysis

In 2007 and 2008, samples were processed at the Stable Isotopes in Nature Laboratory, Canada, and were combusted in an AS128 autosampler. The CO₂and N₂ gases were analysed using a Delta XP isotope-ratio mass spectrometer (Bremen, Germany) using a continuous flow system with every 20 unknowns separated by laboratory standards. In 2011 and 2012, samples were analysed at the Monash University Water Studies Centre, Australia, on an ANCA-GSL 2 elemental analyser. The resultant CO₂ and N₂ gases were analysed using a coupled Hydra 20:22 isotope-ratio mass spectrometer (Sercon Ltd., UK) with every five unknowns separated by laboratory standards. Sample precision was 0.1‰ for both δ¹³C and δ¹⁵N. Stable isotope abundances are expressed in δnotation in per mille units (‰) following the equation:

where R=(¹³C/¹²C or ¹⁵N/¹⁴N) of the sample and standards or where R is the ratio of the heavy (rare) isotope to the light (common) isotope in the sample and standard [60]. The international standards for carbon and nitrogen stable isotope ratios were Pee Dee Belemnite and atmospheric N₂, respectively.

Inter-laboratory variability in stable isotope analysis of animal tissue can lead to discrepancies in δ¹³C/δ¹⁵N results between laboratories and care should be taken to ensure obtaining comparable outcomes [61]. In this study, replicate samples were not sent to both laboratories to ensure result congruency due to logistic, financial and ethical constraints. However, inter-laboratory stable isotope variability does not greatly influence this dietary reconstruction study for two reasons. Firstly, inter-laboratory stable isotope variability does not influence isotopic niche width. Therefore, within and between year changes in isotopic niche width are reflective of consumed prey and are not artefacts of inter-laboratory variability. Secondly, because penguin tissue and corresponding prey samples were processed at the same laboratory, dietary reconstruction models are controlled. However, the isotopic position of penguins and their prey could be influenced by inter-laboratory variability and caution should thus be exercised when comparing penguin isotopic positions between years 2007/2008 and 2011/2012.

3.6 Statistical analysis

All statistical analyses were performed using R software v. 2.14 [62]. Differences in stomach prey composition (of the three main prey species) between year (2007 and 2008) and life-stage (pre-moult: Feb/Mar/Apr; winter: May/June/Jul; breeding 2007: Sep/Oct/Nov, 2008: Aug/Sep/Oct) were tested using a two factor ANOVA with type III sums of squares. Life-stage and inter-annual differences in anchovy sizes were also determined using type III ANOVA. Feather and blood isotopic variations were corrected using regression equations [63], and corrected feather values were used in all statistical analyses. Differences in δ¹³C and δ¹⁵N between years and life-stages were tested using a two factor ANOVA with type III sums of squares to accommodate for the unbalanced sample design. For both δ¹³C and δ¹⁵N, a simple main effects test, using life-stage as a factorial subset, was analysed using MS_Resid from the global model. Tukey's post hoc tests were used to identify differences between life-stages for δ¹³C and δ¹⁵N. Differences in δ¹³C and δ¹⁵N between prey species (and between age classes) and across years were assessed using a multivariate analysis of variance (MANOVA).

Stable Isotope Analysis in R (SIAR) (v. 4.1.3) [64], a Bayesian computing framework, was used to solve mixing models. A non-informative Dirichlet prior distribution, with zero concentration dependencies, and default SIAR MCMC estimation (iterations=2×10⁵, burning=5×10⁴, thinning=15) were included in the model. Stable isotope mixing models were run for each year. Prey obtained in 2007/2008 was applied to mixing models in 2007 and 2008, and prey obtained in 2011/2012 was incorporated to 2011 and 2012 mixing models, respectively. An isotopic mean discrimination factor of 3.9‰ for δ¹⁵N and 0.2‰ for δ¹³C was applied to models, based on fractionation values obtained from little penguins experimentally fed a diet consisting solely of sprats (Sprattus sprattus) [65]. Prey proportion densities (50, 75 and 95% credibility intervals) for the pre-moult, winter and breeding life-stage in 2007, 2008, 2011 and 2012 were assessed to reconstruct the diet of penguins.

The SIAR function SIBER (stable isotope Bayesian ellipses in R) [22] was used to calculate the isotopic niche widths of pre-moult, winter and breeding birds from years 2007, 2008, 2011 and 2012. Standard ellipses represent the isotopic niche width of 40% (SIBER default) of typical individuals within the groups based on bivariate normal distributions. We used the corrected version of the standard ellipse area (SEAc) to account for the loss of an extra degree of freedom when calculating bivariate data and to control for small sample sizes [22]. A Bayesian estimate of the standard ellipse area (SEA_B) was used to compare niche widths between groups. Differences in niche width between groups were compared in a probabilistic manner based on the size of simulated ellipse areas and their estimated posterior distributions [22]. Density plots display 50, 75 and 95% credibility intervals. Additionally, SIBER was used to calculate the isotopic niche widths of anchovy at three age cohorts (less than 1 year, 1–2 years and 2–3 years) to assess if this species displayed ontogenetic shifts in their niche width.

4. Results

4.1 Stomach content analysis

Based on the quantified mass of each prey item for each month, anchovy (E. australis) and southern garfish (H. melanochir) dominated the diet of penguins year-round (table 3). Cephalopods comprised only a small proportion of the diet overall with bay squid (L. noctiluca) being the most common species (table 3). The remaining species predominantly comprised Australian sprat (Sprattus novaehollandiae), blue sprat (S. robustus), sandy sprat (H. vittatus), pilchard (S. sagax) and hardyhead spp. The stomach content sampling showed no significant difference between year (F₁=3.3, p>0.05) or life-stage (F₂=2.3, p>0.05) for the three main species (anchovy, southern garfish and bay squid) consumed. Mean size of anchovy calculated from published otolith-standard length regression equations was approximately 72 mm, corresponding to anchovies of approximately 1–2-year age class. This age class dominated the diet of penguins year-round (figure 1). There was no significant difference in the size of anchovies taken either by year (F₁=3.3, p>0.05, figure 1) or life-stage (F₂=3.1, p>0.05, figure 1).

4.2 Penguin stable isotope ratios

A total of 440 blood and feather samples were obtained from penguins over the four years at three life-stages (pre-moult, winter and breeding) (table 1). The analysis of corrected isotope values revealed a significant interaction between year and life-stage for both δ¹³C signatures (F_6,429=39.7, p<0.001, table 4) and δ¹⁵N signatures (F_6,429=27.2, p<0.001, table 4). We observed inter-annual fluctuations in the isotopic niche position of penguins but found some consistent shifts in the isotopic position of penguins between life-stages. Within years, the winter isotopic position of penguins was consistently more enriched in δ¹⁵N and δ¹³C levels (excepting δ¹³C levels in 2007) compared with pre-moult birds (table 4). During the breeding season, the δ¹³C values of breeding penguins were generally more depleted than winter adults (except in 2007 when they did not differ significantly), while δ¹⁵N signatures shifted in an unpredictable manner between years (table 4).

4.3 Prey stable isotope ratios

In 2007/2008, the δ¹³C and δ¹⁵N isotopic values of prey differed significantly between species (δ¹³C: F_2,41=39.8, p<0.001, δ¹⁵N: F_2,41=11.3, p<0.001). The stable isotope values of prey ranged from −21.5 to −15.2‰ for normalized δ¹³C and from 20.2 to 13.8‰ for δ¹⁵N. Anchovy had the most depleted mean δ¹³C levels, while southern garfish was most enriched (table 2). Bay squid was the most enriched in δ¹⁵N levels between species (table 2). The isotopic values of anchovies differed significantly between ontogenetic stages for normalized δ¹³C values (F_2,20=4.9, p<0.05) but not for δ¹⁵N (F_2,20=1.62, p>0.05). Anchovies in the 2–3 year cohort had the most depleted mean δ¹³C values (−20.7±0.5‰) and those in the 1–2 year cohort had the most enriched mean δ¹³C levels (−20.3±0.1‰). Due to the lack of δ¹⁵N isotopic position distinction between ontogenetic stages, all anchovy results were pooled for stable isotope mixing models.

In 2011/2012, the stable isotope values of prey ranged from −21.7 to −14.6‰ for δ¹³C and from 20.2 to 10.9‰ for δ¹⁵N. Significant differences among species (and between juvenile and adult pilchard) for δ¹³C (F_5,51=205.6, p<0.001) and δ¹⁵N (F_5,51=136.8, p<0.001) were found. Southern garfish had the most enriched mean δ¹³C value, whereas blue sprat had the most depleted mean δ¹³C value (table 2). Anchovy displayed the highest mean δ¹⁵N signature and juvenile pilchard had the most depleted mean δ¹⁵N signature (table 2).

We found significant differences in the stable isotope composition of anchovy between years 2007/2008 and 2011/2012 for δ¹³C (F_1,34=262.7, p<0.001) and δ¹⁵N (F_1,34=50.9, p<0.001). In 2011/2012, anchovies were more enriched in both δ¹³C (2.4‰) and δ¹⁵N (2.8‰) compared with anchovies 2007/2008. In 2011/2012, southern garfish was more enriched in δ¹³C compared with garfish in 2007/2008 (F_2,21=3.7, p<0.005). No significant difference in δ¹⁵N levels was found. Our results show that annual variations in the isotopic niche of penguins are influenced by inter-annual changes in the stable isotopic composition of prey.

4.4 Stable isotope mixing models

Stable isotope mixing model outputs revealed significant differences in the relative proportion of ingested food sources between years. In 2007, the diet of penguins was dominated by anchovies in all life-stages (figure 2a–c). Southern garfish contributed significantly to penguin diet, and bay squid contributed least among resources. Compared to 2007, the relative contribution of anchovy to penguin diet declined in 2008, but anchovy continued to dominate the diet of penguins in all life-stages (figure 2d–f). An increase in the contribution of southern garfish to penguin diet was observed and bay squid continued to contribute little over the entire course of the year.

In 2011 anchovy's contribution to penguin diet decreased in all life-stages compared with years 2007 and 2008 (figure 2g–i). Blue sprat and pilchard dominated the pre-moult diet of penguins and sandy sprat, anchovy and southern garfish contributed marginally. During winter, anchovy's contribution to penguin diet increased and it had a mean proportional contribution of 32%, followed by southern garfish and sandy sprat. The contribution of pilchard and blue sprat to penguin diet was minor. During the 2011 breeding season, penguins displayed a diverse diet with similar contributions of anchovy, sandy sprat, blue sprat and pilchard. Garfish was detected at relatively low levels.