Omega-3 enriched chick diet reduces the foraging areas of breeders in two closely related shearwaters from contrasting marine environments Free

Our planet is currently experiencing rapid climate changes that are affecting biota in different ways (Walther et al., 2002), including the structure and functioning of marine ecosystems (Dorresteijn et al., 2012; Hoegh-Guldberg and Bruno, 2010). As top predators, seabirds are affected by changes in forage fish abundance (Carroll et al., 2017; Cury et al., 2011), thus providing information on lower trophic levels and acting as bioindicators of the marine health status (Furness and Camphuysen, 1997; Iverson et al., 2007; Piatt et al., 2007). Increased seawater temperatures caused by global climate change may alter phytoplankton growth and composition (Winder and Sommer, 2012). As these primary producers synthesise complex biomolecules, such as lipids and fatty acids (FAs), climate warming is expected to have strong directional effects on the amount and quality of FAs in phytoplankton (Hixson and Arts, 2016). Phytoplankton adapts to changing temperatures by modifying the structure of their membrane cells (Guschina and Harwood, 2009). Thus, to maintain cell membrane structural rigidity in response to increasing temperature, phytoplankton decreases polyunsaturated fatty acids (PUFAs) membrane content and, simultaneously, increases saturated fatty acids (SFAs) (Thompson et al., 1992). Therefore, increasing ocean temperature is predicted to reduce omega-3 synthesis by phytoplankton (Gladyshev et al., 2009; Hixson and Arts, 2016), namely eicosapentaenoic (EPA) and docosahexaenoic (DHA) acids. These highly unsaturated fatty acids (HUFAs; i.e. EPA and DHA) are transferred to higher trophic levels through bottom-up consumption (Gladyshev et al., 2009, 2013). Because marine phytoplankton is the main dietary source of essential omega-3, upon which marine organisms depend for optimum physiological function (Dalsgaard et al., 2003; Parrish, 2013), it is crucial to ascertain how marine wildlife will be affected and how they will physiologically respond to such potential reduction in omega-3 in food webs.

In marine ecosystems, the overall PUFA content tends to decrease as latitude decreases (Colombo et al., 2016). Therefore, temperate marine ecosystems, located at higher latitudes, are generally PUFA-enriched environments compared with tropical marine systems, located at lower latitudes (Colombo et al., 2016). Moreover, whereas the mixed layer in the tropical open ocean is relatively constant and unchanged with very limited diffusion of nutrients (i.e. slower nutrient turnover), neritic environments (e.g. upwelling areas) have seasonal changes in the strength of upwelling phenomena, with nutrient turnover especially stronger during spring (Mann and Lazier, 2005), determining higher food availability for marine predators.

Although more pronounced in tropical areas, procellariform seabirds adopt a dual-foraging strategy throughout their range, especially while rearing chicks (Congdon et al., 2005). This involves a repeated alternation of short foraging trips, mostly around the colony to collect food for the chick, with long trips mostly to forage for themselves and restore their own physical condition (Granadeiro et al., 1998; Magalhães et al., 2008; Paiva et al., 2015). Some studies have shown that experimental food supply may affect the foraging behaviour in seabirds (e.g. Bukacińska et al., 1998; Whelan et al., 2020), but those studies have only manipulated food quantity. This limits our comprehension of how a high-quality diet could affect the growing chick and foraging behaviour of parent pelagic seabirds. Indeed, it has become increasingly evident that food quality, in terms of highly unsaturated omega-3 FAs, can be even more relevant than food availability across ecosystems (Twining et al., 2016, 2018).

In this study, we used an experimental approach to determine the role of omega-3 FA supplementation to chicks in chick growth and physiology, and in parental foraging behaviour of Cory's shearwater (Calonectris borealis) and Cape Verde shearwater (Calonectris edwardsii). We studied populations breeding on islands with contrasting marine environments: Cory's shearwaters breeding on Berlenga Island, located on the upwelling system off the west Portuguese coast (Sousa et al., 2008), and Cape Verde shearwaters breeding on Raso Islet, located in a tropical oceanic environment in the Cabo Verde Archipelago (Paiva et al., 2015). Cory's and Cape Verde shearwaters are suitable study species to assess the additional omega-3 supplementation effects because they are both pelagic procellariforms with a diet naturally containing omega-3-rich prey items (e.g. Sardina pilchardus and Sardinella sp. largely represent the diet of Cory's and Cape Verde shearwaters, respectively) and provision their single chick during a long period (Warham, 1990). Because these sibling shearwater species are closely related, we may assume both species to have an identical reaction to internal and external stimuli.

Ricklefs (1992) found that lipid-supplemented chicks retain oil in their proventriculus, hindering further feeding. Furthermore, Granadeiro et al. (2000) demonstrated that shearwaters can adjust their provisioning rate in response to short-term variations in the nutritional status of their chicks, through different begging behaviour by chicks. Therefore, we expect the omega-3 supply to function as a substitute for the nutritional value of high trophic level prey, and to guarantee a good physiological condition of the chicks (i.e. presumably mediated by a decreased begging behaviour). Omega-3 supplementation probably leads to lower pressure on breeders regarding chick provisioning requirements and, consequently, lower foraging effort when compared with breeders from the control group. During chick rearing, Cory's shearwaters at Berlenga Island forage in highly productive waters close to the colony (Pereira et al., 2022), whereas Cape Verde shearwaters exploit the comparatively less productive environment around their breeding colony (Paiva et al., 2015). We expect the influence of omega-3 supplementation on birds of the temperate neritic area (Berlenga Island) to be less marked, given the comparably higher PUFA in prey of this environment. In contrast, birds in the oceanic and less productive tropical area (Raso Islet) should exhibit a more marked difference between supplemented and control groups. Thus, to assess our hypotheses, we incorporated the use of stable isotope (SI) analysis to determine the trophic position of chicks and breeders, FA quantification in plasma and adipose tissue sampled from chicks, as well as physiological measures of chicks, coupled with global positioning system (GPS) tracking of chick-rearing adult shearwaters. For physiological analyses, we used the heterophil:lymphocyte (H:L) ratio and the white blood cell (WBC) count, because they are considered important tools to monitor immune function and health status in birds (Bried et al., 2011; Lentfer et al., 2015; Norte et al., 2009). Additionally, because dietary PUFAs are thought to be involved in energy metabolism and in inflammation and oxidative processes, we also performed oxidative stress analyses. We predicted that: (1) chick growth and fledging biometrics, for both species, would not differ between treatments (omega-3 versus control), because the natural diet of shearwaters was not altered (similarly to Ricklefs et al., 1987); (2) provision of an extra omega-3 FAs to chicks would increase their omega-3 FA content and positively influence their physiological condition, particularly for Cape Verde shearwaters; and (3) parents of the omega-3 group would exhibit a lower foraging effort than those from the control group.

The experiment was conducted on Cory's shearwaters [Calonectris borealis (Cory 1881)] at Berlenga Island, Peniche, Portugal (39°24′N, 9°30′W), and Cape Verde shearwaters [Calonectris edwardsii (Oustalet 1883)] at Raso Islet, São Nicolau, Cabo Verde (16°37′N, 24°36′W), during the chick-rearing period of 2019 (August–October). Berlenga is a neritic island, located within the continental shelf off the west coast of Portugal. This system is characterised by shallow waters (<200 m), high productivity and a strong seasonal upwelling (Sousa et al., 2008), which provides optimal conditions for a wide number of seabirds, including Cory's shearwaters (Pereira et al., 2018). Raso is a small volcanic arid islet located in the tropical region of Cabo Verde, generally surrounded by oligotrophic waters and exploited by Cape Verde shearwaters and other seabird species during the chick-rearing period (Cerveira et al., 2020; Paiva et al., 2015; Semedo et al., 2020).

In each study area, we identified and randomly selected 30 chicks (with similar ages and the absence of a full-time parent in the nest) into two groups of 15 chicks to receive one of two supplemental treatments: an omega-3 supplementation treatment [omega-3 group (OM), through omega-3 pills] or a placebo treatment [control group (CT), through placebo pills]; see the schematic representation in Fig. 1. The age of Cory's shearwaters chicks was estimated using the chick growth curves (wing length, mm) from Granadeiro (1991). At the beginning of the treatment (day 0), the average chick age was 13.9±3.2 days (13.2±3.7 days for the OM group; 14.7±2.5 days for the CT group). For Cape Verde shearwaters, it was possible to register the exact day of hatching of chicks and, at the beginning of the treatment (day 0), the average chick age was 13.2±2.7 days (12.5±2.8 days for the OM group; 14.0±2.5 days for the CT group).

Before the start of the supplementation experiment, on day 0 (Cory's shearwater on 8 August; Cape Verde shearwater on 15 August), blood samples were collected from chicks to assess their health status (i.e. white blood cell count, heterophil:lymphocyte ratio and oxidative stress analyses). Owing to unsuitable freezing conditions at Raso Islet, the oxidative stress analyses were only performed for Cory's shearwaters.

Chicks were monitored throughout the chick-rearing period until fledging (Fig. 1). During the supplementation period (days 0–40), chicks were measured (wing length and body mass) and immediately supplemented with omega-3 (OM group) or placebo pills (CT group), to expose both groups to the same handling conditions. Chicks were measured and supplemented approximately at the same time, and before parents returned to feed the chick at night. Placebo pills were composed of neutral constituents (see below).

The supplementation protocol differed slightly between the two species owing to differences in body size [Cory's shearwaters=605–1060 g (Onley and Scofield, 2007); Cape Verde shearwaters=420–540 g (del Hoyo et al., 2020); Fig. 1]. Furthermore, the treatment period was divided in two phases to adjust for chick size. For Cory's shearwaters, during the first phase (days 0–26), we orally administrated one pill (omega-3 capsule or placebo pill) to chicks every 2 days. In the second phase (days 27–40), we adjusted the omega-3 dosage to two pills, to match the likely amount of omega-3 they are incorporating from parental provisioning of pelagic fish (Fig. 1A). For Cape Verde shearwaters, during the first phase (days 0–24), pill content was extracted from the capsule and orally administrated with a syringe, every 3 days. In the second phase (days 25–40), chicks received the whole pill (capsule included), orally administrated every 2 days, until the end of the treatment period (Fig. 1B).

When chicks were no longer supplemented, blood and adipose tissue samples were collected (collection date, day 42: Cory's shearwater on 17 September; Cape Verde shearwater on 27 September) to assess physiological (white blood cell count, heterophil:lymphocyte ratio and oxidative stress analyses) trophic (SI analysis) and metabolic (FA analyses) parameters.

In the days closer to the expected fledging period for both populations (Berlenga: from 22 October to 1 November, days 77–86; Raso: from 30 October to 13 November, days 75–89), chicks were measured daily, thus allowing us to record morphometric measurements and age of chicks on the day before fledging.

For each chick, we randomly captured one parent for GPS logger deployment to study the foraging patterns of breeders under contrasting chick supplementation conditions (OM and CT). When retrieving the devices, we collected blood samples to perform SI analysis.

For both groups (OM and CT), we did not interfere with the delivery of food provided by the parents to the chicks (i.e. quantity of food). Instead, we increased only the intake of omega-3 by providing the experimental chicks with omega-3 pills.

Both Cape Verde and Cory's shearwaters feed mainly on pelagic fish (Paiva et al., 2010a; Rodrigues, 2014). Thus, we defined the concentration of omega-3 supplementation used according to the omega-3 content of sardine Sardina pilchardus, a pelagic prey that is an excellent source of DHA and EPA content. Sardines contain almost 3 g of ∑omega-3 (sum of all omega-3 fatty acids, not just DHA and EPA) per 100 g of edible portion during the chick-rearing period (i.e. during the months of June–October; Zotos and Vouzanidou, 2012). During the linear chick growth phase, Cory's shearwater chicks receive from their parents up to 68 g of food at night (Ramos et al., 2003). If sardine represents the highest quality diet for Cory's shearwaters, chicks consume approximately 2 g of ∑omega-3 in a meal of approximately 60 g. Similarly, assuming a large meal size of 40 g for Cape Verde shearwaters, they consume approximately 1 g of ∑omega-3 per meal. Therefore, this supplementation experiment represented an extra omega-3 intake to the usual chick meal.

Omega-3 supplementation pills (MorEPA Platinum Smart Fats, MINAMI) consisted of commercial softgel capsules containing a high concentration of omega-3 FAs, mostly EPA and DHA (for further details, see ‘Constituents of the commercial softgel capsules MorEPA Platinum Smart Fats, MINAMI^®’ in the Supplementary Materials and Methods). The chicks in the control group received 1 placebo pill (with neutral constituents) so they were exposed to the same handling conditions. Placebo pills, which were manufactured and provided by the Pharmaceutical Faculty of the University of Coimbra, consisted of microcrystalline cellulose (96%, Avicel PH-101) and magnesium stearate (4%) in a gelatine casing.

During the mid-chick-rearing period (from the first day of treatment), chick wing length and body mass were measured every 2–3 days, using a metal ruler to the nearest 0.1 mm and a Pesola spring balance to the nearest 1 g, respectively. We also measured these biometrics and registered their age at fledging (days). All biometric data were used to calculate chick linear growth rate and asymptotic mass. Chick linear growth rate (g day⁻¹) was defined as the slope of a regression of chick mass on age during the linear growth period (Cory's shearwaters: 12–40 days of age; Cape Verde shearwaters: 12–55 days of age). Asymptotic mass was calculated as the mean chick body mass during the plateau period (Cory's shearwaters: 45–60 days of age; Cape Verde shearwaters: 65–85 days of age).

Blood samples were collected from chicks before the supplementation period, on day 0 (collection date: Cory's shearwater on 8 August; Cape Verde shearwater on 15 September); immediately after chick sampling, we gave the first supplementation. Considering the small size of chicks, approximately 150 μl of blood was collected from the brachial vein using a 25 G needle with the aid of heparinised capillary tubes, to make a thin film smear. The remaining blood was kept cool (0–5°C) until centrifugation and then the plasma was stored at −20°C for later oxidative stress analyses on Cory's shearwaters. We did not perform oxidative stress analyses (before and after supplementation treatment) on Cape Verde shearwaters because there were no suitable freezing conditions at Raso Islet.

At the end of the treatment period, 1 ml of blood was sampled from each chick to assess their health status. Blood remains were collected using heparinised capillary tubes and used to make a thin film smear. Blood was divided into aliquots for SI, FA and oxidative stress analyses, and was kept cold (0–5°C) until centrifugation. Then the plasma was stored with ethanol 70% for later FA and SI analyses, and kept at −20°C for later oxidative stress analyses.

Blood smears were air-dried and fixed by submerging in 100% methanol for 3 min and then air-dried. In the laboratory, blood smears were stained using the Giemsa procedure and scanned under 1000× magnification, allowing the determination of WBC count and H:L ratio. WBC count was determined by counting the number of WBCs in approximately 10,000 red blood cells. H:L ratio was calculated by dividing the number of heterophils by the number of lymphocytes in a total of 100 WBCs (Owen and Moore, 2006).

To assess the oxidative stress of chicks we measured plasma non-enzymatic antioxidants [using the OXY-Adsorbent test (OXY), Diacron, Grosseto, Italy] and plasma reactive oxygen metabolites [through d-ROMs test (ROM), Diacron], according to the manufacturer's protocol with a few modifications (see ‘Laboratory procedures for oxidative stress analysis’ in Supplementary Materials and Methods). Owing to low sample sizes, especially at day 0 because of the collection of reduced blood volumes from chicks at this time, some oxidative parameters could not be measured for all chicks. Thus, we describe the methodology for oxidative stress analyses in the Supplementary Materials and Methods (for further details, see ‘Laboratory procedures for oxidative stress analysis’).

Chick adipose tissue samples were collected after the treatment period, following a non-lethal biopsy technique (see Rocha et al., 2016 for technique details). Samples were stored in 1.5 ml microtubes with ethanol (70%) before gas chromatography-mass spectrometry (GC-MS) analysis.

We randomly captured one parent in each study nest (N=30 in each study area) to track parental at-sea foraging behaviour. The GPS devices (CatLog Gen 2 GPS Loggers, Catnip Technologies, USA) were attached to the four central tail feathers of each individual bird, using waterproof TESA 1641 tape and following the methodology described by Wilson et al. (1997). Tags and attachment tape weighed approximately 19 and 15 g for Cory's and Cape Verde shearwaters, respectively, which in both species represented a maximum of 2.7% of the body mass of the lightest bird tagged, thus reducing potential deleterious effects on the ecology and behaviour of birds (Igual et al., 2005; Phillips et al., 2003). GPS devices were programmed to record fixes every 10 min and were attached (during the chicks' treatment period, ∼days 10–20) and retrieved approximately 2 weeks later at night when adults were returning to feed their chicks (Cory's shearwaters: N=28; Cape Verde shearwaters: N=26). We also collected 0.5–1 ml blood samples from adult metatarsal veins. These samples were subjected to centrifugation and the plasma was stored in ethanol 70% until SI analysis (δ¹⁵N and δ¹³C values). The total handling time of seabirds did not exceed 10 min and individuals were released back into their burrow to minimise stress.

We processed the data recorded by GPS loggers, removing positions within a 1 km radius of the colonies to reduce the influence of rafting and circling behaviours close to the colony (Weimerskirch et al., 2020). Departure and arrival points were used to identify individual trips.

Tracking datasets were analysed separately for short (≤1 day and ≤150 km) and long (>1 day and >150 km) foraging trips for both populations. This classification was designed based on frequency histograms of occurrence of (1) trip duration (days) and (2) maximum distance from the colony for each foraging trip (km). Tracking data allowed us to calculate the following metrics: (1) behaviour within foraging trips (i.e. time spent flying, time spent foraging, time spent resting, and trip length) and (2) spatial ecology and habitat use (i.e. Bhattacharyya's affinity index overlap within the same group, 50% kernel utilization distribution and 95% kernel utilization distribution). We used the expectation-maximization binary clustering (EMbC) algorithm implemented in the R package EMbC to classify the foraging behaviours from the GPS tracks (Garriga et al., 2016). This algorithm classifies four different movement categories based on the travel speed and turning angle between subsequent GPS positions: (1) travelling (high velocity, low turning angles), (2) extensive foraging (high velocity, high turning angles), (3) intensive foraging (low velocity, high turning angles) and (4) resting (low velocity, low turning angles) (Garriga et al., 2016; Louzao et al., 2014).

Kernel density estimates (KDEs) were generated from the foraging positions only (i.e. intensive and extensive search) using the adehabitatHR R package (Calenge, 2006). Birds perform an extensive search, at high-speed movements, when foraging in a large area to locate prey patches, and perform an intensive search, with high turning movements and low speed, when adopting an area-restricted search (ARS) behaviour after prey location (Louzao et al., 2014; Weimerskirch, 2007). The combined use of intensive and extensive search positions as representative of foraging has been reported for procellariiformes such as Cory's shearwater (Ramos et al., 2020) and the Cape Verde little shearwater (Puffinus lherminieri boydi) (dos Santos et al., 2022). Based on our data and on previous studies, although some seabirds, such as gulls and pelicans (Spelt et al., 2019; Zavalaga et al., 2011) use more frequently sit-and-wait foraging strategies, shearwaters forage more actively using soaring flying strategies (Paiva et al., 2010b; Rosén and Hedenström, 2001; Weimerskirch et al., 2020). Therefore, we did not consider positions associated with floating behaviour as indicative of foraging. The most appropriate smoothing factor (h) used in the kernel density analysis was calculated as the average value of ARS behaviour exhibited across short and long foraging trips within each breeding site and treatment (approximately 10 km) (Lascelles et al., 2016). We used Bhattacharyya's affinity (BA) index, a statistical measure of affinity between two groups that assumes they use space independently of another (Fieberg and Kochanny, 2005), to calculate the utilisation distribution overlap between OM and CT groups. We only used the ‘intensive and extensive foraging’ behaviour positions to compute the BA index, which ranges from 0 (no overlap) to 1 (total overlap) (Fieberg and Kochanny, 2005).

We analysed δ¹³C and δ¹⁵N values in chicks (Cory's shearwaters: N=29; Cape Verde shearwaters: N=30) and tracked parents (Cory's shearwaters: N=28; Cape Verde shearwaters: N=26) to study the trophic ecology of chicks and breeders under the omega-3 experimental conditions (for further details, see ‘Laboratory procedures for stable isotope analysis’ in Supplementary Materials and Methods). Plasma retains dietary information over a few days before sampling (Hobson et al., 1993). Thus, it was possible to evaluate the foraging habitat use and the trophic level of consumed prey during the trips in which adult birds were tracked (Jaeger et al., 2010). Chick plasma was also used for SI analysis to confirm the trophic niche of their parents.

Results are expressed in the usual δ notation as parts per thousand (‰) deviation from the international standards atmospheric nitrogen (N₂) for δ¹⁵N and Vienna-PeeDee Belemnite (V-PDB) for δ¹³C.

Additionally, we extracted FAs from plasma and adipose tissue samples of chicks. We followed the modified one-step method of Abdulkadir and Tsuchiya (2008) to perform FA analysis. However, as in Gonçalves et al. (2012), we used a 2.5% H₂SO₄-methanol solution instead of boron trifluoride-methanol (BF₃-methanol) because BF3-methanol can cause artefacts or loss of PUFAs (Eder, 1995) (see ‘Laboratory procedures for fatty acid analysis’ in Supplementary Materials and Methods for more details). Concentrations of individual FAs were quantified as mg g⁻¹ for adipose and mg μl⁻¹ for plasma samples.

All applicable institutional and/or national guidelines for the care and use of animals were followed. All animals were handled in strict accordance with good animal practice as defined by the current European legislation. All animal work was approved by the Portuguese Government (ICNF) under license: 101/2019/CAPT. The ‘National Directorate of the Environment’ of Cabo Verde (DNA) authorised the annual work carried out at Raso Islet, Desertas Islands Natural Reserve. All sampling procedures and/or experimental manipulations have been reviewed and specifically approved as part of obtaining the field license.

Linear models (LMs) were used to assess the effect of experimental treatments (OM versus control) on (1) chick biometrics (i.e. linear growth rate, asymptotic mass and wing length at fledging, body mass at fledging and age at fledging), (2) percentage of main FA in chick plasma and adipose tissue, and (3) trophic ecology (i.e. plasma δ¹⁵N and δ¹³C values) of Cory's and Cape Verde shearwater chicks. We also used LMs to assess the effect of OM versus CT on (5) trip metrics (i.e. percentage of short trips) and (6) trophic ecology (i.e. plasma δ¹⁵N and δ¹³C values) of Cory's and Cape Verde shearwater breeders.

Considering that chicks were sampled in two different sampling periods (day 0 – before the supplementation experiment; and day 40 – after the supplementation experiment), we used linear mixed models (LMMs) to assess the effect of the interaction between treatment (OM versus CT) and sampling period (day 0 versus day 40) on physiological parameters (i.e. H:L, WBC, OXY and ROM) of shearwater chicks. Therefore, chick identity was fitted as a random effect to account for the repeated sampling of chicks (day 0 and day 40). Because we fitted different models using the same dataset for chick biometrics and isotopic parameters, we adjusted P-value thresholds with the Benjamini–Hochberg correction for each set of parameters, to reduce the probability of Type I errors, and results were considered significant at P≤0.01.

Additionally, LMMs were used to evaluate the effect of treatment (OM versus CT) on (1) behaviour within foraging trips (i.e. time spent flying, time spent foraging, time spent resting and trip length) and (2) spatial ecology and habitat use [i.e. BA index within the same group, 50% kernel utilisation distribution (KUD) area and 95% KUD area] of Cory's and Cape Verde shearwater breeders. Individual ID was fitted as a random effect to account for multiple trips per individual. LMMs were computed using the lme4 (Bates et al., 2015) and lmerTest (Kuznetsova et al., 2017) R packages. To compute kernal density estimates and the BA index, we used the adehabitatHR R package. For foraging parameters of parents, results were considered significant at P≤0.05.

SIBER (Stable Isotope Bayesian Ellipses in R; package SIBER; Jackson et al., 2011) was used separately for chicks and breeders to compare isotopic niches between treatments for each population. We calculated the area of the standard ellipse corrected for small sample size that contains approximately 40% of the data (SEA_C) for each treatment (i.e. niche width), which was used to calculate niche overlap between treatments. Bayesian estimate of the standard ellipse and its area (SEA_B) was adopted to test for differences in niche widths between treatments (i.e. the proportion of ellipses in OM groups that were lower than that of CT groups), using the rjags R package (https://CRAN.R-project.org/package=rjags).

We used the performance R package (Lüdecke et al., 2021) to evaluate model assumptions before each statistical test, and response variables were transformed (log, arcsin or sqrt transformations) when necessary to approach a normal distribution. All statistical analyses were performed within the R environment, R v.4.0.4 (https://www.r-project.org/). Data are shown as means±s.d. To calculate the effect size, we used the metric η² to evaluate the proportion of variance associated with ‘treatment’ (OM versus CT) effect in the ANOVA models. The value of η² varies from 0 to 1, and the following rules of thumb are used to interpret values: η²<0.01 represents a very small effect size, 0.01≤η²<0.06 is a small effect size, 0.06≤η²<0.14 is a medium effect size and η²≥0.14 is a large effect size (Field, 2013). Thus, values close to 1 indicate that a specific variable in the model can explain a greater fraction of the variation. To calculate the η² in R we use the etaSquared() function from the effectsize package (Ben-Shachar et al., 2020).

We found that omega-3 supplementation did not influence chick growth measures, i.e. linear growth rate and asymptotic mass for either species (F<0.53, P>0.531, η²<0.02; Table 1; Table S1). Also, there were no significant differences in wing length, body mass and chick age at fledging between CT and OM groups for either study species (F<4.75, P>0.191; Table S1). Nevertheless, the ‘treatment’ variable showed medium (chick age at fledging for Cory's shearwaters and wing length at fledging for Cape Verde shearwaters) and large (chick age at fledging) effect sizes.

There was no significant interaction between treatment group (OM versus CT) and sampling period (i.e. day 0 versus day 40) for WBC and H:L ratio (Cape Verde and Cory's shearwaters) and OXY (Cory's shearwaters) (P>0.075; see Table S2 and Fig. S1).

However, all physiological metrics of Cory's shearwaters differed between sampling periods (day 0 versus day 40) and the effect size was large (η²>0.42) (see Table S2 and Fig. S1A,C,E,F). But WBC count and H:L ratio on Cape Verde shearwaters were not significantly different between sampling periods (Table S2, Fig. S1B,D).

The concentration of the most SFAs, MUFAs, HUFAs and PUFAs in plasma and adipose tissue did not differ significantly between CT and OM-supplemented chicks. However, we found differences in EPA (C20:5n−3) FA composition between Cory's and Cape Verde shearwater chicks from the two treatments (Fig. 2A,B). For Cory's shearwater chicks, the OM group had a significantly higher amount of EPA in the adipose tissue (F_1,28=4.67, P=0.039, s.e.=0.255, η² [confidence intervals, CI]=0.14 mg g⁻¹ [0.00, 1.00]; Fig. 2A) than the CT group. Furthermore, the OM group of Cape Verde shearwater chicks had a significantly higher quantity of EPA in plasma than the CT group (F_1,28=5.54, P=0.026, s.e.=0.005, η²=0.17 mg µl⁻¹ [0.01, 1.00]; Fig. 2B). For Cory's shearwaters, we also found differences in the behenic acid (C22:0) (OM>CT, F_1,28=5.24, P=0.030, s.e.=0.001, η²=0.16 mg µl⁻¹ [0.01, 1.00]) and eicosadienoic acid (C20:2n-6) (OM>CT, F_1,28=8.12, P=0.008, s.e.=0.001, η²=0.22 mg µl⁻¹ [0.04, 1.00]) plasma FA. Additionally, for Cape Verde shearwaters, there were significant differences in behenic acid (C22:0) plasma FA (OM>CT, F_1,28=4.74, P=0.038, s.e.=0.000, η²=0.14 mg µl⁻¹ [0.00, 1.00]). The mean values of all FA analyses for adipose tissue and plasma of chicks are shown in Table S3 for both study species.

The δ¹⁵N and δ¹³C values for chicks and breeders of both populations were similar between treatments (F<3.21, P>0.170; Tables S1 and S4). In addition, the SIBER analysis did not show different isotopic niche spaces between treatments (Table S4) for either populations or ages (chicks and breeders). However, for Cory's shearwaters, the variable ‘treatment’ had a medium effect size (i.e. 0.06≤ES<0.14) for both δ¹⁵N and δ¹³C (Tables S1 and S4).

Cory's and Cape Verde shearwaters had a foraging pattern mainly characterised by short trips around the colony (Table S4). For both species there was relatively high spatial overlap (measured by the BA index) with no significant differences between OM and CT groups, during short and long trips (F<3.86, P>0.063; Table S8 and Table 2). However, there were significant differences for the area sizes of at-sea distribution (i.e. 50% and 95% KUD). Yet, during short trips of Cape Verde shearwaters, parents of OM-supplemented chicks had a smaller 50% KUD compared with parents of CT chicks (F_1,90=6.01, P=0.016; Figs 2C and 3, Table 2). We found the same pattern for Cory's shearwaters during long trips, where parents of OM-supplemented chicks had smaller 50% and 95% KUDs than those of CT chicks (F_1,62=4.70, P=0.042 and F_1,62=5.84, P=0.025, respectively; Figs 2D and 3, Table 2). Furthermore, for Cory's shearwaters, the effect size of ‘treatment’ was large for all spatial ecology metrics during long trips.

Tracked shearwaters of both groups did not differ in the percentage of time spent foraging for both study sites and trip types (short and long) (F<3.64, P>0.060, η²<0.04; see Table 2). However, for Cory's shearwaters, we found that parents of the OM group spent significantly more time resting (during both short and long trips) and less time flying (during long trips) compared with parents of the CT group (‘treatment’ large effect size values match the observed P-values, see Table 2). There were also significant differences between treatments in the trip length for long foraging trips of Cory's shearwaters (F_1,62=10.05, P=0.005, η²=0.32).

We investigated how an omega-3-lipid-enriched diet influenced chick development and parental at-sea foraging behaviour of Cory's and Cape Verde shearwaters breeding in contrasting marine environments. We found differences in the FA composition of chicks between treatments, which suggested that chicks preferentially retained EPA. As expected, omega-3 supplementation had no significant effect on chick development, although it also had no effect on their physiological condition, contrary to our expectations. However, parents of omega-3-supplemented chicks foraged in smaller areas for both studied colonies, during both short and long foraging trips, when compared with parents of control chicks. During long trips of Cory's shearwaters, parents of omega-3-supplemented chicks spent significantly less time flying, more time resting and travelled smaller distances when compared with parents of control chicks, likely because of comparatively higher availability of prey resources at those distant sites.

As expected, omega-3 supplementation had no visible influence on chick development (i.e. linear growth rate, asymptotic mass and fledging biometrics). Similar results were recorded for Leach's storm-petrel (Hydrobates leucorhous) (Ricklefs et al., 1987) and Cape gannet (Morus capensis) (Navarro, 1991), where chicks supplemented with a high lipid and protein diet did not show faster growth or greater body mass when compared with control chicks. As the rate at which parents supply food under natural conditions may be close to the maximum assimilatory capacity of the chick's digestive tract (Hamer and Hill, 1994), our results suggest that omega-3-supplemented chicks regulated their consumption of energy, where food intake was adjusted to accommodate the ‘supplemented energy’.

Several studies have shown that PUFAs and HUFAs are of benefit to the health status of animals, including humans (Arts and Kohler, 2009; Arts et al., 2001; Brenna et al., 2009; Parrish, 2013; Simopoulos, 2011; Swanson et al., 2012). We found no physiological differences (i.e. WBC, H:L, OXY and ROM) between treatments for Cory's shearwaters (breeding in temperate marine ecosystems, generally PUFA-enriched) nor for Cape Verde shearwater chicks (breeding in the tropical marine environment, less PUFA-enriched; i.e. WBC; H:L). This was particularly expected for Cory's shearwaters because omega-3 supplementation effects were presumably less marked by an environment with a high abundance of food resources naturally rich in omega-3. Nevertheless, and contrary to what we expected, we also did not observe significant physiological differences in the leukocyte profile between omega-3-supplemented and control chicks for Cape Verde shearwaters. These results suggest that omega-3 FAs are not a limiting factor within this tropical environment, particularly within the productive area off West Africa.

Our results of oxidative stress for Cory's shearwater chicks (see Tables S1 and S2) were similar to those shown by Costantini et al. (2007) in nestling Eurasian kestrels (Falco tinnunculus). The authors supplemented nestling Eurasian kestrels with carotenoids (i.e. pigments with antioxidant properties) and demonstrated that nestlings did not benefit from the increased intake of carotenoids regarding the reduction in oxidative damage (i.e. ROM) or increase in serum antioxidant capacity (i.e. OXY). Furthermore, OXY and ROM of Eurasian kestrels decreased with age (Costantini et al., 2006, 2007), which is in accordance with our results for Cory's shearwaters (Table S2). A decrease in OXY and ROM with age in chicks is probably related to the maturation of the antioxidant defences and the effects of growth and its associated levels of metabolic activity (Costantini et al., 2006, 2007). Also, an increase in WBC count and H:L ratio in Cory's shearwater chicks with age is in accordance with findings in many other bird species (Howlett et al., 1998, 2002; Jakubas et al., 2015; Quillfeldt et al., 2008). These changes are thought to be related to the maturation of the immune system and increased exposure to pathogens (Apanius, 1998; Klasing and Leshchinsky, 1999), but also to the higher activity and wing exercise of chicks (Jakubas et al., 2015).

Experiments based on birds' omega-3 FA dietary enrichment have demonstrated that there is an overall linear relationship between the concentration of EPA in the diet and the EPA content of tissues (Rymer and Givens, 2005). However, the relationship between dietary and tissue concentrations of DHA is much weaker than that observed with EPA (Rymer and Givens, 2005). Indeed, we found differences in the EPA concentrations between treatments, for both plasma (Cape Verde shearwater chicks) and adipose (Cory's shearwater chicks) tissues, but no differences were found for DHA.

To cope with unpredictable feeding opportunities, seabird chicks can store extremely large lipid reserves (Boersma and Parrish, 1998; Gjerdrum et al., 2003; Ricklefs and Schew, 1994; Williams and Buck, 2010). Adipose tissue functions as the body's main energy reservoir, storing energy in the form of triglycerides. Thus, there is a higher correlation between adipose tissue FA composition and diet than between plasma FA composition and diet (Käkelä et al., 2009). Araújo et al. (2019) observed that fasting black-tailed godwits (Limosa limosa limosa) had high levels of omega-3 PUFAs in the fat tissue and suggested that these levels did not decrease because they are integral lipids, which help to maintain the structure and function of cell membranes. In our study, we hypothesise that the longer trip lengths of breeding Cape Verde shearwaters, when compared with those of Cory's shearwaters, resulted in longer fasting periods for chicks from Raso Islet. In this sense, we suggest that these fasting periods imposed on Cape Verde shearwaters result in similar EPA content of the adipose tissue between omega-3 group and control group. In contrast, plasma triglyceride levels tend to increase after feeding activities (Araújo et al., 2019), so the higher plasma EPA composition found in the omega-3 group suggests that omega-3 supplementation acted as food events for Cape Verde shearwater chicks. Conversely, at Berlenga Island, Cory's shearwater chicks may not be subjected to long-term fasting periods, suggesting that chicks were fed more than the amount required for proper growth and fat accumulation needed for fledging (Ricklefs and Schew, 1994). Thus, the external omega-3 supplementation combined with food naturally given by parents may be responsible for the greater EPA content assimilation in adipose tissues of the omega-3 group. Yet, there were no EPA differences in plasma, probably owing to frequent feeding events in both groups of Cory's shearwaters.

Varying environmental conditions affect the trophic ecology and foraging strategies of seabirds (Paiva et al., 2013; Pereira et al., 2020; Ramos et al., 2018). Previous studies indicate that Cory's shearwaters can respond to experimentally induced changes in the nutritional status of their offspring (e.g. Granadeiro et al., 2000). Parents of omega-3-supplemented chicks performed shorter trips, spent more time resting and less time flying when compared with parents of control chicks. This suggests that they were under lower pressure to invest in chick-provisioning. Also, Bukacińska et al. (1998) provided additional food to lesser black-blacked gull (Larus fuscus) chicks and showed that adult females whose chicks were experimentally fed made shorter trips than females of the control group chicks. In our study, Cory's shearwaters from the two groups extensively overlapped in their foraging areas, but the areas of their at-sea distribution (i.e. 50% and 95% KUDs) were smaller for parents of omega-3-supplemented chicks than for parents of control group chicks. Therefore, for Cory's shearwaters, it appears that parents of omega-3-supplemented chicks reduced their foraging activity, which resulted in a lower foraging effort (Soanes et al., 2021). Contrary to our expectations, for Cape Verde shearwaters, parents of omega-3-supplemented chicks did not change their foraging effort (i.e. trip length, time spent resting and time spent flying) in relation to the control group. Cape Verde shearwaters strongly foraged in distant and highly productive areas over the Canary Current system off the West African coast (Paiva et al., 2015; Ramos et al., 2013). Therefore, it is likely that the effect of supplementation cannot be detected for the farthest and longer trips. When foraging in the colony surroundings, there is the potential to detect a reduction in foraging effort by the metrics of short trips. Thus, Cape Verde shearwater parents of omega-3 supplemented chicks showed smaller 95% KUD foraging areas during short trips when compared with those of control group chicks.

Despite the differences that were detected in foraging patterns, both groups of Cory's and Cape Verde shearwaters showed similar δ¹⁵N and δ¹³C values, suggesting short-term consistency in foraging trophic level and habitats (Ceia et al., 2012). Foraging patterns of breeders suggest that global warming effects (i.e. a decreased amount and quality of FAs in phytoplankton) may lead to an increased foraging effort by shearwaters in temperate marine environments, as observed for the Cory's shearwater control group. Cape Verde shearwaters breeding at Raso Islet usually rely on food sources off the coast of West Africa to satisfy their nutritional requirements (Cerveira et al., 2020; Paiva et al., 2015; Ramos et al., 2013). As a result, a reduced PUFA marine environment (caused by climate changes) in this area could result in higher costs of foraging and reproduction, because even long foraging trips would not be able to provide the necessary nutritional requirements for chicks.

Overall, this study highlights the importance of diet quality for wild seabird populations and suggests that the nutritional status behind chick development might influence the foraging efforts of their parents. Our results suggest that the communication of higher nutritional status by begging chicks does not seem to drive a change in the prey type targeted, but instead might lead to comparably smaller food parcels brought to the chicks. In contrast to our prediction, seabirds breeding on the neritic island of Berlenga showed a more measurable response to the omega-3 supplementation experiment than those breeding on the oceanic Raso Islet. Supplementation of omega-3 to chicks resulted in lower chick-rearing stress for breeder Cory's shearwaters: shorter foraging trips, longer resting periods, reduced flying times and reduced foraging activity in the colony surroundings.

Marine resources in tropical areas are usually unpredictable (Weimerskirch, 2007), but the relative proximity of the oceanic Raso Islet to the neritic region of high prey availability off West Africa (Paiva et al., 2015) may lead to the evolution of a particular life-history strategy. Through localisation and memorisation of these enhanced productivity patches (Nevitt, 2008; Pereira et al., 2022; Soanes et al., 2021), birds may compensate for some extent the less predictable environmental conditions and reduced-quality prey around the breeding colony. Nevertheless, our study was carried out during the chick-rearing period, when Cape Verde shearwaters mainly exploited the tropical environment within the archipelago of Cabo Verde (Paiva et al., 2010a). We found that omega-3 supplementation also reduced the 95% KUD foraging areas during short trips for Cape Verde shearwaters.

Our results suggest that in a future of higher oceanic temperatures caused by climate change, shearwaters may adjust their foraging effort until their physiological limit. A PUFA reduction in the predictable and usual foraging patches for shearwaters will likely lead to longer and unpredictable foraging trips that may result in impaired growth and physiological condition of chicks, and therefore in higher reproductive costs.

Future studies on seabird ecology and the lipid quality of the diet of their chicks are required to assess annual variations in breeding success and foraging patterns. Furthermore, future investigations should monitor the chick's begging behaviour and track both parents (to control for potential sex differences) to better understand the effects of omega-3 supplementation on chick–parental interactions in the at-sea foraging distribution of breeders, as well as parental investment.

We are grateful to Instituto da Conservação da Natureza e Florestas (ICNF) for the logistical support and permission for data collection, especially the wardens of Reserva Natural das Berlengas (Alexandre Bouça, Ana Santos, Eduardo Mourato, Márcio Duarte, Nuno Dias and Paulo Crisóstomo), and their total support and companionship during fieldwork. We thank Biosfera I and its staff for the logistics, namely transport to the colony and all the provided conditions, supplies and companionship during fieldwork. We thank Ana Rita Carreiro and Filipe Rafael Ceia for their help with data collection and companionship during fieldwork. We also are very grateful to Prof. Dr João José Martins Simões Sousa from the Faculty of Pharmacy of the University of Coimbra for the production and supply of the placebo pills used in this study.