Introduction

The marine environment is highly dynamic with physical parameters determining the spatial and temporal distribution of primary productivity, thereby resulting in patchily distributed food resources. Marine predators, therefore, face the challenge of locating the prey which their survival and reproductive success depend on in this heterogeneous environment (Oosthuizen et al. 2015). From an optimal foraging perspective, there may be long-term breeding and survival benefits (Bradshaw et al. 2004) for animals which use prior knowledge about where food is (i.e., predictable) and return to the same foraging area, rather than randomly searching for food (Call et al. 2008). Indeed, many marine species, such as sea birds (Weimerskirch 2007), sharks (Espinoza et al. 2011), whales (Yates et al. 2007), turtles (Tucker et al. 2014), and seals (Oksanen et al. 2014; Arthur et al. 2015; Abrahms et al. 2018a), display individual foraging site fidelity. However, repeated use of the same foraging patch may lead to prey depletion and/or the prey distribution and the density may have changed over time, thereby resulting in site fidelity being a sub-optimal foraging strategy (Pichegru et al. 2010; McIntyre et al. 2017; McHuron et al. 2018). Thus, this illustrates a trade-off between a conservative strategy of sticking to what one already knows and another riskier strategy of switching and searching for new and potentially more profitable foraging patches.

Income-breeding marine predators provisioning offspring (Houston et al. 2007), such as fur seals (Staniland and Boyd 2003), sea lions (Womble et al. 2009), and seabirds (Croll et al. 2006; Rayner et al. 2010), can be considered as central-place foragers, because they alternate between foraging at sea and returning to the colony to feed their nutritionally dependent offspring. When offspring are young, they have a limited ability to fast and restrict the foraging trip duration of parents. As offspring get older and their nutritional demand increases, parents may have to increase their foraging effort by making longer trips to more distant oceanic foraging areas, which may be more energetically profitable than remaining in the coastal foraging areas (Boyd 1999; Kowalczyk et al. 2015). Individual foraging site fidelity may be an important foraging strategy, particularly during the later stages of offspring-provisioning, where energetic demands for self-maintenance and parental investment increase. This crucial period may be important in determining pup-weaning mass which is a significant factor influencing the first year (Georges and Guinet 2000) and the juvenile survival (McMahon et al. 2005; Burkanov et al. 2011), as well as carrying a pup to term as females may also be gestating (Georges and Guinet 2000).

Lactating long-nosed fur seals (LNFS; Arctocephalus forsteri) provide a unique opportunity to investigate individual foraging site fidelity in oceanic environments due to the plasticity of their foraging behaviour over a long 10-month pup-rearing period, and the ability to track seals over consecutive foraging trips. In a typical breeding cycle, adult female LNFS give birth to pups in December (austral summer) and mating takes place shortly after that. Mothers nurse their pups for 8–11 months while simultaneously gestating a foetus (Goldsworthy 2006). During the austral summer, the majority of the lactating females from Kangaroo Island (South Australia) forage on the Australian continental shelf and shelf-break associated with seasonal upwelling (Foo et al. 2019); whereas during the austral winter, the majority of the lactating females forage in oceanic waters associated with the Subtropical Front (Baylis et al. 2008). The marked change from the coastal to the oceanic foraging as lactation progresses is referred to as an alternating foraging strategy (Foo et al. 2019). Foraging site fidelity is generally high in the continental shelf habitats (Baylis et al. 2012; Sommerfeld et al. 2015) and the continental shelf in our study region is narrow, thereby concentrating productivity into a small predictable area. Conversely, the prey in large-scale oceanic habitats are likely to be more dispersed and ephemeral (Kotliar and Wiens 1990) and, thus, foraging site fidelity is generally lower (Baylis et al. 2012). Nonetheless, the greater spatial variability in oceanic regions potentially allows for more intra- and inter-individual variability in foraging strategies to develop and, thus, lead to different levels of individual reproductive success. Indeed, the variability in individual oceanic foraging trip distances and direction has been observed in lactating LNFS (Foo et al. 2019); however, the underlying environmental drivers to this variability are still unclear.

In oceanic regions, several marine predator species (e.g., Cotté et al. 2007; Simmons et al. 2007; Cleeland et al. 2014; Miller et al. 2015) preferentially target mesoscale features, such as eddies and frontal systems, which are relatively productive as they provide intermittent pulses of nutrients to the photic layer for the phytoplankton growth which would attract prey (Williams et al. 2011). Indeed, the oceanic waters associated with the Subtropical Front, located in the south of Australia, are known to have high incidences of eddies, fronts, and filaments (James et al. 2002b; Tomczak et al. 2004), and we expect lactating LNFS foraging to be associated with these oceanographic features. In this study, we investigate the potential environmental drivers to individual foraging site fidelity in lactating LNFS. Specifically, our aims are to (1) determine if sea-surface temperature (SST) and sea-surface height (SSH) gradients are associated with intra-annual individual foraging site fidelity, and (2) quantify any inter-annual differences in foraging site fidelity.

Methods

Study site, animal handling, and instrumentation

The study was undertaken between February and August/September in 2016 and 2017 at Cape Gantheaume, Kangaroo Island, South Australia (36° 04′ S, 137° 27′ E). Forty-five lactating LNFS (2016: n = 15, 2017: n = 30) were randomly selected and captured using a hoopnet. Details of the animal handling and tag attachment methods can be found in a concurrent study by Foo et al. (2019). Seals were weighed (± 0.5 kg), and their body length (nose to tail) and axillary girth were measured (± 1 cm). Geolocation (GLS, Intigeo-C330, 17 × 19 × 8 mm, 3.3 g, Migrate Technology Ltd, Cambridge, UK) loggers were deployed on all individuals and recovered later in the same year between June and August.

The GLS loggers measured ambient light every minute and recorded the maximum value every 4 min. The loggers also sampled sea temperature (0.125 °C resolution; ± 0.5 °C accuracy) after 20 min of being continuously wet and recorded the minimum, maximum, and mean temperature every 4 h. The 2016 loggers sampled ambient temperature every 5 min, and recorded the minimum, maximum, and mean temperature every 4 h. The loggers also sampled the time when an activity (wet or dry)-state change occurred. Each logger was activated and left in an open area at the study site for approximately 5–7 days, either immediately before or after deployment to obtain solar elevation estimates necessary for location calibration.

Location estimation

All data analyses were done using the R program (v3.5.3) (R Core Team 2019). Using the raw light data from GLS loggers, a number of times of twilight period were estimated using the BAStag package (v0.1–3) (Wotherspoon et al. 2016a). Next, we created Markov Chain–Monte Carlo simulations within a Bayesian framework to estimate the final posterior mean of the two primary locations per day while incorporating temperature and land-mask (Sumner et al. 2009; Lisovski et al. 2012) and haul-out constraints; this was done using the SGAT package (v0.1.3) (Wotherspoon et al. 2016b). For the latter, seals were assumed to be back to the colony when the GLS logger was continuously dry for ≥ 4 h as lactating LNFS are not known to haul out at other locations (Page et al. 2006; Baylis et al. 2012). Individual foraging trips and their duration were then extracted from location estimates and checked with wet/dry information from the GLS logger. The accuracy of location estimates, using this method, was 45 ± 29 km (mean ± SD) for a lactating LNFS carrying GLS and GPS tags simultaneously (Foo et al. 2019). As we were only interested in the seal’s lactation period between summer and winter (February–August), the few locations obtained after August from one individual were excluded from further analyses.

Core foraging areas

We assumed that core foraging areas were locations where females spent majority of their time on a foraging trip; we calculated the time spent (h) per 30 × 30 km grid cell, as a proxy for the foraging effort like others have done (Pistorius et al. 2017), for each foraging trip using the trip package (v1.5.0) (Sumner 2016). The size of the grid cell was chosen to account for the error associated with GLS tags while still providing realistic representation of true locations. We found this cell size adequate for creating a measure of foraging site fidelity in subsequent steps (detailed later in “Oceanic Foraging Site Fidelity”). Trips with less than three locations and locations within a 10-km buffer around the colony were removed, because foraging trips are typically, at least, 2 days long (i.e., total of four location points) (Baylis et al. 2008; Foo et al. 2019), and cells near the colony may represent non-foraging periods when females were travelling and/or displaying other kinds of behaviour (e.g., thermoregulation) in waters near the colony (Page et al. 2005). We identified core foraging areas as cells (1) within the 90th percentile of the range of time spent values per foraging trip and (2) within 2° in the latitude north from the southernmost core foraging cell identified for each trip. The latter was done to exclude outlier cells in the 90th percentile that were far away from where many of them tend to cluster. The 90th percentile is a conservative threshold that was chosen to ensure that the selected cells indeed represented foraging behaviour.

Oceanographic data

To describe the physical at-sea environment encountered by the seals in core oceanic foraging areas, we extracted the following environmental variables using the raster (v2.6-7) (Hijmans et al. 2017) and raadtools (v0.5.1) (Sumner 2016) packages: 0.02° bathymetry, daily 0.25° sea-surface temperature gradient (SSTgrad; derived from SST), and daily 0.12° sea-surface height gradient (SSHgrad; derived from SSH). The sources of environmental variables are shown in Table 1. Areas of high SSTgrad and SSHgrad are typically associated with frontal and eddy activities (Sokolov and Rintoul 2007; Bost et al. 2009; Scales et al. 2014). Fronts are horizontal gradients with enhanced water properties (e.g., temperature, salinity, density, etc.) on the water surface that outline transitions between water masses (Belkin et al. 2009). Eddies are circular currents of water that have a change in SSH between its core and boundary (Douglass and Richman 2015), and they can influence chlorophyll-a and SST gradients particularly at their edges (Gaube et al. 2017). All environmental variables were reprojected onto a 0.25 × 0.25° (~ 27–28 km) grid. The mean and standard deviation (SD) of the dynamic environmental variables were calculated from daily values over the period of each foraging trip. SD was used, because the stability of an environment is relevant to fidelity (Arthur et al. 2015). Missing values were removed from calculations. Subsequently, the mean of each dynamic environmental covariate (SSTgrad, SSTgrad_SD, SSHgrad, and SSHgrad_SD) was extracted from cells within a 30-km buffer of each core foraging cell.

Table 1 Characteristics and sources of the environmental variables extracted for data analyses
Full size table

Oceanic foraging site fidelity

Foraging trips were classified as “shelf”, which included the continental shelf and shelf-break (bathymetry ≤ 2000 m), or “oceanic” (bathymetry > 2000 m) depending on the mean bathymetry of its core foraging cells. The 2000-m isobath generally marked the end of the shelf-break in this region (Page et al. 2005). Only data from oceanic foraging trips were used in subsequent analyses. Maximum distance from the colony and the duration of each trip were also calculated. Trips were also classified into early autumn (February–March) or late autumn/winter (April–August), as this separates the upwelling and non-upwelling period on the shelf (Foo et al. 2019).

We noticed that consecutive oceanic trips were sometimes interrupted by a shelf trip, and individuals may not necessarily return to previous foraging areas on successive foraging trips but do so in later foraging trips; hence, oceanic foraging site fidelity was investigated at two timescales: (1) over the entire oceanic foraging period (i.e., from the first to the last oceanic foraging trip made over the deployment period) and (2) over the period of pairs of consecutive oceanic foraging trips for each individual (Fig. 1). For the former timescale, the SD (variability) in the latitude (Lat_SD) and the longitude (Lon_SD) of all core foraging cells, over the observational period for each individual, were calculated and used as a proxy for foraging site fidelity. For the latter timescale, the average location of core foraging cells for each trip was calculated and the great circle distance (km; hereafter distancefidelity) between the average core locations of pairs of consecutive oceanic trips was used as a proxy for foraging site fidelity. For both fidelity measurements, greater values meant lower foraging site fidelity.

Fig. 1
figure 1

Timeline of the types (oceanic or shelf) of foraging trips made by individual seals. Each observation only indicates the start of the foraging trip. The solid rectangle is an example of the entire oceanic foraging period. The dashed rectangle is an example of a pair of consecutive oceanic foraging trips

Full size image

Another measure of site fidelity is the tendency for individuals to repeatedly travel in similar directions on multiple trips (Pettex et al. 2012). We calculated the direction of each foraging trip as the bearing between the colony and the furthest location travelled. We used circular statistics from the circular package (v.0.4-93) for the following tests. The persistence of travelling in the same direction over multiple trips was tested for each individual using Rayleigh’s uniformity test. This test provides an r value between 1 and 0, where values closer to 1 indicate greater directionality. We tested the differences in the mean direction of foraging trips between years using the non-parametric Watson–Wheeler test.

Statistical analyses

For the entire oceanic foraging timescale, we calculated the average of each environmental covariate over all core foraging cells for each individual. We then fitted linear models for Lat_SD and Lon_SD against all the environmental predictors and the year. Model selection for linear models was done by sequentially removing the most non-significant term in the model until only significant terms remained. For the pairs of consecutive oceanic foraging trip timescale, we calculated the average of each environmental covariate over all core foraging cells for the earlier trip, since the outcome of the previous foraging trip is more likely to influence the foraging decisions made in the successive trip. We then fitted linear mixed-effect models for distancefidelity against all environmental predictors, season, year, and the number of core foraging cells (proxy for size of core foraging area of the trip), including trip nested within seal identity as a random effect. For linear mixed-effect models, backward model selection was done using Akaike information criterion that was corrected for small-sample sizes (AICc) (Burnham and Anderson 2002). Least significant fixed-effect terms were removed sequentially from the full model to obtain an optimal model with the lowest AICc. Models with delta AICc < 2 were considered not significantly different, and the more parsimonious model was used as the final model. The final model was evaluated for linearity, homogeneity, and normality using standard graphical methods and autocorrelation using the acf() function on model’s residuals. If a temporal correlation existed, an autoregressive function would be added for linear mixed-effect models. The response variable was transformed, and/or a variance component function (only for linear mixed-effect models) was included if it was necessary to fulfil model assumptions. Trip duration, maximum trip distance travelled from colony, and total time spent in core foraging cells per trip were modelled against year and season with trip nested within seal identity as a random effect. This was analysed using a penalised-quasi likelihood linear mixed-effect model with a Gamma distribution and identity link. Results were reported as mean ± SD unless otherwise stated. Statistical significance was determined at p < 0.05.

Results

Location statistics and track summaries

We recovered GLS loggers from 17 adult female seals. However, based on the estimated tracks derived from light data, four seals did not show central-place foraging behaviour (possibly due to pup death or abandonment) and only one seal made shelf foraging trips. These individuals were excluded from subsequent analyses. From the remaining 12 seals (n2016 = 5; n2017 = 7), a total of 3585 at-sea locations were obtained from 159 foraging trips from the early autumn to the winter in both study years (2016 = 99 trips, 2017 = 60 trips; Table 2). Of the total number of foraging trips, 94 were oceanic trips and 65 were shelf trips (Table 2). Oceanic trip duration (Table 3a) and maximum distance travelled (during oceanic trips) (Table 3b) were generally longer and farther in 2017 and in the late autumn/winter.

Table 2 Summary of individual foraging site fidelity statistics of 12 lactating long-nosed fur seals from Cape Gantheaume, Kangaroo Island
Full size table
Table 3 Final models in relation to (a) trip duration, (b) maximum distance, (c) total time spent in core foraging areas and individual foraging site fidelity during oceanic foraging at (d) the entire oceanic foraging period and (e) the consecutive oceanic foraging trip period
Full size table

Environmental drivers to individual oceanic foraging site fidelity

A total of 534 core foraging cells were detected from all the oceanic foraging trips with an average of six core foraging cells per trip (Table 2). The total time spent in the core foraging cells during an oceanic foraging trip was greater in 2017 than in 2016 (Tables 2 and 3c). The distribution of the oceanic core foraging cells tended to occur near mesoscale eddy edges and fronts (as indicated by SSHgrad; Fig. 2). An animation of individual foraging tracks overlayed on top of SSTgrad and SSHgrad in the study region is shown in Supplementary videos 1–4.

Fig. 2
figure 2

Examples of sequential foraging tracks (trip number above each plot) from an individual seal (#353). Core foraging cells (90th percentile of time spent in cell of each trip) are red and non-core foraging cells are opaque white. The red triangle represents the colony at Cape Gantheaume, Kangaroo Island. Tracks are overlaid onto a sea-surface height gradient (SSHgrad; units: change in m per 0.25 × 0.25° pixel) and b sea-surface temperature gradient (SSTgrad; units: change in °C per 0.25 × 0.25° pixel). Environmental values show the average values during the period of each foraging trip. High values of SSHgrad and SSTgrad represent areas of strong eddy and frontal activity, respectively. See SFig. 1. for additional examples

Full size image

During the entire oceanic foraging timescale (n = 12), Lat_SD increased with increasing SSTgrad_SD, but decreased with increasing SSTgrad (Fig. 3, Table 3d), Lat_SD was lower in 2017 than in 2016 (Fig. 3, Table 3d), and Lon_SD decreased with SSHgrad (Fig. 3, Table 3d). The overall average distancefidelity was 195 ± 162 km, n = 66 (range = 12.5–698 km) (Table 2). The global model for distancefidelity, with just seal identity as a random effect, shows to be the most parsimonious random structure (global model with random effect seal, AIC = 185; global model with random effect trip nested within seal, AIC = 187). At the consecutive oceanic foraging timescale, the best model (Table 4) indicated that distancefidelity increased with increasing SSTgrad_SD experienced by an individual during the earlier foraging trip (Fig. 4, Table 3e). The mean bearing across all oceanic foraging trips was different between years (2016: mean = 165 ± 11°, n = 42; 2017: 177 ± 20°, n = 52; Watson–Wheeler: T = 22.8, N = 94, p < 0.001; Table 2). There was a high directional persistence over all oceanic foraging trips for each individual (r value: mean = 0.986 ± 0.011, range = 0.96–0.998, N = 12, all individual p < 0.05; Table 2).

Fig. 3
figure 3

Effect plots of a SSTgrad SD, b SSTgrad, and c year on latitude SD; and d SSHgrad on longitude SD over the entire oceanic foraging period of individual seals. The solid black lines and dots represent the predicted effect. Shaded bands and error bars represent the 95% confidence intervals. Grey circles represent raw data points

Full size image
Table 4 Candidate linear mixed-effect models fitted during the model selection process using Akaike information criterion corrected for small sample sizes (AICc)
Full size table
Fig. 4
figure 4

Effect of SSTgrad_SD on distancefidelity (distance between mean core foraging areas of consecutive pairs of oceanic foraging trips). The solid black lines and dots represent the predicted effect. Shaded bands and error bars represent the 95% confidence intervals. Grey circles represent raw data points

Full size image

Discussion

Previous studies on LNFS have reported differences in individual foraging site fidelity between shelf and oceanic habitats (Baylis et al. 2012), but a consensus of understanding of environmental drivers to individual foraging site fidelity is still lacking. Here, we focused on foraging in the oceanic region which typically occurs at the later stage of lactation when female energetic demands are the highest. During this time, lactating females are likely to be operating within their limits, so there is pressure on them to forage as optimally as possible to provide their pups at rates that maximise the pups’ survival probabilities, given the huge amount of energy they have already invested in their pups over the preceding months of lactation. Using continuous foraging tracks of individuals from the summer/early autumn to the winter and over two breeding seasons, we found that individual foraging site fidelity is influenced by the strength and variability of SSTgrad and SSHgrad. This knowledge helps us to identify important foraging hotspots and understand how environmental changes of these hotspots may impact the population dynamics of this top marine predator.

Importance of mesoscale oceanographic features

Individual foraging site fidelity over the entire oceanic foraging period increased (lower Lat_SD and Lon_SD) with greater SSHgrad and SSTgrad, which are typically associated with eddy and frontal activities. The oceanic waters in this study region, which encompasses the Subtropical Front, is known to have a strong eddy field (Tomczak et al. 2004) particularly in winter (Middleton and Bye 2007). From video animations of female foraging tracks superimposed onto SSHgrad, foraging locations during oceanic foraging trips can be seen near or within eddy features (Supplementary video 1-2). Spatial heterogeneity (i.e., strong gradients) associated with fronts and eddies leads to a diversity of habitats, nutrient mixing, and, therefore, areas of high productivity that concentrate prey; thus making them good places to feed especially for animals that are highly motivated to acquire lots of resources (Bost et al. 2009). Hence, it follows that females keep returning to the vicinity of the same profitable foraging area on consecutive trips until profitability declines due to prey depletion, or temporal and spatial variability in prey distribution (Charnov 1976). Other marine predators including the Southern elephant seals (Mirounga leonina) (Campagna et al. 2006; Bailleul et al. 2010; Della Penna et al. 2015; Tosh et al. 2015), Northern elephant seals (Mirounga angustirostris) (Abrahms et al. 2018b), King penguins (Aptenodytes patagonicus) (Cotté et al. 2007), Loggerhead sea turtles (Caretta caretta) (Polovina et al. 2004), Bottlenose dolphins (Tursiops truncates) (Bailey and Thompson 2010), and Antarctic fur seals (Guinet et al. 2001) have all been observed to forage in the proximity of the eddies where the prey tend to aggregate. We note that while our sample size was relatively small (12 individuals), it would not be unrealistic to find an association between individual foraging site fidelity and SSHgrad and SSTgrad, as these oceanographic features are commonly targeted by various marine predators (Bost et al. 2009; Chambault et al. 2017; Reisinger et al. 2018). A preference for foraging in areas with strong oceanographic gradients has also been observed in Macaroni penguins (Eudyptes chrysolophus), Chinstrap penguins (Pygoscelis antarctica), Antarctic fur seals (Arctocephalus gazella) (Lowther et al. 2014), and Northern fur seals (Callorhinus ursinus) (Nordstrom et al. 2013).

In oceanic waters, good foraging patches may be spatially and temporally stable (i.e., predictable) (Staniland et al. 2004). Our results support this—at the entire oceanic foraging and consecutive oceanic foraging trip levels; individual foraging site fidelity decreased with increasing SSTgrad_SD. Similarly, Arthur et al (2015) reported that within-year foraging site fidelity in Antarctic fur seals decreased with greater long-term SST variability in foraging areas. That predictability of the oceanographic environment influenced individual foraging site fidelity at both fine and coarser intra-annual timescales highlights its importance on the foraging strategies of lactating females.

The importance of predictability of SSTgrad may be linked to relatively slow-moving eddies which are characteristic of the Great Australian Bight eddy field (this study region) (Oke et al. 2018). Mesoscale eddies can influence the spatial variability of SST through various mechanisms (Gaube et al. 2015). Observations from supplementary videos 1–2 show certain individuals (e.g., seal IDs #072 and #311) foraging at the same eddy on multiple foraging trips more clearly. Other marine mammals, such as fin (Balaenoptera physalus) and minke whales (Balaenoptera acutorostrata) in the Bay of Fundy, focus their foraging in areas of prey aggregation associated with slower velocity regions within an eddy system (Johnston et al. 2005). Predictability of slow-moving eddies may explain the high individual directional fidelity observed during oceanic foraging trips—a trait which is consistent with the previous work on LNFS at this Kangaroo Island study site (Baylis et al. 2012) and across a diversity of species from Antarctic fur seals (Bonadonna et al. 2001) to northern gannets (Morus bassanus) (Pettex et al. 2012). Nonetheless, this hypothesis requires further investigations to be certain.

The predominance of SSTgrad on individual foraging site fidelity in the latitudinal axis might be due to the cyclical north–south movement of the Subtropical Front, a feature which can affect the stability of the mixed layer in the surrounding water column and, thus, influence the spatial variability of eddies and fronts (Tomczak et al. 2004). Conversely, the predominance of SSHgrad in the longitudinal axis might be due to the westward drift of eddies due to strong westward currents from the south of Tasmania in the winter (James et al. 2002a; Oke et al. 2018).

Inter-annual variability in foraging site fidelity

Individual foraging site fidelity in the foraging latitude was greater in 2017 than in 2016, and this may have contributed to heavier winter pup mass in 2017 (12.6 ± 2.25 SD kg) than in 2016 (11 ± 2.11 kg SD) (Foo et al. 2019). Additionally, LNFS pup production at Cape Gantheaume in the 2017–2018 breeding season (following season after the fieldwork for this study ended) was the largest ever recorded at the time (20% increase from the previous season, whereas there was a 3% decrease in pup production between 2015–2016 and 2016–2017 seasons), and the average summer (2 months old) pup mass was greater in the 2017–2018 (6.69 ± 1.31 kg SD) than in 2016–2017 (6.55 ± 1.46 kg SD) breeding season (S. Goldsworthy, pers comm), which has been on a steady decline since the late 1990s (Goldsworthy et al. 2016). These observations suggest that 2017 was, especially, a good year for the pup growth rates (higher winter mass) and carrying the pups to term. The mid-to-late stage of gestation plays a crucial role in determining the reproductive success of LNFS as significant foetal mortality occurs during this period (McKenzie et al. 2005). By returning to the same profitable foraging site repeatedly, females can maximise the time spent on foraging and minimise the time spent on searching for prey. Coincidentally, 2017 females also had a longer time spent in core foraging areas per trip, and longer and farther oceanic foraging trips. Prior knowledge of the profitable areas is particularly beneficial for long foraging trips, which are energetically expensive, as the high variability in the individual foraging trip route and habitat may lead to mass loss for adults (Call et al. 2008) and, thus, lower the parental investment in gestation.

Conclusion

Lactating LNFS in South Australia display higher intra-annual individual foraging site fidelity to areas with strong and stable oceanographic gradients. At the individual level, females also tend to be persistent in the direction of their oceanic foraging trips from the colony. Future research on the persistence of individual foraging site fidelity over multiple years would give us insight into how this strategy might help them to cope with possible severe environmental changes.