Tidal drift removes the need for area-restricted search in foraging Atlantic puffins

Abstract

Understanding how animals forage is a central objective in ecology. Theory suggests that where food is uniformly distributed, Brownian movement ensures the maximum prey encounter rate, but when prey is patchy, the optimal strategy resembles a Lévy walk where area-restricted search (ARS) is interspersed with commuting between prey patches. Such movement appears ubiquitous in high trophic-level marine predators. Here, we report foraging and diving behaviour in a seabird with a high cost of flight, the Atlantic puffin (Fratercula arctica), and report a clear lack of Brownian or Levy flight and associated ARS. Instead, puffins foraged using tides to transport them through their feeding grounds. Energetic models suggest the cost of foraging trips using the drift strategy is 28–46% less than flying between patches. We suggest such alternative movement strategies are habitat-specific, but likely to be far more widespread than currently thought.

1. Introduction

Optimal foraging theory assumes that animals move between patches of prey in order to maximize prey encounter rate and nutritional intake [1,2]. Brownian motion or random walks appear to be optimal when prey is uniformly distributed [3] and are characterized by movement where turning angles between relocations are effectively random [4]. By contrast, when resources are patchy, or animals have a preferred direction or orientation, movement is non-random, leading to correlated random walks (CRWs) [5] or Lévy flights [6]. CRWs incorporate an element of directional persistence [7], while Lévy walks allow random direction, but step lengths between successive locations reflect a power-law distribution [8]. First reported in albatross [6], Lévy flight has been identified across multiple taxa; however, its universality for explaining foraging ecology has stimulated debate [6,9–13] and is now considered one element of a number of strategies used in foraging for resources [14–16]. This foraging component of animal movement is universally characterized by area-restricted search (ARS): the tendency for animals to concentrate search within relatively small areas before continuing wider range exploration [17]. ARS is widely accepted as a core component within foraging theory across a wide array of taxa ranging from grazing herbivores [18] to higher trophic-level predators [19]. ARS is scale-dependent [20,21] and because it occurs within CRWs and Lévy flights [22], has provided a link between movement ecology, optimal foraging theory and energetics.

Seabird research has led much of the development of animal movement theory for Lévy flight and ARS, where individuals travel large distances between prey patches with subsequent fine-scale search behaviour above water [23,24]. The generality of ARS in seabirds is well established and it therefore could be assumed that ARS is energetically optimal. Here we report a lack of ARS in a seabird with a high cost of flight, the Atlantic puffin, Fratercula arctica, (hereafter puffin). Auks, including puffins, have previously been shown to use ARS between prey patches [25–27]. Unusual track patterns led us to hypothesize that puffins were exploiting tidal movement to transport them over prey patches. We predicted that this strategy must be energetically lower than engaging in flight/ARS, suggesting a new energetically optimal foraging strategy in seabirds.

2. Material and methods

Research was approved by University College Cork Animal Ethics Committee and under licences from the British Trust for Ornithology, National Parks and Wildlife Service and the Health Products Regulatory Authority of Ireland. Sixteen chick-provisioning puffins were tracked during the breeding season (2017 n = 12, 2018 n = 4) on Little Saltee, Co. Wexford (52.137 N, −6.590 W), Ireland. Birds were caught using purse nets at burrow entrances, and only individuals returning with fish were tagged to ensure they were provisioning chicks. In 2017, birds were equipped with either an Ecotone Uria solar-charging GPS logger (n = 10), or a PathTrack NanoFix tag with time-depth recorder (CEFAS G5 TDR) (n = 2). In 2018, only Ecotone Uria tags were deployed (n = 4). Ecotone tags downloaded data remotely, while Pathtrack tags required bird recapture.

Both tag types recorded a GPS position every 15 min. TDRs were used to log dives deeper than 1 m, and longer than 5 s duration [28]. Ecotone tags deployed in 2018 were programmed to record times of descent and ascent using manufacturer recommended settings. Devices were attached ventrally to the lower back using strips of Tesa tape to minimize hydrodynamic and aerodynamic drag following Harris et al. [27] and Elliott et al. [29]. Mean deployment weight was 2.73% of body weight. Further information on tag attachment is provided in the electronic supplementary material.

Tracks were atypical to those previously observed in this species, with no obvious ARS, and characterized by three distinct components: (1) straight line commuting from the colony out to sea, (2) slow directed phases in easterly or westerly directions and (3) return to the colony in a straight line (figures 1 and 2; electronic supplementary material, figures S1 and V1). Tracks of multiple puffins were observed to be synchronized as they moved over the core use area, often starting and ending drifting periods at similar times (electronic supplementary material, figure V1). GPS tracks were processed in the AdehabitatLT package [30] using R [31]. Tracks were standardized to a 15-min interval using linear interpolation to account for missed positions and divided into individual foraging trips, using a distance threshold of 350 m from the colony to define the start and end of a trip. We used bootstrapping to determine the goodness of fit to a power-law distribution for relocation step length using the poweRlaw package in R [32] to determine whether trips were consistent with Lévy flight.

Figure 1.

An example GPS track from puffin EX01307 tracked from Little Saltee, County Wexford, Ireland in June 2017. The non-typical movement patterns can be seen clearly with a commuting stage away from the colony before long straight section in broad easterly–westerly directions. Different colours represent different foraging trips. Inset: all puffin trips recorded between 2017 and 2018. The area of interest showing colony location.

Figure 2.

Three puffin trips (top) with corresponding TDR dive data (middle) and trip-level energetic expenditure (TEE) (bottom). (a) Trip EX01312_1. (b) Trip EX01312_2. (c) Trip EA15001_1. Green and red crosses indicate points along the foraging track to illustrate the direction of travel (outward and return respectively) and highlight series of clustered dives occurring during drift periods consistent with foraging on patchily distributed prey. Large gaps in dives (grey boxes) observed in (a,b) indicate overnight periods where foraging did not occur. (Online version in colour.)

Puffin speed was matched to tidal movement by correlating travel speeds with tidal velocity from the Irish Marine Institute's Regional Ocean Modelling System (ROMS) model (http://www.marine.ie/Home/site-area/data-services/marine-forecasts/ocean-forecasts) using linear models. Only puffin speeds below the ROMS maximum tide velocity (2.3 m s⁻¹) were correlated to account for potential periods of flight occurring within 15-min fixes. Speed thresholds produced from movement datasets were used to determine behaviour [33], using the midpoint between mean puffin flight speed (13.2 m s⁻¹) [34–36] and peak current velocity (2.3 m s⁻¹) to differentiate flight and surface behaviours as both could occur within 15-min relocations.

Dive data were retrieved from two TDR deployments (due to GPS tag failure, only one TDR deployment had accompanying location data) in 2017 and from three Ecotone tags in 2018. TDR data were reduced to a time series of dives consisting of dive depth (metres below the surface) and dive duration (seconds). Many dives were not recorded accurately by Ecotone tags, resulting in unfeasibly long dives (mean 29.7 ± 188.7 min). Visual inspection of Ecotone tag dive events identified only a single full trip as suitable for energetic analysis. TDR data were used to build a linear model of dive depth based upon dive duration (electronic supplementary material, figure S4), which was extrapolated onto the Ecotone dive data to produce estimates of dive depth from dive duration.

(a). Trip energetic expenditure

To explore the energetics of tidal drift versus flight, we modified a daily energetic expenditure equation for auks (DEE, equation (2.1); Elliott et al. [37], previously used on puffins [38]) to calculate a trip energetic expenditure (TEE) for individual foraging trips. TEE estimates are calculated in watts but can be considered as relative energetic costs.

$$\mathrm{TEE}\hspace{0.17em}=148(T_{\mathrm{f}})+28(T_{\mathrm{ws}})+27(T_{\mathrm{d}})+28(T_{\mathrm{i}})+8.9(T_{\mathrm{n}}),$$ 2.1

where T_f denotes time spent flying, T_ws indicates time spent sitting on the water's surface (drifting on surface currents), T_d is time spent diving and T_i is the dive interval (in seabirds oxygen stores are replenished during inter-dive intervals [39,40]). While Elliott et al. [37] did not provide a cost for a post-dive recovery period, we assumed it was analogous to the cost attributed to sitting on the water's surface. T_i was calculated by modelling the minimum dive interval with general linear models (GLM) and multiplying by the number of dives in a trip. GLMs predicted dive interval using a subset of the dive data (dive interval less than 200 s), dive depth, dive duration and the interaction between the two. All models were ranked by AIC and the most parsimonious model used to predict dive interval. Puffins are diurnal predators, not feeding at night [28,41,42], and since no dives were recorded at night, T_n is the overnight period, with a coefficient equal to rest [37].

The energetic cost of the observed ‘drift strategy’ (TEE_drift) was compared with a modelled trip following the same route with the same time spent diving, but with travel costed energetically as flight (TEE_flight). This produced an estimate where instead of drifting, puffin behaviour was considered as flight interspersed with dives throughout the trip. We applied a flight speed for puffins of 13.2 m s⁻¹ [34–36], resulting in a faster trip (faster to fly the route than drift) and applied the lowest energetic cost of rest (8.9 [37]) as a conservative estimate to the remaining time to ensure strategies are considered over the same time frame. The resulting TEE_flight is likely to be a large underestimate of the actual cost as ARS involves extensive periods of nonlinear flight.

3. Results

A total of 107 foraging trips were recorded (including 51 overnight trips) from 12 individuals, all of which were atypical of previously published seabird foraging tracks. Average distance from the colony was 15 km with a maximum of 38 km. Mean trip duration was 10 h 6 min (±14 h 45 min). Individuals targeted the same core foraging area (electronic supplementary material, figures S1 and S2) by departing the colony in a more easterly or westerly direction to account for the direction of tidal flow at the beginning of the trip.

Movement did not conform to a power-law distribution consistent with Lévy flight (goodness of fit = 0.0521, p < 0.001) and did not exhibit ARS. Tidal velocity correlated strongly (coefficient 0.723) with puffin speed during the slow, directed phases of trips (F₄₅₀₀ = 429.6, p < 0.001, electronic supplementary material, figure S6).

Dive data showed temporally clustered bouts suggesting an element of underwater search on patchily distributed prey, and no dives occurred during darkness. Mean dive depth from TDRs (n = 2224 dives) was 9.67 m (s.d. = 3.78, range = 2.03–23.43) while mean dive length was 34.4 s (s.d. = 19.072, range = 5–82). Linear models to predict dive interval for TEE analysis found the most parsimonious model consisted solely of dive duration predicting dive interval (coefficient = 0.89, F₇₃₉ = 505.27, p < 0.001; electronic supplementary material, figure S7).

TEE estimates were calculated for three indicative trips (figure 2) to determine whether the drift strategy is energetically optimal for short to long duration trips. TEE for the drift strategy was always lower than for the flight strategy, ranging from 28 to 45% cheaper (table 1) with the greatest energy saving occurring on overnight trips.

Table 1.

Trip energetic expenditure (TEE) costs during observed drift and simulated flight strategies based on the model by Elliott et al. [37]. Here we see the cost differences between the two strategies for the three trips analysed. Energetic savings from drifting ranged between 28.2 and 45.7% though mean saving was 39%.

trip ID	drift TEE (kW)	flight only TEE (kW)	difference	per cent saving by drift strategy
EX01312_1	978	1425	447	46
EX01312_2	1150	1667	517	45
EA15001_1	1167	1496	329	28
MEAN	1098	1529	431	39

4. Discussion

ARS is considered to be a widely used optimal search strategy for encountering patchily distributed prey [3,17], observed across a wide variety of marine and terrestrial taxa [43–46], including puffins [28]. While foraging, puffin movement did not conform to a power-law distribution and did not contain ARS. To our knowledge, this is the first time that non-ARS foraging has been reported in a mobile marine predator. Instead, puffins flew to foraging areas and used tides to transport them across prey patches. Drifting behaviour in seabirds is typically associated with periods of rest [47–49], but our dive data show that puffins were diving during drifts. The core foraging area of the tracked puffins occurs over sand substrate [50], the habitat of the puffins' main prey species, sandeels [51]. Dives were pelagic, likely due to targeting sandeels which school pelagically [52], and chick-provisioning puffins were regularly observed bringing sandeels back to the colony. Our data therefore show that the tracked puffins are employing an alternative foraging strategy by using tidal currents to forage over sandeel habitat. It is interesting to note that puffins may predict the state and direction of the upcoming tide, leaving the colony in a direction to ensure that tides transported them over a relatively small core habitat.

Puffins, and auks in general, are energetically expensive flyers with high wing loading [53–55]. Using currents is likely to drastically reduce locomotion costs associated with travelling between prey patches, making it an energetically optimal strategy. The energetic cost of drift trips was on average 39.2% lower than flight, although this is likely a gross underestimate because our flight costings relied on the assumption of straight line flight, while ARS (by definition) consists of more tortuous travel covering a greater total distance. We also did not consider the high cost of take-off following dives between prey patches because these costs are unknown.

Here we have described a foraging behaviour where puffins use a tidal landscape; however, seabird associations with tides are not unknown. Magellanic penguins change swimming and dive patterns when faced with strong tidal currents [56], while Adélie penguins change foraging locations based upon tidal cycles [57]. Furthermore, black guillemots forage in areas of high tidal activity [58,59] where tides are thought to aggregate prey [60]. However, these examples still contain ARS, whereas these puffins have used tides to remove the need for this ARS entirely. The dive profiles of the puffins do suggest search/pursuit occurring underwater. All diving seabirds undertake search/pursuit during dives, in addition to above surface ARS (that is absent this study), reflecting finer-scale processes associated with prey capture [28,61,62]. We suggest that further studies combining diving behaviour and movement patterns are likely to confirm that non-ARS foraging is more widespread, particularly in areas of high tidal activity. Interestingly, a study on razorbills in Wales, UK, showed extensive periods of drift and little evidence of ARS, and while foraging behaviour was not the focus of that study [48], birds were diving during these trips (E. Owen 2019, personal communication).

The drifting strategy, while energetically cheaper, comes at the expense of time. Observed trips were longer than those reported for puffins at other colonies [35] and likely results in less frequent food deliveries for chicks. How this impacts upon chick development is unknown but it may be that adults coordinate feeding times to ensure healthy chick development [63].

We highlight a previously unreported foraging strategy which departs from the ‘optimal’ ARS for foraging for patchily distributed prey [1,17]. Whether this strategy is stable over time, how it spreads through a population and how common it is in other marine foragers or other systems will be an interesting area of further research.

Ethics

All research was approved by University College Cork Animal Ethics Committee and conducted under licences obtained from the British Trust for Ornithology, National Parks and Wildlife Service (82/2017 C87/2017, 06/2018).

Data accessibility

Data are available from the Dryad Digital Repository: https://doi.org/10.5061/dryad.8gj8kc3 [64].

Authors' contributions

M.J. and A.D. conceived the research topic. A.B. and M.J. undertook fieldwork. A.B., M.J. and J.Q. further developed a theoretical framework and project direction. A.B. led analysis with assistance from M.J. and J.Q. A.B. led writing the manuscript with contributions from M.J. and J.Q. and A.D. All authors agree to be held accountable for the content therein and take responsibility for any questions arising to accuracy or integrity of any part of this work. All authors reviewed and approved the final version of the manuscript.

Competing interests

We declare we have no competing interests.

Funding

This work was funded by the Zoological Society of London through donations from Good Gifts (Charities Advisory Trust). AB was funded by Irish Research Council, Grant GOIPG/2016/503.