Time to eat: measurements of feeding behaviour in a large marine predator, the northern elephant seal Mirounga angustirostris


  • Carey E. Kuhn,

    Corresponding author
    1. National Marine Mammal Laboratory, Alaska Fisheries Science Center/NOAA, 7600 Sand Point Way NE, Seattle, WA 98115, USA
    2. Department of Ecology and Evolutionary Biology, University of California, Santa Cruz, CA 95064, USA
      *Correspondence author. E-mail: carey.kuhn@noaa.gov
    Search for more papers by this author
  • Daniel E. Crocker,

    1. Department of Biology, Sonoma State University, Rohnert Park, CA 94928, USA
    Search for more papers by this author
  • Yann Tremblay,

    1. Department of Ecology and Evolutionary Biology, University of California, Santa Cruz, CA 95064, USA
    Search for more papers by this author
  • Daniel P. Costa

    1. Department of Ecology and Evolutionary Biology, University of California, Santa Cruz, CA 95064, USA
    Search for more papers by this author

*Correspondence author. E-mail: carey.kuhn@noaa.gov


  • 1The at-sea behaviour of marine predators is often described based on changes in behavioural states, such as transit, searching, and feeding. However, to distinguish between these behaviours, it is necessary to know the actual functions of the behaviours recorded. Specifically, to understand the foraging behaviour of marine predators, it is necessary to measure prey consumption. Therefore, the at-sea feeding behaviour of northern elephant seals (N = 13) was examined using satellite transmitters, time-depth recorders, and stomach temperature recorders. In addition, stomach temperature telemetry allowed for the validation of indirect measures of feeding behaviour used for marine predators, including decreases in transit rate and changes in dive shape.
  • 2Feeding data were recorded for the early phase of the migration (2·2–21 days). The first feeding events occurred shortly after animals departed (4·0 ± 1·5 h) and close to the rookery (58·6 ± 21·9 km), but these feedings were followed by extended periods without prey consumption (14·5 ± 2·5 h). Continuous (bout) feeding did not occur until on average 7·5 ± 1·8 days after the females left the rookery. Females showed significant differences in the feeding rate while feeding in a bout (1·3–2·1 feeding events hour−1).
  • 3There was a significant negative relationship between interpolated transit rate and feeding events (r2 = 0·62, P < 0·01). Feeding, which was associated with all dive types, occurred most often during the foraging type dive shape (74·2%). Finally, successful feeding only occurred between 18–24% of the time when females displayed the foraging type dive shape suggesting that the use of dive shape alone, while indicative of behaviours associated with foraging (searching and catching prey) overestimates actual feeding behaviour.
  • 4This study showed females not only feed extensively during the early migration, but there was individual variation in both foraging locations and foraging success. In addition, by combining direct and indirect measures of feeding, this study has provided support for the use of foraging indicators in marine predators.


The movement patterns of animals often result from the integration of multiple life-history requirements including foraging, reproduction, predator avoidance, and dispersal. Each of these behaviours is shaped by both intrinsic and extrinsic factors, which can result in animals having a range of potential strategies to accomplish a goal. For many species, periods of long-distance movements that vary in frequency (daily, seasonally, annually, etc.) and duration are an essential part of their life-history patterns and are often linked to foraging behaviour. These movements can include long-distance travel to either forage over vast areas or as commuting periods between breeding and foraging locations (e.g. Dingle 1996; Alerstam, Hedenstrom & Akesson 2003). This latter type of movement allows animals to meet the differing requirements between satisfactory breeding sites and highly productive foraging areas. This disjunct between suitable habitats is clearly evident in marine predators that display terrestrial breeding, such as sea turtles, sea birds, and pinnipeds (Costa 1991; Alerstam et al. 2003). However, this behaviour is not limited to terrestrial breeding marine predators as it is also found in some cetaceans, fish, and sharks (e.g. Corkeron & Connor 1999; Hulbert et al. 2005).

With the exception of the sea otter, it is almost impossible to directly observe the foraging behaviour and movement patterns of marine predators as these behaviours occur underwater and therefore outside the scope of observational studies (Estes, Jameson & Rhode 1982). With the development of time-depth recorders and satellite telemetry, information on the diving behaviour and movement patterns of an assortment of free-ranging marine predators has become increasingly available (Shaffer & Costa 2006; Hays 2008). The data obtained from these electronic tags have been used to assess different behavioural states such as periods of foraging, resting, dispersal, and migration (e.g. Catry et al. 2004; Insley et al. 2008). However, in some cases, these tools require researchers to make assumptions about the behaviour being recorded. Such is the case when studying foraging behaviour, as distinct changes in diving and movement patterns are assumed to coincide with periods of feeding. Relying on these assumptions without adequate validation could result in a misinterpretation of the function of specific behaviours. Therefore, to truly understand foraging behaviour in marine predators, it is necessary to directly measure prey consumption, making it possible to more accurately classify at-sea behaviour.

Like other pinnipeds (fur seals, sea lions, true seals and walrus), northern elephant seals (Mirounga angustirostris, Gill) exhibit a separation between terrestrial breeding and marine foraging. During two annual foraging migrations, seals have been shown to travel over 3000 km from the rookery (Le Boeuf et al. 2000; Simmons et al. 2007). As with other diving species, these animals exhibit dives that vary in two-dimensional shape and it has been speculated that these dive shapes reflect a variety of behaviours such as transit, foraging, and food processing (Hochscheid et al. 1999; Schreer, Kovacs & Hines 2001). In addition, based on these dive shapes and decreases in transit rate, it has been suggested that foraging can either occur along the transit route or at the distal ends of these migrations (Le Boeuf et al. 2000; Simmons et al. 2007). However, as with most marine predators, the actual extent of foraging that takes place both during transit periods and while in core foraging areas is unknown.

This study directly measured feeding events in migrating northern elephant seals through the use of stomach temperature telemetry. Since elephant seals feed on ectothermic prey, the prey is colder than the seals’ core body temperature. This results in a drop in stomach temperature with consumption (Fig. 1). Captive feeding experiments conducted with northern elephant seals have shown that this technology can accurately identify prey consumption (up to 97·9% of the time) and can be a valuable tool to measure at-sea feeding in free-ranging animals (Kuhn & Costa 2006). When used in conjunction with time-depth recorders and satellite transmitters, stomach temperature telemetry provides information about when and where animals successfully capture prey, making it possible to identify various at-sea behaviours. Therefore, our study objectives were twofold: (i) identify the locations and extent of feeding during the early migration and compare feeding behaviour among individuals; and (ii) examine the validity of indirect measures used to identify feeding in both northern elephant seals and other marine predators. Given that indirect indicators of feeding have been associated with mass gain (Le Boeuf et al. 2000; Crocker et al. 2006), we hypothesized feeding events would be highly associated with changes in both dive shape and transit rate.

Figure 1.

Plot of stomach temperature (top, gray) and dive depth (bottom, black) against time (Greenwich Mean Time) from a free-ranging female northern elephant seal. Stomach temperature record shows two single feeding events denoted by sharp declines in temperature, as indicated by arrows.

Materials and methods

animal handling

Adult female northern elephant seals at Año Nuevo State Reserve (California, 37·116° W, 122·332° N) were equipped during their annual breeding haulout. After females give birth (December–February), they nurse a pup for an average of 28 days before departing on a post-breeding foraging migration. Research was conducted on seven seals in February 2004 and eight in February 2005. Animals were sedated with an initial intramuscular injection of Telazol (Tiletamine hydrochloride and Zolazepam hydrochloride, Fort Dodge Animal Health, Fort Dodge, IA, USA) at 1·0 mg kg−1 based on a visual estimate of mass. Sedation was maintained with intravenous doses of ketamine hydrochloride and valium (Fort Dodge Animal Health, Fort Dodge, IA, USA), as necessary.


Each seal was equipped with a satellite tracking transmitter (PTT; Telonics, Mesa, AZ, USA or Wildlife Computers, Redmond, WA, USA) and a time-depth recorder (TDR; Mk7 or Mk9; Wildlife Computers, Redmond, WA, USA). In addition, a VHF transmitter (Telonics, Mesa, AZ, USA) was attached to locate seals on land after the migration. Stomach temperature was measured using a stomach temperature recorder (HTR) and stomach temperature telemeter (STT; Wildlife Computers, Redmond, WA, USA). The recording instruments (PTT, TDR, HTR) were attached to high-tension, nylon mesh netting with cable ties, and glued to the pelage using 5 min quick set epoxy (Loctite or Devcon epoxy). Recording instruments were wrapped in rubber splicing tape, to facilitate removal when the animals returned for the annual moult. The STT was placed in the stomach via a stomach tube and was modified with foam to increase retention time (Austin et al. 2006; Kuhn & Costa 2006). For instrument recovery, females were sedated following the methods described above.

data processing and analysis

Erroneous satellite locations were filtered based on a maximum transit rate of 10 km hr−1 and a minimum time between satellite locations of 10 min (Tremblay et al. 2006). All Z quality satellite locations were removed and the remaining location qualities were used in the filtering algorithm. Tracks were interpolated at 20-min intervals using a Bezier curve (µ = 0·1) based on the methods described in Tremblay et al. (2006). This allowed each dive and feeding event to be associated with a location along the track. Maximum and total distances travelled were calculated for each female. Maximum distance was defined as the straight-line distance from the rookery to the farthest location. Total distance was defined as the sum of distances travelled between filtered satellite locations along the migration route. Tremblay et al. (2006) found that using this type of linear interpolation almost always underestimates track lengths when compared to curvilinear interpolation, producing the ‘absolute minimum distances’ travelled. Therefore, both maximum distance and total distance were only used to determine if movement patterns differed between years.

Dive data were processed using a purpose-built zero-offset correction algorithm and analysis program written in matlab (IKNOS-DIVE, IKNOS toolbox, Y. Tremblay, unpublished). A dive was defined as a minimum depth of 4 m and minimum duration of 16 s. The analysis program calculated basic parameters for each dive such as maximum depth, dive duration, and surface interval. Bottom time was calculated at 95% of the maximum dive depth (for comparison with Le Boeuf et al. 2000). Percentage of time diving was the percentage of time at sea for which a seal was at depths greater than 4·0 m.

Dive records from previous studies were typed visually using the categories described by Le Boeuf et al. (1988, Fig. 2). Dive parameters from each category were then used to create discriminant functions that were applied to the dive records in this study. Dives assigned to the A category were assumed to represent transit dives as they displayed little to no bottom time. C type (drift) dives were considered putative food processing dives (Crocker, Le Boeuf & Costa 1997) and D dives, defined by vertical excursions in the bottom phase (wiggles), were considered putative foraging dives. Benthic dives, used for either travelling along the continental shelf or benthic foraging, were assigned to the E category. Short, shallow dives (<100 m) that did not fall into the previous categories were defined as I dives. This dive type occurred most often at the beginning and end of the migration as animals transited over the continental shelf, and made up only 5·3 ± 0·9% of the total dive records (N = 12).

Figure 2.

Distribution of dive types for the period of matching stomach temperature records in northern elephant seal females (N = 10). A schematic of each dive type and putative function is presented for each category. Percentage of each dive type is presented above each bar.

To analyse stomach temperature records, the methods from Lesage, Hammill & Kovacs (1999) were modified based on data collected from laboratory feeding experiments with northern elephant seals (Kuhn & Costa 2006). Feeding was defined simply as ingestion of prey and could include consumption of one or more prey items. A feeding event was identified when stomach temperature dropped by a minimum of 1·0 °C and was followed by a warming of a minimum of 0·4 °C within a 10-min period (Fig. 1). Using time between ingestion, feeding events were divided into single feeding events and bout (continuous) feeding. A minimum of four feeding events each separated by less than 1 h was considered bout feeding. The ending criterion for a feeding bout was determined using log-survivorship curves based on time between feeding events following the methods of Gentry & Kooyman (1986). All other feedings were defined as single feeding events (Fig. 1). In addition, to assess differences in foraging success due to the behaviour of vertically migrating prey, comparisons were made between daytime and night-time feedings. Local sunrise and sunset times were used to determine differences in day and night behaviour.

To examine whether the feeding rates measured in the early migration would be appropriate to explain mass gain over the entire foraging trip, a simple model was constructed extrapolating feeding behaviour for the duration of the migration. We used each female's measured mass gain for the migration and assumed a metabolic rate of either Kleiber or two times Kleiber (Kleiber 1961). For simplicity, further assumptions were females had an assimilation efficiency of 94% (Goodman-Lowe, Carpenter & Atkinson 1999) and fed on exclusively gonatid squid (Gonatopsis borealis) of an average size of 249 g (Antonelis et al. 1994) with an energy content of 4·64 MJ kg−1. Using this estimate of energy required, we calculated the number of prey required per feeding event based on the average and maximum feeding rates measured.

indirect measures of feeding

Le Boeuf et al. (2000) used a 2-day average transit rate of less than 0·4 m s−1 to define ‘focal foraging areas’ for female northern elephant seals. Two-day average transit rate was calculated following the methods of Le Boeuf et al. (2000) to examine if seals showed decreased transit rates while feeding. However, 2-day average transit rate may obscure animal movements on a smaller temporal scale. Therefore, transit rates were also calculated between interpolated locations. Each feeding event was associated with a transit rate, making it possible to examine the relationship between short-term transit rates and feeding. Additionally, to examine the relationship between dive types and feeding behaviour, the proportion of feeding events associated with each dive type was calculated.

statistical analysis

Summary data are reported as mean ± SE. Statistical analysis was conducted using systat 10 (©SPSS Inc. 2000) or r 2·2·1 (R. Gentleman and R. Ihaka, http://www.r-project.org). Comparisons between female foraging bout behaviour were conducted using an anova. A general linear mixed model was used to test the relationship between 2-day average transit rate as well as interpolated transit rate and feeding events with individual as the random factor. The model with the lowest Akaike's information criterion (AIC) was selected as the best model, unless there was no significant difference based on the log-likelihood test (Burnham & Anderson 2002). Day and night differences for both dive depths and feeding events were examined using paired t-tests. All data sets were tested for normality and homogeneity of variances. When data were non-normal or displayed unequal variances, they were log10 or square root transformed. Differences were considered significant at P < 0·05.


at-sea behaviour

Of the 15 females tagged, instruments were recovered from thirteen: 12 from Año Nuevo and one from San Miguel Island (California, 34·023° W, 120·308° N). Ten females had complete satellite tracks, two had tracks that lasted into the return phase of the migration, and one female's satellite transmitter failed early along the migration. Of the 13 dive recorders recovered, one instrument failed to record any dive data. There were 12 stomach temperature records that recorded data after the females left for the migration. This resulted in eight females with all three records (satellite, dive, stomach temperature), while the remaining females had some combination of two records.

A total of 7767 satellite locations were acquired, resulting in an average of 7·2 locations day−1 during the migration. Filtering removed 2646 satellite locations resulting in 5·3 locations day−1. Distributions of location qualities were not significantly different after filtering once Z quality locations were removed (χ2 = 2·05, P = 0·84, Table S1, Supporting Information). The average gap duration for the filtered data was 4·6 ± 0·2 h with a range of 3·5 to 6·5 h.

There were no significant differences in diving parameters (depth: t10 = −1·68, P = 0·12; duration: t10 = 0·38, P = 0·71; bottom time: t10 = 1·64, P = 0·13; surface interval: t10 = −0·04, P = 0·97; percentage of time diving: t10 = −0·31, P = 0·76) or movement parameters (maximum distance travelled: t9 = 0·38, P = 0·72; total distance travelled: t9 = 1·1, P = 0·30; trip duration: t11 = 0·73, P = 0·48) between years, so years were pooled for analysis. The post-breeding migration lasted 83·3 ± 2·5 days (range 68·3–105·4 days). Both maximum distance and total distance travelled spanned a large range for the females (1267–4145 km, 4777–11 072 km, respectively). This was a result of the differing foraging habitats used by females as some stayed near the continental shelf and others displayed the more typical pelagic foraging pattern (Le Boeuf et al. 2000; Simmons et al. 2007).

Females displayed diving patterns similar to those previously described for this species (Table 1, Le Boeuf et al. 1988; Naito et al. 1989). Females spent 87·5 ± 0·01% of the time at-sea diving with surface intervals of 2·4 ± 0·09 min. Based on dive shape, during the period of time when stomach temperature records were available, females spent over 43% of the time travelling and only 23·4% of the time foraging (Fig. 2). As this analysis was only conducted for the duration of stomach temperature records, type I (shallow) dives represented a larger proportion of the dive types than when compared to the entire trip records (21·7% vs. 5·3%).

Table 1.  Summary of dive parameters for 12 adult female northern elephant seals during the post-breeding migration. Ranges provided are the minimum and maximum means for all females. Bottom time was defined as the period when the animal was at 95% of the maximum dive depth. Percentage of time diving was the period at sea during which the females were below 4·0 m
 Mean (SE)Range
Dive depth (m)466·4 (23·2)304·9–613·9
Dive duration (min) 20·1 (0·5) 17·8–22·9
Bottom time (min)  8·6 (0·4)  7·0–10·4
Surface interval (min)  2·4 (0·09)  2·0–3·3
Percentage of time diving 87·5 (0·01) 85·8–91·0

feeding behaviour

STTs remained in the stomach for an average of 7·8 ± 1·9 days (range 2·2 to 21 days, N = 12). The median number of feedings per record was 5·5 (range 0 to 191). The first feeding occurred early and close to the rookery for all but one female. On average, the first feeding occurred 4 h after females left the beach and 58·6 km from the rookery (Table 2). One female consumed her first prey 48·8 h after leaving the beach. First feedings were usually single feeding events (1·4 ± 0·3 feeding events) and were followed by extended periods without feeding (14·5 ± 2·5 h, Fig. 3).

Table 2.  Summary of stomach temperature records and feeding events for 12 adult female northern elephant seals during the post-breeding migration. Where N = 3, all values are presented instead of the range.
 NMean (SE)Range
Stomach temperature record length (days)12 7·8 (1·9)2·3–21·1
Number of feedings1243·2 (21·1)  0–191
Time to first feeding (hours)11 4·0 (1·5)0·8–48·8
Time to bout feedings (days) 3 7·5 (1·8)4·1, 8·1, 10·3
Distance to first feeding (km)1158·6 (21·9)2–180
Distance to bout feeding (km) 3 610 (167)281, 728, 821
Figure 3.

Satellite tracks for two females with feeding areas highlighted to show the spatial and temporal differences in feeding between individuals. Tracks are plotted for the duration of the stomach temperature records (female A, black line: 12·1 days, female B, grey line: 10·3 days). Grey circles (female A) and triangles (female B) represent feeding events for each female. Female B fed 191 times during the length of the stomach temperature record while female A fed only two times. Inset: satellite tracks for the entire foraging migration for the two females. Dashed box highlights area of feeding records.

Three females retained STTs into areas where continuous (bout) feeding occurred (n = 35 bouts). These females averaged between 7·9 and 13·4 feeding events per day and maximum feeding events per day ranged from 20 to 30. Based on log-survivorship curves, the feeding bout ending criterion was determined to be 105 min, meaning a feeding bout was considered over when a female did not show additional feeding within this amount of time. Feeding bout duration was calculated from the first to the last feeding event. The first feeding bout occurred on average 7·5 ± 1·8 days after leaving the rookery, at an average distance of 610 km. Feeding bout duration did not differ significantly among females (6·7 ± 1·2 h, F2,32 = 0·27, P = 0·76). On average, females displayed 1·5 feeding bouts per day. There was no significant difference in the number of feeding events per bout, with females feeding on average less than nine times per bout (F2,32 = 0·01, P = 0·38). There was a significant difference among the three females in feeding rate when bout feeding (1·3 ± 0·5, 1·5 ± 0·5, 2·1 ± 0·6 feeding events hour−1, F2,32 = 5·27, P = 0·01).

For the females that displayed bout feeding behaviour, there was a significant diel pattern in feeding events (Fig. 4, t2 = 4·8, P = 0·04) as females displayed increased feeding events just after local sunrise which continued until sunset. For all females, only 35·0% of feeding events occurred during the night (range 25–39%). Greater number of feeding events during the day was not a result of increased dive frequency as females displayed a significantly greater number of dives during the night (t7 = −3·0, P = 0·02).

Figure 4.

Top: distribution of feeding events by hour for female northern elephant seals that displayed bout feeding behaviour (N = 3). Bottom: diel cycle in dive depths for the same females with matching dive records (N = 2). Error bars are SE. Shaded area indicates period of local night. Females showed significantly higher number of feeding events during the daytime hours, as feeding rate increased just after sunrise and dropped sharply after sunset (t2 =  4·8, P = 0·04).

Based on the simple bioenergetic model created, using the average feeding rates measured (7·9–13·4 feeding events day−1), these seals would need to eat 2·3–5·2 prey per feeding event to gain the observed average of 68·3 ± 18·2 kg. If feeding continued at the maximum rate for the remainder of the migration (20–30 feeding events day−1), then only 1·0–2·1 prey per feeding event would be required. As expected, when using the higher metabolic rates, intake increased to 3·9–7·6 prey feeding−1 event at the average feeding rate and 1·8–3·1 prey feeding−1 event at the maximum rate.

indirect measures of feeding

During the period of stomach temperature records, females displayed a 2-day average transit rate of 1·0 ± 0·02 m s−1, an average interpolated transit rate of 1·43 ± 0·01 m s−1, and exhibited feeding behaviour at all transit rates measured. For the three females that displayed bout feeding, there was a significant negative relationship between 2-day average transit rate and daily number of feeding events (Fig. 5a, r2 = 0·42, P < 0·01). When interpolated transit rate was used, the model explains 20% more variance (Fig. 5b, r2 = 0·62, P < 0·01). There was no significant effect of individual for both of these relationships (2-day average transit rate: log-likelihood ratio = 2·3, P = 0·13, interpolated transit rate: log-likelihood ratio = 0·01, P = 0·93). Two-day average transit rate did not identify any areas of ‘focal foraging’ for the three females that displayed bout foraging behaviour because transit rate did not fall below 0·4 m s−1.

Figure 5.

(a) Relationship between 2-day average transit rate (m s−1) and number of feedings per day for female northern elephant seals displaying bout feeding behaviour (N = 3). As the 2-day average transit rate decreased, the number of feedings per day increased (r2 = 0·42, P < 0·01). Dashed line represents cut-off used by Le Boeuf et al. (2000) to identify ‘focal foraging areas’ based on a transit rate of less than 0·4 m s−1. (b) Relationship between interpolated transit rate (m s−1) and number of feedings (r2 = 0·62, P < 0·01).

Females displayed D (putative foraging) type dives 24% of the time when stomach temperature was recorded (Fig. 2). Although most feeding was associated with D dives (74·2%), the stomach temperature records indicated that feeding occurred with all dive types. Type A (transit) dives accounted for the second highest dive type associated with feeding at 16·4%. Drift (C), benthic (E), and shallow (I) dives were associated with a minimal amount of feeding at 2·1%, 3·0%, and 4·2%, respectively. When all D dives for the length of the stomach temperature record were examined, feeding only occurred on 18·5% of the dives. This relationship improved to 23·5% when only bout feeding females with matching dive records were examined.


By directly measuring the feeding behaviour of northern elephant seals, this study was able to more accurately characterize the at-sea behaviour of a large marine predator. Using stomach temperature technology, it was possible to identify when and where feeding occurred, provide an estimate of foraging success, and show the extent of feeding that occurred during the early part of the foraging migration. In addition, significant relationships were identified between feeding events and indirect measures of foraging behaviour, such as changes in transit rate and dive shape.

feeding behaviour

Spatial patterns

Given that female northern elephant seals travel over 4000 km from the rookery during a foraging migration (Le Boeuf et al. 2000; this study), it has been suggested that either the prey quality or quantity in the area close to the rookery is not sufficient to meet energetic requirements after the extended breeding fast. From this, it would be expected that seals would feed little between the rookery and their primary foraging grounds. Yet, for all but one female, the first feeding events occurred early and close to the rookery (Table 2). Early feedings were often single feeding events and were possibly opportunistic prey captures as females migrated to more productive foraging grounds in the North Pacific. In grey seals (Halichoerus grypus), where feeding behaviour has been examined in detail using stomach temperature telemetry, feeding also began as early as 1 day after leaving the beach (Austin et al. 2006). However, depending on the foraging strategy used, the main foraging grounds of grey seals can be as close as 39 km from the rookery or up to 259 km away (Austin, Bowen & McMillan 2004). For other pelagic predators such as wandering (Diomedea exulans) and grey-headed albatross (Thalassarche chrysostoma), opportunistic feeding has also been identified, as animals were found to consume single, potentially low-quality prey items in between bout feeding (Weimerskirch, Doncaster & Cuenotchaillet 1994; Catry et al. 2004).

While opportunistic feeding occurred shortly after departure, continuous or bout feeding behaviour did not begin until females travelled on average over 600 km from the rookery (Table 2). This extended travel time resulted in the post-breeding fast lasting an additional week for some females. For one female, the stomach temperature record lasted 13 days at sea with only two, single feeding events occurring (Fig. 6, female: O176). For this seal, the breeding fast essentially lasted at least 43 days. This previously unmeasured fasting at sea represents an increase in the total length of this female's breeding fast by 39%. Females lose on average 36% of their body reserve while on land during the lactation period (Crocker et al. 2001) and this additional at-sea fasting time could significantly change our understanding of the breeding and foraging energetics of northern elephant seals and potentially other capital breeders.

Figure 6.

Daily feeding rates of four female northern elephant seals as they transited away from the rookery (Año Nuevo State Reserve, California). Daily location was calculated as the straight-line distance from the rookery to the interpolated position closest to noon on each day. Each symbol represents a unique animal.

As females travelled away from the rookery, there was an overall increase in daily feeding rate (Fig. 6). Although not directly measured in many species, this increase in foraging success away from breeding areas has been widely suggested in a variety of marine taxa including other pinnipeds, sea turtles, and seabirds. Using changes in body composition as a proxy for foraging success, Biuw et al. (2007) also found in southern elephant seals (Mirounga leonina) that feeding increased as animals transited from the rookery. Leatherback sea turtles (Dermochelys coriacea) show an increase in dive duration once they move into the Atlantic after departing from breeding areas in the Caribbean. Hays et al. (2006) proposed this demonstrated increasing foraging success based on the theory that animals increase dive time with increased foraging success. In a tracking study of another long-distance traveller, the northern fur seal (Callorhinus ursinus), females showed short periods of slower transit and diving behaviour during the transit to winter foraging grounds (Ream, Sterling & Loughlin 2005). This behaviour is thought to be opportunistic feeding as the females move towards more profitable areas including the North Pacific Transition Zone, an area also extensively used by female northern elephant seals (Le Boeuf et al. 2000).

Foraging success and dive patterns

As females travelled away from the rookery, the rate of increased foraging success was not consistent and foraging success varied among females (Fig. 6). The patchiness of prey resources in the marine environment could be one factor shaping this individual variation. As marine predators have incomplete knowledge of the distribution of prey, it is possible some individuals fortuitously encounter prey patches during transit while others do not. This hypothesis was supported by a study examining the feeding behaviour of wandering albatross (Weimerskirch, Gault & Cherel 2005). Extensive variation was measured in the distances travelled between feeding events suggesting animals encountered prey at varying frequencies.

Another reason for variation in foraging success could include differences in individual characteristics, foraging strategies or prey preferences. For bout feeding females, individual variation was found in the timing, frequency, and locations of feeding events (Fig. 3). Although female northern elephant seals showed some similarities in feeding bout duration and number of bouts per day, there was a significant difference among females in feeding events per hour while feeding in a bout. This could be a result of polymorphism in foraging behaviour, which is widely recognized in both terrestrial and marine species (e.g. Bolnick et al. 2003; Hatase, Omuta & Tsukamoto 2007). For example, California sea otters (Enhydra lutris) that foraged on varying prey assemblages had different diving patterns and time spent foraging (Tinker et al. 2007). If female northern elephant seals are consuming different prey species, we would expect to see similar variation in foraging behaviours. Hatase et al. (2007) related polymorphism in the foraging behaviour of loggerhead sea turtles (Caretta caretta) to body size in adult females. Due to the limited sample size in this study, it is difficult to draw relationships between differences in feeding success and other variables such as age, mass, foraging location or patch quality. However, these differences in bout foraging behaviour may help explain the variation in mass gain as measured when females return to the beach for the annual moult (Le Boeuf & Crocker 2005; Crocker et al. 2006).

Based on diving behaviour, it was previously hypothesized that female northern elephant seals foraged throughout the day and night (Le Boeuf et al. 1988, 2000). Dive depths varied with time of day, suggesting females followed the vertical migrating layer; and putative foraging (type D) dives were displayed repeatedly for up to several days (Le Boeuf et al. 1988, 2000). The feeding behaviour of these females supports this hypothesis, but also shows significant differences in the number of feeding events based on time of day (Fig. 4, t2 = 4·8, P = 0·04). All three females who reached areas of bout feeding before stomach temperature telemeters were lost showed an increase in the number of feeding events during the daylight hours (Fig. 4). This is in contrast to some species that show a diel pattern in foraging behaviour where foraging increases during the night when it is hypothesized that vertically migrating prey move to the surface and are easily captured [such as northern fur seals (Gentry & Kooyman 1986); and Antarctic fur seals, Arctocephalus gazella, (Boyd, Lunn & Barton 1991)].

Studies with blue whales (Balaenoptera musculus), king penguins (Aptenodytes patagonicus), and crabeater seals (Lobodon carcinophaga) found that the behaviour of vertically migrating prey influences both the timing of foraging and foraging success (Croll et al. 1998; Pütz et al. 1998;Burns et al. 2004). For these species, foraging on vertically migrating prey appears to be most profitable when they are clumped at depth during the daylight hours. It has been suggested that blue whales and other large cetaceans can only meet their energy needs by foraging on whole schools of prey while they are aggregated at depth during the day (Croll et al. 1998). Pütz et al. (1998) hypothesized the decrease in foraging efficiency of king penguins during the night was a combination of the lack of light and also the dispersion of prey as it moves to the surface. For crabeater seals, it was suggested daytime hunting was more profitable, not only because of the clumping behaviour, but also because the prey moved to the sea floor. The use of this bottom barrier may have increased prey capture efficiency for the crabeater seals (Burns et al. 2004). The assistance of bottom topography is not available to female northern elephant seals as they mostly forage over deep pelagic waters, but clumping of prey could explain the increase of foraging success during the day.

indirect measures of feeding behaviour

Transit rate

Feeding events measured by stomach temperature telemetry provide an opportunity to validate the use of indirect measures of feeding to identify foraging behaviour in elephant seals as well as other marine predators (e.g. Weimerskirch et al. 2005; Hays et al. 2006). Based on the hypothesis that animals will spend more time in areas of increased foraging success, it is expected that decreased transit speed or increased residence time in an area can be used to identify foraging locations (Le Boeuf et al. 2000; Fauchald & Tveraa 2003; Robinson et al. 2007). This was supported by the findings of Le Boeuf et al. (2000), where mass gain over the migration increased as the percentage of time at slow transit rates increased. In this study, both 2-day average transit rate and interpolated transit rate showed a significant relationship with feeding events (Fig. 5a and b). The link between slower transit or increased residence time and increased foraging success has been directly shown in king penguins and bottlenose dolphins (Tursiops truncatus; Bost et al. 1997; Bailey & Thompson 2006). Although this hypothesis is supported, for many species the extent of foraging that occurs when animals are in these key foraging areas when compared to periods defined as transit is still unknown. For this species, it is clear considerable feeding occurs outside of ‘focal foraging areas’ and during the early migration (first 12–27% of the trip; Fig. 5a). Similarly, Horsburgh et al. (2008) recorded feeding behaviour of southern elephant seals outside of the main foraging zones and early along the migration. For wandering albatross, over 75% of foraging occurred outside areas identified as area-restricted search zones (Weimerskirch 2007). Recent work with southern bluefin tuna (Thunnus maccoyii) found no relationship between feeding success and increased residency periods (Bestley et al. 2008). These authors suggest southern bluefin tuna may spend time in alternative areas for reasons other than feeding such as sociality, resting, or predator avoidance. Therefore, it is important to consider that extensive feeding may occur outside locations identified as ‘focal foraging areas’ and the amount of feeding within and outside these areas cannot be determined by transit rate or area-restricted search behaviour alone.

Dive shape

A second foraging proxy used extensively with diving air-breathing vertebrates to identify foraging behaviour is changes in dive shape (Hochscheid et al. 1999; Schreer et al. 2001). For species that forage within the water column, dives distinguished by vertical excursions (wiggles) during the bottom phase have been hypothesized as pursuit of prey. The present study found that of the four dive types previously described for northern elephant seals, feeding is highly associated with the putative foraging dive type (Le Boeuf et al. 1988). However, over 25% of feedings were associated with other dive shapes and transit (A type) dives accounted for 16% of feeding events. This discrepancy between dive shape and feeding events was also recently reported in leatherback sea turtles, where beak openings were associated with all of the dive shapes described (Fossette et al. 2008). In a study examining feeding behaviour of harbour seals (Phoca vitulina), feeding activity was also associated with multiple dive shapes (Lesage et al. 1999). However, these authors suggest harbour seals may use several foraging tactics (active searching vs. sit and wait) resulting in multiple dive shapes being associated with successful feeding. Thus, a potentially significant proportion of feeding events may be missed when using solely putative foraging type dives to examine foraging behaviour. Consequently, care must be taken when attributing feeding behaviour to the various dive shapes.

Furthermore, when using putative foraging (type D) dives as a proxy for feeding behaviour, D dives appear to overestimate successful feeding. However, since these dives are defined as foraging dives, it is important to consider the ultimate differences between feeding (consumption of prey) and foraging behaviour. Foraging encompasses other behaviours such as searching, handling, or unsuccessfully pursuing prey. Stomach temperature measurements can only provide information about when animals successfully capture prey. Therefore, if dive behaviour and specifically putative foraging dives are used to identify all aspects of foraging (searching, pursuing, and capture), then females on average are successfully capturing prey between 18% and 24% of the time during foraging periods. Visual observations of foraging California sea otters showed that they capture prey between 67·2–88·4% of their foraging dives (Tinker et al. 2007). Using a prey encounter index based on spikes in swim velocity, Horsburgh et al. (2008) estimated that female southern elephant seals have a foraging success rate within the range reported here at approximately 20%. However, the values from this study are significantly lower than the beak opening rates measured for foraging leatherback sea turtles, where 38·7–72·1% of dives had grouped beak opening events suggestive of feeding (Fossette et al. 2008). Nonetheless, it is important to note that these rates do not reflect actual foraging success but potential foraging attempts. In addition, this dissimilarity in foraging success may also reflect the differing diets of these species (squid/fish vs. jellyfish or shallow benthic invertebrates).


Since this research recorded stomach temperature during the early phase of the foraging migration (max: 27% of the total migration), this may limit the ability to extrapolate the data obtained through the entire post-breeding migration. Le Boeuf et al. (2000) found diving behaviour changes when animals reach ‘focal foraging areas’. When dive shape distributions were compared for females with bout feeding behaviour and matching dive records, there was a significant difference (χ2 = 1357·4, P < 0·001), with an overall increase in foraging type dives over the entire migration (18·3–27·2%). In addition, as the distribution of dive types change during the migration, this may alter the associations between feeding and dive type. As a result, this could also influence the number of feeding bouts per day, length of feeding bouts, and/or number of feeding events per bout. However, it is also important to consider that intake rates may be influenced by feeding motivation (Tolkamp et al. 1998). Early in the migration, we expect feeding motivation to be high as female elephant seals are recovering from the breeding fast that lasts over 1 month. As the migration progresses, it is unknown how feeding motivation changes and how this impacts feeding rates.

Although bout feeding did not begin until animals travelled away from the rookery, it is important to note extensive feeding occurred during the early period of the foraging migration. In fact, based on the simple model created, if the maximum feeding rates measured were maintained throughout the migration, it could realistically explain mass gain for these females. Given that the diet of female northern elephant seals includes other, more energetically rich species (fish) and dive behaviour changes when females reach ‘focal foraging areas’, this exercise suggests the feeding behaviour measured in this study could reflect feeding behaviour throughout the migration. Furthermore, because of this considerable amount of early foraging, the at-sea behaviour of female northern elephant seals does not meet the typical definition of migration as described by Dingle (1996). Dingle suggested migratory behaviour involves a suppression of characteristic responses to available resources, including a suppression of feeding behaviour. Instead, this study was able to support the proposition of Le Boeuf et al. (2000) that female at-sea behaviour is better described as ranging since females are foraging as they move through the environment.

By directly measuring the feeding behaviour of northern elephant seal females, it is now possible to gain greater insight into their at-sea behaviour. With the ability to quantify foraging success, it is now possible to measure the impacts of environmental change on the foraging success of this species and potentially others. Finally, since direct measures of feeding are available in relatively few marine species, further investigations examining the environmental variables associated with feeding in this species may help elucidate how marine predators in general locate patchy prey resources in the dynamic marine environment.


This work would not have been possible without the support of numerous researchers and volunteers, especially J. Hassrick, B. McDonald, P. Morris, H. Mostman, P. Robinson, M. Rutishauser, S. Simmons, and S. Villegas. Research was supported by grants from the Mildred E. Mathias Foundation, Friends of Long Marine Lab, and the California Sea Grant College Program. This research was conducted as a part of the Tagging of Pacific Pelagics (TOPP) program, funded by the National Ocean Partnership Program, the Moore, Packard, and Sloan Foundations, and the Office of Naval Research. Animal handling procedures were approved by the UCSC Chancellors Animal Research Committee and permitted under NMFS permit #87-143.