This study took place at Nyrøysa (54.41° S, 03.29° E) on the west coast of Bouvetøya, South Atlantic, during three austral summer seasons (2000/01, 2001/02, 2007/08). A total of 46 lactating AFS were instrumented with various combinations of biotelemetry gear during the study (Table 1). Females with newborn pups were selected, assuming that the presence of an umbilicus/placenta indicated that mothers were still in their perinatal period, that is, still ashore after giving birth, before departing for their first trip to sea. The mothers were manually restrained using a large cone-shaped hoop net (2.2 m long×1 m in diameter) attached to an aluminium frame and handle (see David et al. 1990 for further details). Body mass was measured to the nearest 0.1 kg using a 100 kg Salter scale prior to instrument attachment and upon recovery (Table 2). All instruments were glued to the fur of the adult females in the mid-dorsal region with two-component industrial-grade epoxy (Huntsman AW2101/HW2951, Intertronics, Kidlington, Oxfordshire, UK, or 5-Cure, Industrial Formulators Inc., Port Coquitlam, BC, Canada). During their mothers’ instrument deployment period the pups were also weighed several times by suspending them briefly in a canvas bag attached to a mechanical spring; the first weight was taken when the mothers’ instrument(s) were deployed.

Different combinations of instruments were used in the different seasons (Table 1); all had at least one instrument with a protruding antenna. During 2000/01 and 2001/02, a total of six satellite relay data loggers (SRDLs; Sea Mammal Research Unit, St. Andrews, UK) and 20 Mk6 time–depth recorders (TDRs; Wildlife Computers, Redmond, WA, USA) in combination with very high frequency transmitters (VHFs; Model 5955, Advanced Telemetry Systems, Isanti, MN, USA) were deployed. The TDRs were removed after five–six weeks of deployment, prior to the expedition's departure from the island, while the SRDLs were left on the animals to fall off naturally during the annual moult. The SRDL data used in the present study covered only the first five–six weeks of lactation, during which both SRDL and TDR data were collected. Instrument dimensions and weights are summarized in Table 3. The SRDLs sampled depth every 4 s while diving and the on-board processor was set to define the start of a dive to occur when the tag was wet and below 5 m for 8 s. A dive ended when the tag either returned to within 5 m of the surface or became dry. The Mk6 TDRs sampled depth every 10 s while the tag was wet and had a depth resolution of 2 m. Raw data files obtained from the Mk6 TDRs were extracted using purpose-built software provided by the manufacturer (Dive Analysis, Zero Offset Correction, Minimum-Maximum-Mean, Wildlife Computers). Any excursion from the surface to a depth ≥5 m for at least 8 s was considered a dive, matching the settings of the SRDLs.

During 2007/08, three instruments were simultaneously deployed on each of 20 adult female AFS: (1) a 0.5 W Sirtrack Kiwisat 101 platform terminal transmitter (PTT; Sirtrack Inc., Havelock North, New Zealand); (2) a TDR (Mk9 Wildlife Computers); and (3) a Sirtrack VHF to facilitate recapture. All three instruments were glued to the fur along the main axis of the animal in the mid-dorsal region (behind the scapulae), with the TDR and the VHF in front of the PTT (see Table 3 for details on the instruments’ characteristics). The TDRs were programmed to record depth at a resolution of 0.5 m, and light level every second, both when wet and dry in order to record attendance patterns precisely. Any excursion from the surface to a depth=5 m for at least 8 s was defined as a dive to match the definition of the instruments from the two preceding study periods.

Start and end times of trips were defined based on dive and conductivity switch records from the TDRs, or based upon distance from the island for animals with SRDLs and PTTs. To obtain the best possible representation of the real path followed by the animals, a speed–distance–angle filter (see Freitas et al. 2008) was applied to the tracks recorded by the SRDLs and PTTs. All statistical analyses were conducted using R, version 2.12.1 (R Development Core Team 2009). All distances from the island refer to great circle distances and the maximum corresponds to the most distant point reached within a trip. When statistical models were used, model fitting was based on the examination of the residuals and Kolmogorov–Smirnov tests were used to assess whether or not data were normally distributed.

Summary statistics for trip parameters for trips 1 and 2 are presented in Table 4. None of the parameters were significantly different for the two trips (Table 5). Date was not a significant variable in any of the models, but year influenced the mean dive duration and the mean dive depth. Both variables were significantly greater in 2007/08 compared to the two first seasons.

A total of 189 trips and 168 haulout periods were used to compare the attendance patterns and the trip parameters between the two instrument groups (Table 6). A summary of the LMEs used to explore these data is presented in Table 7. The mean trip duration was significantly greater for SRDL animals (mean_SRDL=103.9±33.7 h) compared to TDR instrumented females (mean_TDR=65.4±21.5 h). Instrument type and date were significant variables in the LME model, while year was not significant. The mean haulout duration was not significantly different between the two instrument groups (SRDL animals 45.3±10.9 h; TDR animals 37.7±17.0 h), nor were any of the other variables tested in the model. Dive duration and dive depth were only influenced by date and year but not by the instrument type carried by the female. The data sub-sampling performed by the SRDLs did not affect the results, as the estimates and the p-values were similar between the model using the complete TDR data set or a matching random 30% sample of the TDRs data records, simulating sampling similar to the SRDL (Table 7).

The females instrumented with SRDLs had similar masses to those instrumented with TDRs when capture and instrument deployments took place (mean_SRDL=42.2±6.6 kg; mean_TDR=37.8±6.4 kg; ANOVA F_1,23=2.02; p=0.17). The percentage mass loss per day (mean_SRDL=−0.39±0.15%; mean_TDR=−0.18±0.28%; ANOVA F_1,23=2.59; p=0.12) and the total mass loss over the period of instrumentation (mean_SRDL=−7.2±4.4 kg; mean_TDR=−2.8±4.4 kg; ANOVA F_1,23; p=0.055) were not statistically different for the two groups, although the mean values for the SRDL group were more than twice that of the TDR group for both parameters.

Pup start masses were not significantly different between the two instrument groups (mean_SRDL=5.3±0.7 kg; mean_TDR=5.0±0.5 kg; ANOVA F_1,20=1.19; p=0.28). Over the study period, pup growth was close to linear in each treatment group (Fig. 1). The simplest LME model fitted to explore pup growth rate used pup age, instrument type borne by the mother and the interaction between age and instrument. Pup sex and the interaction between age and pup sex were tested but there was no significant differences (Table 7). However, pups with SRDL-instrumented mothers grew significantly slower than those with TDR-instrumented mothers (growth rate_SRDL=0.078±0.011 kg day⁻¹; growth rate_TDR=0.104±0.025 kg day⁻¹).

There were no detectable differences in the recorded trip parameters (trip duration, maximum distance, total distance, mean dive duration and mean depth duration) between the first trip performed following capture and handling and the subsequent trip. Nor were there any obvious behavioural changes observed upon the release of instrumented mothers and their pups in the treatment groups; all pairs showed apparently normal behaviour when reunited, as has been reported in previous studies (Doidge et al. 1986, but also see Geertsen et al. 2004; Hooker et al. 2007; Walker et al. 2012). Despite having undergone what is likely a moderately stressful experience (the capture and instrument deployment) immediately prior to the first trip, behaviour during this first trip to sea did not differ from the subsequent trip in any manner that was measured. Consequently, using data from the first trip after instrumentation or deploying instruments on a large number of animals for only one or two trips in order to increase the sample size (and minimize the risk of longer-term deployments) seems to be a reasonable protocol for AFS studies.

AFS females instrumented with larger instruments (SRDLs+VHFs) in this study performed substantially longer trips than those carrying smaller instruments (TDRs+VHFs). Similar findings have been shown for various other marine mammal species (Baker & Johanos 2002; Walker et al. 2012) and penguins (Croll et al. 1991; Ropert-Coudert, Wilson et al. 2007). In previous studies, AFSs carrying VHFs +TDRs showed extended attendance patterns compared to those carrying only VHFs (Walker & Boveng 1995). Boyd et al. (1997) purposely studied the consequences of adding drag and increasing the cost of transport on lactating females of this species by attaching a wooden block on the animals’ backs, mimicking the shape and position of an instrument similar to the SRDLs used in this study. These authors also found that animals carrying the largest and heaviest load performed longer trips. In other species, such as the Juan Fernández fur seal (Arctocephalus philipii), the first trip of females carrying a TDR was up to four times longer than trips performed by females carrying no instrument (Francis et al. 1998). But, in contrast Boyd et al. (1991) did not find any difference in attendance patterns between AFS with and without TDRs.

Foraging trip and haulout durations are usually correlated in AFS activity records (Costa et al. 1989; Boyd et al. 1991). But, in the present study, no significant difference in haulout duration was observed between the two instrument groups, despite the significant difference in at-sea trip duration. It is possible that an existing difference went undetected because of our relatively small sample sizes (see Walker & Boveng 1995), but it seems more likely that additional foraging costs were met via spending more time at sea, but this fact did not impact how long it takes to provision the pup with available milk stores before needing to again access food to restart the cycle. The increased trip duration observed could be related to lower prey capture rates as a consequence of the added drag, but is more likely due to increased costs to the female of swimming and diving. Hydrodynamic shape, density and position of a tag are known to affect streamlining and therefore the efficiency with which the animals move in a fluid (Wilson et al. 1986; Agnew 2004; Hooker et al. 2007; Walker et al. 2012). Instrument mass is likely of relatively minor importance for most marine mammal deployments because the instruments are close to being neutrally buoyant and are usually <5% of the animal's body mass; in this study equipment mass was a maximum of 1% of body mass (see Hawkins 2004; Walker et al. 2012). Boyd et al. (1997) showed that animals with larger, heavier tags swam slower both during surface swimming and diving compared to animals with smaller tags, although the present study did not detect any difference in dive depth or dive duration between the two groups with SRDLs and TDRs. If the net energy gain per dive is decreased due to increased drag of larger instruments, the animals might have to stay longer at sea in order to meet their energetic requirements before returning to the colony to feed their pup if they do not increase the time spent in individual dives. Increased foraging costs do occur naturally in fur seals in periods of reduced prey abundance (Boyd et al. 1991; Boyd et al. 1994) with the consequence of increased time spent at sea under such circumstances, comparable to the situation observed in the present study with the largest tags.

Foraging in female fur seals is influenced by two contrasting mechanisms: the amount of energy the female can store, while meeting her own energy needs, and the amount of time she can spend at sea, which is dependent on the pup's fasting abilities (Costa et al. 1989; Boyd et al. 1991; Arnould et al. 1996). If the female's energy expenditure at sea increases, the time needed to replenish her energy stores and accumulate new reserves to produce milk will increase but she will still be constrained by the pup's fasting limits. If the females cannot spend the necessary time at sea to recoup their metabolic and milk production costs, they will be forced to use more body reserves or lower energy delivery to the pup (or a combination). In the present study, SRDL animals lost on average twice as much mass per day (and total mass) compared to TDR animals, although this difference was not statistically significant. Small sample size (n _SRDL=5) and high individual variance typically observed in AFS foraging behaviour (Walker & Boveng 1995) limit the power to detect differences between groups in this study. Boyd et al. (1997) reported a mass loss of 2 kg more for treatment animals (large instrument n=6) compared to control animals (small instrument n=8), but similarly failed to detect a statistically significant difference. It is difficult to assess if larger instruments would have any negative long-term effects on the carriers. The present study and that of Boyd et al. (1997) lasted for only the first third of the lactation period, though this does cover the most energetically demanding period for the females. Milk production decreases approximately 80 days after birth (Arnould 1996; Arnould et al. 1996) and the females might then be able to allocate more energy to replenish their own energy reserves regardless of the added foraging cost.

Several factors are known to influence AFS pup growth rates, such as sex, prey availability (Vargas et al. 2009) and mass loss rates during the fasting periods while their mother is at sea (Guinet et al. 1999). In the present study, sample sizes were too small to assess potential effects of sex and year but a significant difference in growth rate was detected with pups of SRDLs mothers growing more slowly. However, it should be noted that both sets of pups remained within the range of pup growth rates described in the scientific literature for AFS from South Georgia and Macquarie Island (Lunn et al. 1993; Guinet et al. 1999; Vargas et al. 2009). Both sexes were represented within each treatment group but, due to the unbalanced number of SRDLs and TDRs, there were more females in the TDR group. However, this lack of balance is counter to the trend observed between the equipment treatment groups because females normally have slower growth rates than males (Guinet et al. 1999; Vargas et al. 2009).

In conclusion, this study showed that data from the first trip after instrumentation is representative of subsequent trips and that handling does not appear to have any measurable effect on the parameters explored. Therefore, deploying instruments on a large number of animals for only one or two trips in order to increase the sample size (and minimize the risk of longer-term deployment costs) seems to be a reasonable protocol for AFS studies. Secondly, larger instruments seem to have at least temporary negative effects on lactating AFSs and their pups. At-sea time was significantly longer for animals with larger instruments and pup growth was significantly lower, suggesting that larger instruments should be used for only short periods, or that the shape of the instruments should be altered prior to being used again on this species. Lastly, instrument effects should be taken into account when comparing time series collected under different equipment deployment regimes in order to avoid false conclusions regarding interannual variation or other explanatory variables.