Loughlin et al.: Diving behavior of immature Eumetopias /ubatus 



569 



grams were summarized separately for dive depth and dive 

 duration for each of the four time periods. The SLTDRs re- 

 corded dive depth information in six separate "bins": 4—10 m, 

 10-20 m, 20-50 m, 50-100 m, 100-250 m, and >250 m. We 

 used 4 m as the minimum depth for a dive based on ear- 

 her studies in Alaska (Merrick et al., 1994). Dive-duration 

 bins were 0-60 sec, 60-120 sec, 120-180 sec, 180-240 sec, 

 240-360 sec, and >360 sec. 



The ST-10 and ST-16 units used the same 6-h periods as 

 the ST-6. However, the ST-10 and ST-16 SDKs subdivided 

 dive depth information into 14 bins: 4 m; 4-6 m, 6-10 m, 

 10-20 m, 20-34 m, 34-50 m, 50-74 m, 74-100 m, 100- 

 124 m, 124-150 m, 150-174 m, 174-200 m, 200-250 m, 

 and >250 m. Dive duration also contained 14 bins at one- 

 minute intervals (e.g. 1-2 min, 2-3 min, 3-4 min, etc.). The 

 14 time-at-depth bins coincided with dive-depth bins (e.g. 0, 

 4, 4-6, 6-10, etc. and the last was >200). However, the first 

 bin was set to zero to determine if an animal was on land 

 based on the proportion of dry readings of the salt-water 

 switch during a 6-hour period. Time-at-depth was calcu- 

 lated as the proportion of time that dives occurred within 

 a particular depth bin of a 6-h period while the sea lion was 

 at sea (e.g. if an animal was at sea for 3 hours during a 6-h 

 period and spent half its dive time in bin 50-74, the value 

 in bin 50-74 would be 25'7f ). 



We deployed ten ST-10 and ST-16 SDKs (Table 1) which 

 transmitted time-line messages in bins of 20-min periods 

 (there are 72 periods of 20-min each in a 24-h day). These 

 messages provide information on whether the instrument 

 was wet or dry >10 min of a 20-min period for each of the 

 72 periods. Time-line messages thus allow calculation of 

 time spent at sea and on land. 



Maximum dive depth in a 24-h period, from midnight 

 GMT to midnight GMT. was provided in the status mes- 

 sage. This is a separate message that provides information 

 on transmitter status, including a pressure offset, battery 

 status, number of transmissions to date, at-surface data, 

 date, time, ID of message, and a saltwater conductivity 

 reading. All 25 transmitters that we deployed transmitted 

 a status message. 



The ST-6 SLTDRs were on 24 h/day and transmitted a 

 maximum of 400 transmissions/day. To save battery power 

 the instrument had a 6-h haul-out period; that is, it would 

 turn off only if the transmitter was "dry" for 6 hours, indi- 

 cating that the animal was on land. The ST-10 and ST-16 

 SDRs had 3-h haul-out periods; the ST-10 had a maximum 

 of 250 transmissions/day, and the ST-16 had a maximum of 

 325/day Both the ST-10 and ST-16 had duty cycles of 4 h on 

 and 2 h off during a 24-h period to distribute transmissions 

 during different times of the day and to ensure recording of 

 information in all bins. All duty cycles started at midnight, 

 with an offset of + 13 h from GMT for Alaska. 



Location data 



Locations were estimated by the Service-Argos, Inc. clas- 

 sification scheme, where location class (LC ) 3 is accurate to 

 <150 m, LC 2 is accurate 150 m-s350 m, LC 1 is accurate 

 350 m-1000 m, and LC is accurate >1000 m. LCs A and 

 B have no assigned accuracy range (Service-Argos, 1984; 



Keating, 1994). However, after our analysis, Vincent et 

 al. (2002) used an algorithm published by McConnell et 

 al. (1992) to filter satellite locations and found that both 

 filtered and unfiltered LC A locations were of a similar 

 accuracy to LC 1 locations for four gray seals iHalichoerus 

 grypus). Because of the large variance in our samples asso- 

 ciated with LC A locations, we excluded them (and the LC 

 Bs) from our analyses. We sorted location data by date and 

 time line to determine the locations for each trip. 



Data analysis 



Data analysis followed that of Merrick et al. (1994) and 

 Merrick and Loughlin (1997). Analysis of the number of 

 dives was prepared by summing counts of dives from the 

 histograms. Median depths and durations of dives were 

 calculated by using the range midpoint of a bin (e.g. 7 m for 

 a 4—10 m bin) as the depth for all dives in the bin. We rec- 

 ognize that this approach invokes a possible error for dive 

 profiles in large increment bins (e.g. 50-100 m) where the 

 mean dive depth is the same, 75 m, regardless of whether 

 the animal made most of its dives between 51 and 60 m or if 

 it made most of its dives between 90 and 100 m. This error 

 is inherent in the data collection process and could not be 

 eliminated with the instruments used in our study. We also 

 recognize that more locations will be recorded when the 

 animals are at the surface for long periods or when transit- 

 ing to different locations. However, because of the repetitive 

 transmission of the histogram data and the usual short 

 duration of short-range trips, there should be no inherent 

 behavior-based bias in the dive data reported. Differences 

 in dive depth and duration between locations were tested 

 by using the Pearson chi-square tests or analysis of vari- 

 ance (ANOVA) (F-statistic), and P-value differences less 

 than 0.05 were considered significant. Analysis of trip 

 distance and duration were analyzed by using a repeated 

 measures ANOVA, and because the distances were skewed, 

 they were log-trEinsformed to examine differences among 

 groups. 



Trips were defined and measured for distance by us- 

 ing an integrated process of the SDR data. For animals 

 deployed with ST6 SDRs, which did not contain time line 

 data sets (/! = 15), trip distances were extracted by using a 

 combination of the dive histogram, duration histogram, 

 and land or sea data sets to estimate arrival and departure 

 times as well as locations calculated at sea or on land. 

 Once arrival and departure times were estimated, the 

 location data were examined to confirm that all locations 

 calculated during that trip were wet locations. We then 

 had all locations for an individual trip and from those 

 locations we filtered out all A and B locations and im- 

 posed a swim speed filter (3 m/s). Finally, we reported the 

 maximum straight-line distance from the departure site. 

 For animals with STlO/16 SDRs, we were able to extract 

 arrival and departure times from the time-line messages. 

 However, if a day of time-line data was not received, we 

 referenced the time-at-depth data, depth, and duration 

 histograms to reconstruct the missing day of data. Once 

 arrival and departure times were calculated we then fol- 

 lowed the protocol stated above. 



