Frequently Asked Questions

Transcription

1 Frequently Asked Questions Does the elephant seal research impact individuals or the population? The short answer is: we probably have a small influence on their short-term behavior, but no long-term impacts on health or survival. Before we get into the details, let s first highlight a few of the reasons the elephant seal research is so important. Elephant seal research conducted at Año Nuevo has resulted in important findings in areas such as foraging behavior, disease ecology, and metabolic physiology, just to name a few. These studies help us understand not only how elephant seals make a living in the open ocean, but they also have important implications for the conservation of other species. For example, by modeling the potential effects of human disturbance on elephant seals, we can better understand how we might be impacting other more elusive species of greater conservation concern (such as beaked whales). Even more amazing, the results from studies at Año Nuevo may help us solve some of the mysteries of human diseases, such as diabetes. Researchers always try to minimize their impact, but these studies do require researchers to interact with a small number of seals in four ways: (1) attaching and reading flipper tags, (2) Obtaining measurements (such as length, girth, and weight), (3) collecting small samples such as blood, blubber, and fur (a bit like a physical exam), and (4) temporarily attaching instruments to track their movements and record their diving behavior. When conducting these studies, researchers minimize the impact of their work by studying only a small subset of the population, using established protocols, and pooling data. One of the most effective ways researchers minimize their impact is to coordinate their efforts. For example, when a seal is anesthetized, one researcher can attach a tracking tag to monitor its movements at sea, another researcher can collect a whisker to investigate the seal s diet, and a third researcher can collect a blood sample to measure DDT and other pollutants. Each project must also obtain several permits from the federal government, a local ethics committee, and the state park. Even with these efforts, it is important to demonstrate that researchers are not impacting the seal population. Fortunately, several studies have investigated this exact question! In short, careful study has shown that standard research activities do not impact the foraging success or survival rates of elephant seals, nor do they impact the weaning mass of pups. The only measurable effect was a brief stress response when physically restraining young seals (probably similar to a human baby during the first trip to a doctor). We are always interested in speaking with anyone who is interested in our studies and methods. Please feel free to me with any questions, comments, or suggestions: robinson@biology.ucsc.edu On the following page, you can find a brief summary of the studies mentioned above and their full references.

2 Summaries and References: Champagne et al (2012) found that anesthetic restraint does not typically cause a cortisol stress response in any age or sex class, whereas physical restraint alone may result in a stress response. Champagne CD, Houser DS, Costa DP, Crocker DE (2012) The Effects of Handling and Anesthetic Agents on the Stress Response and Carbohydrate Metabolism in Northern Elephant Seals. PLoS ONE 7(5): e doi: /journal.pone Engelhard et al (2001) showed that daily research activities such as weighing pups or recording tagged individuals did not influence the weaning mass of pups (an important indicator of future survival). Engelhard, G. H., van den Hoff, J., Broekman, M., Baarspul, A. N., Field, I., Burton, H. R., & Reijnders, P. J. (2001). Mass of weaned elephant seal pups in areas of low and high human presence. Polar Biology, 24(4), Engelhard et al (2002a) investigated the blood chemistry of seals handled several times and found no differences to seals handled just once. Engelhard, G. H., Hall, A. J., Brasseur, S. M., & Reijnders, P. J. (2002). Blood chemistry in southern elephant seal mothers and pups during lactation reveals no effect of handling. Comparative Biochemistry and Physiology-Part A: Molecular & Integrative Physiology, 133(2), Englehard et al (2002b) found a very short-term increase in the stress response of physical handling, but no long-term effects. Also found that chemically immobilizing a seal multiple times (3 or 4) in a single breeding season may cause longer-term effects. Engelhard, G., Brasseur, S. M. J. M., Hall, A., Burton, H., & Reijnders, P. (2002). Adrenocortical responsiveness in southern elephant seal mothers and pups during lactation and the effect of scientific handling. Journal of Comparative Physiology B: Biochemical, Systemic, and Environmental Physiology, 172(4), Englehard et al (2002c) investigated the effect of human disturbance on suckling behavior of motherpup pairs. Alertness was elevated in the presence of researchers, but quickly returned to baseline levels when researchers left the area. Researchers did not impact other behavioral parameters such as the total time spent suckling. Engelhard, G. H., Baarspul, A. N., Broekman, M., Creuwels, J. C., & Reijnders, P. J. (2002). Human disturbance, nursing behaviour, and lactational pup growth in a declining southern elephant seal (Mirounga leonina) population. Canadian Journal of Zoology, 80(11), McMahon et al (2005) found no difference in the first year survival rates of pups that were simply marked versus those that were handled several times during the breeding season. McMahon, C., Hoff, J. V. D., & Burton, H. (2005). Handling intensity and the short-and long-term survival of elephant seals: addressing and quantifying research effects on wild animals. AMBIO: A Journal of the Human Environment,34(6), McMahon et al (2008) found that attaching tracking devices to the head of seals does not result in reduced foraging success or long-term survival. This is true even when individuals were instrumented on several migrations. McMahon, C. R., Field, I. C., Bradshaw, C. J., White, G. C., & Hindell, M. A. (2008). Tracking and data logging devices attached to elephant seals do not affect individual mass gain or survival. Journal of Experimental Marine Biology and Ecology, 360(2),

3 Polar Biol 2001) 24: 244±251 DOI /s ORIGINAL PAPER Georg H. Engelhard á John van den Ho Martijn Broekman á Antonie N. J. Baarspul á Iain Field Harry R. Burton á Peter J. H. Reijnders Mass of weaned elephant seal pups in areas of low and high human presence Accepted: 20 October 2000 / Published online: 15 December 2000 Ó Springer-Verlag 2000 Abstract On sub-antarctic Macquarie Island, we examined pup weaning mass of southern elephant seals in relation to human presence. Pup weaning mass was previously found to be positively associated with 1styear survivorship. Weaned pups were weighed in a remote area, Middle Beach, and in an area of relatively high human presence, Isthmus East. The areas were reasonably similar in beach topography, wind and surf conditions, numbers of seals present per kilometre of coastline, and numbers of males and females present in harems. For a sub-sample of measured pups, data on the respective maternal size were collected using photogrammetry. Both male and female weaned pups on Middle Beach were signi cantly heavier than those on Isthmus East. Estimated length of mothers was signi cantly higher on Middle Beach. In proportion to their own size, mothers in both areas produced weaners of similar mass, indicating no direct e ect of human disturbance on the e ciency of lactation. It remained unclear whether the area di erences in maternal and pup size were due to natural or human-related factors. Introduction Several studies report direct behavioural responses of marine mammals to human presence e.g. Salter 1979; G. H. Engelhard &) á M. Broekman á A. N. J. Baarspul Centre for Ecological and Evolutionary Studies, Zoological Laboratory, Groningen University, P.O. Box 14, 9750 AA Haren, The Netherlands G.H.Engelhard@biol.rug.nl Tel.: ; Fax: J. van den Ho á I. Field á H. R. Burton Australian Antarctic Division, Channel Highway, Kingston 7050, Tasmania, Australia G. H. Engelhard á P. J. H. Reijnders Alterra Institute, Marine and Coastal Zone Research Team, P.O. Box 167, 1790 AD Den Burg, The Netherlands Brasseur 1993; Reijnders et al. 1993; Born et al. 1999; Suryan and Harvey 1999). However, it is complicated to assess the long-term impact of disturbance on reproduction and survival chances, and hence at the population level. In long-lived species such as marine mammals, examining human impact at the population level necessarily implies collection of data over a very) long time span. Thus the implementation of the appropriate conservation measures may also take a long time. Therefore it is of special importance to quickly detect those measurable impacts that are likely to a ect tness. Antarctica and the sub-antarctic attract growing numbers of visitors. Scienti c, commercial and tourist activities are increasing, in particular near pinniped and penguin breeding sites. Thus there is the potential for impact of disturbance on breeding success and survival in populations of these species. Human disturbance was found to a ect Antarctic penguins e.g. Wilson et al. 1991; Culik and Wilson 1995; Giese 1996, 1998; Regel and PuÈ tz 1997; Cobley and Shears 1999). However, little is known on e ects of disturbance on sub-)antarctic pinnipeds e.g. Wilkinson and Bester 1988). This is remarkable, as several of these species are relatively well studied and currently the subject of considerable scienti c investigations. Pinnipeds are major top predators in Southern Ocean ecosystems, and studies on the impact of humans on their breeding biology might contribute to conservation. Moreover, such studies may also support the assessment of impact of human activity on pinniped populations elsewhere in the world Trites 1991; Reijnders et al. 1993, 1997). For the southern elephant seal, Mirounga leonina L.), we examine whether human disturbance can be detected through an easily measurable parameter related to both reproduction and survival: mass of juveniles at weaning. Pinniped females usually nurse a single young per season, and pup weaning mass can be interpreted as an index of maternal investment Costa 1991; Arnbom et al. 1997). Most importantly, mass of juveniles at nutritional independence is likely to in uence their survival northern fur seals, Callorhinus ursinus L.): Baker and

4 245 Fowler 1992; Hawaiian monk seals, Monachus schauinslandi Matschie: Craig and Ragen 1999), although in an increasing northern elephant seal [Mirounga angustirostris Gill)] population, no evidence was found that pup weaning mass was related to 1st and 2nd year survivorship Le Boeuf et al. 1994). However, for the species and population examined in the present paper, previous work has shown that heavier weaned southern elephant seal pups were indeed more likely to survive their 1st year than their light counterparts; 1st year survival of heavy and light quartile weaners was estimated around 71% and 54%, respectively McMahon et al. 2000). In addition, juvenile size is likely to a ect adult size, and larger male and female seals appear to have higher reproductive success females: Reiter et al. 1981; Arnbom et al. 1994; Pomeroy et al. 1999; males: McCann 1981; Anderson and Fedak 1985; Bartsh et al. 1992; Haley et al. 1994). Because of these long-term implications, we believe that pup weaning mass is an interesting parameter to study in relation to human activity. Southern elephant seals forage widely in the Southern Ocean and reproduce on traditional breeding locations; most of these are on sub-antarctic islands, and one is in temperate Patagonia Laws 1994). During the breeding season September/November), female elephant seals aggregate on beaches, forming harems that are competed for by males Laws 1956). Pregnant females give birth to a single pup about 4 days after arrival at the beach and nurse it for about 23 days, whereafter it is weaned abruptly McCann et al. 1989; Arnbom et al. 1997; McMahon et al. 1997). Throughout the nursing period, mothers neither feed nor drink, while their pups rapidly accumulate fat reserves through milk intake. At weaning, pups are left by their mothers and are nutritionally independent. Male pups have, on average, higher birth and weaning masses than female pups, and in both sexes there is much variation in weaning mass; most of this is related to variation in the mass of mothers at parturition Arnbom et al. 1993, 1997; Fedak et al. 1996; Carlini et al. 1997; McMahon et al. 1997). Most populations of southern elephant seals have been harvested commercially in the past, but all have been protected for several decades. In spite of this, populations in the Southern Paci c and Southern Indian Oceans either have since been, or currently are, in decline Hindell and Burton 1987; Hindell et al. 1994; Guinet et al. 1999; McMahon et al. 1999; Pistorius et al. 1999). The causes of this are not fully understood, but are most likely related to unfavourable conditions in the ocean environment, probably in combination with equilibration processes after the cessation of sealing, and potentially other local factors that also regulate populations Hindell 1991; Hindell et al. 1994). Is disturbance by humans among these additional factors? To our knowledge, in the past only a single study on disturbance in elephant seals has been carried out. Within the declining breeding population of Marion Island, Wilkinson and Bester 1988) found similar rates of decrease in the numbers of seals over a 10-year period, regardless of local di erences in onshore human activity. In addition, the authors reported that the rates of decline in elephant seal numbers were not di erent among islands in the Southern Ocean where there had either been permanent research stations or far less frequent human visitation over the past decades. However, disturbance in the earlier stages may not directly a ect the total numbers of animals, but rather their condition. Indirect impacts on reproduction and survival as a result of lowered condition may ultimately be detectable by counts, but may not necessarily be identi ed as a result of disturbance per se. We hypothesized that during the breeding season of southern elephant seals, human disturbance could have negative impacts on the e ciency of energy transfer from lactating females to pups, thereby a ecting mass of pups at weaning. This would have consequences for their survivorship and/or future reproductive success McMahon et al. 2000). This was examined by comparing a sample of elephant seal pups nursed in two study areas that were similar in geography but di erent in the levels of human presence. We tested if pup weaning mass was lower in the area of higher human activity. As elephant seal weaning mass is highly dependent upon maternal mass at onset of lactation Arnbom et al. 1997), we collected data on maternal size for a sub-sample of mother-pup pairs using photogrammetry, and examined the relative e ects of maternal size and study area. Materials and methods Location During September/November 1998, eldwork was carried out on sub-antarctic Macquarie Island S, E), which has the third-largest breeding population of southern elephant seals in the world Laws 1994). Since 1948, the Australian Antarctic Division has run a permanent research station on the island, situated at the Isthmus; for most of this period the elephant seal population has been monitored Carrick and Ingham 1962; Carrick et al. 1962a; Hindell and Burton 1987; Hindell et al. 1994; McMahon et al. 1999). Traditionally, most human activities scienti c, maintenance, sight-seeing, other) have been on the Isthmus near the station, which continuously accommodates around 10±15 persons in winter and around 30±40 persons in summer; other areas of the island receive relatively infrequent human visitation. Area of high human presence The Isthmus was chosen as representing an area with a relatively high level of human presence. Here, elephant seal harems are formed on beaches adjacent to the research station, which are within the o cial ``Station Limits'' directly accessible to expedition members and vehicles at the base. Harems on the east and west coasts of the Isthmus were visited each day of the 1998 pupping season by four to ve persons in order to continue long-term monitoring studies of the population. Elephant seal pups were weighed and marked with plastic ipper tags within 24 h of birth, a procedure that implied some physical handling and a brief separation from the mother approximately 1±3 min if carried out routinely). Tagged pups were re-weighed at weaning. Methods of weighing and tagging were similar to those described by McMahon

5 246 et al. 1997). Weather and surf conditions, as well as characteristics of beach topography, were rather di erent between the wind-swept west coast and the more sheltered east coast of the island Streten 1988; Selkirk et al. 1990). Furthermore, harem size tended to be far more variable on the west side Australian Antarctic Division, unpublished data; Carrick et al. 1962b). For these reasons, only data for the east coast of the Isthmus were used in the present study; hereafter this area is referred to as Isthmus East. Its shingle beaches were 20±40 m wide at locations where elephant seal harems formed. Locally, there were sites of regular strong winds within the area. In total, 459 day-old pups were weighed on Isthmus East 50% of all pups in the area) during 28 days; on average, 16 animals were weighed each day median 17.5, range 0±32). Of these animals, 429 were re-weighed at weaning. Area of low human presence Middle Beach was chosen as the control area with low level of human presence. The site is also on the east coast, 2 km south of the station but separated from that area by a 1-km zone of steep rocky shore at the foot of a scree slope. Middle Beach is reasonably similar to Isthmus East in beach topography, with shingle beaches 15±40 m wide at sites where elephant seal harems formed. Wind and surf conditions were also comparable; during the study period, we observed similar wind forces Beaufort scale) in the two areas [for both sites, median 3; Mann-Whitney N = 127), U = )0.869, P = 0.385]. Three harems in the control area were, as in Isthmus East, visited each day of the pupping season; however, this was by only up to three persons on foot. All other human presence was excluded during the observation period. No day-old pups were weighed or tagged at Middle Beach. For individual recognition during behavioural studies G.H. Engelhard, unpublished data), 24 study pups 4% of all pups in the area) were marked with a paint dot within 24 h of birth, by touching them brie y with a paint sponge attached to a long pole; they were not otherwise physically handled or separated temporarily from the mother. Pups observed outside harems and considered to be weaners were weighed using similar methods as on the Isthmus, and marked with plastic ipper tags for future recognition. Data on weaning mass were collected for 168 elephant seal pups on Middle Beach. Photogrammetry For a randomly selected sub-sample of pups in both study areas, data were collected on the size of their respective mothers. Rather than sedating and physically weighing females, we used photogrammetric techniques modi ed from Bell et al. 1997), in order to minimize handling disturbance. Females were photographed repeatedly on di erent days of lactation, preferably at earlier stages. Standardized lateral photographs were made, under conditions when the subject was lying on a at surface, well visible from the side, at right angles to the body axis. A black and white 100-mm scale was held above the animal's midline and images taken at distances of 5±20 m, depending on circumstances. We used a Canon Eos 100QD body tted with a lens of 50 mm Canon EF, 1:1.8 II) or 200 mm Canon EF, 1:2.8). Prints were digitized and analysed using Optimas ±1994 Optimas, Seattle, Wash.). From the images, body length L metres, from snout tip to base of hind ipper; Haley et al. 1991; Bell et al. 1997) and body side area SA square metres) were measured Bell et al. 1997). Maternal length For 46 females 22 Middle Beach, 24 Isthmus East), maternal snoutto-tail length was estimated from the photographic variable L measured on all images available per female N = 151), as follows: ML ˆ R 1:074 L =N r 2 ˆ 0:998; N ˆ 36Š The multiplication factor was based on a set of calibration images of females taken during an earlier eld season 1996) for the 1 same project on Macquarie Island. In that year, direct data on snout-to-tail length and mass were collected on 32 females at different stages of the lactation period. Maternal postpartum mass For estimating mass M of a female from an image taken n days after parturition, model [18] in Bell et al. 1997) was modi ed based on data collected during our earlier 1996) eld season, as follows: M ˆ 506:99 SA 1:193 r 2 ˆ 0:707; N ˆ 39Š 2 Reasons for modi cation were: 1) for mass estimation of lactating females, only images of postlactation females were available in Bell et al.'s 1997) study; the present study, however, examines females at all stages of lactation; 2) it was found that models [18], [19], and [20] Bell et al. 1997) underestimated female mass when applied to our 1996 set of known-mass females. As lactating females lose on average 35% of mass from parturition to departure Arnbom et al. 1997), maternal mass is highly variable throughout lactation. For each of our 1998 study females, we established one estimate of maternal postpartum mass MPPM) from multiple images per individual taken over the rst 12 days of lactation most of these within a week of parturition). This was done by adding the number of days n that the image was taken after parturition, multiplied by the average mass loss per day for lactating females, as measured during our 1996 study season kg day )1 ; cf. McCann et al. 1989): MPPM ˆ R M 7:615 n =N 3 Postpartum mass was estimated for 44 females 21 Middle Beach, 23 Isthmus East), based on a total of 91 photographs. Maternal condition An index of the condition of females at onset of lactation was obtained using the postpartum mass:length ratio derived from photographic variables unit: kg m ±1 ). Parameters examined Parameters examined were: 1) maternal length, 2) maternal postpartum mass, and 3) maternal condition index, derived from photographic measures, and 4) pup weaning mass directly measured. Statistics were carried out following Zar 1984), using the SPSS Windows software package SPSS 1999). Results At the peak of the breeding season counted 15 October 1998), 509 females were distributed over 3 harems on Middle Beach, and 866 females distributed over 6 harems on Isthmus East. Peak numbers of females present per kilometre of coastline were similar for the two study areas respectively, 727 vs 722 females km ±1 of coastline; v 2 = , P = 0.683). Harem size, de ned as the number of females present in a harem on 15 October, was comparable for the study areas Middle Beach, from south to north 172, 319 and 18 females; Isthmus East, from south to north 29, 104, 113, 303, 310 and 7 females; Mann-Whitney, U = 7.00, exact P = 0.714). For two harems on Middle Beach and three harems on Isthmus East, where the numbers of adult males were monitored regularly over the breeding season, there

6 247 was no evidence for di erences in male presence between the study areas. The number of males situated in a harem, expressed as average per week, was not signi cantly di erent between the two areas repeated measures ANOVA, F 1,2 = 0.078, P = 0.806); neither was the number of males in or within 30 m of a harem, averaged per week repeated measures ANOVA, F 1,2 = 0.040, P = 0.861). Adult sex ratio, de ned as the proportion of adult males to the total of both sexes in or within 30 m of a harem, was similar for the two areas repeated measures ANOVA, F 1,2 = 0.536, P = 0.540). Maternal size Estimated snout-to-tail length of all study females averaged m mean SD; N = 46, range 2.00±2.99 m). Estimated maternal postpartum mass averaged kg N = 44, range 355±769 kg). Maternal condition index or mass:length ratio averaged kg m ±1 N = 46, range 164±269 kg m ±1 ). Fig. 1a±c shows length, postpartum mass and condition index for mothers of male and female pups on Middle Beach and Isthmus East. Length estimates were similar for mothers of male or female pups [respectively, m N = 15) vs m N = 29), independent samples t-test, t = 0.120, P = 0.905]. Mothers of male or female pups were not signi cantly di erent in either estimated postpartum mass [ kg N = 15) vs kg N = 27), t = ±0.786, P = 0.436] or condition index [ kg m ±1 N=15) vs kg m ±1 N=27), t = ±1.448, P=0.155]. Size di erences were detected when females were compared between areas. Study females on Middle Beach were signi cantly longer than on Isthmus East [ m N = 22) vs m N = 24), t = 2.798, P = 0.008]. There appeared to be a similar tendency in estimates of maternal postpartum mass for the two areas [respectively, kg N = 21) vs kg N = 23)], but the di erence was not statistically signi cant t = 1.535, P = 0.132). Condition index was similar for females on Middle Beach and Isthmus East [ kg m ±1 N = 21) vs kg m ±1 N = 23), t = 0.528, P = 0.600]. For the sample of photographed females, there was a tendency of somewhat earlier parturition date on Middle Beach median 25 September 1998, range 9±30 September, N = 22) when compared to Isthmus East median 26 September 1998, range 16±30 September, N = 24; Mann-Whitney, U = 175.5, P = ). Pup size Mass at weaning of male and female elephant seal pups in both study areas is shown in Fig. 1d. Over the whole sample, weaning mass averaged kg mean SD; N = 597, range 38±208 kg). On Middle Beach, average weaning mass was kg N = 87) and kg N = 81) for male and female pups, respectively; on Isthmus East; mass of males and females averaged kg N = 205) and kg N = 224). Weaning mass was signi cantly di erent between the two sexes, and between pups on Middle Beach and Isthmus East. This was shown with a 2-way ANOVA of weaning mass with the factors sex and area. Both factors enhanced the explained variance signi cantly sex: F 1,593 = 5.146, P = 0.024; area: F 1,593 = 7.371, P = 0.007), while no signi cant interaction term was found sex area: F 1,593 = 0.035, P = 0.851). On average, male and female weaners, respectively, were 7.8 kg 6.4% of overall mean for males) and 6.8 kg 5.8% of overall mean for females) heavier on Middle Beach than on Isthmus East. There was no signi cant di erence in the sex ratio of weighed weaners between both areas males:females, Middle Beach 87:81, Isthmus East 205:224; Pearson's v 2 = 0.77, P = 0.379). Fig. 1a±d Maternal and pup size parameters mean SE) for Middle Beach and Isthmus East, Macquarie Island. a Maternal length. b Maternal postpartum mass. c Maternal condition index. d Pup weaning mass. Un lled symbols mothers of) female pups; lled symbols mothers of) male pups. Numbers indicate sample sizes Pup size relative to maternal size For all studied mother-pup pairs, the factor maternal length alone explained 69.4% of the variation in pup weaning mass linear regression, F 1,41 = , P < 0.001). Fig. 2 shows pup weaning mass as a func-

7 248 P = 0.435, g 2 = 1.7%); inclusion of this factor into the model enhanced the explained total variation only slightly 76.5%). There were no signi cant e ects of the factor pup sex or of any of the interactions of area and sex with terms of maternal size. Discussion Fig. 2 The relation between maternal length and pup weaning mass for elephant seals on Middle Beach un lled symbols, dotted lines) and Isthmus East lled symbols, solid lines). Regression lines with 95% con dence intervals are shown tion of estimated maternal length for Middle Beach and Isthmus East. The factor estimated maternal postpartum mass accounted for 67.6% of the variation in pup weaning mass linear regression, F 1,39 = , P < 0.001). Using backward general linear model analysis, we tested for the relative e ects of maternal size, pup sex and study area Middle Beach or Isthmus East) on pup weaning mass. Maternal size was included as the terms maternal length, postpartum mass and squared postpartum mass Table 1; see note for explanation). The nal model consisted of the constant and each of the maternal size terms, and explained 76.1% of the variation in pup weaning mass. The factor area was rejected Our working hypothesis was that in southern elephant seals, human disturbance may reduce energy transfer from mothers to pups during lactation, resulting in lower masses of pups at weaning. At rst sight, the results of this study seem to support the hypothesis: weaning mass was indeed, both for male and female pups, higher in the remote area of Middle Beach when compared to the human-accessible area of Isthmus East; an average di erence of 7.5 kg 6.3% of mean weaning mass) was recorded. However, pup weaning mass is largely determined by maternal postpartum mass Arnbom et al. 1993, 1997; Fedak et al. 1996; McMahon et al. 1997). The photographic data set on the size of adult females showed evidence of maternal size di erences between the two areas consistent with pup weaning mass di erences. Length estimates of adult females were signi cantly higher on Middle Beach than on Isthmus East mean 2.53 m vs 2.33 m). A similar tendency was found in estimated maternal postpartum mass, but the e ect was not signi cant even though there was a 51 kg di erence in average mass estimates mean 543 kg vs 492 kg). That discrepancy could be explained as follows. Maternal length remains constant over lactation; the parameter could be estimated using all images available per female Table 1 Analysis of pup weaning mass with maternal length ML), maternal postpartum mass MPPM) and squared maternal postpartum mass MPPM 2 ) by pup sex and study area. N=41 8 males and 11 females on Middle Beach, 7 males and 16 females on Isthmus East). The nal model explained 76.1% of the variation in pup weaning mass. Areas are Middle Beach remote) and Isthmus East accessible to people). ML and MPPM were derived from di erent photographic variables. Squared MPPM is included in the model since pup weaning mass relative to maternal postpartum mass decreases with maternal postpartum mass Arnbom et al. 1997). Null model includes the constant only. Final model includes all signi cant parameters. Changes in SS and df indicate the changes when parameters are dropped from the nal model one at a time or added to the nal model in case of rejected terms). When testing the signi cance of higher order interaction terms, the lower order terms were included in the model, regardless of their signi cance Change in) SS type III) Change in) df F P Coe cients Null model Final model Constant ±296.3 ML MPPM MPPM ± Rejected terms Area ± Sex ± Area Sex ± Area ML ± Area MPPM ± Sex ML ± Sex MPPM ±

8 249 N=151). However, as females show considerable mass loss over lactation McCann et al. 1989), estimates of postpartum mass could be derived only from images taken during the initial stage of lactation N=91). In cases where females were photographed on later days than the parturition date, postpartum mass had to be extrapolated using the average mass loss rate for females as measured previously. Thus, length estimates were probably of higher accuracy than postpartum mass estimates, which could explain the nding of a signi cant di erence in length but only a trend in postpartum mass. There was no signi cant di erence in maternal postpartum condition index between areas. In the sample of study females, there was a trend of slightly earlier parturition date on Middle Beach when compared to Isthmus East median 25 vs 26 September 1998), but this date tendency of 1 day only cannot explain the di erence in female size between areas; moreover, younger and therefore smaller) pregnant females Laws 1953; Carrick et al. 1962a) are expected to arrive earlier on breeding beaches Kirkman 1999). For the pups with known maternal data, 76.1% of the variation in weaning mass could be explained by maternal length, postpartum mass and squared postpartum mass Table 1). Once these parameters were taken into account, there was no e ect of area remote or disturbed) on pup weaning mass. This implies that in proportion to their own size, study females on Middle Beach and Isthmus East produced weaners of similar masses. If this nding based on the smaller set of mother-pup pairs with both maternal and juvenile data N=44) can be extrapolated to the larger pool of measured weaners N=597), then it may be concluded that the area di erences in pup size found here were due to di erences in maternal size, which were already present at the stage that pregnant females arrived on breeding beaches. Signi cant di erences in weaning mass of elephant seal pups between populations in the Atlantic, Indian and Paci c sectors of the Southern Ocean are probably partly attributable to variation in food availability between these regions Burton et al. 1997). All animals in the present study came from breeding sites less than a few kilometres apart, distances negligible to the scale of foraging ranges of elephant seals Hindell et al. 1991; Slip et al. 1994); for example, adult female elephant seals previously tagged on the Isthmus of Macquarie Island foraged in moderately warm sub-antarctic as well as cold Antarctic waters between around 50 Sand over 70 S Slip et al. 1994). With these extensive foraging ranges, it seems unlikely that variation in local food availability can explain the di erences in maternal and pup size found here between adjacent breeding beaches. Rather, it seems that upon arrival, there was some degree of site selection by females depending on their size. Several causes might be adduced to explain why larger females apparently chose relatively more often to breed on Middle Beach than on Isthmus East. Female elephant seals continue to grow after reaching sexual maturity, and length is proportional to age Laws 1953; Carrick et al. 1962a); size and age are associated with dominance in female elephant seals Reiter et al. 1981; McCann 1982). Older, more experienced and/or dominant females might be expected to be more selective in breeding site choice. Middle Beach may have been preferred above Isthmus East as a breeding area either because of subtle) di erences in natural characteristics, or because of its remoteness from human activity. Although beach and surf conditions appeared similar, wind force was variable at a local scale especially on Isthmus East. Harem sizes, numbers of adult males present in or near harems, and adult sex ratios were not signi cantly di erent between the areas. However, there could have been di erences in harem characteristics not quanti ed, such as aggressive activities of adult males potentially causing natural disturbance Le Boeuf and Briggs 1977; Modig 1996). High-quality males may have been unequally distributed, causing di erential site selectivity among females Cox 1981). Aggression is likely to a ect reproductive success especially in smaller females, possibly forcing these to move to sites where they can more e ectively compete, thus causing size segregation Le Boeuf and Briggs 1977; McCann 1982). Alternatively, as there has been a research station on the Isthmus for several decades, there could be some degree of avoidance by older females of breeding in that area, caused by human disturbance during a previous breeding season cf. Thiel et al. 1992). Breeding site delity of female southern elephant seals is high Nicholls 1970; Hindell and Little 1988), but females may show increased breeding dispersal and reduced site delity in the case of low reproductive success in the previous season, as described in several other species Wauters et al. 1995; Murphy 1996; Haas 1998; Schjùrring et al. 2000). We underline the importance of quickly measurable parameters with known or expected links to survivorship in studies of human impact on wildlife, especially in long-lived animals. The present study examines such a survival parameter, pup weaning mass in southern elephant seals McMahon et al. 2000), in relation to human presence on Macquarie Island. It shows no signi cant di erence in pup weaning mass between sites other than that due to the size of the mothers, indicating no direct e ect of human presence on the e ciency of lactation in southern elephant seals. This is in line with the results of Wilkinson and Bester 1988): onshore human activity appears not to be a signi cant factor in the drastic decline of the species in the southern Indian and Paci c Oceans. Population declines are most likely attributable to changes in ocean productivity Hindell et al. 1994; Pistorius et al. 1999). Acknowledgements This study is a collaboration of the Netherlands Antarctica Program of NWO Project ), the Alterra Institute Marine and Coastal Zone Research Team Project ) and the Australian Antarctic Division Human Impacts Project 1007 and Biological Sciences Project 2265). Fieldwork was conducted as part of the 51st Australian National Antarctic Research Expeditions, and would have been impossible without the

9 250 ANARE sta and participating expeditioners on Macquarie Island, in particular Michael Carr and Lloyd Fletcher. We thank Maria Clippingdale, Paul Denne and Adam Jagla for their support in the eld. Valuable comments on the manuscript were given by Marthan Bester, Rudi Drent, Clive McMahon, Wim Wol and an anonymous referee. At earlier stages of this project, support and suggestions were given by Sophie Brasseur, Pirie Conboy, Jeroen Creuwels, Mike Fedak, Ailsa Hall and David Slip. The Tasmanian National Parks and Wildlife Service supplied permits to work on Macquarie Island. Our activities involving the handling of animals followed the Guidelines of the Antarctic Animal Ethics Committee. References Anderson SS, Fedak MA 1985) Grey seal males: energetic and behavioural links between size and sexual success. Anim Behav 33: 829±838 Arnbom T, Fedak MA, Boyd IL, McConnell BJ 1993) Variation in weaning mass of pups in relation to maternal mass, postweaning fast duration, and weaned pup behaviour in southern elephant seals Mirounga leonina) at South Georgia. Can J Zool 71: 1772±1781 Arnbom T, Fedak MA, Rothery P 1994) O spring sex ratio in relation to female size in southern elephant seals, Mirounga leonina. Behav Ecol Sociobiol 35: 373±378 Arnbom T, Fedak MA, Boyd IL 1997) Factors a ecting maternal expenditure in southern elephant seals during lactation. Ecology 78: 471±483 Baker JD, Fowler CW 1992) Pup weight and survival of northern fur seals Callorhinus ursinus. J Zool Lond 227: 231±238 Bartsh SS, Johnston SD, Sini DB 1992) Territorial behavior and breeding frequency of male Weddell seals Leptonychotes weddelli) in relation to age, size, and concentrations of serum testosterone and cortisol. Can J Zool 70: 680±692 Bell CM, Hindell MA, Burton HR 1997) Estimation of body mass in the southern elephant seal, Mirounga leonina, by photogrammetry and morphometrics. Mar Mamm Sci 13: 669±682 Born EW, Riget FF, Dietz R, Andriashek D 1999) Escape responses of hauled out ringed seals Phoca hispida) to aircraft disturbance. Polar Biol 21: 171±178 Brasseur SMJM 1993) Tolerance of harbour seals to human related disturbance sources abstract). Tenth Biennial Conference on the Biology of Marine Mammals, Galveston, Texas, 11±15 November 1993 Burton HR, Arnbom T, Boyd IL, Bester M, Vergani D, Wilkinson I 1997) Signi cant di erences in weaning mass of southern elephant seals from ve sub-antarctic islands in relation to population declines. In: Battaglia B, Valencia J, Walton DWH eds) Antarctic communities: species structure and survival. Cambridge University Press, Cambridge, pp 335±338 Carlini AR, Daneri GA, Marquez MEI, Soave GE, Poljak S 1997) Mass transfer from mothers to pups and mass recovery by mothers during the post-breeding foraging period in southern elephant seals Mirounga leonina) at King George Island. Polar Biol 18: 305±310 Carrick R, Ingham SE 1962) Studies on the southern elephant seal, Mirounga leonina L.). I. Introduction to the series. CSIRO Wildl Res 7: 89±101 Carrick R, Csordas SE, Ingham SE 1962a) Studies on the southern elephant seal, Mirounga leonina L.). IV. Breeding and development. CSIRO Wildl Res 7: 161±197 Carrick R, Csordas SE, Ingham SE, Keith K 1962b) Studies on the southern elephant seal, Mirounga leonina L.). III. The annual cycle in relation to age and sex. CSIRO Wildl Res 7: 119±160 Cobley ND, Shears JR 1999) Breeding performance of gentoo penguins Pygoscelis papua) at a colony exposed to high levels of human disturbance. Polar Biol 21: 355±360 Costa DP 1991) Reproductive and foraging energetics of pinnipeds: implications for life history patterns. In: Renouf D ed) Behaviour of pinnipeds. Chapman & Hall, London, pp 300±344 Cox CR 1981) Agonistic encounters among male elephant seals: frequency, context, and the role of female preference. Am Zool 21: 197±209 Craig MP, Ragen TJ 1999) Body size, survival, and decline of juvenile Hawaiian monk seals, Monachus schauinslandi. Mar Mamm Sci 15: 786±809 Culik BM, Wilson RP 1995) Penguins disturbed by tourists. Nature 376: 301±302 Fedak MA, Arnbom T, Boyd IL 1996) The relation between the size of elephant seal mothers, the growth of their pups, and the use of maternal energy, fat, and protein during lactation. Physiol Zool 69: 887±911 Giese M 1996) E ects of human activity on AdeÂlie penguin Pygoscelis adeliae breeding success. Biol Conserv 75: 157±164 Giese M 1998) Guidelines for people approaching breeding groups of AdeÂlie penguins Pygoscelis adeliae). Polar Rec 34: 287±292 Guinet C, Jouventin P, Weimerskirch H 1999) Recent population change of the southern elephant seal at IÃles Crozet and IÃles Kerguelen: the end of the decrease? Antarct Sci 11: 193±197 Haas CA 1998) E ects of prior nesting success on site delity and breeding dispersal: an experimental approach. Auk 115: 929±936 Haley MP, Deutsch CJ, Le Boeuf BJ 1991) A method for estimating mass of large pinnipeds. Mar Mamm Sci 7: 157±164 Haley MP, Deutsch CJ, Le Boeuf BJ 1994) Size, dominance and copulatory success in male northern elephant seals, Mirounga angustirostris. Anim Behav 48: 1249±1260 Hindell MA 1991) Some life-history parameters of a declining population of southern elephant seals, Mirounga leonina. J Anim Ecol 60: 119±134 Hindell MA, Burton HR 1987) Past and present status of the southern elephant seal Mirounga leonina) at Macquarie Island. J Zool Lond 213: 365±380 Hindell MA, Little GJ 1988) Longevity, fertility and philopatry of two female southern elephant seals Mirounga leonina) at Macquarie Island. Mar Mamm Sci 4: 168±171 Hindell MA, Burton HR, Slip DJ 1991) Foraging areas of southern elephant seals, Mirounga leonina, as inferred from water temperature data. Aust J Mar Freshwater Res 42: 115±128 Hindell MA, Slip DJ, Burton HR 1994) Possible causes of the decline of southern elephant seal populations in the Southern Paci c and Southern Indian Oceans. In: Le Boeuf BJ, Laws RM eds) Elephant seals: population ecology, behavior, and physiology. University of California Press, Berkeley, pp 66±84 Kirkman SP 1999) The seasonal haulout cycle of the declining southern elephant seal, Mirounga leonina, population at Marion Island. MSc Thesis, University of Pretoria Laws RM 1953) The elephant seal Mirounga leonina Linn.). I. Growth and age. Falkland Isl Depend Surv Sci Rep 8: 1±62 Laws RM 1956) The elephant seal Mirounga leonina Linn.). II. General, social and reproductive behaviour. Falkland Isl Depend Surv Sci Rep 13: 1±88 Laws RM 1994) History and present status of southern elephant seal populations. In: Le Boeuf BJ, Laws RM eds) Elephant seals: population ecology, behavior, and physiology. University of California Press, Berkeley, pp 49±65 Le Boeuf BJ, Briggs KT 1977) The cost of living in a seal harem. Mammalia 41: 167±195 Le Boeuf BJ, Morris P, Reiter J 1994) Juvenile survivorship of northern elephant seals. In: Le Boeuf BJ, Laws RM eds) Elephant seals: population ecology, behavior, and physiology. University of California Press, Berkeley, pp 121±136 McCann TS 1981) Aggression and sexual activity of male southern elephant seals, Mirounga leonina. J Zool Lond 195: 295±310 McCann TS 1982) Aggressive and maternal activities of female southern elephant seals Mirounga leonina). Anim Behav 30: 268±276 McCann TS, Fedak MA, Harwood J 1989) Parental investment in southern elephant seals, Mirounga leonina. Behav Ecol Sociobiol 25: 81±87 McMahon CR, Ho J van den, Burton HR, Davis PD 1997) Evidence for precocious development in female pups of the southern elephant seal Mirounga leonina at Macquarie Island.

10 251 In: Hindell M, Kemper C eds) Marine mammal research in the southern hemisphere, vol 1. Status, ecology and medicine. Surrey Beatty, Chipping Norton, pp 92±96 McMahon CR, Burton HR, Bester MN 1999) First-year survival of southern elephant seals, Mirounga leonina, at sub-antarctic Macquarie Island. Polar Biol 21: 279±284 McMahon CR, Burton HR, Bester MN 2000) Weaning mass and the future survival of juvenile southern elephant seals, Mirounga leonina, at Macquarie Island. Antarct Sci 12: 149±153 Modig AO 1996) E ects of body size and harem size on male reproductive behaviour in the southern elephant seal. Anim Behav 51: 1295±1306 Murphy MT 1996) Survivorship, breeding dispersal and mate delity in eastern kingbirds. Condor 98: 82±92 Nicholls DG 1970) Dispersal and dispersion in relation to birth site in the southern elephant seal, Mirounga leonina L.), of Macquarie Island. Mammalia 34: 598±616 Pistorius PA, Bester MN, Kirkman SP 1999) Survivorship of a declining population of southern elephant seals, Mirounga leonina, in relation to age, sex and cohort. Oecologia 121: 201±211 Pomeroy PP, Fedak MA, Rothery P, Anderson S 1999) Consequences of maternal size for reproductive expenditure and pupping success of grey seals at North Rona, Scotland. J Anim Ecol 68: 235±253 Regel J, PuÈ tz K 1997) E ect of human disturbance on body temperature and energy expenditure in penguins. Polar Biol 18: 246±253 Reijnders PJH, Brasseur SMJM, Van der Toorn J, Van der Wolf P, Boyd I, Harwood J, Lavigne D, Lowry L 1993) Seals, fur seals, sea lions, and walrus: status survey and conservation plan. IUCN, Gland Reijnders PJH, Verriopoulos G, Brasseur SMJM 1997) Status of pinnipeds relevant to the European Union. IBN Sci Contrib 8. DLO Institute for Forestry and Nature Research IBN-DLO), Wageningen Reiter J, Panken KJ, Le Boeuf BJ 1981) Female competition and reproductive success in northern elephant seals. Anim Behav 29: 670±687 Salter RE 1979) Site utilization, activity budgets, and disturbance responses of Atlantic walruses during terrestrial haul-out. Can J Zool 57: 1169±1180 Schjùrring S, Gregersen J, Bregnballe T 2000) Sex di erence in criteria determining delity towards breeding sites in the great cormorant. J Anim Ecol 69: 214±223 Selkirk PM, Seppelt RD, Selkirk DR 1990) Subantarctic Macquarie Island: environment and biology. Cambridge University Press, Cambridge Slip DJ, Hindell MA, Burton HR 1994) Diving behavior of southern elephant seals from Macquarie Island: an overview. In: Le Boeuf BJ, Laws RM eds) Elephant seals: population ecology, behavior, and physiology. University of California Press, Berkeley, pp 253±270 Streten NA 1988) The climate of Macquarie Island and its role in atmospheric monitoring. Pap Proc R Soc Tasmania 122: 91±106 Suryan RM, Harvey JT 1999) Variability in reactions of Paci c harbour seals, Phoca vitulina richardsi, to disturbance. Fish Bull 97: 332±339 Thiel M, Nehls G, BraÈ ger S, Meissner J 1992) The impact of boating on the distribution of seals and moulting ducks in the Wadden Sea of Schleswig-Holstein. In: Dankers N, Smit CJ, Scholl M eds) Proceedings of the 7th International Wadden Sea Symposium, Ameland, The Netherlands, 22±26 October Netherlands Institute for Sea Research, pp 221±233 Trites AW 1991) Does tagging and handling a ect the growth of northern fur seal pups Callorhinus ursinus)? Can J Fish Aquat Sci 48: 2436±2442 Wauters LA, Lens L, Dhondt AA 1995) Variation in territory delity and territory shifts among red squirrel, Sciurus vulgaris, females. Anim Behav 49: 187±193 Wilkinson IS, Bester MN 1988) Is onshore human activity a factor in the decline of the southern elephant seal? SAfr J Antarct Res 18: 14±17 Wilson RP, Culik BM, Dannfeld R, Adelung D 1991) People in Antarctica ± how much do AdeÂlie penguins Pygoscelis adeliae care? Polar Biol 11: 363±370 Zar J 1984) Biostatistical analysis. Prentice-Hall, Englewood Cli s, NJ

11 Comparative Biochemistry and Physiology Part A 133 (2002) Blood chemistry in southern elephant seal mothers and pups during lactation reveals no effect of handling Georg H. Engelhard *, Ailsa J. Hall, Sophie M.J.M. Brasseur, Peter J.H. Reijnders a,b,1, c a a a Alterra Institute, Marine and Coastal Zone Research, P.O. Box 167, 1790 AD Den Burg, The Netherlands b Centre for Ecological and Evolutionary Studies, Groningen University, P.O. Box 14, 9750 AA Haren, The Netherlands c Sea Mammal Research Unit, Gatty Marine Laboratory, University of St Andrews, Fife KY16 8LB Scotland, UK Received 5 February 2002; received in revised form 25 May 2002; accepted 27 May 2002 Abstract Serum clinical chemistry parameters were examined in lactating southern elephant seal Mirounga leonina mothers and their pups from the declining Macquarie Island population. There were significant changes in serum values from 2 to 21 days postpartum in both nursing mothers (increase: inorganic phosphate; decrease: creatinine, potassium, chloride, cholesterol, total protein, albumin, globulin, aspartate aminotransferase, creatine kinase) and suckling pups (increase: inorganic phosphate, globulin, cholesterol; decrease: albumin, alkaline phosphatase, gammaglutamyl transferase; increase followed by decrease: triglycerides, iron). We found no evidence that changes were due to chronic stress effects caused by repeated chemical immobilisations (mothers) or physical restraint (pups): at late lactation, clinical chemistry values were similar for mother pup pairs of a control group (not handled previously), moderate treatment group (previously handled twice) and high treatment group (previously handled three to four times). We were not able to detect differences in clinical chemistry values between mother pup pairs distributed over two areas differing in the frequency of human visits. The clinical chemistry values presented here can serve as reference ranges to allow future comparison with other southern elephant seal populations to investigate factors, e.g. food limitation, suspected to be involved in population declines Elsevier Science Inc. All rights reserved. Keywords: Chemical immobilisation; Clinical chemistry; Health status; Human disturbance; Lactation; Mirounga leonina; Physical restraint; Pinnipedia 1. Introduction Many studies on marine mammals require some degree of animal handling, including capture and restraint (Gales, 1989; Lynch et al., 1999; McMahon et al., 2000). Since such procedures are potentially stressful for the animals involved (St. Aubin et al., 1979; Dierauf, 1990; Engelhard et *Corresponding author. Tel.: q ; fax: q address: georg.engelhard@imr.no (G.H. Engelhard). 1 Present address: Institute of Marine Research, P.O. Box 1870 Nordnes, N-5817 Bergen, Norway. al., 2002), it is important to examine their consequences for the health status and condition of individuals and, ultimately, on survival and reproduction. Effort can then be put into minimising any negative impacts. One way of examining health status is by assessing clinical chemistry values in the blood (Reijnders, 1988; Bossart and Dierauf, 1990; Roletto, 1993; de Swart et al., 1995; Fadely, 1997; Hall, 1998). Since the 1950s, populations of the southern elephant seal Mirounga leonina have declined over portions of their range in the Southern Ocean, particularly in the Indian and Pacific sectors (Bar /02/$ - see front matter 2002 Elsevier Science Inc. All rights reserved. PII: S Ž

12 368 G.H. Engelhard et al. / Comparative Biochemistry and Physiology Part A 133 (2002) rat and Mougin, 1978; Hindell and Burton, 1987; Guinet et al., 1999; Slip and Burton, 1999). Despite substantial research effort, the factors causing these declines remain unidentified. Food limitation in the pelagic foraging areas appears to be a possible cause (Burton, 1986); in addition, local factors influencing populations could be involved (Hindell et al., 1994b; Pistorius et al., 1999). This study is part of a project examining to what extent human disturbance, including the activities of researchers, might be a factor in the declines (Engelhard et al., 2001a), or might bias the baseline population values being studied (e.g. physiology, behaviour) through stress effects. The ultimate goal is to obtain indicators for impact of human disturbance and use those for nature management purposes by assessing and monitoring human impact in Antarctica. Based on census data, human disturbance at elephant seal breeding locations had previously been rejected as a significant factor in elephant seal population declines (Wilkinson and Bester, 1988). However, since that study, human presence in the (sub) Antarctic has increased substantially (Enzenbacher, 1992; IAA- TO, 2000) and southern elephant seals have continued to be the subject of considerable scientific investigation (e.g. Ling and Bryden, 1992; Le Boeuf and Laws, 1994). Much of that research was carried out during breeding seasons (September November) and necessitated either physical restraint or chemical immobilisation of study animals. Here, we examine whether different degrees of handling disturbance may affect the health status in lactating elephant seal mothers and their pups, as indicated by a set of standard clinical blood chemistry parameters. The handling procedures consisted of chemical immobilisation of adult females and physical restraint of pups. These activities are representative of a number of studies on this and several other pinniped species (Gales, 1989; Le Boeuf and Laws, 1994; Lynch et al., 1999). A companion paper examines the direct cortisol stress response to handling activities (Engelhard et al., 2002). During the lactation period, which on average lasts 24 days (McMahon et al., 1997), elephant seal mothers do not feed or drink, relying entirely on stored energy reserves (McCann et al., 1989; Fedak et al., 1996). In the same period, their pups may grow rapidly solely through the intake of milk, often gaining over 100 kg from birth to weaning (Arnbom et al., 1997; McMahon et al., 1997). In the light of these important, natural changes in the physiology of both mothers and pups during the lactation period, firstly, we examine how clinical chemistry parameters may normally vary over the course of lactation. Secondly, we examine whether different degrees of research handling procedures may affect the clinical chemistry values of mothers and pups. Thirdly, we compare clinical chemistry values between mother pup pairs from two areas similar in natural features but widely different in the general levels of human presence including research activities. 2. Materials and methods 2.1. Location Fieldwork was carried out on Macquarie Island (548309S, E), the breeding location for a southern elephant seal population that is in longterm decline (Hindell and Burton, 1987; McMahon et al., 1999). The Australian Antarctic Division established a research station on the island in 1948; since that year the station has been permanently manned by members of the Australian National Antarctic Research Expeditions (A.N.A.R.E.) Areas of high and low human presence During September November 1996 we monitored eight elephant seal harems situated on the east coast of the northern portion of the island. Four harems were on the eastern beaches of the Isthmus (Isthmus East); and four were 2 km to the south on Middle Beach. The peak numbers of females present in study harems (counted 16 October 1996) were not significantly different between these areas (Middle Beach, from south to north 137, 161, 54 and 28 females; Isthmus East, from south to north 78, 260, 360 and 150 females; Mann Whitney, Us3.00, Ps0.149). In addition, the two areas were similar in beach topography, wind and surf exposure. However, the levels of human presence were very different between Isthmus East and Middle Beach, being areas of high and low human presence, respectively (Engelhard et al., 2001a). Since the permanent research station is located on the Isthmus, most human activity (scientific, maintenance, sightseeing and other) has traditionally been at or near that site. Other parts

13 G.H. Engelhard et al. / Comparative Biochemistry and Physiology Part A 133 (2002) of the island, including Middle Beach, are visited far less frequently. In addition, during the study season elephant seal harems on the Isthmus were visited daily by a second, four-person team carrying out long-term population monitoring studies. Team members weighed and tagged pups within a day of birth and re-weighed these at weaning (methods as in McMahon et al., 1997); this sample included approximately 500 pups on Isthmus East. These procedures were not carried out on Middle Beach. The amount of disturbance in harems due to research activities (see below) of our own, separate team comprising four persons, was approximately equal for the two study areas Control, moderate and high treatment groups Twenty-six mother pup pairs distributed randomly over the eight study harems and the areas of low and high human presence, were either given low, moderate, or high degrees of experimental disturbance, which consisted of chemical restraint of mothers and physical restraint of pups (Engelhard et al., 2002). As a control group, 13 mother pup pairs were only captured once at approximately the 21st day postpartum, towards the very end of the lactation period. As an experimental group, 13 pairs were handled repeatedly (three to five times) at early, middle and late stages of lactation. Within this experimental group, five pairs captured three times (approx. 2, 11 and 21 days postpartum) are considered as the moderate treatment group; and eight pairs captured four to five times (approx. 2, 11, 14, w18x and 21 days postpartum) are considered as the high treatment group. For individual recognition, each of the study animals (control and experimental) was given a paint mark on one of the first days of lactation Handling procedures Adult females were chemically immobilised with a 1:1 tiletamine zolazepam mixture (Zoletil 100, Virbac), following Baker et al. (1990). We used a combined dose rate of 1 mg per kg of estimated seal body mass (up to a maximum of 450 mg). The anaesthetic was prepared at a site distant from the harem; one person then quickly approached the animal and, by means of an anaesthetic dart delivered by blowpipe, administered the drug intramuscularly into the pelvic region (Fedak et al., 1996). Once the animal was sedated, blood samples were collected from the extradural vein (Geraci and Smith, 1975), by means of G Yale spinal needles into 6 ml untreated Plain Plus vacutainers (Beckton Dickinson UK, Ltd.). After the completion of blood sampling we measured nose-to-tail length and weighed the female in a net attached to a load cell balance (Salter DC2, West Bromwich, UK), slung by a tripod. Simultaneously, the pup was physically restrained in either a pup bag or (if heavier than approx. 75 kg) a pole net (Laws 1953). Blood samples were collected as in mothers, using G needles. In addition, nose-to-tail length and mass were measured. After sampling, both mother and pup were released. They were monitored until the mother had fully recovered from the anaesthesia and mother pup contact was reestablished Analysis of blood samples Blood samples were allowed to clot for approximately 5 15 h at 4 8C while kept in the dark. The serum was then removed and stored in 2-ml cryotubes at y20 8C. The samples were kept frozen during transport until laboratory analysis, which was carried out by BCO Analytical Services, Breda, The Netherlands. Clinical chemistry parameters were analysed in serum using a Kodak Ectachem dry-chemistry analyser (Vitros 750 XRC). The machine measured different clinical parameters in one sample, and was run at standard conditions for human samples, using an incubation temperature of 37 8C for the enzymatic analyses and 25 8C for the colorimetric analyses. Analytes q measured were: urea; creatinine; sodium (Na ); q y potassium (K ); ionic chloride (Cl ); calcium 2q 3y (Ca ); inorganic phosphate (PO 4 ); bilirubin; triglycerides; cholesterol; total protein; albumin; globulin (calculated by subtraction of albumin from total protein); glucose; aspartate aminotransferase (ASAT); alanine aminotransferase (ALAT); lactate dehydrogenase (LD); alkaline phosphatase (AP); gammaglutamyl transferase (GGT); creatinine kinase (CK); and iron (Fe ). The interpre- 3q tation of the values for some of these analytes requires some caution, as the samples were allowed to clot for different lengths of time before removal of serum due to practical limitations; this could q 3y have resulted in levels of K, PO 4, LD and GGT y artifactually elevated and levels of Cl and glu-

14 370 G.H. Engelhard et al. / Comparative Biochemistry and Physiology Part A 133 (2002) cose artifactually reduced to some extent, due to exposure to red blood cells andyor haemolysis (Bossart and Dierauf, 1990; Morgan et al., 1998). However, comparisons with previously published values indicated no serious biases in our results (e.g. Bossart and Dierauf, 1990; Roletto, 1993; de Swart et al., 1995; Fadely, 1997). Moreover, blood samples representing different treatment groups and different study areas were analysed using consistent methods Statistical analyses Statistical analyses followed Zar (1984) and were carried out using the SPSS package (SPSS Inc., ). The effects of stage of lactation, treatment and area on blood composition were tested using multivariate general linear models. As the dependent variables, all examined clinical chemistry parameters were included, after log transformation of the data to increase the homogeneity of the variances. We used multivariate tests with Pillai s trace statistic for examining effects on overall blood composition, and univariate tests for examining effects on single blood parameters. In multivariate tests, values of P were considered to denote statistical significance. In assessing the separate effects on each single blood parameter, a partial Bonferroni correction was applied since no less than 21 parameters were tested. The mean correlation among all blood parameters was 0.21 for the mothers and 0.24 for the pups. Therefore, in univariate tests, values of P were considered to denote statistical significance. 3. Results 3.1. Change in blood parameters over the course of lactation The lactation period for the elephant seal mothers and pups in this study averaged 23.5"2.3 days (mean"s.d.; range days, ns26). Changes in blood parameters over the course of lactation were examined for mother pup pairs in the experimental group, where longitudinal data were available (ns13 pairs). During this period, the overall composition of clinical chemistry values changed significantly both in the mothers (Table 1; Pillai s trace F40,12s3.736, Ps0.015) and in the pups (Table 2; Pillai s trace F s7.049, Ps0.001). 40,12 In nursing mothers, the levels of creatinine, q y K, Cl, cholesterol, total protein, albumin, glob- ulin, ASAT and CK decreased during lactation, although many of these parameters only changed during either the first or second half of this period (Table 1). There was an increase in the levels of 3y PO4 towards the end of lactation. The other parameters examined showed no significant change. 3y In suckling pups, the levels of PO 4, globulin and, in particular, cholesterol increased over the same period (Table 2). Those of albumin, AP and GGT decreased. Serum triglycerides and Fe 3q levels were highest during the second week of lactation. None of the remaining parameters changed significantly during this period. Male and female pups were not significantly different in any of the studied blood parameters Comparison between treatment groups Table 3 shows clinical chemistry values for mothers at late lactation that either belonged to the control or to the experimental (moderate and high treatment) groups. None of the blood parameters, when examined separately, was significantly different between the mothers in the control and experimental groups; neither was there a difference in overall blood composition between these groups (Pillai s trace F21,4s4.36, Ps0.081). When the control group was compared with the high treatment group only, again no significant differences in any of the parameters were found. In addition, the overall blood composition was similar for control and high treatment mothers (Pillai s trace F19,1s0.893, Ps0.697). We also tested for differences in clinical chemistry values between 21 day-old pups belonging to the different treatment groups (Table 4). When the control group was compared with the experimental group, no significant differences in single blood parameters nor in overall blood composition (Pillai s trace F21,4s1.106, Ps0.521) were found. When the comparison was restricted to the control and high treatment group only, again no differences in single blood parameters nor in overall blood composition (Pillai s trace F19,1s1.053, Ps 0.658) were found Comparison between areas of high and low human presence Clinical chemistry values were compared between 11 mother pup pairs in the remote area

15 G.H. Engelhard et al. / Comparative Biochemistry and Physiology Part A 133 (2002) Table 1 Clinical chemistry values (mean"s.d., with total ranges in brackets) in adult female elephant seals during early, middle and late stages of the lactation period (respectively, 2, 11 and 21 days postpartum) Parameter Unit Early stage Middle stage Late stage Test statistics (ns13) (ns13) (ns26) F 2,24 P Contrasts Univariate statistics Urea mmolyl 11.0" " " ( ) ( ) ( ) Creatinine mmolyl 138"21 137"15 116" efm)l (89 169) ( ) (95 142) Na q mmolyl 146"1 147"1 145" ( ) ( ) ( ) K q mmolyl 4.0" " " efm)l ( ) ( ) ( ) Cl y mmolyl 102"1 100"1 99" e)m)l ( ) (99 102) (97 103) Ca 2q mmolyl 2.35" " " ( ) ( ) ( ) PO 3y 4 mmolyl 1.61" " " efm-l ( ) ( ) ( ) Bilirubin mmolyl 16"9 12"3 14" (9 37) (8 18) (7 34) Triglycerides mmolyl 0.5" " " ( ) ( ) ( ) Cholesterol mmolyl 8.8" " " e)mfl ( ) ( ) ( ) Total protein gyl 77"6 78"5 71" efm)l (69 95) (71 89) (63 80) Albumin gyl 34"2 35"2 33" efm)l (31 39) (31 38) (29 38) Globulin gyl 43"5 43"4 38" efm)l (37 56) (38 54) (33 46) Glucose mmolyl 7.2" " " ( ) ( ) ( ) ASAT Uyl 73"13 52"9 43" e)m)l (49 94) (40 76) (24 63) ALAT Uyl 14"11 29"20 20" (3 33) (3 73) (3 51) LD Uyl 1249" " " ( ) ( ) ( ) AP Uyl 76"27 63"24 63" (31 120) (26 95) (25 131) GGT Uyl 11"5 10"5 14" (5 21) (5 19) (5 31) CK Uyl 1096" " " e)mfl ( ) ( ) ( ) Fe 3q mmolyl 33.6" " " ( ) ( ) ( ) Multivariate statistics F 40,12 P Combined parameters Data on early and middle lactation represent the experimental group only; data on late lactation represent the combined experimental and control groups (see Table 3 for comparison between treatment groups). Within the experimental group, the effect of lactation stage was analysed using repeated measures ANOVA, after log transformation of data. Univariate tests show the effects for single parameters; multivariate tests show the combined effect for all parameters. Significant values of P are shown in bold type.

16 372 G.H. Engelhard et al. / Comparative Biochemistry and Physiology Part A 133 (2002) Table 2 Clinical chemistry values (mean"s.d., with total ranges in brackets) in elephant seal pups during early, middle and late stages of the lactation period (respectively, 2, 11 and 21 days postpartum) Parameter Unit Early stage Middle stage Late stage Test statistics (ns13) (ns13) (ns26) F 2,24 P Contrasts Univariate statistics Urea mmolyl 11.1" " " ( ) ( ) ( ) Creatinine mmolyl 95"9 87"12 91" (79 112) (54 98) (62 126) Na q mmolyl 147"2 145"2 145" ( ) ( ) ( ) K q mmolyl 4.3" " " ( ) ( ) ( ) Cl y mmolyl 99"2 98"2 98" (97 102) (94 101) (96 104) Ca 2q mmolyl 2.71" " " ( ) ( ) ( ) PO 3y 4 mmolyl 2.09" " " e-m-l ( ) ( ) ( ) Bilirubin mmolyl 8"4 7"4 10" (4 20) (4 18) (4 58) Triglycerides mmolyl 0.8" " " e-m)l ( ) ( ) ( ) Cholesterol mmolyl 3.3" " " e-m-l ( ) ( ) ( ) Total protein gyl 62"3 62"6 69" (58 68) (55 72) (53 96) Albumin gyl 39"3 33"3 33" e)mfl (35 45) (29 37) (24 47) Globulin gyl 23"1 29"4 36" e-m-l (21 26) (23 35) (25 50) Glucose mmolyl 7.9" " " ( ) ( ) ( ) ASAT Uyl 63"12 67"19 67" (45 85) (47 110) (41 134) ALAT Uyl 31"16 32"15 35" (10 61) (3 57) (3 76) LD Uyl 992" " " ( ) ( ) ( ) AP Uyl 466" " " e)m)l ( ) ( ) (93 431) GGT Uyl 17"6 20"8 12" efm)l (7 27) (5 39) (5 46) CK Uyl 441" " " ( ) ( ) ( ) Fe 3q mmolyl 37.3" " " e-m)l ( ) ( ) ( ) Multivariate statistics F 40,12 P Combined parameters Data on early and middle lactation represent the experimental group only; data on late lactation represent the combined experimental and control groups (see Table 4 for comparison between treatment groups). Within the experimental group, the effect of lactation stage was analysed using repeated measures ANOVA, after log transformation of data. Univariate tests show the effects for single parameters; multivariate tests show the combined effect for all parameters. Significant values of P are shown in bold type.

17 G.H. Engelhard et al. / Comparative Biochemistry and Physiology Part A 133 (2002) Table 3 Comparison of clinical chemistry values (mean"s.d.) between adult female elephant seals of the control and experimental (moderate and high treatment) groups Parameter Unit Control Experimental Control vs. Control vs. (ns13) (moderateq high high) Moderate tr. High tr. F P F P (ns5) (ns8) Univariate statistics Urea mmolyl 10.1" " " Creatinine mmolyl 118"12 116"5 112" Na q mmolyl 145"2 145"1 146" K q mmolyl 3.8" " " Cl y mmolyl 100"2 98"1 99" Ca 2q mmolyl 2.34" " " PO 3y 4 mmolyl 1.77" " " Bilirubin mmolyl 13"4 12"3 16" Triglycerides mmolyl 0.5" " " Cholesterol mmolyl 7.8" " " Total protein gyl 70"4 69"6 73" Albumin gyl 33"2 31"2 33" Globulin gyl 38"4 38"5 40" Glucose mmolyl 8.0" " " ASAT Uyl 44"9 44"7 42" ALAT Uyl 20"13 24"17 18" LD Uyl 1483" " " AP Uyl 60"18 64"20 67" GGT Uyl 15"7 15"3 11" CK Uyl 225"98 205"68 260" Fe 3q mmolyl 45.9" " " Multivariate statistics F 21,4 P F 19,1 P Combined parameters All females were at the late stage of lactation (approx. 21 days postpartum). The effect of previous treatment was tested by comparing the control group with the combined moderate and high treatment groups; and by comparing the control and high treatment groups. Data were log transformed for statistical analysis. Univariate tests show the effects for single parameters; multivariate tests show the combined effect for all parameters. of Middle Beach and 15 pairs in the frequently visited area of Isthmus East (Table 5). Mothers in areas of high and low human presence showed no significant differences in any of the blood parameters examined; they were similar in overall blood composition (Pillai s trace F21,4s1.072, Ps 0.536). In accordance, pups in the two areas showed no significant differences in any of the blood parameters examined and were similar in overall blood composition (Pillai s trace F21,4s 2.187, Ps0.234). 4. Discussion The clinical chemistry values reported in this study are comparable to those described previously for this and other phocid species (southern elephant seal: Lane et al., 1972; Ramdohr et al., 1998; northern elephant seal Mirounga angustirostris: Roletto, 1993; harbour seal Phoca vitulina: Engelhardt, 1979; Roletto, 1993; de Swart et al., 1995; Schumacher et al., 1995; Fadely, 1997; harp seal Phoca groenlandica: Vallyathan et al., 1969; Engelhardt, 1979; Worthy and Lavigne, 1982; Bossart and Dierauf, 1990; grey seal Halichoerus grypus: Greenwood et al., 1971; Hall, 1998; Weddell seal Leptonychotes weddellii: Schumacher et al., 1992). However, the levels of creatinine reported here for adult female elephant seals (range mmolyl) were relatively high compared to those published for closely related species (; mmolyl: Worthy and Lavigne, 1982; Roletto, 1993; de Swart et al., 1995; Fadely, 1997). Creatinine is formed in the muscle tissue during the metabolism of creatine and phosphocreatine and enters the circulation only for transportation to the

18 374 G.H. Engelhard et al. / Comparative Biochemistry and Physiology Part A 133 (2002) Table 4 Comparison of clinical chemistry values (mean"s.d.) between elephant seal pups of the control and experimental (moderate and high treatment) groups Parameter Unit Control Experimental Control vs. Control vs. (ns13) Moderate tr. High tr. (moderateq high high) (ns5) (ns8) F 1,24 P F 1,19 P Univariate statistics Urea mmolyl 10.6" " " Creatinine mmolyl 86"13 98"20 96" Na q mmolyl 145"2 146"3 146" K q mmolyl 4.6" " " Cl y mmolyl 98"2 99"1 99" Ca 2q mmolyl 2.59" " " PO 3y 4 mmolyl 2.98" " " Bilirubin mmolyl 13"15 8"4 7" Triglycerides mmolyl 1.3" " " Cholesterol mmolyl 7.7" " " Total protein gyl 71"12 72"15 65" Albumin gyl 34"6 33"7 31" Globulin gyl 36"7 38"8 34" Glucose mmolyl 8.7" " " ASAT Uyl 67"14 70"36 66" ALAT Uyl 28"17 39"25 44" LD Uyl 1230" " " AP Uyl 188"81 252" " GGT Uyl 13"11 11"8 11" CK Uyl 611" " " Fe 3q mmolyl 63.0" " " Multivariate statistics F 21,4 P F 19,1 P Combined parameters All pups were at the late stage of lactation (age approx. 21 days). The effect of previous treatment was tested by comparing the control group with the combined moderate and high treatment groups; and by comparing the control and high treatment groups. Data were log transformed for statistical analysis. Univariate tests show the effects for single parameters; multivariate tests show the combined effect for all parameters. kidneys. Increased levels are used as an indicator of kidney disease (Bossart and Dierauf, 1990), but it must be considered unlikely that renal dysfunction occurred in all study females which would account for this finding. Contrary to this study, the other studies on creatinine in pinnipeds did not involve lactating females. The high levels could be the result of increased muscle catabolism during lactation; elevated creatinine levels during late pregnancy or lactation have been documented for other mammals (Doornenbal et al., 1988; Mbassa and Poulsen, 1991; El-Sherif and Assad, 2001). In the moderate and high treatment groups of mother pup pairs, many blood parameters changed significantly between 2 and 11 days postpartum, which coincided with the first and second handling events in these animals. Again, there were significant changes at 21 days postpartum, which in the moderate treatment group coincided with the third handling event, and in the high treatment group with either the fourth or fifth handling event (Tables 1 and 2). These changes in the parameters cannot be attributed to any possible stress effects due to repeated handling, or to a form of adaptation to the handling procedure: if this would be the case, then differences in blood values between the treatment groups at late lactation would be expected. But no such differences were found (Tables 3 and 4). Thus it is likely that the observed longitudinal trends were natural changes related to the stage of the lactation period. Lactation in elephant seals is short but intensive. Nursing mothers, without access to food or water, have to allocate their stored energy reserves between their own physiological maintenance and the synthesis of sufficient milk to allow rapid

19 G.H. Engelhard et al. / Comparative Biochemistry and Physiology Part A 133 (2002) Table 5 Comparison of clinical chemistry values (mean"s.d.) between elephant seal mother pup pairs in areas of low and high human presence Mothers Area of low Area of high F 1,24 P Area of low Area of high F 1,24 P human human human human presence presence presence presence (ns11) (ns15) (ns11) (ns15) Univariate statistics Urea a 9.8"2.2 10" " " Creatinine b 113"10 117" "12 87" Na qa 145"2 146" "1 145" K qa 3.6" " " " Cl ya 99"2 99" "2 99" Ca 2qa 2.34" " " " PO 3a " " " " Bilirubin b 15"7 13" "15 9" Triglycerides a 0.4" " " " Cholesterol a 7.1" " " " Total protein c 70"4 72" "10 65" Albumin c 32"3 33" "5 31" Globulin c 37"2 39" "5 34" Glucose a 7.7" " " " ASAT d 45"6 42" "17 68" ALAT d 15"11 23" "22 37" LD d 1607" " " " AP d 54"18 69" "87 208" GGT d 12"8 15" "7 12" CK d 223"51 238" " " Fe 3qb 50.5" " " " Multivariate statistics F 21,4 P F 21,4 P Combined parameters Mothers: Pups: All animals were at the late stage of lactation (approx. 21 days postpartum). Area differences were tested for mothers and pups separately using multiple ANOVA, after log transformation of the data. Univariate tests show the effects for single parameters; multivariate tests show the combined effect for all parameters. a In mmolyl. b In mmolyl. c In gyl. d In Uyl. Pups weight gain in the pup; in this process the pup may double to quadruple its birth weight at weaning (Costa et al., 1986; McCann et al., 1989; McMahon et al., 1997). Hence, changes in blood parameters over the lactation period, as found in both mothers and pups in this study, are not surprising. The mass loss in mothers includes approximately 61% of the fat, and 24% of the proteins initially present (Fedak et al., 1996), here reflected in decreasing serum levels of cholesterol at the early stage and of proteins at the late stage of lactation (Table 1). The earlier change in cholesterol than in protein levels indicates that the contribution of lean tissue catabolism to energy metabolism, when compared to adipose tissue catabolism, increases towards the end of the lactation fast. This is also suggested by the reduction in serum creatinine from middle to late lactation (Table 1), which may be due to a reduction in muscle mass at that stage. Previously, an increase in lean tissue catabolism over the nursing period has been shown for northern elephant seal females by Crocker et al. (1998), who suggested that the females have a reduced ability to spare proteins as lactation progresses due to the gradual depletion of fat reserves. The change in clinical chemistry values in the pups is in accordance with previous work on developmental changes in phocid species (e.g. Hall, 1998; Ramdohr et al., 1998). Southern elephant seal pups are born with almost no blubber; but, by the end of the lactation period, blubber

20 376 G.H. Engelhard et al. / Comparative Biochemistry and Physiology Part A 133 (2002) makes up nearly half of the body mass (Bryden, 1968; Hindell et al., 1994a). In the pups investigated, this was mirrored by a doubling of serum cholesterol concentrations from birth, when values were comparable to those in humans, to weaning (Table 2; Ramdohr et al., 1998). Proteins contribute a far lower proportion of the mass gain than fat (11 vs. 48%; Carlini et al., 2000). Although there was no trend in total serum protein levels, we observed a drop in albumin from 2 days to 11 days postpartum. This may have been attributable to thermoregulatory demands shortly after birth when the insulating blubber layer had not yet developed. Globulin levels increased throughout the suckling period indicating development of the immune system. The relative changes in albumin and globulin levels were comparable to those in suckling grey seal pups (Hall, 1998). The AP levels in the pups were high, particularly shortly after birth; this likely reflected the high osteoblastic activity related with bone development, characteristic of healthy growing juvenile mammals (Bossart and Dierauf, 1990). We were not able to detect any chronic effects of moderate to high degrees of handling on clinical chemistry values in southern elephant seal mothers and pups. However, the conclusion that these animals may tolerate high levels of human disturbance without negative health consequences should be drawn with caution. First, the sample sizes in this study were fairly small, so that possible, small-scale effects on serum values may have remained undetected that nevertheless could have significant health implications. Second, given the sampling intervals, it is possible that changes in clinical chemistry values did occur but homoeostatic mechanisms resolved these changes prior to the next sampling. Third, the absence of handling effects on clinical chemistry values agreed only partly with our findings on the adrenocortical stress response to capture and handling for the same study animals (Engelhard et al., 2002). Correspondingly, there was no difference in the cortisol response to the stress of physical restraint between pups of the three treatment groups; and there was no difference in the cortisol response to the stress of chemical immobilisation between control and moderate treatment mothers at late lactation. However, mothers of the high treatment group showed significantly attenuated cortisol responses; this suggested that the 3 4 immobilisations given previously to these females may have resulted in reduced adrenocortical responsiveness at late lactation (Engelhard et al., 2002). Despite this, the general health status of these animals as indicated by their serum chemistry values (this study) and, in addition, by body weight measures (Engelhard et al., 2002), appeared not to be significantly impaired due to the relatively high degree of treatment. Clinical chemistry values were similar for elephant seal mother pup pairs in the frequently visited area of Isthmus East and the remote area of Middle Beach. Cortisol responses were also similar for seals in these two areas (Engelhard et al., 2002). Based on a larger sample size, we found no difference in the mass of weaned pups other than that due to the size of the respective mothers (Engelhard et al., 2001a). No behavioural differences were observed between mother pup pairs in the two areas (Engelhard et al., 2001b). Overall, these findings give an indication that the human presence and ongoing research activities carried out at the station situated on the Isthmus do not pose a serious threat to the elephant seal population on Macquarie Island through increased disturbance (Wilkinson and Bester, 1988). There is suspicion that limited availability of food in the Indian and Pacific Sectors of the Southern Ocean, when compared to the Atlantic Sector, might be a cause of southern elephant seal population declines in that area (Burton, 1986; Hindell et al., 1994b; Pistorius et al., 1999). If poor nutritional status due to food limitation is indeed a factor, then this is likely to be reflected by differences in clinical chemistry values between declining and stable populations (Geraci et al., 1979; Worthy and Lavigne, 1982; Fadely, 1997; Rea et al., 1998). Extensive data on clinical chemistry are not yet available for any of the other southern elephant seal populations. The values presented here for lactating mothers and their pups from a population in decline may thus serve as reference ranges to allow future comparisons with other populations. Acknowledgments The study was a collaboration of the Alterra Institute (Marine and Coastal Zone Research, Project ), the Sea Mammal Research Unit and the Australian Antarctic Division (Human Impacts Project 1007 and Biological Sciences Project 2265); it was made possible by a grant from

21 G.H. Engelhard et al. / Comparative Biochemistry and Physiology Part A 133 (2002) the Netherlands Antarctica Program (NWO Project ). Fieldwork was carried out as part of the 49th Australian National Antarctic Research Expeditions. We thank the A.N.A.R.E. staff and participating expeditioners on Macquarie Island: Pirie Conboy, Rupert Davies, Paul Davis, Dave Holley, Frans Jonker, Mary Anne Lea, Margaret Morrice, and David Slip supported us in the field. John van den Hoff organised overseas transport of the samples. We thank Harry Burton, Jeroen Creuwels, Rudi Drent, Clive McMahon, Jan Verhees, Wim Wolff and two anonymous referees for valuable comments on the manuscript. The Tasmanian National Parks and Wildlife Service supplied permits to work on Macquarie Island. Our activities involving the handling of animals followed the Guidelines of the Australian Antarctic Animal Ethics Committee. References Arnbom, T., Fedak, M.A., Boyd, I.L., Factors affecting maternal expenditure in southern elephant seals during lactation. Ecology 78, Baker, J.R., Fedak, M.A., Anderson, S.S., Arnbom, T., Baker, R., Use of a tiletamine zolazepam mixture to immobilise wild grey seals and southern elephant seals. Vet. Rec. 126, Barrat, A., Mougin, J.L., L Elephant de mer Mirounga leonina de l ıle ˆ de la Possession, archipel Crozet ( S, E). Mammalia 42, Bossart, G.D., Dierauf, L.A., Marine mammal clinical laboratory medicine. In: Dierauf, L.A. (Ed.), CRC Handbook of Marine Mammal Medicine: Health, Disease and Rehabilitation. CRC Press, Boca Raton, Florida, pp Bryden, M.M., Growth and function of the subcutaneous fat of the elephant seal. Nature 220, Burton, H., A substantial decline in numbers of the southern elephant seal at Heard Island. Tasm. Nat. 96, 4 8. Carlini, A.R., Panarello, H.O., Marquez, M.E.I., Daneri, G.A., Soave, G.E., Energy gain and loss during lactation and postweaning in southern elephant seal pups (Mirounga leonina) at King George Island. Polar Biol. 23, Costa, D.P., Le Boeuf, B.J., Huntley, A.C., Ortiz, C.L., The energetics of lactation in the northern elephant seal, Mirounga angustirostris. J. Zool. Lond. 209, Crocker, D.E., Webb, P.M., Costa, D.P., Le Boeuf, B.J., Protein catabolism and renal function in lactating northern elephant seals. Physiol. Zool. 71, de Swart, R.L., Ross, P.S., Vedder, L.J., et al., Haematology and clinical chemistry values for harbour seals (Phoca vitulina) fed environmentally contaminated herring remain within normal ranges. Can. J. Zool. 73, Dierauf, L.A., Stress in marine mammals. In: Dierauf, L.A. (Ed.), CRC Handbook of Marine Mammal Medicine: Health, Disease and Rehabilitation. CRC Press, Boca Raton, Florida, pp Doornenbal, H., Tong, A.K., Murray, N.L., Reference values of blood parameters in beef cattle of different ages and stages of lactation. Can. J. Vet. Res. 52, El-Sherif, M.M.A., Assad, F., Changes in some blood constituents of Barki ewes during pregnancy and lactation under semi arid conditions. Small Rum. Res. 40, Engelhard, G.H., van den Hoff, J., Broekman, M., Baarspul, A.N.J., Field, I., Burton, H.R., Reijnders, P.J.H., Mass of weaned elephant seal pups in areas of low and high human presence. Polar Biol. 24, Engelhard, G.H., Creuwels, J.C.S., Broekman, M., Baarspul, A.N.J., Reijnders, P.J.H., Effect of human disturbance on lactation behaviour in the southern elephant seal. Abstract: VIII SCAR International Biology Symposium: Antarctic Biology in a Global Context, 27 August 1 September 2001, Vrije Universiteit, Amsterdam, The Netherlands. Engelhard, G.H., Brasseur, S.M.J.M., Hall, A.J., Burton, H.R., Reijnders, P.J.H., Adrenocortical responsiveness in southern elephant seal mothers and pups during lactation and the effect of scientific handling. J. Comp. Physiol. B. 172, Engelhardt, F.R., Haematology and plasma chemistry of captive pinnipeds and cetaceans. Aq. Mam. 7, Enzenbacher, D.J., Tourists in Antarctica: numbers and trends. Polar Rec. 28, Fadely, B.S., Investigations of harbour seal (Phoca vitulina) health status and body condition in the Gulf of Alaska. Ph.D Diss., Univ. Alaska, Fairbanks, Alaska. Fedak, M.A., Arnbom, T., Boyd, I.L., The relation between the size of southern elephant seal mothers, the growth of their pups and the use of maternal energy, fat and protein during lactation. Physiol. Zool. 69, Gales, N.J., Chemical restraint and anesthesia of pinnipeds: a review. Mar. Mamm. Sci. 5, Geraci, J.R., Smith, T.G., Functional hematology of ringed seals (Phoca hispida) in the Canadian Arctic. J. Fish. Res. Board Can. 410, Geraci, J.R., St. Aubin, D.J., Smith, T.G., Influence of age, condition, sampling time and method on plasma chemical constituents in free-ranging ringed seals, Phoca hispida. J. Fish. Res. Board Can. 36, Greenwood, A.G., Ridgway, S.H., Harrison, R.J., Blood values in young gray seals. J. Am. Vet. Med. Ass. 159, Guinet, C., Jouventin, P., Weimerskirch, H., Recent population change of the southern elephant seal at ˆIles Crozet and ˆIles Kerguelen: the end of the decrease? Antarct. Sci. 11, Hall, A.J., Blood chemistry and hematology of gray seal (Halichoerus grypus) pups from birth to postweaning. J. Zoo Wildl. Med. 29, Hindell, M.A., Burton, H.R., Past and present status of the southern elephant seal (Mirounga leonina) at Macquarie Island. J. Zool. Lond. 213, Hindell, M.A., Bryden, M.M., Burton, H.R., Early growth and milk composition in southern elephant seals (Mirounga leonina). Aust. J. Zool. 42, Hindell, M.A., Slip, D.J., Burton, H.R., Possible causes of the decline of southern elephant seal populations in the Southern Pacific and Southern Indian Oceans. In: Le Boeuf, B.J., Laws, R.M. (Eds.), Elephant Seals: Population Ecology, Behavior and Physiology. University of California Press, Berkeley, Los Angeles, pp

22 378 G.H. Engelhard et al. / Comparative Biochemistry and Physiology Part A 133 (2002) IAATO, Tourism Statistics: Antarctic Tourist Trends. International Association of Antarctica Tour Operators, stats.html. Lane, R.A.B., Morris, R.J.H., Sheedy, J.W., A haematological study of the southern elephant seal Mirounga leonina (Linn.). Comp. Biochem. Physiol. A 42, Laws, R.M., The elephant seal (Mirounga leonina Linn.). I. Growth and age. Falkl. Isl. Depend. Surv. Scient. Rep. 8, Le Boeuf, B.J., Laws, R.M., Elephant Seals: Population Ecology, Behavior and Physiology. University of California Press, Berkeley and Los Angeles. Ling, J.K., Bryden, M.M., Mirounga leonina. Mammalian Species 391, 1 8. Lynch, M.J., Tahmindjis, M.A., Gardner, H., Immobilisation of pinniped species. Aust. Vet. J. 77, Mbassa, G.K., Poulsen, J.S., Influence of pregnancy, lactation and environment on some clinical chemical reference values in Danish Landrace dairy goats (Capra hircus) of different parity. II. Plasma urea, creatinine, bilirubin, cholesterol, glucose and total serum proteins. Comp. Biochem. Physiol. B 100, McCann, T.S., Fedak, M.A., Harwood, J., Parental investment in southern elephant seals, Mirounga leonina. Behav. Ecol. Sociobiol. 25, McMahon, C.R., van den Hoff, J., Burton, H.R., Davis, P.D., Evidence for precocious development in female pups of the southern elephant seal Mirounga leonina at Macquarie Island. In: Hindell, M., Kemper, C. (Eds.), Marine Mammal Research in the Southern Hemisphere. Status, Ecology and Medicine, Vol. 1. Surrey Beatty, Chipping Norton, pp McMahon, C.R., Burton, H.R., Bester, M.N., First-year survival of southern elephant seals, Mirounga leonina, at sub-antarctic Macquarie Island. Polar Biol. 21, McMahon, C.R., Burton, H., McLean, S., Slip, D., Bester, M., Field immobilisation of southern elephant seals with intravenous tiletamine and zolazepam. Vet. Rec. 146, Morgan, L., Kumaresan, S., Thomas, C., MacWilliams, P., Hematology and chemistry reference values for freeranging harbor seals (Phoca vitulina) and the effects of hemolysis on chemistry values of captive harbor seals. J. Zoo Wildl. Med. 29, Pistorius, P.A., Bester, M.N., Kirkman, S.P., Survivorship of a declining population of southern elephant seals, Mirounga leonina, in relation to age, sex and cohort. Oecologia 121, Ramdohr, S., Plotz, J., Bornemann, H., Engelschalk, C., Thiery, J., Eisen, R., Studies on the lipoproteins of the southern elephant seal (Mirounga leonina) during the breeding season at King George Island. Ber. Polarf. 299, Rea, L.D., Castellini, M.A., Fadely, B.S., Loughlin, T.R., Health status of young Alaska Steller sea lion pups (Eumetopias jubatus) as indicated by blood chemistry and hematology. Comp. Biochem. Physiol. A 120, Reijnders, P.J.H., Ecotoxicological perspectives in marine mammalogy: research principles and goals for a conservation policy. Mar. Mamm. Sci. 4, Roletto, J.M.A., Hematology and serum chemistry values for clinically healthy and sick pinnipeds. J. Zoo Wildl. Med. 24, St. Aubin, D.J., Austin, T.P., Geraci, J.R., Effects of handling stress on plasma enzymes in harp seals, Phoca groenlandica. J. Wildl. Dis. 15, Schumacher, U., Rauh, G., Plotz, J., Welsch, U., Basic biochemical data on blood from Antarctic Weddell seals (Leptonychotes weddelli): ions, lipids, enzymes, serum proteins and thyroid hormones. Comp. Biochem. Physiol. A 102, Schumacher, U., Heidemann, G., Skırnisson, K., Schumacher, W., Pickering, R.M., Impact of captivity and contamination level on blood parameters of harbour seals (Phoca vitulina). Comp. Biochem. Physiol. A 112, Slip, D.J., Burton, H.R., Population status and seasonal haulout patterns of the southern elephant seal (Mirounga leonina) at Heard Island. Antarct. Sci. 11, Vallyathan, N.V., George, J.C., Ronald, K., The harp seal, Pagophilus groenlandicus (Erxleben, 1777). V. Levels of haemoglobin, iron, certain metabolites and enzymes in the blood. Can. J. Zool. 47, Wilkinson, I.S., Bester, M.N., Is onshore human activity a factor in the decline of the southern elephant seal? S. Afr. J. Antarct. Res. 18, Worthy, G.A.J., Lavigne, D.M., Changes in blood properties of fasting and feeding harp seal pups, Phoca groenlandica, after weaning. Can. J. Zool. 60, Zar, J., Biostatistical Analysis. Prentice Hall, Englewood Cliffs, New Jersey.

23 J Comp Physiol B (2002) 172: DOI /s ORIGINAL PAPER G.H. Engelhard Æ S.M.J.M. Brasseur Æ A.J. Hall H.R. Burton Æ P.J.H. Reijnders Adrenocortical responsiveness in southern elephant seal mothers and pups during lactation and the effect of scientific handling Accepted: 29 January 2002 / Published online: 11 April 2002 Ó Springer-Verlag 2002 Abstract We examined the cortisol responses to chemical and physical restraint stress in southern elephant seal Mirounga leonina females and their pups at three stages during lactation. In anaesthetised females the serum cortisol levels changed moderately during the 45-min sampling period following restraint, with average peaks at 23 min after anaesthetic administration. Overall, cortisol was relatively low 2 days postpartum and increased throughout lactation. In physically restrained pups serum cortisol increased rapidly after capture; the response was milder at age 2 days than at 11days and 21days. Levels were higher in female pups than in males. In order to test whether cortisol levels and/or responses became chronically (i.e. days to weeks) altered due to restraint, we compared the cortisol response at a late stage of lactation between three groups of mother-pup pairs previously given different levels of chemical (mothers) or physical (pups) restraint stress: control (not handled previously), moderate treatment (previously handled twice), and high treatment (previously handled 3 4 times). Pups of the three treatment groups showed similar adrenocortical responses suggesting no chronic effect of repeated physical restraint, despite the clear acute effects. Mothers of the control and moderate treatment groups showed similar cortisol responses; however, mothers of the high treatment group showed significantly attenuated responses. This indicated that elephant seals tolerated moderate degrees of handling disturbance; however, repeated (3 4) chemical immobilisations in lactating females may reduce their adrenocortical responsiveness for a period of days or weeks. Keywords Cortisol Æ Hypothalamic-pituitary-adrenal axis Æ Anaesthesia Æ Lactation Æ Mirounga leonina Abbreviation HPA hypothalamic-pituitary-adrenal Communicated by G. Heldmaier G.H. Engelhard (&) Centre for Ecological and Evolutionary Studies, Zoological Laboratory, Groningen University, P.O. Box 14, 9750 AA Haren, The Netherlands Georg.Engelhard@imr.no Tel.: Fax: G.H. Engelhard Æ S.M.J.M. Brasseur Æ P.J.H. Reijnders Alterra Institute, Marine and Coastal Zone Research, P.O. Box 167, 1790 AD Den Burg, The Netherlands A.J. Hall Sea Mammal Research Unit, Gatty Marine Laboratory, University of St Andrews, Fife KY16 8LB Scotland, UK H.R. Burton Australian Antarctic Division, Channel Highway, Kingston 7050, Tasmania, Australia Present address: G.H. Engelhard Institute of Marine Research, P.O. Box 1870 Nordnes, 5817 Bergen, Norway Introduction In pinniped research a certain degree of animal handling can in many cases not be avoided. For the collection of physiological data, or for the attachment of telemetry devices needed for studies on diving behaviour, either physical restraint by means of nets or chemical restraint by means of anaesthetics is usually required (reviewed by Gales 1989; Lynch et al. 1999). It is important to understand the stressful effects of research handling activities, both direct or acute (during or immediately after handling) and indirect or chronic (days to months after the event). It will then be easier to assess to what extent the base-line population values being studied (e.g. blood parameters, behaviour) may be biased through impact of stress. Moreover, the success of management and conservation activities could be improved if any negative fitness consequences due to anthropogenic stress were minimised (Hofer and East 1998). Such knowledge will be particularly valuable in species with declining or vulnerable populations (Reijnders et al. 1993).

24 316 The southern elephant seal Mirounga leonina, a top predator in the Southern Ocean ecosystem, is among the most extensively studied pinnipeds (e.g., Le Boeuf and Laws 1994). Despite this, it remains poorly understood why populations of the species in the southern Indian and Pacific Oceans have been declining over the past decades or still are in decline (Hindell and Burton 1987; Guinet et al. 1999; McMahon et al. 1999; Pistorius et al. 1999), while South Atlantic populations have been stable (Laws 1994). Declines appear to be primarily due to unfavourable food conditions in the pelagic foraging ranges of the seals, but other factors that regulate populations may also be involved (Hindell 1991; Hindell et al. 1994). Southern elephant seals breed and moult on (sub)antarctic islands. During the reproductive season (September November), females aggregate on beaches and form harems that are competed for by males (Carrick et al. 1962). Females give birth to a pup some days after arrival, and nurse it intensively for approximately 24 days (McMahon et al. 1997) while not feeding themselves. Mothers then return to sea, weaning the pup abruptly when it becomes independent. Lactation is a period of maximum energetic demand for female mammals; hence, it might be a time when animals are more vulnerable to human disturbance. In particular, reproductive success is likely to be more directly affected during this period. Until recently, only a single survey studying disturbance in elephant seals had been carried out, where no effect was found based on population census data (Wilkinson and Bester 1988). Since that study southern elephant seals have continued to be the subject of considerable scientific investigation. This often includes physical or chemical restraint of the animals. We consider it important to examine, at an individual level, how handling activities may affect the physiology of these animals. Endocrinology is a possible way to study levels of stress in wild animals, and in many cases the response of the hypothalamic-pituitary-adrenal (HPA) axis is a reliable indicator of a stressed state (Sapolsky 1990; Wingfield et al. 1997; von Holst 1998). Stressful events are followed by increased levels of circulating glucocorticoids, such as cortisol in pinnipeds (Liggins et al. 1979). Elevated glucocorticoid levels cause many changes in the body such as increased heart rate, blood pressure and peripheral blood supply, analgesia, suppression of digestion, and mobilisation of glucose, nutrients and oxygen (Hadley 1996). Normal short-term adrenocortical responses usually are highly adaptive in unpredictable circumstances. However, prolonged high glucocorticoid levels can trigger harmful effects including immune suppression and neural degeneration (Axelrod and Reisine 1984). Single stressful events, if severe, may in some cases have long-lasting effects on physiology and behaviour (Meerlo et al. 1996; Koolhaas et al. 1997). Prolonged stress responses are often maladaptive, ultimately resulting in a decrease in fitness. In a field experiment, we studied the cortisol stress response to typical research handling activities in southern elephant seals during the reproductive season. The animals were distributed over two geographically similar, almost adjacent study areas that were different in the general levels of human presence (Engelhard et al. 2001). The research handling activities consisted of chemical immobilisation of adult lactating females and physical restraint of pups. Such handling procedures are also representative of studies on several other pinniped species. First, we investigated the acute effects of restraint on the levels of circulating cortisol during the 45-min time spans following the start of handling. Second, we examined natural factors that may affect levels of cortisol, including the stage of lactation, the mass of the animals, and the sex of the pup; we also studied whether there was evidence of circadian and/or seasonal variation in glucocorticoids, which is known to exist in a range of species including the harbour seal Phoca vitulina (Gardiner and Hall 1997). Third, we studied whether different levels of restraint stress may lead to chronically altered cortisol levels or responses (i.e. over a period lasting days to weeks): to test this we compared, at a final stage of the lactation period, the stress response between mother-pup pairs that at earlier stages of lactation had either experienced relatively high, moderate, or no handling. Fourth, we compared cortisol levels and responses between the seals located in the areas of low and high human presence. Based on results in the foregoing investigations, we assessed whether there is reason to assume an impact of stress on blood parameters obtained in earlier seal studies. Materials and methods The study was carried out on sub-antarctic Macquarie Island ( S, E), which has the third largest breeding population of southern elephant seals in the world (Laws 1994). Since 1948, the Australian Antarctic Division has run a permanent research station on the island situated on the Isthmus. For most of this time the elephant seal population has been monitored, and was found to be in long-term decline (Hindell and Burton 1987; McMahon et al. 1999). Areas of high and low human presence During September November 1996 we monitored eight elephant seal harems, which at the peak of the breeding season (around 16 October) on average numbered 154 females (range females). Four harems were on the east coast of the Isthmus (Isthmus East) adjacent to the research station; and four were on Middle Beach, an east coast site 2 km south of the station. The two areas were similar in natural features such as beach topography, wind and surf exposure, and characteristics of elephant seal harems, but mainly different in the levels of human presence (Engelhard et al. 2001). Most human activities on Macquarie Island (scientific, maintenance, sightseeing, and other) have been at or near the research station on the Isthmus (area of high human presence); there is far less frequent human visitation at any other site of the island, including Middle Beach (area of low human presence). Moreover, during the study season, elephant seal harems on the Isthmus were visited daily by a second, five-person team carrying out long-term population monitoring studies. Team members weighed and tagged pups within a day of birth, and later re-weighed these at weaning (methods as in McMahon et al. 1997);

25 317 this sample included approximately 500 pups on Isthmus East. These activities were not carried out on Middle Beach. The amount of disturbance in harems caused by our own, separate team (comprising four persons) was about equal for all eight study harems. Levels of experimental disturbance: treatment groups Twenty-six mother-pup pairs distributed over these harems were marked with a paint dot on one of the first days of lactation to allow individual recognition. Of these, 10 pairs were early breeders (pup born September) and 16 pairs were late breeders (pup born 9 23 October). Within each harem, some study animals were given low and others high levels of experimental disturbance. As a control group, 13 pairs were captured once around 21 days postpartum, i.e., towards the end of the lactation period. As an experimental group, 13 pairs were captured 3 5 times at early, middle, and late stages of lactation, i.e. around 2 days, 11 days, (14 days, 18 days), and 21days postpartum. Within the experimental group, five pairs captured three times during lactation are here considered a moderate treatment group; and eight pairs captured four or five times throughout lactation are considered a high treatment group. The variability of the number of captures within the experimental group was due to practical circumstances resulting in not all planned captures being carried out. In several cases, the study animal was situated in the centre of a large harem for a number of consecutive days; this made it impossible to approach the individual without causing unacceptable levels of disturbance in the harem. All handling procedures were in the light phase of the day between 8.00 a.m. and 3.00 p.m. Handling procedure of adult females Adult females were immobilised with a mixture of tiletamine and zolazepam in a 1:1 ratio (Zoletil 100, Virbac). At present this is one of the more commonly used anaesthetic combinations in wild pinnipeds (Gales 1989; Baker et al. 1990; Lynch et al. 1999; McMahon et al. 2000b); it provides a safe, predictable and reliable means of immobilising elephant seals (McMahon et al. 2000b). The approximate combined dose was 1mg Zoletil (dry mass) per kg seal body mass; this was based on visual estimation of the mass of a study animal. As a result, the average dosage was 386±79 mg (mean±sd, range mg, n=52); based on body mass measured afterwards, this was equivalent to 0.97±0.13 mg kg 1 of seal body mass (range mg kg 1 ). Preparation of anaesthetics was at a site distant from the harem; the animal was then quickly approached and within 1min the drug was administered intramuscularly into the pelvic region by means of an anaesthetic dart delivered by blowpipe (Fedak et al. 1996). Depth of anaesthesia was recorded on a scale of 1 5 (light sedation to surgical anaesthesia; Baker et al. 1990). Once the degree of sedation was an estimated 3 5, a timed series of four blood samples (1 5 ml each) was collected from the extradural vein by means of a G Yale spinal needle in 6 ml untreated Plain Plus Plastic vacutainers (Beckton Dickinson UK). The timing of the first blood sample in relation to the start of experimental disturbance (i.e. anaesthetic administration) ranged from 6 min to 59 min (median 18 min); the three consecutive samples were taken 7, 14 and 21 min after the initial sample. Simultaneously, additional blood samples were collected for further studies on blood parameters. Once blood sampling was complete, nose-to-tail length was measured with a tape measure. The female was weighed in a net attached to a load cell (Salter DC2, West Bromwich, UK), suspended under a tripod. Finally, the seal was released and monitored until fully recovered from anaesthesia. Handling procedure of pups Pups were immobilised by physical restraint in either a pup bag or, if heavier than approximately 75 kg, in a pole net (Laws 1953). Blood samples were collected as in adult females, however, restricted to a series of three samples and by means of G needles. The interval between the start of experimental disturbance (i.e. pup capture) and the first blood sample was in the range 1 44 min (median 7 min); the two consecutive samples were taken 10 min and 20 min after the initial sample. Next, nose-to-tail length and mass were measured. The pup was released near its mother and monitored until re-establishment of mother-pup contact. Laboratory analysis Blood samples were allowed to clot, and the serum was then removed and stored in 2 ml cryotubes at 20 C. Samples were later analysed in duplicate for cortisol concentrations using a radio Immunoassay kit (Coat-A-Count Cortisol Kit, Diagnostic Products Corporation), at an incubation temperature of 37 C. The inter- and intra-assay coefficients of variation were 5% and 6%, respectively. Laboratory analysis was carried out by BCO Analytical Services, Breda, The Netherlands. Statistics Our dataset on cortisol measurements was characterised by a hierarchically nested or clustered structure. First, during each handling event repeated blood samples were collected from one animal, so that such serial data points were not fully independent of each other. Second, at three stages of lactation series of blood samples were taken from animals of the experimental group, so that in addition, different sets of samples collected from one individual were not entirely independent of each other. Third, individual study animals could be situated, and hence clustered, within the same breeding harem. The structure of the data was further complicated since it was unbalanced due to unequal sample sizes per handling event, per individual or per harem; therefore conventional repeated measures analysis could not be applied. Because of these complications, we used hierarchical linear modelling (Goldstein et al. 1998; Loonen et al. 1999) for analysing factors affecting cortisol, while taking into account the dependency or clustering among the measurements. A hierarchical linear model is a special type of regression model designed for data with a hierarchical structure. By incorporating a random error term for each level of clustering, it allows analysis of variances and covariances of factors acting at different hierarchical levels. We applied the MLwiN 1.10 software (Multilevel Models Project, Institute of Education: Goldstein et al. 1998) based on iterative generalised least-squares regression analysis. The models used here included three or four hierarchical levels: 1. Restraint time level; single blood samples collected during 45 min restraint or one handling event (mothers n=155, pups n=139 samples). 2. Lactation stage level; sets of blood samples collected either at early, middle, or late stages of the lactation period (2 days, 11days, 21days postpartum; mothers n=50, pups n=52 sets of samples). 3. Individual level; individual mothers (n=26) and pups (n=26). 4. Harem level; harems (n=8) over which individuals were distributed. Where comparisons between the different treatment groups were made, only data for the late stage of lactation were included (no data on earlier stages being available for the control group); accordingly, the respective models did not contain the hierarchical level of lactation stage. The dependent variable was cortisol (nmol l 1 ). In modelbuilding, a forward stepwise selection procedure was used (see appendix in Loonen et al. 1999). We first tested for a significant linear change in cortisol due to the direct stress response, by incorporating the covariate restraint time (min) since initiation of treatment. Next, we tested for a quadratic change in cortisol levels (such as an increase followed by a decrease), by including the

26 318 covariate squared restraint time (min 2 ). We then examined the effects of other factors and covariates on cortisol levels. Categorical factors included the stage of lactation (early, middle, or late), treatment group (control, moderate or high treatment), pup sex, time of day, season (pup born either in September or October), and area (Middle Beach or Isthmus East). Covariates included body mass (kg), nose-to-tail length (m), body condition index (mass: length ratio, kg m 1 ), mass loss in mothers (kg), and mass gain in pups (kg). Results Mothers of experimental group at three stages of lactation For the 13 adult female elephant seals comprising the experimental group, changes in serum cortisol during immobilisation with Zoletil at three stages of lactation are shown in Fig. 1. Direct response to immobilisation stress Due to the interval between anaesthetic administration and the first possible collection of a blood sample, basal levels of cortisol were not directly measured. Over the sampled time-spans (in most cases, min following initial disturbance) there were moderate, but significant, changes in the levels of serum cortisol. Hierarchical linear model analysis (Table 1) indicated a quadratic effect (P=0.004) of immobilisation time (min) on the levels of serum cortisol (nmol l 1 ), following the relationship (±SE): cortisol=initial value+8.16 Fig. 1. Cortisol response during 45 min of chemical immobilisation in adult female elephant seals of the experimental group, sampled repeatedly at early, middle and late stages of lactation (2 days, 11days, 21days postpartum). Marked and open symbols represent samples taken from early or late breeding females (pup born either in September or October), respectively. Thin lines connect sequential samples taken from the same individual. Heavy lines represent regressions, based on hierarchical linear modelling (Table 1). Letters a c refer to significant differences in cortisol levels (±3.05) restraint time (±0.059) restraint time 2. Cortisol levels peaked on average 23 min after anaesthetic administration (see regression lines in Fig. 1). Effect of the stage of lactation There was, regardless of the direct effect of chemical restraint, a strong increase in the levels of serum cortisol throughout the lactation period. Around 2 days, 11 days and 21 days, postpartum cortisol values (±SD) averaged 126±64 nmol l 1 (n=35), 280±99 nmol l 1 (n=35), and 358±100 nmol l 1 (n=41), respectively. Both the difference in cortisol between early and middle lactation (P=0.0005) and that between middle and late lactation (P=0.009) were significant (Table 1). While overall levels of cortisol increased throughout lactation, the shape of the quadratic response-curve was not significantly different between the three stages of lactation (Table 1, see rejected interactions of restraint time and squared restraint time with early and late lactation stage). Diurnal and seasonal effects The longitudinal data set for the experimental group showed no evidence for an effect of time of day (P=0.669) or season (P=0.783) on the levels of serum cortisol (Table 1). Mass change Around 2 days postpartum, the 13 experimental females weighed 498±79 kg (mean±sd). Body mass averaged 433±64 kg around 11 days postpartum and 346±53 kg around 21days postpartum. Throughout lactation, females lost, on average, 190±30 kg (37.0±2.2% of initial body mass). There was no association between the total loss of mass during lactation and the levels of serum cortisol (P=0.849; Table 1).

27 Table 1. Hierarchical linear model describing the effect of chemical restraint time (min since Zoletil administration) on serum cortisol (nmol l 1 ) in adult female elephant seals at early, middle and late stages of lactation (around 2 days, 11days and 21days postpartum, respectively), based on the data set for the experimental group (compare with Fig. 1). Note: the factors restraint time and squared restraint time describe the change over time in cortisol levels. The decrease in deviance from the null model including the Deviance df Estimate SE P Null model Final model Random parameters Variance harem level Variance individual level Variance lactation stage level Variance restraint time level Fixed parameters Constant < Restraint time (min) (Restraint time) Early lactation (2 days postpartum) Late lactation (21days postpartum) Rejected fixed parameters Early lactation restraint time Early lactation (restraint time) Late lactation restraint time Late lactation (restraint time) Time of day (0 24 h) Season (birth in Sep=0, birth in Oct=1) Sex of pup (male=0, female=1) Mass loss during lactation (kg) Area (Middle Beach=0, Isthmus East=1) constant only, to the final model including all parameters significantly affecting cortisol was significant (change in deviance 43.7, change in df=4, P<0.0001). A forward selection process was used. The significance of parameters was tested by adding these to the model one at a time; when testing the significance of higher order terms (interactions, squares), the lower order terms were included into the model, regardless of their significance. P values <0.05 are shown in bold type Mothers of different treatment groups at late lactation Figure 2 shows the cortisol response to immobilisation stress for mothers at late lactation (around 21days postpartum), that either belonged to the control, moderate, or high treatment group. There was no difference in the dosages of Zoletil anaesthetics given to these females during the handling event (mean±sd: control 321±45 mg, moderate treatment 340±55 mg, high treatment 344±75 mg, 1-way ANOVA, F 2,23 =0.466, P=0.633) or in the dosage of Zoletil given per kilogram seal body mass (mean±sd: control 0.98±0.18 mg kg 1, moderate treatment 1.02±0.07 mg kg 1, high treatment 0.97±0.08 mg kg 1, F 2,23 =0.286, P=0.754). Moreover, there was no difference in the degree of sedation between the treatment groups (scaled 1 5, mean±sd: control 4.1±0.7, moderate treatment 3.7±0.8, high treatment 3.8±0.9, F 2,23 =0.659, P=0.527). Direct response to immobilisation stress Basal serum cortisol levels remained unknown due to the time lag between anaesthetic administration and the first possible collection of a blood sample. Over the sampled time-spans (in most cases, min after anaesthetic administration), cortisol levels generally decreased according to linear relationships with the time of restraint (P<0.0001, Table 2; see regression lines in Fig. 2). In this data set restricted to late lactation, there was no evidence of a quadratic relationship with time (squared restraint time, P=0.590). Diurnal and seasonal effects We did not detect an effect of time of day on levels of serum cortisol at late lactation (P=0.361). However, there was a significant effect of season; females that gave birth in September showed lower cortisol levels at late lactation than females that gave birth in October (average difference±se: 109±32 nmol l 1 ; P=0.003). No effect of moderate, but significant effect of high previous treatment There was no evidence indicating an effect of moderate previous treatment on the cortisol response to immobilisation stress (Table 2). At late lactation, females of the control group (not handled previously) and moderate treatment group (immobilised twice before) showed similar levels of cortisol (P=0.509) and similar linear declines of cortisol with immobilisation time (P=0.622). In contrast, both the levels of cortisol (P=0.024) and the cortisol change during restraint time (P=0.039) were significantly different in the high treatment group (after 3 4 previous immobilisations) when compared to the

28 320 Fig. 2. Comparison of the cortisol response during 45 min of chemical immobilisation between adult female elephant seals at late lactation of the control, moderate and high treatment groups. Thin lines connect sequential samples taken from the same individual. Marked and open symbols represent early or late breeding females (pup born either in September or October), respectively. Regression lines are based on hierarchical linear modelling (Table 2); continuous regression lines represent early breeders, dotted regression lines represent late breeders. Letters a d refer to significant differences in the cortisol response or October, although there was a trend (P=0.072) indicating an interaction between high previous treatment and season. Within the high treatment group, the four early breeding females tended to have relatively lesser reduced levels of cortisol than the four late breeding females (see also Fig. 2). There was no evidence that the degree of sedation affected the levels of cortisol (P=0.946) or the cortisol response (P=0.920; Table 2). control and moderate treatment groups. The effect of milder responsiveness to immobilisation stress in the high treatment group appeared to be consistent for the subsets of females that gave birth either in September Maternal size and condition Body length was not significantly different between mothers of the three treatment groups (mean±se: Table 2. Hierarchical linear model comparing the effect of chemical restraint time (min since Zoletil administration) on serum cortisol (nmol l 1 ) between adult females of the control, moderate and high treatment groups at late lactation (21days postpartum; compare with Fig. 2). The model shows: (1) that the cortisol response in the high treatment group was significantly lower than in both the moderate treatment and control groups and (2) that cortisol levels were higher for late-breeding when compared to early breeding females. See note in Table 1; the change in deviance from the null model to the final model was 31.1 (change in df=4, P<0.0001) Deviance df Estimate SE P Null model Final model Random parameters Variance harem level Variance individual level Variance restraint time level Fixed parameters Constant < Restraint time (min) < Season (birth in Sep=0, birth in Oct=1) High treatment ( 3 previous captures) High treatment restraint time Rejected fixed parameters (Restraint time) Moderate treatment ( 2 previous captures) Moderate treatment restraint time Moderate treatment season n.a. n.a. n.a. n.a. High treatment season Season restraint time Degree of sedation (scaled 1 5) Sedation restraint time Time of day (0 24 h) Sex of pup (male=0, female=1) Length (m) Mass (kg) Condition index (kg m 1 ) Pup weaning mass (kg) Area (Middle Beach=0, Isthmus East=1) Area restraint time

29 321 control 2.42±0.04 m, moderate treatment 2.54±0.06 m, high treatment 2.53±0.06 m; 1-way ANOVA, F 2,23 = 1.501, P=0.244) and did not co-vary with levels of serum cortisol (Table 2, P=0.857). At 21days postpartum, neither body mass nor condition index were significantly different between females of the three treatment groups (mean±se, mass: control 332±13 kg, moderate treatment 332±20 kg, high treatment 355±21kg, F 2,23 = 0.532, P=0.594; condition index: control 137±4 kg m 1, moderate treatment 130±5 kg m 1, high treatment 140±5 kg m 1, F 2,23 =0.676, P=0.519). There was no evidence that body mass (P=0.671) or body condition index (P=0.430) affected levels of cortisol (Table 2). Mothers of the moderate and high treatment groups showed similar loss of mass over the entire lactation period (mean±se: moderate treatment 183±11 kg, high treatment 194±12 kg, independent samples t-test, t 11 = 0.637, P=0.537). The mass loss during lactation was unknown for control females, since no data on mass at early lactation had been collected for this group. Pup sex and maternal cortisol Mothers of male or female pups did not differ in the levels of serum cortisol (Table 2, P=0.287). Comparison between areas different in frequency of human visits At late lactation, adult females on Isthmus East (area of high human presence; n=15) and Middle Beach Fig. 3. Cortisol response during 45 min of physical restraint in elephant seal pups of the experimental group at early, middle and late lactation (2 days, 11days, 21days postpartum). Thin lines connect sequential samples taken from the same individual. Closed and open symbols refer to male and female pups, respectively. Regression lines are based on hierarchical linear modelling (Table 3); continuous and dotted lines represent male and female pups. Letters a d refer to significant differences in the cortisol response (area of low human presence; n=11) showed similar levels of serum cortisol (P=0.847) and similar changes in cortisol during immobilisation time (P=0.376; Table 2). Pups of experimental group at three stages of lactation For nine female and six male pups comprising the experimental group, changes in serum cortisol during physical restraint at three stages of lactation are illustrated in Fig. 3. Direct stress response at early, middle and late lactation Capture and handling generally caused rapid increases in the levels of serum cortisol. Over the sampled intervals (in most cases, 5 35 min post-capture) the cortisol responses followed linear patterns (Table 3: effect of restraint time, P<0.0001), with no evidence of a quadratic effect (squared restraint time, P=0.665). Pups showed a weaker (P=0.003) cortisol response at early lactation (age 2 days) than at the middle or late stages of lactation (age 11 days, 21 days, respectively); the response was not significantly different between the middle and late stages (interaction late lactation restraint time, P=0.660). At early lactation, serum cortisol increased on average (±SE) with 1.95±1.07 nmol l 1 min 1 over the sampled intervals following restraint; at middle and late lactation, the average increase rates were 5.60±1.10 nmol l 1 min 1 and 6.33±1.13 nmol l 1 min 1, respectively. Diurnal and seasonal effects Within the experimental group of pups there was no evidence for an effect of time of day (P=0.371) or season (P=0.417) on the levels of serum cortisol (Table 3).

30 322 Table 3. Hierarchical linear model describing the direct effect of physical restraint time (min since capture) on serum cortisol (nmol l 1 ) in male and female pups at early, middle and late stages of lactation (age around 2 days, 11 days and 21days, respectively), based on the data set for the experimental group (compare with Fig. 3). Cortisol levels increased linearly during restraint, more rapidly at middle and late stages of lactation than at early lactation. See note in Table 1; the change in deviance from the null model to the final model was 70.3 (change in df=4, P<0.0001) Deviance df Estimate SE P Null model Final model Random parameters Variance harem level Variance individual level Variance lactation stage level Variance restraint time level Fixed parameters Constant < Restraint time (min) < Early lactation (2 days postpartum) Early lactation restraint time Sex of pup (male=0, female=1) Rejected fixed parameters (Restraint time) Late lactation (21days postpartum) Late lactation restraint time Sex of pup restraint time Time of day (0 24 h) Season (birth in Sep=0, birth in Oct=1) Mass gain during lactation (kg) Area (Middle Beach=0, Isthmus East=1) Sex difference Effect of previous treatment Increase rates of cortisol during physical restraint were highly similar for the pups of the two sexes (Table 3, P=0.925). However, absolute levels of serum cortisol were significantly higher in females than males, regardless of the time since capture (Table 3, P=0.023). The difference averaged throughout lactation was 72±27 nmol l 1. Mass change The 13 pups of the experimental group weighed 40±4 kg and 117±27 kg at birth and at weaning, respectively, on average gaining 77±25 kg of mass throughout the lactation period (means±sd). We detected no effect of serum cortisol on total mass gain during lactation (Table 3, P=0.860). Pups of different treatment groups at late lactation We compared serum cortisol responses to physical restraint between pups aged 21days of the control, moderate and high treatment groups (Fig. 4). Hierarchical linear modelling comparing the treatment groups (Table 4) confirmed the sex difference in absolute cortisol levels already found within the experimental group (females greater than males; P=0.0002). The rate of increase of serum cortisol during handling was similar for male and female pups (P=0.369). The average difference (±SE) regardless of the effect of restraint time was 123±28 nmol l 1. There was no evidence of diurnal (P=0.145) or seasonal (P=0.860) effects on serum cortisol in pups. There was no evidence suggesting an effect of either a moderate or high level of handling of pups during early and middle lactation on their levels of serum cortisol at late lactation (Table 4). When compared to the control group, pups of the moderate treatment group showed similar absolute levels (P=0.814) and rates of increase (P=0.299) of serum cortisol during physical restraint at late lactation. Neither did pups of the high treatment group differ from the other groups in absolute levels (P=0.612) or increase rates (P=0.169) of serum cortisol during restraint at late lactation. Pup size and condition There was no evidence that body length (P=0.943), mass (P=0.882) or condition index (P=0.988) affected levels of serum cortisol in pups aged 21days (Table 4). Pups of the three treatment groups showed similar body lengths (means±se: control 1.36±0.02 m, moderate treatment 1.36±0.03 m, high treatment 1.30±0.06 m; 1-way ANOVA, F 2,20 =0.834, P=0.449), similar body masses (control 112±9 kg, moderate treatment 120±9 kg, high treatment 108±12 kg, F 2,23 =0.262, P=0.771), and similar body condition indices (control 79±6 kg m 1,moderate treatment 88±5 kg m 1, high treatment 81±7 kg m 1, F 2,20 =0.379, P=0.689). The mass of pups at weaning was not significantly different between the treatment groups (control 119±8 kg, moderate treatment 122±7 kg, high treatment 114±12 kg, F 2,23 =0.157, P=0.856). The gain of mass from birth to weaning was similar for pups of the moderate and high treatment groups (moderate treatment 81±7 kg, high treatment 75±11 kg, independent samples t-test, t 11 =0.391,

31 323 Fig. 4. Comparison of the cortisol response during 45 min of physical restraint between elephant seal pups aged around 21days of the control, moderate, and high treatment groups. Thin lines connect sequential samples taken from the same individual. Closed and open symbols represent male and female pups, respectively. Regression lines are based on hierarchical linear modelling (Table 4); continuous and dotted lines represent male and female pups, respectively. Letters a b refer to significant differences in the cortisol response P=0.703); the total mass gain was unknown for control pups due to absence of data at early lactation. The efficiency of mass transfer from mother to pup, expressed as the mass gained by the pup as a percentage of the mass lost by the mother throughout lactation, averaged 44.0± 1.1% and 37.8±3.5% (±SE) for mother-pup pairs of the moderate and high treatment groups; the difference was not significant (Mann-Whitney, U=12.0, P=0.242). Comparison between areas different in frequency of human visits There was no difference in cortisol levels (P=0.796) or in cortisol responses during restraint (P=0.227) between the 11 pups in the remote area of Middle Beach and 15 pups in the frequently visited area of Isthmus East (Table 4). Discussion All levels of serum cortisol in southern elephant seal mothers and pups on sub-antarctic Macquarie Island (range nmol l 1 ) found in this study, were lower than the plasma cortisol levels found previously in two Table 4. Hierarchical linear model comparing the effect of physical restraint (min since capture) on serum cortisol (nmol l 1 ) between pups of the control, moderate, and high treatment groups at age 21days (compare with Fig. 4). Both the factors moderate and high treatment and the interactions of moderate and high treatment with restraint time were rejected. See note in Table 1; the change in deviance from the null model to the final model was 47.0 (change in df=2, P<0.0001) Deviance df Estimate SE P Null model Final model Random parameters Variance harem level Variance individual level Variance restraint time level Fixed parameters Constant < Restraint time (min) < Sex of pup (male=0, female=1) Rejected fixed parameters (Restraint time) Moderate treatment ( 2 previous captures) Moderate treatment restraint time High treatment ( 3 previous captures) High treatment restraint time Season (birth in Sep=0, birth in Oct=1) Season restraint time Time of day (0 24 h) Sex of pup restraint time Length (m) Mass (kg) Condition index (kg m 1 ) Pup weaning mass (kg) Area (Middle Beach=0, Isthmus East=1) Area restraint time

32 324 male subadult southern elephant seals sampled near Palmer Base, Antarctica (Liggins et al. 1993: 1420 nmol l 1 and 820 nmol l 1 ). The levels were considerably lower than the plasma cortisol levels of typical Antarctic phocid seals (about nmol l 1 ), such as Weddell seals Leptonychotes weddellii (Liggins et al. 1979; Barrell and Montgomery 1989; Liggins et al. 1993), crabeater seals Lobodon carcinophagus and leopard seals Hydrurga leptonyx (Liggins et al. 1993). Levels were comparable to those in northern temperate and Arctic phocids, such as grey seals Halichoerus grypus (Engelhardt and Ferguson 1980; Nordøy et al. 1990; Boily 1996), harbour seals Phoca vitulina (Riviere et al. 1977; Gardiner and Hall 1997; P.J.H. Reijnders, unpublished results), larga or spotted seals Phoca largha (Ashwell-Ericksson et al. 1986), harp seals Phoca groenlandica (Engelhardt and Ferguson 1980) and ringed seals Phoca hispida (St. Aubin and Geraci 1986). In these species, base-line plasma or serum levels are generally in the range of nmol l 1. The very high cortisol levels in Antarctic Weddell, crabeater and leopard seals have been related to their high diving capabilities, as potentially being protective against high pressure nervous syndrome (Liggins et al. 1993). Our data do not provide correlative evidence supporting the hypothesis. Diving performance of adult female southern elephant seals generally exceeds that of typical Antarctic phocid seals (Hindell et al. 1991; Jonker and Bester 1994), while the cortisol levels we found were not exceptionally high. It should be remarked that our results do not concern the pelagic phase of the annual cycle of the seals. Mild cortisol response in immobilised lactating females No direct data on base-line cortisol in lactating female elephant seals were available for this study, due to the time lag between administration of Zoletil and sedation (10 min, Baker et al. 1990). This is unfortunate since the initial rise in cortisol from base-line levels is an important component of the adrenocortical response. Over the measured time-spans (about min after anaesthetic administration) we found significant but moderate changes in the levels of serum cortisol. Mild cortisol responses in anaesthetised adult female elephant seals are consistent with a comparative study on African land mammals (Morton et al. 1995), where in six out of seven artiodactyl species the plasma cortisol levels were relatively low in chemically immobilised animals when compared to physically restrained individuals. Rather inconsistent and generally moderately variable cortisol profiles were also found in spotted hyenas Crocuta crocuta immobilised with Zoletil (Van Jaarsveld and Skinner 1992). However, the moderate cortisol responses to immobilisation stress in lactating female elephant seals found in this study do not necessarily imply that the procedure itself was a mild stressor. It is likely that during lactation a variety of physiological responses to stress are suppressed, including responsiveness of the adrenocortical axis; hyporesponsiveness during lactation has been described for rats Rattus norvegicus (e.g. Windle et al. 1997) and humans (Altemus et al. 1995). If the adrenocortical axis is also toned down in the nursing elephant seal, which is likely since lactation is particularly intensive in this species and the adults fast throughout this period (Fedak et al. 1996), then our results on the changes in cortisol levels may underestimate the magnitude of the entire stress response. A more appropriate description of the full impact of disturbance might be given by measures on the sympathetic-adrenal-medullary response to stress, or by measures of immunocompetence. Cortisol increase during lactation in mothers Cortisol levels in elephant seal mothers, regardless of the effect of restraint, increased over the 24-day lactation period (Fig. 1). Glucocorticoids have profound effects on metabolism in fasting animals, and on lactogenesis (Hadley 1996). Engelhardt and Ferguson (1980) found high plasma cortisol levels in nursing female harp seals when compared to non-nursing females, suggesting this stimulates extrahepatic lipid mobilisation necessary for the production of fat-rich milk. During the first 20 days of lactation, southern elephant seal milk becomes increasingly rich in fat content, from about 12% to approximately 52% (Carlini et al. 1994). This may correspond with increasing cortisol levels. Glucocorticoids are specifically involved in lipid and protein degradation to produce glucose, therefore increasing basal cortisol from initially low levels may be involved in maintaining adequate lactation in a fasting mother. However, glucocorticoid levels that continue to increase are also likely, in the end, to facilitate and induce foraging behaviour and food intake (e.g. Honma et al. 1984; Leibowitz et al. 1984; Wingfield 1994). In emperor penguins Aptenodytes forsteri which fast for up to 4 months while breeding on the sea ice, body stores are gradually depleted while low plasma corticosterone levels are maintained and the birds show little activity; at a threshold body condition however, both plasma corticosterone rises and activity increases, which then may act as a refeeding signal stimulating birds to depart from the breeding group, return to sea and start foraging again (Robin et al. 1998). In lactating and fasting elephant seal mothers, cortisol levels may similarly increase up to a threshold level, and may at that stage be involved in motivating the female to wean the pup and enter the pelagic foraging stage. In addition to, or alternatively, the increasing levels documented here may be related to oestrus, sexual activities, and harem dynamics. Mating usually occurs during the last days of the lactation period (Carrick et al. 1962), when we observed an increase in cortisol levels. In harems, sexual activities are far more frequent late in the

33 325 breeding season than at early stages (McCann 1981); as a result, harassment of females due to male sexual activity is more frequent during late October and early November than earlier in the season (McCann 1981; Galimberti et al. 2000). This increase in natural disturbance over the season may be a crucial factor explaining why cortisol levels increased during lactation (Fig. 1); in particular, it may explain why late breeding females showed higher cortisol levels when compared to early breeding females (Fig. 2). It is remarkable that in grey seals, exhibiting harem systems, lactational oestrus and an approximate lactation period of 18 days, cortisol in females decreases over the lactation period (P.J.H. Reijnders, unpublished results). In phocid seals, cortisol might be involved in the regulation of moult, although the evidence of its role is not fully clear (Riviere et al. 1977; Ashwell-Erickson et al. 1986; Boily 1996). Changes in cortisol over the lactation period (September November) in southern elephant seals are unlikely to be associated with moult, since the females undergo a distinct moulting fast in January February (Hindell and Burton 1988). Direct stress response in pups While in chemically immobilised mothers, blood samples usually could not be collected within 10 min of anaesthetic administration and, hence, the initial rise in cortisol from baseline levels was likely to be missed, physically restrained pups could mostly be blood sampled fairly rapidly after the start of handling. Thus the cortisol profiles reported here for pups (Figs. 3, 4) in most cases are likely to include the initial rise from baseline levels. The profiles are similar to typical adrenocortical responses to capture and handling stress described in a number of studies (e.g. Thomson and Geraci 1986; Harlow et al. 1987; Gardiner and Hall 1997). The adrenocortical responses of pups were stronger around ages 11days and 21days than around 2 days after birth (Fig. 3). This may reflect the development of the adrenocortical system during the first few weeks of life (Walker et al. 1991) and suggests a hyporesponsive period of the HPA axis shortly after birth; the existence of postnatal hyporesponsiveness has been well documented in rodents (Sapolsky and Meaney 1986). If cortisol is involved in regulation of moult, the increasing responsiveness may in addition be related with the onset of the first moult, or the replacement of the black lanugo by the grey hairs typical of weaned pups (Ling and Thomas 1967). In suckling harp seal pups, increasing cortisol levels during lactation have been related to the first moult (Engelhardt and Ferguson 1980). Pup sex There is a sex difference in the activity of the HPA axis of many species (Kitay 1961; Montano et al. 1991; Vierhapper et al. 1998; Canny et al. 1999). In line with this, female elephant seal pups showed significantly higher cortisol values than males; it is interesting to note that a substantial difference in levels was already present within days to weeks of birth (Figs. 3, 4). Although adult southern elephant seals are among the most sexually dimorphic mammal species (Laws 1953), this dimorphism is still extremely small during the suckling period (Campagna et al. 1992; McMahon et al. 1997). However, at the stage of weaning there are small but significant differences in mean weight and tooth development between the sexes; higher cortisol levels in female pups might be involved in their more precocious development of teeth when compared to male pups (McMahon et al. 1997). Effects of repeated handling In laboratory rat pups, repeated handling can cause long-term changes in the adrenocortical responsiveness to stress that persist into adulthood (Liu et al. 1997; Durand et al. 1998; Meerlo et al. 1999). In the present study on wild southern elephant seals in their natural environment, no chronic effect of repeated physical restraint of pups on their adrenocortical axis was found for the duration of the lactation period. At late lactation, cortisol responses to physical restraint were similar for elephant seal pups of the control, moderate and high treatment groups (Fig. 4). Despite the absence of a chronic effect of repeated handling of juveniles over a period of weeks, the possibility of a long-term effect on adrenocortical responsiveness during adulthood cannot be fully excluded (Meerlo et al. 1999). By contrast, the study shows that even though it is likely that the HPA axis is naturally toned down in the lactating elephant seal, repeated chemical immobilisations in lactating females may nevertheless, over a period of weeks, result in chronic changes in adrenocortical responsiveness. First, this was shown by the finding that although at late lactation there was no significant difference in the cortisol stress response between females of the control and moderate treatment groups, this response was significantly attenuated in females of the high treatment group (Fig. 2; Table 2). Second, altered responsiveness may also be indicated by the discrepancy between the comparison of lactational stages which was limited to the experimental group (quadratic responses indicative of mild cortisol peaks after around min: Fig. 1; Table 1), and the comparison of treatment groups at late lactation (linearly decreasing trends despite partial overlap of data: Fig. 2; Table 2). Over the measured time-spans (10 40 min after anaesthetic administration), the negative trends were particularly substantial for the control and moderate treatment groups; this suggests that these females may have been characterised by a relatively rapid cortisol response such that blood sampling missed the initial increase and only caught the subsequent decline. However, the trend for

34 326 the high treatment group was almost horizontal; this may indicate that in this group not only was the plateau of the response reduced, but the speed of the response was also delayed. There was no evidence that the dosage of anaesthetics used during the handling event, or the resulting degree of sedation were different in the high treatment group; therefore these factors are unlikely to explain the attenuated cortisol responses. This indicates that the high degree of treatment previously given to these females may have dampened their HPA responsiveness to stress (Dhabhar et al. 1997), either due to a general desensitisation of the adrenals as a by-effect of the anaesthetic, or to a degree of habituation to the repeated handling procedure, or through other mechanisms. A reduction in physiological condition is likely to have adverse effects on chances of survival and reproductive success (e.g. Fedak et al. 1996; Guinet et al. 1998; Pomeroy et al. 1999; McMahon et al. 2000a; Hall et al. 2001). We compared a set of mass- and conditionrelated measures between the treatment groups, that may be interpreted as indicating potential or likely fitness consequences of disturbance. There was no effect of either a moderate or a high degree of treatment on the mass and condition of mothers and pups at late lactation, the mass lost by mothers throughout lactation, the total mass gained by pups from birth to weaning, and the mass gain of pups proportional to the mass loss of mothers. Moreover, there was no effect on the mass of the pup at weaning; this is of particular significance, since a positive association of pup weaning mass with 1st-year survival has been shown for the southern elephant seal (McMahon et al. 2000a) and for several other pinniped species (Baker and Fowler 1992; Craig and Ragen 1999; Hall et al. 2001). Hence, based on proxy measures there was no indication of an adverse fitness effect of disturbance. However, negative consequences for true Darwinian fitness cannot be directly excluded by this study, since no data on survival of study animals, or on reproductive success of study females during later breeding seasons were available. Area comparison Cortisol levels and responses were similar for elephant seals in the two areas different in the frequency of human visits, i.e. Isthmus East and Middle Beach (cf. Engelhard et al. 2001), indicating that the human presence and activities involved with the research carried out at the station on the Isthmus of Macquarie Island do not pose a long-term stress effect on the residing elephant seals, as measured by cortisol. Absence of a difference in cortisol levels gave no reason to assume that earlier collected chemical and physical blood parameter baseline values are seriously biased by stress-related artefacts. Moderate to high levels of physical restraint in pups, and moderate levels of chemical immobilisations in lactating mothers, caused direct cortisol stress responses during the period of handling, but no chronic changes in the adrenocortical response over a period of days to weeks. Hence, elephant seals appear to tolerate moderate levels of handling disturbance. However, higher levels of handling, especially immobilisation treatments in lactating females, may result in altered adrenocortical responsiveness, and should be minimised either for scientific or other purposes. Acknowledgements This study is a collaboration of the Netherlands Antarctica Program of NWO (Project ), the Alterra Institute (Marine and Coastal Zone Research Team Project ), and the Australian Antarctic Division (Human Impacts Project 1007 and Biological Sciences Project 2265). Fieldwork was conducted as part of the 49th Australian National Antarctic Research Expeditions. We thank the ANARE staff and participating expeditioners on Macquarie Island, in particular Pirie Conboy, Paul Klemes and Jeremy Smith. Rupert Davies, Paul Davis, Frans Jonker, Margaret Morrice, David Slip, and John van den Hoff supported us in the field. Maarten Loonen and Popko Wiersma contributed in discussions on statistics. The manuscript was greatly improved through the comments of Leo Bruinzeel, Jeroen Creuwels, Rudi Drent, Katalin M. Horvath, Clive McMahon, Lucia Privitera, Jeroen Reneerkens, Jan Verhees, Wim Wolff, and an anonymous referee. The Tasmanian National Parks and Wildlife Service supplied permits to work on Macquarie Island. Our activities involving the handling of animals followed the Guidelines of the Antarctic Animal Ethics Committee. References Altemus M, Deuster PA, Galliven E, Carter CS, Gold PW (1995) Suppression of hypothalamic-pituitary-adrenal axis responses to stress in lactating women. J Clin Endocrinol Metab 80: Ashwell-Erickson S, Fay FH, Elsner R, Wartzok D (1986) Metabolic and hormonal correlates of molting and regeneration of pelage in Alaskan harbor and spotted seals (Phoca vitulina and Phoca largha). Can J Zool 64: Axelrod J, Reisine TD (1984) Stress hormones: their interaction and regulation. Science 224: Baker JD, Fowler CW (1992) Pup weight and survival of northern fur seals Callorhinus ursinus. J Zool (Lond) 227: Baker JR, Fedak MA, Anderson SS, Arnbom T, Baker R (1990) Use of a tiletamine-zolazepam mixture to immobilise wild grey seals and southern elephant seals. Vet Rec 126:75 77 Barrell GK, Montgomery GW (1989) Absence of circadian patterns of secretion of melatonin and cortisol in Weddell seals under continuous natural daylight. J Endocrinol 122: Boily P (1996) Metabolic and hormonal changes during the molt of captive gray seals (Halichoerus grypus). Am J Physiol 270:R1051 R1058 Campagna C, Le Boeuf BJ, Lewis M, Bisioli C (1992) Equal investment in male and female offspring in southern elephant seals. J Zool (Lond) 226: Canny BJ, O Farrell KA, Clarke IJ, Tilbrook AJ (1999) The influence of sex and gonadectomy on the hypothalamo-pituitary-adrenal axis of the sheep. J Endocrinol 162: Carlini AR, Márquez MEI, Soave G, Vergani DF, Ronayne de Ferrer PA (1994) Southern elephant seal, Mirounga leonina: composition of milk during lactation. Polar Biol 14:37 42 Carrick R, Csordas SE, Ingham SE (1962) Studies on the southern elephant seal, Mirounga leonina (L.). IV. Breeding and development. CSIRO Wildl Res 7: Craig MP, Ragen TJ (1999) Body size, survival, and decline of juvenile Hawaiian monk seals, Monachus schauinslandi. Mar Mamm Sci 15:

35 327 Dhabhar FS, McEwen BS, Spencer RL (1997) Adaptation to prolonged or repeated stress comparison between rat strains showing intrinsic differences in reactivity to acute stress. Neuroendocrinology 65: Durand M, Sarrieu A, Aguerre S, Mormède P, Chaouloff F (1998) Differential effects of neonatal handling on anxiety, corticosterone response to stress, and hippocampal glucocorticoid and serotonin (5-HT) 2A receptors in Lewis rats. Psychoneuroendocrinology 23: Engelhard GH, Hoff J van den, Broekman M, Baarspul ANJ, Field I, Burton HR, Reijnders PJH (2001) Mass of weaned elephant seal pups in areas of low and high human presence. Polar Biol 24: Engelhardt FR, Ferguson JM (1980) Adaptive hormone changes in harp seals, Phoca groenlandica, and gray seals, Halichoerus grypus, during the postnatal period. Gen Comp Endocrinol 40: Fedak MA, Arnbom T, Boyd IL (1996) The relation between the size of southern elephant seal mothers, the growth of their pups, and the use of maternal energy, fat, and protein during lactation. Physiol Zool 69: Gales NJ (1989) Chemical restraint and anesthesia of pinnipeds: a review. Mar Mamm Sci 5: Galimberti F, Boitani L, Marzetti I (2000) The frequency and costs of harassment in southern elephant seals. Ethol Ecol Evol 12: Gardiner KJ, Hall AJ (1997) Diel and annual variation in plasma cortisol concentrations among wild and captive harbour seals (Phoca vitulina). Can J Zool 75: Goldstein H, Rasbash J, Plewis I, Draper D, Browne W, Yang M, Woodhouse G, Healey M (1998) A User s Guide to MLwiN. Institute of Education, University of London Guinet C, Roux JP, Bonnet M, Mison V (1998) Effect of body size, body mass, and body condition on reproduction of female South African fur seals (Arctocephalus pusillus) in Namibia. Can J Zool 76: Guinet C, Jouventin P, Weimerskirch H (1999) Recent population change of the southern elephant seal at Iˆ les Crozet and Iˆ les Kerguelen: the end of the decrease? Antarct Sci 11: Hadley ME (1996) Endocrinology, 4th edn. Prentice-Hall, New Jersey Hall AJ, McConnell BJ, Baker RJ (2001) Factors affecting firstyear survival in grey seals and their implications for life history strategy. J Anim Ecol 70: Harlow HJ, Thorne ET, Williams ES, Belden EL, Gern WA (1987) Adrenal responsiveness in domestic sheep (Ovis aries) to acute and chronic stressors as predicted by remote monitoring of cardiac frequency. Can J Zool 65: Hindell MA (1991) Some life-history parameters of a declining population of southern elephant seals, Mirounga leonina. J Anim Ecol 60: Hindell MA, Burton HR (1987) Past and present status of the southern elephant seal (Mirounga leonina) at Macquarie Island. J Zool (Lond) 213: Hindell MA, Burton HR (1988) Seasonal haul-out patterns of the southern elephant seal (Mirounga leonina), at Macquarie Island. J Mammal 69:81 88 Hindell MA, Slip DJ, Burton HR (1991) The diving behaviour of adult male and female southern elephant seals, Mirounga leonina (Pinnipedia: Phocidae). Aust J Zool 39: Hindell MA, Slip DJ, Burton HR (1994) Possible causes of the decline of southern elephant seal populations in the Southern Pacific and Southern Indian Oceans. In: Le Boeuf BJ, Laws RM (eds) Elephant seals: population ecology, behavior, and physiology. University of California Press, Los Angeles, pp Hofer H, East ML (1998) Biological conservation and stress. Adv St Behav 27: Holst D von (1998) The concept of stress and its relevance for animal behavior. Adv St Behav 27:1 131 Honma K-I, Honma S, Hiroshige T (1984) Feeding-associated corticosterone peak in rats under various feeding cycles. Am J Physiol 246:R721 R726 Jonker FC, Bester MN (1994) The diving behaviour of adult southern elephant seal, Mirounga leonina, cows from Marion Island. S Afr J Antarct Res 24:75 93 Kitay JI (1961) Sex differences in the adrenal corticol secretion in the rat. Endocrinology 68: Koolhaas JM, Meerlo P, Boer SF de, Strubbe JH, Bohus B (1997) The temporal dynamics of the stress response. Neurosci Biobehav Rev 21: Laws RM (1953) The elephant seal (Mirounga leonina Linn.). I. Growth and age. Falkl Isl Depend Surv Sci Rep 8:1 62 Laws RM (1994) History and present status of southern elephant seal populations. In: Le Boeuf BJ, Laws RM (eds) Elephant seals: population ecology, behavior, and physiology. University of California Press, Los Angeles, pp Le Boeuf BJ, Laws RM (1994) Elephant seals: population ecology, behavior, and physiology. University of California Press, Los Angeles Leibowitz SF, Roland CR, Hor L, Squillari V (1984) Noradrenergic feeding elicited via the paraventricular nucleus is dependent upon circulating corticosterone. Physiol Behav 32: Liggins GC, France JT, Knox BS, Zapol WM (1979) High corticosteroid levels in plasma of adult and foetal Weddell seals (Leptonychotes weddelli). Acta Endocrinol 90: Liggins GC, France JT, Schneider RC, Knox BS, Zapol WM (1993) Concentrations, metabolic clearance rates, production rates and plasma binding of cortisol in Antarctic phocid seals. Acta Endocrinol 129: Ling JK, Thomas CDB (1967) The skin and hair of the southern elephant seal, Mirounga leonina (L.). II. Pre-natal and postnatal development and moulting. Aust J Zool 15: Liu D, Diorio J, Tannenbaum B, Caldji C, Francis D, Freedman A, Sharma S, Pearson D, Plotsky PM, Meaney MJ (1997) Maternal care, hippocampal glucocorticoid receptors, and hypothalamicpituitary-adrenal responses to stress. Science 277: Loonen MJJE, Bruinzeel LW, Black JM, Drent RH (1999) The benefit of large broods in barnacle geese: a study using natural and experimental manipulations. J Anim Ecol 68: Lynch MJ, Tahmindjis MA, Gardner H (1999) Immobilisation of pinniped species. Aust Vet J 77: McCann TS (1981) Aggression and sexual activity of male southern elephant seals, Mirounga leonina. J Zool (Lond) 195: McMahon CR, Hoff J van den, Burton HR, Davis PD (1997) Evidence for precocious development in female pups of the southern elephant seal Mirounga leonina at Macquarie Island. In: Hindell M, Kemper C (eds) Marine mammal research in the Southern Hemisphere, vol 1. Status, ecology and medicine. Surrey Beatty, Chipping Norton, pp McMahon CR, Burton HR, Bester MN (1999) First-year survival of southern elephant seals, Mirounga leonina, at sub-antarctic Macquarie Island. Polar Biol 21: McMahon CR, Burton HR, Bester MN (2000a) Weaning mass and the future survival of juvenile southern elephant seals, Mirounga leonina, at Macquarie Island. Antarct Sci 12: McMahon C, Burton H, McLean S, Slip D, Bester M (2000b) Field immobilisation of southern elephant seals with intravenous tiletamine and zolazepam. Vet Rec 146: Meerlo P, Boer SF de, Koolhaas JM, Daan S, Van den Hoofdakker RH (1996) Changes in daily rhythms of body temperature and activity after a single social defeat in rats. Physiol Behav 59: Meerlo P, Horvath KM, Nagy GM, Bohus B, Koolhaas JM (1999) The influence of postnatal handling on adult neuroendocrine and behavioural stress reactivity. J Neuroendocrinol 11: Montano MM, Wang MH, Even MD, Saal FS vom (1991) Serum corticosterone in fetal mice: sex differences, circadian changes, and effect of maternal stress. Physiol Behav 50: Morton DJ, Anderson E, Foggin CM, Kock MD, Tiroan EP (1995) Plasma cortisol as an indicator of stress due to capture and translocation in wildlife species. Vet Rec 136:60 63 Nordøy ES, Ingebretsen OC, Blix AS (1990) Depressed metabolism and low protein catabolism in fasting grey seal pups. Acta Physiol Scand 139:

36 328 Pistorius PA, Bester MN, Kirkman SP (1999) Survivorship of a declining population of southern elephant seals, Mirounga leonina, in relation to age, sex and cohort. Oecologia 121: Pomeroy PP, Fedak MA, Rothery P, Anderson S (1999) Consequences of maternal size for reproductive expenditure and pupping success of grey seals at North Rona, Scotland. J Anim Ecol 68: Reijnders PJH, Brasseur SMJM, Van der Toorn J, Van der Wolf P, Boyd I, Harwood J, Lavigne D, Lowry L (1993) Seals, fur Seals, sea lions, and walrus: status survey and conservation action plan. IUCN, Gland Riviere JE, Engelhardt FR, Solomon J (1977) The relationship of thyroxine and cortisol to the moult of the harbor seal Phoca vitulina. Gen Comp Endocrinol 31: Robin J-P, Boucontet L, Chillet P, Groscolas R (1998) Behavioral changes in fasting emperor penguins: evidence for a refeeding signal linked to a metabolic shift. Am J Physiol 274:R746 R753 St Aubin DJ, Geraci JR (1986) Adrenocortical function in pinniped hyponatremia. Mar Mamm Sci 2: Sapolsky RM (1990) Stress in the wild. Sci Am 262: Sapolsky RM, Meaney MJ (1986) Maturation of the adrenocortical stress response: neuroendocrine control mechanisms and the stress hyporesponsive period. Brain Res Rev 11:65 76 Thomson CA, Geraci JR (1986) Cortisol, aldosterone, and leucocytes in the stress response of bottlenose dolphins, Tursiops truncatus. Can J Fish Aquat Sci 43: Van Jaarsveld AS, Skinner JD (1992) Adrenocortical responsiveness to immobilization stress in spotted hyenas (Crocuta crocuta). Comp Biochem Physiol A 103:73 79 Vierhapper H, Nowotny P, Waldhäusl W (1998) Sex-specific differences in cortisol production rates in humans. Metab Clin Exp 47: Walker CD, Scribner KA, Cascio CS, Dallman MF (1991) The pituitary adrenocortical system of neonatal rats is responsive to stress throughout development in a time-dependent and stressor-specific fashion. Endocrinology 128: Wilkinson IS, Bester MN (1988) Is onshore human activity a factor in the decline of the southern elephant seal? S Afr J Antarct Res 18:14 17 Windle RJ, Wood S, Shanks N, Perks P, Conde GL, Da Costa APC, Ingram CD, Lightman SL (1997) Endocrine and behavioural responses to noise stress: comparison of virgin and lactating female rats during non-disrupted maternal activity. J Neuroendocrinol 9: Wingfield JC (1994) Modulation of the adrenocortical response to stress in birds. In: Davey KG, Peter RE, Tobe SS (eds) Perspectives of comparative endocrinology. National Research Council of Canada, Ottawa, pp Wingfield JC, Hunt K, Breuner C, Dunlap K, Fowler GS, Freed L, Lepson J (1997) Environmental stress, field endocrinology, and conservation biology. In: Clemmons JR, Buchholz R (eds) Behavioural approaches to conservation in the wild. Cambridge University Press, Cambridge, pp

37 1876 Human disturbance, nursing behaviour, and lactational pup growth in a declining southern elephant seal (Mirounga leonina) population Georg H. Engelhard, Antonie N.J. Baarspul, Martijn Broekman, Jeroen C.S. Creuwels, and Peter J.H. Reijnders Can. J. Zool. Downloaded from by Santa Cruz (UCSC) on 11/15/12 For personal use only. Introduction Abstract: We studied lactation behaviour in relation to pup growth in southern elephant seals (Mirounga leonina) at Macquarie Island, and compared harems in areas of high and low human presence to determine if there is an effect attributable to human activities, including scientific research. Pup weaning mass, a known correlate of first-year survival, was positively influenced by suckle bout durations during early and middle lactation and by maternal aggression during late lactation; no other behavioural variables were associated with weaning mass. In the area of high human presence, we observed from a distance the behaviour of mother pup pairs directly before, during, and after visits to harems by other researchers. Alertness was raised threefold in the presence of people but quickly returned to predisturbance levels after their departure; there were no significant short-term effects on other behavioural variables. In the areas of high and low human presence, we observed the undisturbed behaviour of the seals in the absence of other people. No significant differences in any behavioural variables examined were found, indicating no long-term changes in behaviour resulting from human presence. Human disturbance therefore appears not to have significantly contributed to the population decline observed at Macquarie Island, but the conclusion requires caution given the fairly low power of our analyses. Résumé : Nous avons étudié le comportement d allaitement en relation avec la croissance des petits chez l éléphant de mer du sud (Mirounga leonina) à l île Macquarie et comparé les harems des zones fortement utilisées par les humains à ceux des zones de faible densité humaine afin de déterminer si les activités humaines, y compris les recherches scientifiques, ont un effet. La masse des petits au moment du sevrage, que l on sait être en corrélation avec la survie durant la 1 ère année, est influencée par la durée des tétées au début et au milieu de la période d allaitement et par l agressivité maternelle à la fin. Aucune autre variable du comportement n est associée à la masse des petits au sevrage. Dans la zone de forte densité humaine, nous avons observé à distance le comportement des groupes mère petit avant, pendant et après la visite de chercheurs. La vigilance est trois fois plus élevée en présence d humains, mais elle revient au niveau initial après leur départ et la présence humaine ne produit pas d effet à court terme sur d autres variables du comportement. Nous avons observé le comportement normal, non perturbé, des phoques en l absence d humains, dans les zones de forte densité humaine aussi bien que dans les zones de faible densité. Nous n avons trouvé aucune différence dans les variables du comportement, ce qui indique que la présence d humains n a pas d effet à long terme sur le comportement des phoques. L activité humaine n a donc pas contribué de façon significative au déclin de la population enregistré à l île Macquarie, mais c est là une conclusion qu il faut envisager avec prudence, étant donné la faible puissance de nos analyses. [Traduit par la Rédaction] Engelhard et al The impact of anthropogenic disturbance on free-living marine mammals, a topic of current concern, can be assessed by observations on their behavioural responses. Indeed, several studies documented changes in marine mammal behaviour resulting from human activity (e.g., Salter 1979; Renouf et al. 1981; Allen et al. 1984; Reijnders et al. 1993; Born et al. 1999; Suryan and Harvey 1999). In comparison with data collection on physiological disturbance parameters (e.g., Received 22 February Accepted 30 September Published on the NRC Research Press Web site at on 4 December G.H. Engelhard 1,2 and J.C.S. Creuwels. Alterra Institute, Marine and Coastal Zone Research, P.O. Box 167, 1790 AD Den Burg, the Netherlands, and the Centre for Ecological and Evolutionary Studies, Zoological Laboratory, Groningen University, P.O. Box 14, 9750 AA Haren, the Netherlands. A.N.J. Baarspul and M. Broekman. Centre for Ecological and Evolutionary Studies, Zoological Laboratory, Groningen University, P.O. Box 14, 9750 AA Haren, the Netherlands. P.J.H. Reijnders Alterra Institute, Marine and Coastal Zone Research, P.O. Box 167, 1790 AD Den Burg, the Netherlands. 1 Corresponding author ( Georg.Engelhard@imr.no). 2 Present address: Institute of Marine Research, P.O. Box 1870 Nordnes, N-5024 Bergen, Norway. Can. J. Zool. 80: (2002) DOI: /Z NRC Canada

38 Engelhard et al Can. J. Zool. Downloaded from by Santa Cruz (UCSC) on 11/15/12 For personal use only. blood values, stress hormones), which often necessitates animal handling or immobilisation (Thomson and Geraci 1985; Gardiner and Hall 1997; Engelhard et al. 2002), behavioural observations incur only minimal disturbance created by the researcher himself. However, in many cases it remains unclear whether behavioural responses to human presence may have any negative consequences for survivorship, reproductive success, and other parameters that have direct implications for population status. Therefore behavioural changes should be considered in a fitness context (Hofer and East 1998). Southern elephant seal (Mirounga leonina (Phocidae)) populations in the southern Atlantic Ocean appear stable (Laws 1994; Boyd et al. 1996), whereas those in the southern Indian and Pacific oceans have been in decline over the past decades, or currently still are (Barrat and Mougin 1978; Hindell and Burton 1987; Guinet et al. 1999; McMahon et al. 1999; Pistorius et al. 1999a; Slip and Burton 1999). The causes of population decline remain poorly understood; poor food conditions in the pelagic foraging ranges may be a principal factor, but other local factors that also regulate populations might also be involved (Hindell et al. 1994; Pistorius et al. 1999b). This paper contributes to a project examining to what extent human disturbance (including researcher activity) might be a factor in population decline (cf. Engelhard et al. 2001). Until recently, only one study on disturbance in elephant seals had been carried out; no effect was found based on population census data (Wilkinson and Bester 1988). Little is known on the effect of human activities on (sub-)antarctic pinnipeds (Wilkinson and Bester 1988; Engelhard et al. 2001). This is remarkable in that research effort on several of these species has been intensive (e.g., Le Boeuf and Laws 1994; Hindell and Kemper 1997), and the southern elephant seal has been described as one of the most widely and exhaustively studied pinnipeds (Ling and Bryden 1992). In addition, presence of people in Antarctica and the sub- Antarctic is increasing and is expected to increase further in the near future (Enzenbacher 1992, 1994; IAATO 2000). Lactation is a period of maximum energetic drain for female mammals and intake by their offspring. This might be a time when animals are most vulnerable to disturbance, where either extra energy has to be spent or where intake is hampered. In both ways, reproductive success is likely to be affected. Lactating female elephant seals spend about 3 4 weeks on land without feeding, fuelled exclusively by energy from stored reserves (Fedak et al. 1996). On the breeding beaches, they cluster into harems that are competed for by males (Carrick et al. 1962). Soon after arrival the females give birth to a single pup and nurse it for 23 or 24 days, on average (Arnbom et al. 1997; McMahon et al. 1997). Pups are then weaned abruptly and left by their mothers to be nutritionally independent (Carrick et al. 1962). Mothers lose, on average, about one-third of their initial mass during lactation; their pups may double to quadruple their mass over that time (Arnbom et al. 1997). There is considerable variation in the mass of the pups at weaning, largely because of variation in the mass of their mothers at the start of lactation (Arnbom et al. 1993, 1997; Fedak et al. 1996). Because females usually nurse a single young per season, pup weaning mass can be interpreted as an index of maternal expenditure (Costa 1991; Trillmich 1996). Heavier weaned southern elephant seal pups have higher chances of first-year survival than light weaned pups (McMahon et al. 2000); thus weaning mass may also be considered a proxy for fitness. We address two questions related to human disturbance and mother pup behaviour during lactation. First, which types of behaviour during lactation might (in addition to maternal postpartum mass) affect the mass of the pup at weaning, a parameter associated with survivorship? Second, which aspects of lactation behaviour are affected by human activity, either directly when people are present at elephant seal harems and (or) indirectly over a longer time span when people are absent again from harems? Materials and methods Study population Field investigations were carried out during September to November 1998 on Macquarie Island (54 30 S, E), located in the Pacific sector of the Southern Ocean. The island houses the world s third-largest southern elephant seal population (Laws 1994), which is one of those in decline (Hindell and Burton 1987; McMahon et al. 1999). Since 1948, the Australian Antarctic Division has run a research station on the island, permanently occupied by the Australian National Antarctic Research Expeditions (ANARE). Human activity We studied elephant seals in two study areas similar in natural features but widely different in the levels of human activity: (1) the eastern beaches of the Isthmus on the north of the island, referred to as Isthmus East (cf. Carrick et al. 1962); and (2) a site 2 km south of Isthmus East named Middle Beach. Both study areas were located on the east coast of the island, which is leeward of the prevailing westerly winds. The sites were similar in beach topography; elephant seal harems formed on m wide shingle beaches on Isthmus East and on m wide shingle beaches on Middle Beach. Wind and surf conditions were also comparable; during the study period we observed similar wind forces (Beaufort scale) in the two areas (median 3 in both sites, n = 127; Mann Whitney U test, U = 0.869, P = 0.385). Previously, we found that the average length of adult females and mass of weaned pups were significantly higher on Middle Beach than on Isthmus East; however, in proportion to their own size, mothers in both areas produced pups of similar mass (Engelhard et al. 2001). Moreover, the numbers of adult male and female elephant seals present in breeding harems were comparable in both areas; at the peak of the breeding season (counted 15 October 1998), 866 females were distributed over six harems on Isthmus East, and 509 females were distributed over three harems on Middle Beach. Peak numbers of females present per kilometre of coastline were not significantly different between Isthmus East and Middle Beach (722 vs. 727 females/km of coastline, respectively; χ 2 = , P = 0.683). Both adult and juvenile sex ratios were comparable in the two sites (Engelhard et al. 2001). Isthmus East and Middle Beach are here considered areas of high and low levels of human presence, respectively, on the following grounds. First, Macquarie Island s permanent station, which typically accommodates persons in 2002 NRC Canada

39 1878 Can. J. Zool. Vol. 80, 2002 Can. J. Zool. Downloaded from by Santa Cruz (UCSC) on 11/15/12 For personal use only. summer and persons in winter, is situated on the Isthmus. Most human activity (scientific, maintenance, sightseeing, and tourism) has therefore been at or near that site over the past decades. Other areas of the island, including Middle Beach, generally receive far less frequent human visitation. Second, for continuation of long-term monitoring studies of the elephant seal population, harems on the Isthmus were visited each day of the 1998 pupping season by four to five persons with a small tractor. On Isthmus East, data on birth mass were collected for 459 pups, or about 50% of all pups born in the area (methods as in McMahon et al. 1997). These newborn animals were dragged just outside the harem, captured in a pup bag, lifted for weighing, and given two flipper tags before being brought back into the harem. This handling procedure, carried out routinely, usually implied a brief mother pup separation for about 1 3 min. On average this means a disturbance of 24 min per harem repeated over all days of the pupping season. By contrast, no newborn pups were weighed, flipper-tagged, or otherwise physically handled on Middle Beach. The only exception to this were 24 pups (about 4% of all pups born in the area) that were marked with a paint dot by touching them briefly with a paint sponge attached to a long pole to allow individual recognition. The moderate disturbance created thus lasted 1 3 min and did not incur any temporary mother pup separations. There was no vehicle access to Middle Beach, and the area was closed off for all other people except for our own three-person team that visited both areas daily by foot to carry out behavioural observations. Thus there was a reasonably high difference in the degree of human disturbance between Isthmus East and Middle Beach (Engelhard et al. 2001). Indirect effects of human activity We examined whether human disturbance may indirectly affect normal elephant seal lactation behaviour by comparing the behaviour between mother pup pairs in the areas of low and high human presence, during intervals when no people were present near harems. Throughout lactation, 24 female elephant seals with previously handled pups distributed over three harems on Isthmus East and 24 females with previously unhandled pups distributed over two harems on Middle Beach were monitored. All animals were marked with a paint dot to allow the recognition of individuals. Three observers recorded behaviour of study animals from remote viewpoints overlooking the harems. Behavioural protocols were made from 11 September through 19 October Observation sessions lasted about 2 h and were distributed throughout the day between 8:45 and 18:00 local time. There was no significant difference in the onset of observation sessions carried out on Middle Beach or Isthmus East (Mann Whitney U test: U = 1085, n MB = 58, n IE = 42, NS). Because general activity levels in harems were quite low, we were able to follow several mother pup pairs in a harem simultaneously (each session, 1 5 pairs, median 5). Pairs were observed repeatedly on different days distributed throughout the lactation period; for week 1, week 2, and weeks 3 4 of lactation, the total observation time was 264, 225, and 292 h, respectively. Direct effects of human activity On Isthmus East we examined whether human presence may directly affect lactation behaviour by remotely observing the behaviour of the focal elephant seal mother pup pairs over the time immediately before, during, and after research visits to the harem by other investigators. Typically, four or five persons were present for the weighing and flippertagging of 1-day-old pups born to non-focal females within the same harem where focal individuals were situated (as described above). Duration (mean ± SD) of visits to these harems was 24 ± 19 min (median 20 min, range 2 68 min, n = 17). In observation sessions lasting about 2 h, three observers located on viewpoints overlooking harems monitored the behaviour of focal mother pup pairs (each session, 2 5 pairs, median 5). In total, pairs were observed for 71, 30, and 45 h before, during, and after human visits, respectively. Behavioural variables We made continuous visual recordings of the behaviour of mother pup pairs as focal individuals (Altmann 1974). The duration of behavioural states and timing of events were monitored with the aid of small handheld computers (Psion Organiser LZ-64, Psion PLC, London, U.K.), programmed as event recorders using the software package The Observer 3.0 (Noldus Information Technology b.v., 1994, Wageningen, the Netherlands). The following behavioural variables were recorded (modified from Fogden 1971; McCann 1982, 1983; Anderson and Harwood 1985; Haller et al. 1996): (1) Presenting time percentage of observation time during which the adult female is lying on the side with the venter directed towards her pup, presenting the nipples; the pup may or may not be in oral contact with the nipple (definition similar to suckling session in Oftedal et al. 1987). (2) Suckling time percentage of observation time during which the pup is in oral contact with its mother s nipple (on-teat time). Any interruptions of mouth nipple contact, or any instances where it was impossible to see whether the pup was really in oral contact with the nipple or not, were excluded. (3) Suckle bout duration (s) the mean duration of suckling (on-teat) periods per observation session. (4) Frequency of suckling (h 1 ) the number of suckling (on-teat) periods per hour of observation. (5) Frequency of maternal calls (h 1 ) the number of maternal vocalisations produced by the female per hour of observation, usually in response to the pup s call; often a long, drawn-out falsetto sound, almost a whine (Matthews 1929). (6) Frequency of alertness (h 1 ) the number of times per hour that the female s head is raised with the eyes open and gaze directed. (7) Frequency of flippering (sand flipping; h 1 ) the number of times per hour that the female uses the foreflippers to scoop sand or shingle backwards and upwards, usually onto the back (Laws 1956). (8) Frequency of moves (h 1 ) the number of changes of the female s location per hour of observation. (9) Frequency of aggressive interactions with other females (h 1 ) the number of agonistic encounters per hour di NRC Canada

40 Engelhard et al Can. J. Zool. Downloaded from by Santa Cruz (UCSC) on 11/15/12 For personal use only. rected towards or received from other adult females including threat vocalisations, lunges, high rears, bites, and submissions (McCann 1982). Mass variables In both study areas and for the same mother pup pairs that were also focal individuals in behavioural studies, we collected data on the mass of mothers at the start of lactation (maternal postpartum mass) and the mass of pups at the end of lactation (pup weaning mass). For measuring maternal postpartum mass in a noninvasive way, we applied photogrammetric techniques modified from Haley et al. (1991) and Bell et al. (1997). In this way, disturbance resulting from physical handling necessary for taking direct weight measurements was avoided. Methods were described in detail in a previous paper (Engelhard et al. 2001) and therefore are only briefly outlined here. We made several calibrated, lateral photographs of females, preferably at earlier stages of lactation, under conditions when animals were seen lying quiet and well visible from the side. From the digitized images, body side area (m 2 ) was measured using image analysis software. From these measurements of body side area taken from multiple images available per female, we derived a single estimate of postpartum mass (kg) for each of 44 study females by applying the previously described conversion calculations (Engelhard et al. 2001). Postpartum mass could not be estimated for the other four study females, because insufficient photographic material was available for these individuals. Data on pup weaning mass were collected by weighing weaned pups, which were observed outside harems and therefore were considered to have completed the full lactation period. A net, an aluminum tripod, and 300-kg Salter spring balance (Satter Weigh-Tronix Ltd., West Bromwich, Birmingham, U.K.) were used (McMahon et al. 1997). On Isthmus East, a total of 429 weaned pups previously weighed and tagged at birth were reweighed; on Middle Beach, 168 weaners (previously unhandled) were weighed. These included 45 out of 48 pups observed in detail throughout lactation; three focal pups on Middle Beach could not be relocated after weaning, because they had lost their individual paint marks during the first moult shortly before their mother s departure. As weaned pups were weighed on sites away from the harems, these handling procedures may have caused only minimal degrees of disturbance to mothers inside harems still nursing their young. Statistics Statistical analyses were carried out following Zar (1996), using the SPSS Windows package (SPSS Inc ). For testing if behavioural variables changed in the course of lactation, the mean values of behavioural variables averaged for week 1, week 2, and weeks 3 4 of the lactation period were calculated for individual mother pup pairs: 7-day intervals were considered necessary to obtain sufficiently large sample sizes for statistical analysis while including the majority of study animals. We grouped observations for the 3rd and 4th weeks of lactation, because there is variability in the total duration of lactation (cf. Arnbom et al. 1997; McMahon et al. 1997). In addition, the first, second, and later weeks of lactation are characterized by different rates of growth of the pup (McCann et al. 1989). In parametric statistical analysis, behavioural data were transformed to improve homogeneity of variance if these were the dependent variables. For percent data, we used arcsine transformation; for bout durations and frequencies, we used log transformation. When testing the hypothesis that human presence has an effect (direct or indirect) on elephant seal behaviour, we performed power analyses to examine the possibility of accepting the null hypothesis when in fact it was false. Power tests were based on values of α set to Results Study females gave birth between 9 and 30 September 1998 (median 26 September; n = 48). The duration of lactation ranged from 20 to 32 days (24.2 ± 2.5 days, n = 46; two females were not monitored until the final stage of lactation, because their paint marks faded and became unrecognizable). The dates of weaning ranged from 4 to 29 October 1998 (median 19 October; n = 46). Each female raised her own offspring during the study period, as no adoptions were observed. However, the incidence of adoptions cannot be fully excluded if it occurred in the brief interval (<1 day) between birth and marking of study animals. Allo-suckling was recorded rarely. Throughout all 781 undisturbed observation hours, none of the 48 focal females was observed with certainty to have suckled another pup but their own. Among focal pups, only a single allo-suckle bout (duration 483 s) was observed out of a total of 1390 suckle bouts recorded during undisturbed observation hours. Behavioural changes over the course of lactation Changes in behavioural variables from the early (week 1) to the middle (week 2) and late stages of lactation (weeks 3 4) are shown in Table 1. The proportion of time females spent presenting increased significantly from early to middle lactation; thereafter this proportion remained constant. Over the course of lactation, pups spent an increasing proportion of time sucking. This resulted from an increase in the durations of suckle bouts throughout lactation, as well as from an increase in the frequency of suckling during the first 2 weeks of lactation (Table 1). The frequency of alertness and maternal calls was highest during the first week postpartum. Over the course of lactation, there was no significant change in the frequencies of flippering, moves, and aggressive encounters among females (Table 1). Relationship between behaviour during lactation and the mass of the pup at weaning We examined to what extent different types of behaviour during lactation, in addition to maternal postpartum mass, were associated with the mass of the pup at weaning. In our data set, as expected, there was a strong relationship between maternal postpartum mass (estimated from photographs) and pup weaning mass (linear regression, t = 9.47, P < ); the variable maternal postpartum mass alone explained 67.6% of the variation in pup weaning mass. A linear regression analysis with backward selection procedure (Table 2) showed that pup weaning mass was also 2002 NRC Canada

41 1880 Can. J. Zool. Vol. 80, 2002 Table 1. Effect of lactation stage on behavioural variables (mean ± SE), quantified during undisturbed observation sessions carried out on Middle Beach and Isthmus East in the absence of other people. Can. J. Zool. Downloaded from by Santa Cruz (UCSC) on 11/15/12 For personal use only. Behavioural variable a Week 1 (n = 46) Week 2 (n = 46) significantly and positively influenced by the mean durations of suckle bouts during the first (P = 0.002) and second (P = 0.019) weeks of lactation and by the frequency of aggressive encounters by the respective mothers during the last week of lactation (P = 0.042). A regression model including these three behavioural variables in addition to maternal postpartum mass explained 83.1% of the variation in pup weaning mass. All other behavioural variables were rejected, indicating that these variables were not linked to lactational pup growth (Table 2). It should be noted, however, that a linear regression model including all behavioural variables except for mean suckle bout durations would reveal effects approaching significance of the proportion of time spent suckling during the first (P = 0.063) and second (P = 0.082) week of lactation on pup weaning mass. Mean suckle bout durations and proportion suckling time were correlated during the first (r S = 0.853, P < ) and second week of lactation (r S = 0.297, P = 0.045). By contrast, there was no effect of the frequency of suckling (i.e., the spread of suckling) during any of the weeks of lactation on pup weaning mass. Weeks 3 4 (n = 47) Test statistic P Post-hoc test (Wilcoxon) Time lying on side (%) 38.7± ± ±2.5 F [2,84] = e<m~l Time suckling (%) b 3.3± ± ±1.7 F [2,84] = < e<m<l Suckle bout duration (s) b 151±24 c 276±22 362±27 F [2,66] = < e<m<l Frequency of suckling (h 1 ) b 1.2± ± ±0.2 χ 2 2 = e<m~l Frequency of maternal calls (h 1 ) 2.4± ± ±0.1 χ 2 2 = e>m~l Frequency of alertness (h 1 ) 2.4± ± ±0.1 χ 2 2 = e>m~l Frequency of flippering (h 1 ) 0.8± ± ±0.1 χ 2 2 = Frequency of moves (h 1 ) 1.4± ± ±0.1 χ 2 2 = Frequency of aggressive interactions with other females (h 1 ) 1.2± ± ±0.1 χ 2 2 = Note: Stages include early, middle, and late lactation (respectively, week 1, week 2, and weeks 3 4 postpartum). Effect of lactation stage on percentages and bout durations was tested using repeated measures ANOVA, after arcsine transformation of percentages and log transformation of bout durations. Effect of lactation stage on frequency variables was tested using the Friedman test. If a significant effect was found, Wilcoxon s signed-rank test was applied for post-hoc comparisons: e, early lactation (week 1); m, middle lactation (week 2); l, late lactation (weeks 3 4); symbols < and > indicate significant increase or decrease (P < 0.05); symbol ~ indicates increase or decrease not distinguishable from coincidence. Values of P < 0.05 are shown in boldface type. a Mother pup pairs were, on average, observed for 5 h 44 min during week 1, for 4 h 53 min during week 2, and for 6 h 13 min during weeks 3 4. Not all pairs were observed during each stage of the lactation period. b Because of unequal sample sizes and positively skewed distributions in these variables, multiplication of the mean suckle bout duration with the mean suckling frequency does not result in the mean proportion of time spent suckling. c In some mother pup pairs, suckling was not recorded in week 1 of lactation if observations were few; hence, n = 37 for suckle bout duration during week 1. Direct behavioural responses to human visits The behaviour of elephant seal mother pup pairs immediately before, during, and after visits to seal harems by field investigators was compared (Table 3). Most noticeably, presence of people resulted in a significant change (P < ) in the alertness of elephant seal mothers; frequency of alertness was, on average, raised to a threefold level during human visits when compared with the periods directly before or after human visits. Alertness after visitation was similar to alertness before human visits (Table 3). Moreover, there was a decrease in the frequency of maternal calls directly after people departed the harems compared with the time immediately before human visits. There was no immediate effect of human presence on the proportion of time females spent presenting, the proportion of time pups spent sucking, the duration and frequency of suckle bouts, or the frequencies of flippering, moves, or agonistic encounters among females. It should be noted that between periods before, during, and after human presence there was much variability within individuals in these behavioural variables. This resulted in only low or moderate power to detect significant effects (P < 0.05) on these variables with the current sample sizes (power ranging from to for rejected behavioural variables; see Table 3). Behaviour in areas of low and high human presence Behavioural variables, quantified in the absence of people near elephant seal harems, were compared between mother pup pairs in the areas of low human presence (Middle Beach) and high human presence (Isthmus East) for three stages of the lactation period (Table 4). Of nine variables examined, none was significantly different between seals on Middle Beach and Isthmus East. In addition, there was no significant interaction of week of lactation with area for any of the behavioural variables examined. This indicated that changes in behavioural variables over the course of lactation were similar in the areas of low and high human presence. However, power analysis indicated only low power to detect significant differences (P < 0.05) in single behavioural variables between elephant seals in the two areas (power ranging from to 0.248). Therefore, we further examined if mother pup pairs in the two areas showed differences in overall behaviour using a multivariate repeated-measures ANOVA on the effect of area (remote or visited) on all be NRC Canada

42 Engelhard et al Table 2. Linear regression model examining the effects of maternal postpartum mass and different aspects of lactation behaviour on pup weaning mass. Can. J. Zool. Downloaded from by Santa Cruz (UCSC) on 11/15/12 For personal use only. Behavioural variables Coefficient SE t P Final model Constant Maternal postpartum mass (kg) < Suckle bout duration, week 1 (s) Suckle bout duration, week 2 (s) Frequency of aggressive interactions with other females, week 3 (h 1 ) Rejected terms Time lying on side, week 1 (%) Time lying on side, week 2 (%) Time lying on side, week 3 (%) Time suckling, week 1 (%) Time suckling, week 2 (%) Time suckling, week 3 (%) Suckle bout duration, week 3 (s) Frequency of suckling, week 1 (h 1 ) Frequency of suckling, week 2 (h 1 ) Frequency of suckling, week 3 (h 1 ) Frequency of maternal calls, week 1 (h 1 ) Frequency of maternal calls, week 2 (h 1 ) Frequency of maternal calls, week 3 (h 1 ) Frequency of alertness, week 1 (h 1 ) Frequency of alertness, week 2 (h 1 ) Frequency of alertness, week 3 (h 1 ) Frequency of flippering, week 1 (h 1 ) Frequency of flippering, week 2 (h 1 ) Frequency of flippering, week 3 (h 1 ) Frequency of moves, week 1 (h 1 ) Frequency of moves, week 2 (h 1 ) Frequency of moves, week 3 (h 1 ) Frequency of aggressive interactions with other females, week 1 (h 1 ) Frequency of aggressive interactions with other females, week 2 (h 1 ) Note: Weekly averages of behavioural variables are included as covariates. A model including only maternal postpartum mass as explanatory variable explained 67.6% of the variation in pup weaning mass; the final model including, in addition, all significant behavioural variables explained 83.1% of the variation in pup weaning mass. Initially, a backward selection procedure was used; the significance of rejected terms was then reexamined by adding these to the final model one at a time. Values of P < 0.05 are shown in boldface type. havioural variables combined. The inclusion of all behavioural variables into a single analysis still yielded, with increased power of 0.477, no significant difference in overall behaviour between seals in areas of low and high human presence (F [9,25] = 1.281, P = 0.297). Discussion Significance of different types of behaviour In the context of human disturbance, (i) natural variation in behavioural parameters should be taken into account as a possible confounding factor in assessing which changes are due to human presence; and (ii) it should be considered which possible behavioural impacts are of highest biological significance, i.e., may have consequences for survival or reproductive success (Hofer and East 1998; Gill et al. 2001). There was a significant effect of the stage of lactation on six out of nine variables examined in this study (Table 1). Increases in suckling frequency and suckle bout duration over the lactation period, resulting in an increase in the percentage suckling time, have been recorded previously for southern elephant seals (Bryden 1968) and northern elephant seals (Mirounga angustirostris; Le Boeuf et al. 1972). Suckling time also varied or tended to vary over the course of lactation in Weddell seals (Leptonychotes weddellii; Tedman and Bryden 1979), grey seals (Halichoerus grypus; Kovacs 1987b), harbour seals (Phoca vitulina; Arts and Rijniers 1986; Hedd et al. 1995), and harp seals (Phoca groenlandica; Kovacs 1987a). In addition, we report that presenting time increased from the first to the second week of lactation in southern elephant seals, whereas alertness and frequency of maternal calls decreased over this time. Because of these natural changes in behavioural parameters over the weeks of lactation (see also Tedman and Bryden 1979; Kovacs 1987a, 1987b), the stage of lactation should be considered as an additional factor in assessing the effect of human presence on behaviour. Because the mass of pups at nutritional independence from their mothers is associated with their chances of survival (McMahon et al. 2000), we used this measure as a proxy for fitness in assessing the biological significance of 2002 NRC Canada

43 1882 Can. J. Zool. Vol. 80, 2002 Table 3. Comparison of elephant seal behaviour directly before, during, and after human visits to seal harems, monitored in 24 mother pup pairs on Isthmus East (area of high human presence). Can. J. Zool. Downloaded from by Santa Cruz (UCSC) on 11/15/12 For personal use only. Behavioural variable a Before (n = 24) During (n = 24) different types of mother pup behaviour during lactation (Table 2). Although weaning mass of pups is primarily determined by the mass of their mothers at parturition (Arnbom et al. 1993, 1997; Fedak et al. 1996), the present study shows that it is also linked with some aspects of nursing behaviour during lactation. Mean suckle bout durations during the first and second week of lactation were significantly and positively associated with pup weaning mass, and similar but weaker tendencies were found for the proportion of time spent suckling in the same weeks; there was no evidence of such an association during later weeks (Table 2). Thus we provide correlational evidence that during the initial stages of lactation, suckle bouts are a factor limiting the growth rate of the pup. This implies that if disturbance leads to a reduction in average durations of suckle bouts, a negative effect on weaning mass and, hence, probability of survival is also to be expected. At first hand, suckling time may appear a straightforward behavioural measure of milk intake and, as a result, growth rate of the pup. However, Cameron (1998), who reviewed this relation, concluded that suckling time is not a useful predictor of milk intake in many mammalian species. She discussed a number of possible confounding factors, including variation in the suckle ability of offspring (e.g., Higgins et al. 1988), the motivation for suckling (hunger) of juveniles (Hall et al. 1978; Mendl and Paul 1989), the mother s experience and ability to release milk (Green 1990), and the energy content of the milk (Oftedal et al. 1987). Some factors of influence in other species do not apply in elephant seals, because mothers fast throughout lactation and pup growth over this period is entirely dependent on milk intake (Ortiz et al. 1984; Fedak et al. 1996). Thus, there is no effect of variation in maternal diet during the lactation period itself (e.g., bighorn sheep (Ovis canadensis; Berger 1979)), and Repeated measures ANOVA After (n = 21) b F P Power Post-hoc test (Wilcoxon) Time lying on side (%) 53.2± ± ± Time suckling (%) 7.8± ± ± Suckle bout duration (s) 227±40 c 183±23 c 242±51 c Frequency of suckling (h 1 ) 1.6± ± ± Frequency of maternal calls (h 1 ) 1.8± ± ± b~d~a,b>a Frequency of alertness (h 1 ) 1.7± ± ± < b<d>a,b~a Frequency of flippering (h 1 ) 0.2± ± ± Frequency of moves (h 1 ) 1.2± ± ± Frequency of aggressive interactions with other females (h 1 ) 1.8± ± ± Note: During visits, four or five investigators weighed newborn pups born to other females than those observed (see text). Means (± SE) represent individually monitored mother pup pairs. Behavioural variables were analyzed using repeated measures ANOVA, after arcsine transformation of percentages and log transformation of durations and frequencies. If a significant difference was found, Wilcoxon s signed-rank test was applied for posthoc comparisons: symbols b, d, and a refer to periods immediately before, during, and after human visits, respectively; symbols < and > refer to directions of significant differences (P < 0.05); symbol ~ refers to absence of a significant difference. Values of P < 0.05 are shown in boldface type. a Mean duration of investigator visits, 24 min; mean observation duration before and after visits, 61 and 35 min, respectively. b For three mother pup pairs, no observations directly following human visits were available. c Suckling was not recorded for all individuals during each observation session. Sample sizes for suckle bout durations before, during, and after visits were n = 15, 14, and 17, respectively. Sample size in the test, n = 12. there is no possibility of additional feeding of solid food by juveniles during lactation (e.g., cat (Felis catus; Martin 1986)). Our analysis suggests that after we account for maternal mass and suckling behaviour, the aggression of elephant seal mothers during the late stage of lactation is also positively associated with the weaning mass of their pups. Although earlier studies on elephant seal mothers and related species indicated that maternal aggression serves to protect and successfully wean the pup (Boness et al. 1982; Christenson and Le Boeuf 1978; McCann 1982), these did not report a possible effect on pup growth. We found no evidence that other types of behaviour during lactation either positively or negatively influenced the pup s weaning mass. We conclude that suckle bouts during early and middle lactation and, to a lesser extent, maternal aggression during late lactation (perhaps in addition to the proportion of time spent suckling) are the most robust measures of lactation behaviour to study in the context of maternal reproductive performance in elephant seals. Direct and indirect effects of human presence On Isthmus East, visits by research investigators to elephant seal harems resulted in directly observable behavioural responses by lactating adult females, as revealed by the significantly increased frequency of alert behaviour (P < ; see Table 3). In the presence of people, seal alertness was elevated to threefold levels in comparison with periods immediately before human presence. Directly after departure of researchers from harems, alertness returned to predisturbance levels; we found no difference in alertness between periods before or after human presence. The frequency of maternal calls was apparently lower after human visits than before them (Table 3). Nevertheless, it is unlikely that a negative indirect impact on pup growth occurred because of temporary 2002 NRC Canada

44 Engelhard et al Table 4. Comparison of behaviour between 24 elephant seal mother pup pairs on Middle Beach (remote study area) and 24 pairs on Isthmus East (area of high human presence) during three stages of the lactation period. Can. J. Zool. Downloaded from by Santa Cruz (UCSC) on 11/15/12 For personal use only. changes in these aspects of behaviour: neither the frequency of alertness nor that of maternal calls was significantly associated with pup weaning mass (Table 2). Other behavioural variables were not significantly different directly before, during, and after human visits (Table 3). We detected no effect of human presence on flippering (sand flipping), a behaviour previously described as a displacement activity in response to conditions of stress (Laws 1956; Heath and Schusterman 1975; Lewis and Campagna 1998). Observations carried out in the absence of people yielded no significant difference in any of the behavioural parameters examined between 24 mother pup pairs in the frequently visited area of Isthmus East and an equal number of pairs in the remote area of Middle Beach (Table 4). In addition, natural changes in behaviour over the course of lactation were similar for mother pup pairs in the two areas (Table 4, interaction of week with area). This indicated no indirect changes in mother pup behaviour over a period of weeks as a result of the more intense levels of human activity on the Isthmus. Caution is required, however, when drawing this conclusion; based on all examined behavioural variables, the power to detect a significant effect was only moderate (power = 0.477). Therefore, possible small-scale changes in behaviour that may nevertheless be relevant may have remained undetected. Mean±SE(n) Effect of area Interaction week area Behavioural variable Stage (week(s)) Middle Beach Isthmus East F [1,41] P Power F [2,82] P Power Time lying on side (%) ±4.6 (24) 38.4±5.9 (22) ±4.2 (24) 52.2±5.1 (22) ±3.5 (23) 52.7±3.6 (24) Time suckling (%) 1 3.8±0.6 (24) 2.8±0.9 (22) ±2.2 (24) 13.1±1.5 (22) ±2.4 (23) 16.6±2.5 (24) Suckle bout duration (s) 1 175±35 (21) 119±29 (16) a b ±29 (24) 299±33 (22) ±34 (23) 368±41 (24) Frequency of suckling (h 1 ) 1 1.2±0.2 (24) 1.1±0.3 (22) ±0.4 (24) 2.2±0.4 (22) ±0.1 (23) 2.1±0.3 (24) Frequency of maternal calls (h 1 ) 1 2.1±0.4 (24) 2.6±0.5 (22) ±0.2 (24) 0.9±0.2 (22) ±0.1 (23) 0.7±0.2 (24) Frequency of alertness (h 1 ) 1 2.6±0.4 (24) 2.2±0.3 (22) ±0.2 (24) 1.4±0.3 (22) ±0.2 (23) 1.2±0.1 (24) Frequency of flippering (h 1 ) 1 1.0±0.4 (24) 0.6±0.2 (22) ±0.1 (24) 0.2±0.1 (22) ±0.1 (23) 0.4±0.1 (24) Frequency of moves (h 1 ) 1 1.3±0.3 (24) 1.4±0.2 (22) ±0.2 (24) 1.2±0.2 (22) ±0.1 (23) 1.2±0.2 (24) Frequency of aggressive interactions with other females (h 1 ) 1 1.2±0.3 (24) 1.2±0.3 (22) ±0.2 (24) 0.9±0.3 (22) ±0.1 (23) 0.9±0.2 (24) Note: Differences in behavioural variables between areas were tested using repeated measures ANOVA, after arcsine transformation of percentages and log transformation of durations and frequencies. a Error df = 32. b Error df = 64. Implications at population level We found that disturbance to elephant seal harems caused by visits by researchers resulted in direct but transient changes in some types of behaviour; we found no long-term changes in behaviour (over a period of weeks) as implied from the comparison made between the areas of high and low human presence. From a conservation perspective, fitness consequences of impacts should be considered (e.g., on the effects of disturbance on breeding success in Antarctic penguins, see Woehler et al. 1994, Giese 1996, and Cobley and Shears 1999). Thus, we examined how different types of behaviour during lactation are related to a proxy for fitness in elephant seals, as indexed by the mass of the pup at weaning. For those aspects of behaviour that were affected by human disturbance, including alertness and maternal call frequencies, no association with pup weaning mass was found. By contrast, for the behavioural variables apparently of highest significance for the growth of the pup suckle bout durations during early and middle lactation and frequency of maternal aggression during late lactation we were not able to detect any direct or indirect changes resulting from the presence of people. We previously reported that, on average, mothers were longer and weaned pups heavier on remote Middle Beach than on human-accessible Isthmus East; however, there was 2002 NRC Canada

45 1884 Can. J. Zool. Vol. 80, 2002 Can. J. Zool. Downloaded from by Santa Cruz (UCSC) on 11/15/12 For personal use only. no difference in the mass of pups at weaning in proportion to their mothers s mass (Engelhard et al. 2001). Moreover, the area difference in the size of mothers was already present at the start of the breeding season before the higher degree of human disturbance on Isthmus East took place (Engelhard et al. 2001). It remained unclear whether the apparent preference of larger females to breed on Middle Beach was due either to human activity near the research station in earlier years (cf. Thiel et al. 1992) or to natural differences between the areas, such as in the distribution of high-quality males or of male aggression (Cox 1981; Galimberti et al. 2000a, 2000b). However, the absence of any differences in pup weaning mass between sites other than that due to the size of mothers corresponded with the absence of any significant differences in behaviour during lactation and indicated no direct effect of human presence on lactational pup growth. Hence, we find no evidence that human activity of the type and intensity investigated here will result in a decrease in fitness by affecting mother pup behaviour and pup growth during the present lactation period. This study therefore lends credibility to the notion that the population decline observed at Macquarie Island is not due to human disturbance on land, including researcher activities, given that the absence of any significant effects is not due to the fairly moderate power of our analyses. As this southern elephant seal population has been investigated more extensively than any other of the species, it is postulated that the finding may also be applicable to other declining populations in the southern Indian and Pacific oceans (in accordance with Wilkinson and Bester 1988 and Engelhard et al. 2001). Guidelines for disturbance research in pinnipeds Behavioural studies investigating anthropogenic impacts on wildlife should focus on parameters with known or expected links to survival and (or) reproductive success (Hofer and East 1998; Gill et al. 2001). In southern elephant seals, suckling behaviour in young pups correlates with their mass at weaning (this study), and weaning mass influences their chances of survival (McMahon et al. 2000). Both of these links may also be present in other pinniped species, although the association between suckle bout duration and pup mass gain may well be restricted to phocid seals where females fast throughout lactation (see Oftedal et al. 1987). Associations between weaning mass and survival have also been shown for northern fur seals (Callorhinus ursinus; Baker and Fowler 1992), Hawaiian monk seals (Monachus schauinslandi; Craig and Ragen 1999), and grey seals (Hall et al. 2001). In the impact assessment of human disturbance on populations of pinniped species characterized by lactation fast, we suggest that detailed recordings on suckle bout durations are a more sensitive behavioural indicator than observations on the most conspicuous behavioural responses, such as the levels of alertness. Acknowledgements This work is a collaboration of the Alterra Institute (Marine and Coastal Zone Research, Project ) and the Australian Antarctic Division (Human Impacts Project 1007 and Biological Sciences Project 2265). The study has been made possible by a grant from the Netherlands Antarctica Program (NWO Project ). We acknowledge the expeditioners participating in the 51st ANARE on Macquarie Island, in particular Maria Clippingdale, Paul Denne, Iain Field, Adam Jagla, and John van den Hoff, who supported in the collection of data on the mass of pups. Human presence on Middle Beach was minimal thanks to the effort of Michael Carr and Lloyd Fletcher. At earlier stages of this project, support and suggestions were given by Sophie Brasseur, Harry Burton, Pirie Conboy, Mike Fedak, Ailsa Hall, Bernie McConnell, Clive McMahon, David Slip, and Joost Tinbergen. Rudi Drent, Fritz Trillmich, Wim Wolff, and two anonymous reviewers gave valuable comments on earlier versions of the manuscript. The Tasmanian National Parks and Wildlife Service supplied the permits required for us work on Macquarie Island. Our activities involving animal handling were approved by the Antarctic Animal Ethics Committee. References Allen, S.G., Ainley, D.G., Page, G.W., and Ribic, C.A The effect of disturbance on harbor seal haul out patterns at Bolinas Lagoon, California. Fish. Bull. 82: Altmann, J Observational study of behavior: sampling methods. Behaviour, 49: Anderson, S.S., and Harwood, J Time budgets and topography: how energy reserves and terrain determine the breeding behaviour of grey seals. Anim. Behav. 33: Arnbom, T., Fedak, M.A., Boyd, I.L., and McConnell, B.J Variation in weaning mass of pups in relation to maternal mass, postweaning fast duration, and weaned pup behaviour in southern elephant seals (Mirounga leonina) at South Georgia. Can. J. Zool. 71: Arnbom, T., Fedak, M.A., and Boyd, I.L Factors affecting maternal expenditure in southern elephant seals during lactation. Ecology, 78: Arts, B., and Rijniers, J De invloed van verstoringen op de zeehondenpopulatie in de Nederlandse Waddenzee. De broedbiologie van de gewone zeehond (Phoca vitulina) in gevangenschap. Int. Rep., Rijksinstituut voor Natuurbeheer, Arnhem, Leersum, and Texel, the Netherlands. Baker, J.D., and Fowler, C.W Pup weight and survival of northern fur seals Callorhinus ursinus. J. Zool. (Lond.), 227: Barrat, A., and Mougin, J.L L éléphant de mer Mirounga leonina de l île de la Possession, archipel Crozet (46 25 S, E). Mammalia, 42: Bell, C.M., Hindell, M.A., and Burton, H.R Estimation of body mass in the southern elephant seal, Mirounga leonina, by photogrammetry and morphometrics. Mar. Mamm. Sci. 13: Berger, J Weaning conflict in desert and mountain bighorn sheep (Ovis canadensis): an ecological interpretation. Z. Tierpsychol. 50: Boness, D.J., Anderson, S.S., and Cox, C.R Functions of female aggression during the pupping and mating season of grey seals, Halichoerus grypus (Fabricius). Can. J. Zool. 60: Born, E.W., Riget, F.F., Dietz, R., and Andriashek, D Escape responses of hauled out ringed seals (Phoca hispida) toaircraft disturbance. Polar Biol. 21: Boyd, I.L., Walker, T.R., and Poncet, J Status of southern elephant seals at South Georgia. Antarct. Sci. 8: NRC Canada

46 Engelhard et al Can. J. Zool. Downloaded from by Santa Cruz (UCSC) on 11/15/12 For personal use only. Bryden, M.M Lactation and suckling in relation to early growth of the southern elephant seal, Mirounga leonina (L.). Aust. J. Zool. 16: Cameron, E.Z Is suckling behaviour a useful predictor of milk intake? A review. Anim. Behav. 56: Carrick, R., Csordas, S.E., and Ingham, S.E Studies on the southern elephant seal, Mirounga leonina (L.). IV. Breeding and development. Wildl. Res. 7: Christenson, T.E., and Le Boeuf, B.J Aggression in the female northern elephant seal, Mirounga angustirostris. Behaviour, 64: Cobley, N.D., and Shears, J.R Breeding performance of gentoo penguins (Pygoscelis papua) at a colony exposed to high levels of human disturbance. Polar Biol. 21: Costa, D.P Reproductive and foraging energetics of pinnipeds: implications for life history patterns. In Behaviour of pinnipeds. Edited by D. Renouf. Chapman Hall, London. pp Cox, C.R Agonistic encounters among male elephant seals: frequency, context, and the role of female preference. Am. Zool. 21: Craig, M.P., and Ragen, T.J Body size, survival, and decline of juvenile Hawaiian monk seals, Monachus schauinslandi. Mar. Mamm. Sci. 15: Engelhard, G.H., van den Hoff, J., Broekman, M., Baarspul, A.N.J., Field, I., Burton, H.R., and Reijnders, P.J.H Mass of weaned elephant seal pups in areas of low and high human presence. Polar Biol. 24: Engelhard, G.H., Brasseur, S.M.J.M., Hall, A.J., Burton, H.R., and Reijnders, P.J.H Adrenocortical responsiveness in southern elephant seal mothers and pups during lactation and the effect of scientific handling. J. Comp. Physiol. B, 172: Enzenbacher, D.J Tourists in Antarctica: numbers and trends. Polar Rec. 28: Enzenbacher, D.J Antarctic tourism: an overview of 1992/93 season activity, recent developments, and emerging issues. Polar Rec. 30: Fedak, M.A., Arnbom, T., and Boyd, I.L The relation between the size of elephant seal mothers, the growth of their pups, and the use of maternal energy, fat, and protein during lactation. Physiol. Zool. 69: Fogden, S.C.L Mother young behaviour at grey seal breeding beaches. J. Zool. (Lond.), 164: Galimberti, F., Boitani, L., and Marzetti, I. 2000a. The frequency and costs of harassment in southern elephant seals. Ethol. Ecol. Evol. 12: Galimberti, F., Boitani, L., and Marzetti, I. 2000b. Harassment during arrival on land and departure to sea in southern elephant seals. Ethol. Ecol. Evol. 12: Gardiner, K.J., and Hall, A.J Diel and annual variation in plasma cortisol concentrations among wild and captive harbor seals (Phoca vitulina). Can. J. Zool. 75: Giese, M Effects of human activity on Adelie penguin Pygoscelis adeliae breeding success. Biol. Conserv. 75: Gill, J.A., Norris, K., and Sutherland, W.J Why behavioural responses may not reflect the population consequences of human disturbance. Biol. Conserv. 97: Green, W.C.H Reproductive effort and associated costs in bison (Bison bison): do older mothers try harder? Behav. Ecol. 1: Guinet, C., Jouventin, P., and Weimerskirch, H Recent population change of the southern elephant seal at Îles Kerguelen: the end of the decrease? Antarct. Sci. 11: Haley, M.P., Deutsch, C.J., and Le Boeuf, B.J A method for estimating mass of large pinnipeds. Mar. Mamm. Sci. 7: Hall, A.J., McConnell, B.J., and Baker, R.J Factors affecting first-year survival in grey seals and their implications for life history strategy. J. Anim. Ecol. 70: Hall, R.D., Leahy, J.P., and Robertson, W.M The effects of protein malnutrition on the behavior of rats during the suckling period. Devel. Psychobiol. 12: Haller, M.A., Kovacs, K.M., and Hammill, M.O Maternal behaviour and energy investment by grey seals (Halichoerus grypus) breeding on land-fast ice. Can. J. Zool. 74: Heath, M.E., and Schusterman, R.J Displacement sand flipping in the northern elephant seal (Mirounga angustirostris). Behav. Biol. 14: Hedd, A., Gales, R., and Renouf, D Use of temperature telemetry to monitor ingestion by harbour seal mother and her pup throughout lactation. Polar Biol. 15: Higgins, L.V., Costa, D.P., Huntley, A.C., and Le Boeuf, B.J Behavioral and physiological measurements of maternal investment in the Steller sea lion, Eumetopias jubatus. Mar. Mamm. Sci. 4: Hindell, M.A., and Burton, H.R Past and present status of the southern elephant seal (Mirounga leonina) at Macquarie Island. J. Zool. (Lond.), 213: Hindell, M., and Kemper, C Marine mammal research in the Southern hemisphere. Vol. 1. Status, ecology and medicine. Surrey Beatty and Sons, Chipping Norton, Australia.. Hindell, M.A., Slip, D.J., and Burton, H.R Possible causes of the decline of southern elephant seal populations in the southern Pacific and southern Indian Oceans. In Elephant seals: population ecology, behavior, and physiology. Edited by B.J. Le Boeuf and R.M. Laws. University of California Press, Berkeley and Los Angeles. pp Hofer, H., and East, M.L Biological conservation and stress. Adv. Study Behav. 27: IAATO Tourism statistics: Antarctic tourist trends. International Association of Antarctica Tour Operators ( iaato.org/tour_stats.html). Kovacs, K.M. 1987a. Maternal behaviour and early behavioural ontogeny of harp seals, Phoca groenlandica. Anim. Behav. 35: Kovacs, K.M. 1987b. Maternal behaviour and early behavioural ontogeny of grey seals (Halichoerus grypus) on the Isle of May, U.K. J. Zool. (Lond.), 213: Laws, R.M The elephant seal (Mirounga leonina Linn.). II. General, social and reproductive behaviour. Falkl. Isl. Depend. Surv. Sci. Rep. 13: Laws, R.M History and present status of southern elephant seal populations. In Elephant seals: population ecology, behavior, and physiology. Edited by B.J. Le Boeuf and R.M. Laws. University of California Press, Berkeley and Los Angeles. pp Le Boeuf, B.J., and Laws, R.M Elephant seals: population ecology, behavior, and physiology. University of California Press, Berkeley and Los Angeles. Le Boeuf, B.J., Whiting, R.J., and Grant, R.F Perinatal behavior of northern elephant seal females and their young. Behaviour, 43: Lewis, M.N., and Campagna, C Flipping sand in elephant seals. Aquat. Mamm. 24: Ling, J.K., and Bryden, M.M Mirounga leonina. Mamm. Species, 391: NRC Canada

47 1886 Can. J. Zool. Vol. 80, 2002 Can. J. Zool. Downloaded from by Santa Cruz (UCSC) on 11/15/12 For personal use only. Martin, P An experimental study of weaning in the cat. Behaviour, 99: Matthews, L.H The natural history of the elephant seal with notes on other seals found at South Georgia. Discov. Rep. 1: McCann, T.S Aggressive and maternal activities of female southern elephant seals (Mirounga leonina). Anim. Behav. 30: McCann, T.S Activity budgets of southern elephant seals, Mirounga leonina, during the breeding season. Z. Tierpsychol. 61: McCann, T.S., Fedak, M.A., and Harwood, J Parental investment in southern elephant seals, Mirounga leonina. Behav. Ecol. Sociobiol. 25: McMahon, C.R., van den Hoff, J., Burton, H.R., and Davis, P.D Evidence for precocious development in female pups of the southern elephant seal Mirounga leonina at Macquarie Island. In Marine mammal research in the Southern hemisphere. Vol. 1. Status, ecology and medicine. Edited by M. Hindell and C. Kemper. Surrey Beatty and Sons, Chipping Norton, Australia. pp McMahon, C.R., Burton, H.R., and Bester, M.N First-year survival of southern elephant seals, Mirounga leonina, at sub- Antarctic Macquarie Island. Polar Biol. 21: McMahon, C.R., Burton, H.R., and Bester, M.N Weaning mass and the future survival of juvenile southern elephant seals, Mirounga leonina, at Macquarie Island. Antarct. Sci. 12: Mendl, M., and Paul, E.S Observation of nursing and sucking behaviour as an indicator of milk transfer and parental investment. Anim. Behav. 37: Oftedal, O.T., Boness, D.J., and Tedman, R.A The behavior, physiology, and anatomy of lactation in the pinnipedia. Curr. Mammal. 1: Ortiz, C.L., Le Boeuf, B.J., and Costa, D.P Milk intake of elephant seal pups: an index of parental investment. Am. Nat. 124: Pistorius, P.A., Bester, M.N., and Kirkman, S.P. 1999a. Dynamic age-distributions in a declining population of southern elephant seals. Antarct. Sci. 11: Pistorius, P.A., Bester, M.N., and Kirkman, S.P. 1999b. Survivorship of a declining population of southern elephant seals, Mirounga leonina, in relation to age, sex and cohort. Oecologia, 121: Reijnders, P., Brasseur, S., van der Toorn, J., van der Wolf, P., Boyd, I., Harwood, J., Lavigne, D., and Lowry, L Seals, fur seals, sea lions, and walrus. Status Survey and Conservation Action Plan. International Union for Conservation of Nature and Natural Resources, Gland, Switzerland. Renouf, D., Gaborko, L., Galway, G., and Finlayson, R The effect of disturbance on the daily movements of harbour seals and grey seals between the sea and their hauling grounds at Miquelon. Appl. Anim. Ethol. 7: Salter, R.E Site utilization, activity budgets, and disturbance responses of Atlantic walruses during terrestrial haul-out. Can. J. Zool. 57: Slip, D.J., and Burton, H.R Population status and seasonal haulout patterns of the southern elephant seal (Mirounga leonina) at Heard Island. Antarct. Sci. 11: SPSS Inc SPSS for Windows. SPSS Inc., Chicago. Suryan, R.M., and Harvey, J.T Variability in reactions of Pacific harbor seals, Phoca vitulina richardsi, to disturbance. Fish. Bull. 97: Tedman, R.A., and Bryden, M.M Cow pup behaviour of the Weddell seal, Leptonychotes weddelli (Pinnipedia), in McMurdo Sound, Antarctica. Aust. Wildl. Res. 6: Thiel, M., Nehls, G., Bräger, S., and Meissner, J The impact of boating on the distribution of seals and moulting ducks in the Wadden Sea of Schleswig-Holstein. In Proceedings of the 7th International Wadden Sea Symposium, Ameland, the Netherlands, October Edited by N. Dankers, C.J. Smit, and M. Scholl. Netherlands Institute for Sea Research, Texel, the Netherlands, Publ. Ser. 20. pp Thomson, C.A., and Geraci, J.R Cortisol, aldosterone, and leucocytes in the stress response of bottlenose dolphins, Tursiops truncatus. Can. J. Fish. Aquat. Sci. 43: Trillmich, F Parental investment in pinnipeds. Adv. Study Behav. 25: Wilkinson, I.S., and Bester, M.N Is onshore human activity a factor in the decline of the southern elephant seal? S. Afr. J. Antarct. Res. 18: Woehler, E.J., Penney, R.L., Creet, S.M., and Burton, H.R Impacts of human visitors on breeding success and long-term population trends in Adélie penguins at Casey, Antarctica. Polar Biol. 14: Zar, J Biostatistical analysis. 3rd ed. Prentice Hall Inc., Upper Saddle River, N.J NRC Canada

48 Report Clive McMahon, John van den Hoff and Harry Burton Handling Intensity and the Short- and Long-term Survival of Elephant Seals: Addressing and Quantifying Research Effects on Wild Animals This study addresses the consequences of repeated human handling on the survival of an endangered phocid, the southern elephant seal and the implications for wildlife research. Southern elephant seal pups were repeatedly handled during the first six weeks of their lives. The possibility that such anthropogenic research may have altered the very parameters that were being investigated is a topical and relevant study area that we address here. Our results show that there were no measurable effects on pups that were repeatedly handled and subjected to invasive research methods with respect to survivorship in the short term (the 24-day nursing period) nor in the long term (the first year of life and beyond) and hence fitness one year after handling. In support of this conclusion we were unable to detect any significant differences in the survival rates of the most intensively handled seals and the least intensively handled seals. INTRODUCTION Sound wildlife management is necessarily underpinned by sound scientific study of the basic biology and demography of the species in question (1, 2). Research directed at determining these basic parameters for large, free-ranging vertebrate species often requires the use of invasive methods such as marking, implanting or attaching radio tracking devices, taking blood samples, and anesthetizing and/or manually restraining the subjects (3). Therefore, it is important that the consequences (e.g. survival of handled animals) of such techniques be investigated so that the animals welfare and the resultant findings are not affected or compromised by the research techniques being employed (1). This is especially pertinent when studying endangered animal populations (2). As a result, research conducted in Antarctica, as elsewhere, is subject to approval from various scientific and ethical review bodies that assess the impacts of research on the biotic and abiotic components of ecosystems, but little is known about the Antarctic ecosystem (4) and even less about the impacts of research activities upon it and its inhabitants. For this reason bodies charged with regulating research activities tend to take a precautionary approach to permitting research that is often i) adopted by word of mouth, ii) not supported by experimental evidence, and iii) subjective (5). Indeed, researchers themselves are often to blame because they rarely address issues related to their own impacts because i) handling effects are typically assumed to be negligible, ii) cost often precludes such studies, and iii) appropriate controls to test for the effects are often unavailable (6). Clearly, in the interests of the animals welfare and sound research, there is a need for structured research that objectively addresses both the ethical and scientific objectives of the research undertaken. Customarily, the assessment and measurement of anthropogenic disturbance as a result of scientific activities on freeranging animals is rare, and this is especially true for marine mammals (6). A germane example is the southern elephant seal (Mirounga leonina), one of the most widely and exhaustively studied of all pinnipeds (7). Despite the southern elephant seal being so well studied, it was not until recently that studies were undertaken to assess the effect of human activities upon this species. Engelhard (8) reviewed the immediate and short-term responses and impacts of invasive scientific research on elephant seals at Macquarie Island and concurred with a single earlier study (9) that onshore human disturbance did not contribute to the long-term population decreases observed at the various breeding locations. Burton and van den Hoff (10) reviewed a number of different human interactions with the southern elephant seal and found the seals appeared to be remarkably unaffected by these interactions. However, other studies have shown adverse short-term changes in behavior and physiology resulting from human activity (11 16), but none has determined the long-term effects of such handling. We hypothesized that as the intensity of the researchassociated handling increased during the lactation period and soon after weaning, there might be an adverse response shown by the seals. The seals response to handling can be measured in their survivorship because their ultimate survival prospects are initially solely reliant on the resources (blubber or fat) they accumulate during nursing and have available at the end of their postweaning fast (17, 18). Therefore, activities that disrupt the accumulation of resources during this period may be reflected in the measured first-year survival estimates. We believe that for the first time the effect of cumulative research disturbance is quantified against the survival of an endangered (in Australia) marine mammal. METHODS Southern elephant seal pups were studied at Macquarie Island ( S, E) as part of a long-term demographic study to collect life history information. Morphometric data (19, 20) could not be collected without first capturing and restraining the seals either manually or chemically (19, 20). Therefore, some degree of human handling was unavoidable. We recognized two handling intensity groups that were based on the number of captures prior to weaning (nursing period) and four groups during the postweaning period. During the 24-d nursing period, 32 pups, which were double tagged in their rear flippers, were handled more than once; we placed those seals that were handled three times or more (n ¼ 14) into an intensively handled group and those that were handled less than three times into a low intensity group (n ¼ 18). These two groups of seals were of comparable size (t 26 ¼ 0.164, p ¼ 0.9) in terms of mass ( kg and kg, respectively). Seals that were captured after weaning were assigned to four groups depending on the handling intensity, i.e. the number of handling episodes. These seals were permanently marked by hot-iron branding (18). Group 1 (n ¼ 934) had the lowest handling intensity, and group 2 (n ¼ 857) and group 3 (n ¼ 28) had moderate handling intensity. Group 4 (n ¼ 94) was the most 426 Ó Royal Swedish Academy of Sciences 2005 Ambio Vol. 34, No. 6, August

49 intensely handled group and was likely to be most affected by research activities and so their survival probability might be expected to be lower than other groups. This is because each handling episode would initiate a flight or fight response (21) and thus necessitate the mobilization of proportionally greater amounts of blubber reserves compared with the lowest handling intensity group (group 1). Premature mobilization of these blubber stores necessarily means less blubber reserves are available to the seals when they depart their natal island for the first time, and this reduction in reserves may compromise their survival (17). We assumed that the energy expended in flight or fight responses might be a function of the number of exposures so that seal survival in the moderately handled groups would fall between the two extremes. Because it is well known that mass at weaning is a significant predictor of survival in the first year of life (17, 18), it is important to know a priori and test ad hoc whether the wean masses were different between the handing groups. We tested the weaning masses of the animals in our four groups and found that there were no differences (F 3, 399 ¼ , p ¼ 0.89) in weaning masses. Capture-history matrices were constructed from the resight history of individually marked seals. These matrices were used as input files for the capture-mark-recapture (CMR) program MARK (22) to estimate capture probabilities and survival of the study population. MARK provides survival (/) and recapture (q) estimates under the Cormack-Jolly-Seber (CJS) model and under several models that appear as special cases of the CJS model (23). We tested our data using the RELEASE program (24) to determine the validity of applying the CJS model in MARK and to test whether the data conformed to the CJS-model assumptions (i.e. goodness of fit, GOF). Starting from a general model that fit the data, we used the Akaike information criteria scores (AIC) and DAIC scores to select the most parsimonious model(s) (25). Likelihood ratio tests within MARK were used to test specific hypotheses (23). As a guide to model selection, models with DAIC 2 have substantial support, those with DAICs of 4 to 7 have some support, and those models with DAIC. 10 have no support (25). RESULTS The GOF test within MARK showed that the data we used here conformed to the general CJS-model assumptions (v 3 2 ¼ , p ¼ ) so that there was equality in the survival and recapture probabilities among each of the treatment groups. Therefore, all data were analyzed in MARK under CJS. Seal recapture probabilities were similar for all categories (intensities) of handled seals (p 0.05, v 1 2 ¼ 0.00), and the model that included constant recaptures [q(c)] performed as well as the model that included variable annual recapture rates [q(t)]. Figure 1. The mean (6 95% confidence intervals) first-year survival (closed diamonds) and recapture (open circles) probabilities for seals that were handled prior to weaning and divided into two intensity groups low (n ¼ 18) and high (n ¼ 14). Figure 2. The mean (6 95% confidence intervals) first-year survivorship (closed diamonds) estimates and recapture (open squares) estimates for four groups of southern elephant seal pups. Seals in intensity group 1 were those that were handled least and those in intensity group 4 were the seals that were handled most. Ambio Vol. 34, No. 6, August 2005 Ó Royal Swedish Academy of Sciences

50 We observed (by counting dead pups) the preweaning mortality of pups in i) our intensively handled group (n ¼ 32) and recorded a mortality rate of 3.2%, ii) in our marked cohort of seals (n ¼ 982), where the mortality rate was 4.2%, and iii) in a control harem of seals (n ¼ 225), where the preweaning mortality was 3.6%. The mean first-year survival estimates for the subset of seals that were handled at a low intensity during the nursing period was (n ¼ 18) and for seals that were handled at a high intensity, the survival estimate was (n ¼ 14) (Fig. 1). No differences were demonstrable in these estimates (v 2 2 ¼ 2.172, p ¼ ). The mean estimate for first-year survival for all seals marked in 1998 was (n ¼ 1000). The inclusion of a gender component to the general survival model did not improve model performance, and accordingly it was concluded that first-year survival estimates were unaffected by seal gender (v 1 2 ¼ 0.627, p¼0.43). The survival estimates and recapture estimates for seals in the four handling groups were similar (Fig. 2). No statistical differences were detectable in the first-year survival or recapture estimates for seals subjected to low intensity handling and seals subjected to high intensity handling (v 9 2 ¼ 15.16, p ¼ 0.087). The most parsimonious model did not include the handling intensity groups but did include age-specific survival for the first three years of life (Fig. 3), and annual differences in recapture rates but not handling group differences (Table 1). DISCUSSION The survival probabilities for southern elephant seals that were handled intensely were found to be no different than those for seals that were handled less often or not handled at all. To make such a statement is rare because few scientific studies have so specifically monitored and addressed the effect of anthropogenic research on free-living wildlife. This is an important and pertinent area of investigation in ecology because there is always the possibility of a sampling artifact or harmful occurrences when animals are handled, which in turn can or may influence the very parameters that are being investigated. Survival is one of these parameters, and it has been identified as a fundamental determinant of population growth (1). We estimated first-year survival in our study and could find no differences in the survival of seals due to handling intensity. In the absence of any detectable increased mortality in response to repeated bouts of invasive research techniques, it followed that elephant seal population growth was unlikely to be affected by human research activities. These findings lend quantitative support to Burton and van den Hoff s (10) earlier suggestion that elephant seals appeared remarkably unaffected by close human contact. Elephant seals are not the only pinnipeds to show such a response. A previous (6) study on the endangered (1400 remaining individuals) monk seal (Monachus schauinslandi) also found no negative affect of Figure 3. The age-specific survival probability for four groups of handled elephant seals in the first three years of life. Each of the estimates is bound by the 95% confidence intervals. Table 1. Akaike information criteria (AICc), delta AIC (DAICc), Akaike weights, model likelihood, number of parameters, and the deviance for the candidate survival (/) and recapture (q) models. Model AICc DAICc AICc weights Model likelihood Num. par Deviance /(age 1 3 specific survival) q(annual) /(age 1 3 specific survival* handling group) q(annual) /(handling group*annual non age-specific ) q(annual) /(age 1 3 specific survival) q(annual*handling group) /(constant) q(annual) /(age 1 only specific survival) q(annual) /(handling group) q(annual) /(age 1 only specific survival*handling group) q(annual) /(handling group*annual) q(handling group*annual) /(constant) q(handling group*annual) /(handling group) q(handling group*annual) /(annual) q(constant) /(annual) q(handling group) /(handling group*annual) q(constant) /(handling group*annual) q(handling group) /(constant) q(constant) /(handling group) q(constant) /(constant) q(handling group) /(handling group) q(handling group) Ó Royal Swedish Academy of Sciences 2005 Ambio Vol. 34, No. 6, August

51 tagging, instrument attachment, and blood sampling on the survival, migration, or body condition of the seals. The effects of anthropogenic disturbance in most mammals have only been studied in the immediate to short term timescales, i.e. hours or days (26); however, we studied the effects (mortality) in both the short and long terms. We hypothesized that increased mortality would probably occur as a result of i) decreased suckling time caused by disrupting mother and pup activities resulting in a smaller than average weaning weight, ii) increased metabolic rates in response to the hormone-induced action of the flight or fight response such that valuable resources are prematurely utilized and not available during the first foraging trip, and iii) altering physiological states such as immune responses. Our observed preweaning mortality rates (3.2% to 4.2%) are similar to those reported from Marion Island (3.8%) (27) where elephant seals were not handled prior to weaning, and showed that our research activities did not add to the already naturally occurring mortality. Our data support previous findings (6, 14 16) that there were few detectable effects of handling seals in the short term. But our data are novel because we assessed the long-term survival of the study seals at ages 1, 2, and 3 years. These are important findings because they show that the consequences of single or repeated handling during lactation and the postweaning fast of elephant seal pups have neither a short-term effect, and therefore support previous work (14 16), nor a long-term effect that compromises survival (this study). However, we recognize that this study lacks an unhandled control group, which represents a potential weakness in the present study. This is because unlike some pinnipeds (e.g. Hawaiian Monk seals), elephant seals lack any readily distinguishing phenotypic features that can be used to confidently identify individuals; this means that elephant seals must necessarily be marked and therefore handled at least once. In the absence of individual natural markings, we cannot determine the survival of unmarked and thus unhandled seals. We compensated for this by relating the probability of survival of seals to a gradientdependent variable (handling intensity). We assumed a decreased survival probability would be apparent with increased handling frequency because in penguins (28) it was found that anthropogenic disturbance was gradient-dependent such that stress and heart rate responses were highest in those penguins that received the highest level of disturbance. In the absence of such a gradient-dependant response in elephant seal pups, we concluded that handling does not affect first-year survival, and that handled seals have similar survival probabilities to nonhandled animals. Even though we recognize that each species has the potential to respond differently to human interactions, the evidence is beginning to point toward a rather unexpected finding that, in the short and long term, some endangered and presumably sensitive species are quite resilient to handling for research purposes. The incorporation, continuation, and expansion of the type of investigation we undertook here at the individual and population level will contribute in a direct manner toward the formulation of sound animal research policy. References and Notes 1. Caughley, G Analysis of Vertebrate Populations. John Wiley and Sons, London. 2. Caughley,G.and Gunn,A Conservation Biology in Theory and Practice. Blackwell Science, Cambridge. 3. Grigg, G Wildlife research-why is it conducted? In: The Use of Wildlife for Research. Mellor, D. and Monamy, V. (eds). Australian and New Zealand Council for the Care of Animals in Research and Teaching (ANZCCART), Adelaide, pp Anderson, C., Coles, P., Aldhous, P. and Dickman, S Exploring the still unexplored: research in Antarctica. Nature 350, Plous, S. and Herzog, H Reliability of protocol reviews for animal research. Science 293, Baker, J.D. and Johanas, T.C Effects of research handling on the endangered Hawaiian monk seal. Mar. Mammal Sci. 18, Ling, J.K. and Bryden, M.M Mirounga leonina. Mammal. Species 391, Engelhard, G.H Southern Elephant Seal Population Declines: The Human Onshore Disturbance Hypothesis. PhD Thesis, Groningen University, AA Haren, The Netherlands. 9. Wilkinson, I.S. and Bester, M.N Is onshore human activity a factor in the decline of the southern elephant seal? S. Afr. J. Antarctic Res. 18, Burton, H.R. and van den Hoff, J Humans and the southern elephant seal (Mirounga leonina). Aust. Mammal. 24, Reijnders, P., Brasseur, S., van der Torn, J., van der Wolf, P., Boyd, I., Harwood, J., Lavigne, D. and Lowry, L. (eds.) Seals, Fur Seals, Sea Lions and Walrus. Status Survey and Conservation Action Plan. IUCN/SSG Seal Specialist Group. IUCN, Gland, Switzerland, 88 pp. 12. Suryan, R.M. and Harvey, J.T Variability in reactions of pacific harbor seals, Phoca vitulina richardsi, to disturbance. Fish. Bull. 97, Born, E.W., Riget, F.F., Dietz, R. and Andiashenk, D Escape responses of hauled out ringed seals (Phoca hispida) to aircraft disturbance. Polar Biol. 21, Engelhard, G.H., van den Hoff, J., Broekman, M., Baarspul, A.N.J., Field, I., Slip, D.J., Burton, H.R. and Reijnders, P.J.H Mass of weaned elephant seal pups in areas of low and high human presence. Polar Biol. 24, Engelhard, G.H., Brasseur, S., Hall, A.J., Burton, H.R. and Reijnders, P.J.H Adrenocortical responsiveness in southern elephant seal mothers and pups during lactation and the effect of scientific handling. J. Comp. Physiol. B 172, Engelhard, G.H., Hall, A.J., Brasseur, S.M.J.M. and Reijnders, P Blood chemistry in southern elephant seal mothers and pups during lactation reveals no effects of handling. Comp. Biochem. Physiol., A 133, McMahon, C.R., Burton, H.R. and Bester, M.N Weaning mass and the future survival of juvenile southern elephant seals, Mirounga leonina, at Macquarie Island. Antarctic Sci. 12, McMahon, C.R., Burton, H.R. and Bester, M.N A demographic comparison of two southern elephant seal populations. J. Anim. Ecol. 72, McMahon, C.R., Burton, H., McLean, S., Slip, D. and Bester, M Field immobilisation of southern elephant seals with intravenous tiletamine and zolazepam. Vet. Rec. 146, Field, I.C., Bradshaw, C.J.A., McMahon, C.R., Harrington, J. and Burton, H.R Effects of age, size and condition of elephant seals (Mirounga leonina) on their intravenous anaesthesia with tiletamine and zolazepam. Vet. Rec. 151, Cannon, W.B The Wisdom of the Body. Norton, New York. 22. White, G.C. and Burnham, K.P Program MARK: Survival estimation from populations of marked animals. Bird Study 46, Lebreton, J.D., Burnham, K.P., Clobert, J. and Anderson, D.R Modeling survival and testing biological hypotheses using marked animals: a unified approach with case studies. Ecol. Monogr. 62, Burnham, K.P., Anderson, D.R., White, G.C., Brownie, C. and Pollock, K.H Design and analysis methods for fish survival experiments based on release-recapture. Am. Fish. Soc. Monogr. 5, Burnham, K.P. and Anderson, D.R Kullback-Leibler information as a basis for string inference in ecological studies. Wildl. Res. 28, Hofer, H. and East, M.L Biological conservation and stress. Adv. Study Behavior 27, Pistorius, P.A., Bester, M.N., Kirkman, S.P. and Taylor, F.E Pup mortality in southern elephant seals at Marion Island. Polar Biol. 24, Culik, B., Adekung, D. and Woaks, A.J The effect of disturbance on the heart rate and behavior of Adelie penguins (Pygoscelis adeliae) during the breeding season. In: Antarctic Ecosystems. Ecological Change and Conservation. Kerry, K. and Hempel, G. (eds.). Springer-Verlag, Berlin and Heidelberg, pp We acknowledge the assistance so freely given by our colleagues of the 52nd and 53rd ANARE expeditions to Macquarie Island. The Australian Antarctic Animal Ethics Committee (ASAC 2265) and the Tasmanian Parks Service approved and permitted our research at Macquarie Island. We thank an anonymous reviewer for providing a constructive critique of the final draft of the manuscript. 30. First submitted 3 Dec Revised manuscript received 4 May Accepted for publication 25 May Clive McMahon, PhD, is currently a postdoctoral fellow at the University of Wales, Swansea. This study was undertaken when he worked with the Australian Antarctic Division, 203 Channel Highway, Kingston, 7050, Tasmania, Australia, where he specialized in the study of elephant seal demographics for 10 years. His address: School of Biological Sciences, University of Wales Swansea, Singleton Park, Swansea SA2 8PP, Wales. C.R.McMahon@swansea.ac.uk Wales.C.R.McMahon@swansea.ac.uk John van den Hoff (MSc) is a marine mammal ecologist. His main interests are the relationships between foraging behavior and diet of elephant seals. His address: Australian Antarctic Division, Channel Highway, Kingston Tasmania, Australia. John.Vandenhoff@aad.gov.au Harry Burton BSc (Agric), is a senior research scientist at the Australian Antarctic Division. He has been researching various aspects of Antarctic biology for 30 years, including the longterm study of elephant seals. Harry has contributed significantly to the study of mammals in Antarctica and is a member of the SCAR seal specialist group. His address: Australian Antarctic Division, Channel Hwy, Kingston, 7050, Tasmania, Australia. Harry.Burton@aad.gov.au Ambio Vol. 34, No. 6, August 2005 Ó Royal Swedish Academy of Sciences

52 REVIEWS REVIEWS REVIEWS Measuring devices on wild animals: what constitutes acceptable practice? 147 Rory P Wilson * and Clive R McMahon In a world that is increasingly perturbed by humans, the need to understand ecosystems is urgent. Attaching measuring devices to wild animals is often the only way to acquire vital life-history information on larger, charismatic species, and on cryptic species that do not lend themselves to observation. However, the ethics of acceptable practice for attached devices are poorly defined. Here, we consider the need for further research and attempt to identify a system that allows animal restraint practices and device-induced effects to be quantified and monitored, so that ethics committees can have a defined scale on which to base decisions. Front Ecol Environ 2006; 4(3): Currently, regulatory bodies charged with overseeing research activities on wild animals tend to take a precautionary approach, because little has been published that quantifies the behavioral and life-history effects of capturing and attaching devices to such animals (Plous and Herzog 2001). Indeed, researchers themselves are often to blame for this lack of information, because they rarely address issues related to their own impacts, and assume that handling effects are negligible. Cost may also preclude handling effects studies and appropriate controls on animals to test for the effects are often unavailable (Baker and Johanas 2002). Clearly, in the interests of animal welfare and sound research there is a need for structured research that objectively discusses and addresses both the ethical implications and scientific objectives of the research being undertaken (Minteer and Collins 2005). We suggest that the paucity of assessment and measurement of anthropogenic disturbance caused by measuring devices on free-ranging animals is partly due to the lack of guidelines for researchers on appropriate quantitative measures to use when examining the effects of their research on animals. We therefore urgently need to formulate definitions of what is acceptable, in an unbiased way that is both easily measurable and transparent. Customarily, In a nutshell: As most of the world's fauna are directly affected by humans, there is an urgent need for further study of animal biology to understand the consequences of this impact While attaching transmitters or loggers to animals is often the only way to acquire necessary data, the ethics of animal handling and the attachment of such devices is poorly defined and controversial Quantifying the effects of handling and of device attachment could clarify issues and help facilitate decisions made by ethics committees 1 Institute of Environmental Sustainability, School of Environment and Society, University of Wales Swansea, Singleton Park, Swansea SA2 8PP, Wales, UK *(r.p.wilson@swansea.ac.uk) scientists rely on quantification of data; we argue that animal welfare considerations can, and should, also be couched in numerical terms, rather than expressed in language that is open to interpretation. Here, we present some practical protocols for quantifying and minimizing the negative effects of capture and tagging. To apply numerical techniques to animal welfare questions, we must identify features that indicate the wellbeing of the animals and then quantify the extent to which these features vary in study animals. Ideally, we should attempt to determine the norm in animal controls, but because measurement affects performance (Wilson et al. 1986) this is not always possible. In particular, however, quantification of the degree of departure from the norm will allow us to identify precise boundaries for what is acceptable and what is not. Features that indicate levels of animal welfare include measures of, for example, fear, pain, or reproductive disturbance. These features can be considered in terms of three different numerical scales: (1) the extent of the detriment; (2) the number of animals involved in the procedure (eg individuals, family groups, local accumulations, populations, etc); and (3) the length of time over which the detrimental effect occurs (from seconds to decades). We advocate attempting to quantify these three features in study animals and multiplying them together to obtain a measure of the overall detriment index. Even if initial estimates are largely based on guesswork, this should still serve as a useful point of departure from which ethics committees can work, being less prone to bias, and will function as a baseline to which relevant data can be added as they are acquired. Although this approach will not satisfy all parties interested in animal welfare (eg see Jabour-Green and Bradshaw 2004), transparency can help to ease tensions. Defining human impacts Humans now number approximately 6.5 billion, so it is inevitable that we should impact our environment sub- The Ecological Society of America

53 Measuring devices on wild animals RP Wilson and CR McMahon 148 return to the breeding site (Phillips et al. 2003). However, these are rather crude assessments, which do not allow for the ability of the animals to correct for disadvantages by increased effort (Hamel et al. 2004). Indeed, a real idea of the potential consequences of device-induced effects can only be gained by modeling experiments, using data from both models and free-living animals (Panel 1). The main dilemma is that we need to assess the effects of the measuring devices in free-living animals, while the very means of measurement are themselves affecting performance (Wilson et al. 1986). Figure 1. A juvenile hawksbill turtle equipped with a time depth recorder is released off Arnhem Land, Northern Australia. Since turtles spend most of their lives underwater, they are not easy to observe. This turtle was equipped as part of a conservation program that aims to understand the local movements and behavior of turtles and to teach local aboriginal communities about wildlife research by getting them to work with experienced researchers. stantially, perhaps catastrophically (Burney and Flannery 2005). We are very much a part of the environment that we need to conserve for our collective good (Ehrlich and Ehrlich 1992) and therefore have a responsibility to determine animal needs and their roles in their chosen habitats. However, the importance of a particular animal in an ecosystems is not a function of its visibility (Andelman and Fagan 2000; Carignan and Villard 2002), so shy or cryptic species must also be monitored. This necessitates methods that involve handling and the attachment of devices that allow us to document behavior and life history in places and at times where direct observations are not possible (Ropert-Coudert and Wilson 2005; Figure 1). In short, we have no option but to equip animals with devices in order to study them, yet by doing so we generate ethical questions. There are numerous ways in which animals can be affected by the attachment of devices (Wilson et al. 1986; Marsh and Kenchington 2004), ranging from the problems that arise as a result of capture and handling (McMahon et al. 2005) to the psychological and physical stress of carrying a foreign body (Wilson et al. 1986; Hawkins 2004). Fortunately, there is increasing interest in documenting the deleterious effects of external devices on their animal carriers. For example, the seabird literature contains examples of alterations in duration of foraging trips, meal masses, breeding success, and rates of Detrimental effects The discomfort index Ethics committee statutes commonly use words like suffering and pain that are likely to incite polarized views. This is because they represent a broad array of sentiments and are expressions that may be perceived differently by various sections of society. We suggest that discomfort be used in such instances, because it can represent a broad dynamic range without inherent bias and can be qualified by adjectives such as mild or extreme. It therefore fulfils the scientist s need for a scalar quantification of the parameter indeed, we would argue for a discomfort index ranging from, for example, 0 to 10. Discomfort also seems particularly appropriate when Panel 1. The cost to penguins of having to carry external devices Attempts to determine possible deleterious effects of external devices on penguins are complicated by the difficulty of monitoring birds at sea. However, measurements on the drag resulting from an external antenna fitted to a model (Magellanic) penguin in a water channel showed that antennae that constitute just 0.16% of a real penguin s mass increase drag at normal swimming speeds of 1.77 m s -1 by about 80%, so that energy expenditure increases by a similar amount.theoretically, birds in the wild could mitigate against this by consuming more food. However, studies using devices attached to free-living Magellanic penguins show that their prey (anchovies) in at least one area in their breeding range, are only encountered approximately once every 30 minutes or so of searching, at which time birds are able to feed on them during an average of 2.7 dives before the shoal escapes. Prey capture necessitates speeds of 2.25 m s -1, which increases energy-expenditure in antenna-carrying birds by 147%. This, in turn, uses body oxygen reserves faster, limiting dive duration and the number of fish that can be caught per dive. Thus, antenna-carrying birds use more energy and acquire less fish. Modelling studies indicate that antennacarrying birds are only about one-fifth as efficient as those not carrying this device (Wilson et al. 2004). The Ecological Society of America

54 RP Wilson and CR McMahon Measuring devices on wild animals referring to animal-attached devices as well as for procedures used to capture and handle animals prior and subsequent to the addition of the device. Finally, discomfort is a term that describes an instantaneous state and covers everything from mild discomfort to extreme pain. The value of a discomfort index is complicated by the process of attempting to define what the study animals actually perceive as discomfort. There is a danger in basing everything on human sentiments. Consider, for example, the case of an animal having to ingest a recording device (Ropert-Coudert and Wilson 2005). Many humans will easily ingest pills of up to about 2 g and could therefore theoretically swallow a 2 g stomach-temperature logger with minimal discomfort, although a logger weighing 100 g would be highly problematic. Such an act would probably be trivial for an albatross, however, that regularly swallows whole prey items exceeding 500 g (Cooper et al. 1992). The degree of perceived discomfort for the albatross could easily be put into perspective by someone working on albatross diet, so we suggest that ethics committees could benefit by asking experts in the relevant field for an opinion on possible discomfort levels. This would be somewhat akin to the peer-review process used for scientific publications, and would counterbalance the lack of specialist knowledge within the committee, thereby placing the animal welfare issue into a more informed context. The expert need not (and perhaps should not) be involved in any process other than ranking the discomfort level of the procedure, thereby minimizing their involvement in the decision making regarding the suitability of the technique. Energetics Animal discomfort caused by a procedure can often be documented by observing their behavior. However, being fitted with devices may cause physical handicaps that are less obvious. Perhaps the most important of these is energy expenditure, given that the balance of energy expenditure with intake is what ultimately ensures survival. The presence of an attached device may incur increased energy expenditure for a given activity or may even prevent the activity from being performed (Weimerskirch et al. 2002). Some measure of the effect of such devices on the energetics of the animal is highly desirable and this also should be in the form of numerical data (eg Weimerskirch et al. 2002). An understanding of biomechanics, considered in tandem with the costs of locomotion (using published values based on indirect calorimetry, for example) can help to estimate the cost to the animal of carrying a device. An appropriate expert panel could set out general guidelines (to be used by ethics committees), based on the extra effort that animals need to exert when carrying a device. The first step would be to calculate the power output of an animal with no device, using Newtonian rules to approximate elements such as drag. The same calculations would then be made for an animal with a device. The literature includes many values for conversion efficiency (eg Bishop 2005), so that the (1) (2) (3) (4) (5) (6) (7) (8) Obtain an estimate for power used by animal to travel (from standard animal types, allometry, and movement types): eg P in = f(velocity) Calculate the force involved: eg F animal = drag = f(vel) Compute P in derived from the force: eg P in = f(drag) Estimate extra force required to transport attached unit: F unit Add forces together; F unit + F animal Derive new P in for equipped animal from (3) Compare with P in for unequipped Consider with respect to times spent moving Figure 2. Procedure for estimating the impact of animal-attached devices on the energetics of movement of free-living animals. link between power output and power input could almost be calculated generically. Values for device-related power requirements can then be compared to overall animal power requirements (which depend, among other things, on the size of the animal) to put the effects of the device into perspective (Figure 2). The derived values for increased energy expenditure need to be multiplied by the probable time that the animal spends moving, in order to gain an overall estimate of the energy cost. Such an approach would allow us to do away with the simplistic 5% rule, which says, for example, that birds should not carry devices that exceed their mass by more than 5% (Hawkins 2004), irrespective of wing-loading, type of flight used (powered flight or gliding), or the percentage time that a bird actually spends in the air (Figure 3). Another method for estimating energy expenditure in animals equipped with devices involves the use of gas respirometry (Culik et al. 1996), but such experiments are costly, and contentious for welfare reasons. Nonetheless, it is appropriate that we work towards quantification of device-induced effects and a potential solution lies in the use of accelerometers to quantify animal effort. Research on humans has shown that energy expenditure and body acceleration are highly correlated (Foster et al. 2003). This is hardly surprising, since muscular activity requires energy and also tends to bring about acceleration. In wildlife applications, animals can be fitted with very 149 The Ecological Society of America

55 Measuring devices on wild animals RP Wilson and CR McMahon 150 Courtesy of H Weimerskirch Figure 3. Frigate bird fitted with a satellite-tracking system to determine foraging habits and area use. This species forages in flight and thus has a number of specialized features, including an extraordinarily light skeleton to reduce wing loading. The mass of the attached systems is therefore critical. small accelerometers to quantify this aspect as a function of standardized motion (Sawai et al. 2005). During different applications, where devices of varying sizes or masses are attached to the animal, the degree of overall acceleration (which correlates with energy expenditure) exhibited by the animal during standard motion (such as flight) can be compared with device size. A regression analysis of acceleration versus device size would allow extrapolation to the intercept, thereby providing an estimate of the acceleration signal (energy) for the unequipped animal (Wilson et al. 1986). In this way, the effect of any attached device could subsequently be quantified (Figure 4). Survival probabilities and group problems Increased energy expenditure as a result of an attached device may affect life-history traits and even survival. Relative acceleration Device size/mass/drag... Difference between equipped and unequipped { Predicted intercept for unequipped conspecific Figure 4. Example of how increased mass affects the acceleration achieved by a bird during flight and how recording this acceleration may be used to determine the effect of attached devices. Values for increased acceleration associated with more costly movement can be compared to measures of attached device size to calculate the cost of movement in individuals without devices. Animals that cannot forage successfully cannot provision themselves or their offspring efficiently and are therefore at a disadvantage. The same applies when devices are the wrong color or shape, or have been attached using an inappropriate method. As a result, the animal may expend more energy in preening or grooming (Wilson and Wilson 1989) or become less vigilant against predators or conspecific attacks. Alternatively, a feature such as the color of the device may expose tagged individuals to higher levels of predation by making them stand out (Hawkins 2004; Marsh and Kenchington 2004). Consequently, ethics committees should recognize the importance of ensuring that devices generally blend in with the animal s body color and form. Realistically, even the act of capturing wild animals is likely to compromise their survival to some extent (though the effect may be very small). It is important to recognize that no research of this type can be carried out without the animals being affected in some way. Again, approximate survival probabilities, ranged according to a ranking procedure, would be a valuable resource for ethics committees. Examples include increased mortality (for example, over a year) as compared to non-equipped conspecifics of < 1%, < 5%; and < 10%, etc. Although actual measurement of survival will tend to suffer from the measurement affects performance problem, two recent developments may help. First, where animal survival is assessed using multiple tags (Froget et al. 1998), survival can be plotted against tag number, which will allow the survival of non-tag wearers to be estimated (cf Figure 4). Second, researchers are becoming increasingly aware that individual animals can often be recognized on the basis of natural markings (Stonehouse 1978; Vincent et al. 2001; MacDiarmid et al. 2005). The survival, fate, and wellbeing of individuals can be determined by autonomous computer/film systems, set up in the field, thus avoiding the necessity of capturing or marking the animals at all (Heilbrun et al. 2003). In any event, the likelihood of survival of handled animals should also be extended to associated individuals. The most obvious example of this is in breeding animals, where handled animals may be affected to the point The Ecological Society of America

56 RP Wilson and CR McMahon Measuring devices on wild animals (a) (b) 151 Photos a c courtesy of H Perau (c) Figure 5. The manual capture of a southern elephant seal. (a) Prior to restraining the seal; (b, c) the intravenous administration of an anesthetic agent, in this case a commercially available mixture of teletamine and zolazepam (Zoletil). This drug typically induces anesthesia within 30 seconds (see McMahon et al. 2000). some years, have a lower breeding probability, and produce fewer chicks. In addition, non-banded chicks are twice as likely to survive to 2 3 years as banded individuals (Gauthier-Clerc et al. 2004). How many is enough? that they provision dependents less well (cf Hamel et al. 2004). However, conspecifics that are spatially proximate (eg fur seals; Stirling 1970) may also be affected, and this needs due consideration. The time factor Estimates of survival probabilities and discomfort indices need to be considered with respect to time. How long should animals carry measuring devices? The calculation is simple and involves multiplying the discomfort index, the energy compromise index, and the likelihood of survival index by the time over which the animal has to wear the device. This leads to an overall acceptable practice index, even if it is couched in terms of separate numbers. From these calculations, it is clear that shortterm attachment of devices with quite deleterious effects can be considerably less harmful than long-term attachment with supposedly harmless devices. The banding of penguins shows this clearly (Jackson and Wilson 2002; Panel 1). Those who look for day to day changes will see nothing untoward within the huge variability displayed by the birds carrying small tags. However, 5-year monitoring of a king penguin colony showed that banded birds arrive late for courtship in The case of the penguin bands illustrates the point that the question of how many animals to equip with devices also needs to be addressed (Cuthill 1991; Farnsworth and Rosovsky 1993; Rickard 2004). It is difficult to assess the true numbers required to obtain meaningful answers to research questions because variability in the parameter to be measured may not be known. It is not until these parameters have been quantified that it is possible to judge the number of animals required to answer specific questions (eg Bradshaw et al. 2002). This topic is covered in some detail in the literature (eg Hindell et al. 2003). Nevertheless, the acceptable practice index described above needs to be multiplied by the number of animals used in the study (and the number of associated animals that might be affected) and this needs to be put in the context of both the meta-population and the population. The capture The issue of animal capture, for the purpose of attaching devices, is an important one. Although some animals can become trap-happy, particularly if given food rewards when they are trapped (Tuyttens et al. 1999), animal capture and restraint without reward (Figure 5) is probably one of the most stressful situations that a wild animal can experience, as it may be likened to being captured by a predator. The way animals react to being captured is, therefore, usually a matter of life or death. It is thus not The Ecological Society of America

57 Measuring devices on wild animals RP Wilson and CR McMahon 152 Acceleration (dynamic and static) (G) Hall sensor output (mv x 1000) Time (s) Figure 6. To measure the degree of stress experienced by a restrained animal, an Imperial cormorant is fitted with an elastic band, clipped around the animal s thorax. A dedicated logging system monitors parameters via sensors embedded in the band. The bird was fitted with the unit during time = 0 to 3 seconds (s) (see black bar in lowest graph), after which it was held by a researcher for approximately 80 s. A triaxial accelerometer (measuring surge, heave, and sway graphs a,b,c) showed extensive struggling at time = 5 to 9 s. A Hall sensor and adjacent magnet (cf Wilson et al. 2004), placed in the elastic band, measures changes in the bird s body circumference due to respiration; respiratory frequency and tidal volume decreased during the quiet period following the struggle (graph d). During the quiescent period, all accelerometers show wave patterns corresponding to breathing. The data from each channel were recorded at 32 Hz. (a) (b) (c) (d) surge heave sway surprising that there is a substantial literature on the radical physiological changes that accompany capture (eg Giese 1996; Weimerskirch et al. 2002). Measurable parameters include increases in blood adrenaline, raised metabolic rate, and increased heart rate, respiratory frequency, and body temperature (Giese 1996; Weimerskirch et al. 2002). Although the period of restraint may be short, stress levels during capture are extreme. It is therefore vital that the capture and handling procedure should be examined carefully. Obvious measures to reduce stress include ensuring rapid sensory deprivation (eg covering the animal s eyes; Figure 5b). In keeping with the proposition that the effects of capture and tagging require documentation, we suggest that researchers monitor easily measurable indicators of stress, such as respiration rate, heart rate, and body temperature, either individually or in combination. A simple logging system embedded within an elastic belt can be rapidly placed around the thorax (the implementation of a stress measuring system should take a negligible time to deploy and should, in itself, not contribute significantly to stress levels). Within the elastic belt, acceleration transducers (cf Yoda et al. 1999, 2001) monitor struggling, while magnetometry (eg Wilson et al. 2004), piezoelectric systems, and heat flux sensors (Willis and Horning 2005) are used to measure respiration rates, tidal volumes, heart beat, and temperature, respectively. All sensors can be linked to a single logger, recording at high frequencies (Figure 6). The use of a logger-based system ensures continuous monitoring and produces a high quality record of changes during the period of restraint. Such records can be made available to appropriate animal welfare bodies at the completion of the field work and, aside from serving as a measure of worker competence, should prove very useful to the researchers themselves, since they will be able to assess what procedures minimized stress. The intensity of stress resulting from capture and restraint has longer term consequences for the animals because learned behavior, particularly as it relates to avoidance, occurs most efficiently when the stimulus is most intense (Schwartzkopf-Genswein et al. 1997). This underscores the need to minimize stress, but also makes a case for the use of sedatives, particularly ones such as Zoletil (teletamine and zolazepam) that have an amnesic effect (Lin et al. 1993). Although there are many important issues related to the use and effects of sedatives, due care, proper training, and the administration of low levels of proven products should be seriously considered as part of the restraint policy (Figure 5; for reviews where many seals were successfully and safely anesthetized see McMahon et al. 2000; Field et al. 2002). No pain, no gain? Once a comprehensive assessment of animal welfare has been completed, the extent of the impacts on the animals involved should have been approximately quantified. Critically, then, do these impacts justify the purported gains? Few would argue that a bird with a broken wing should not be treated, even though this will involve discomfort. From there it is only a short step to The Ecological Society of America

58 RP Wilson and CR McMahon Measuring devices on wild animals the view that it is acceptable for individual animals to experience discomfort for the good of the species (eg a telemetry study of an endangered species, to identify areas that should be protected). Should studies involving attached devices always be linked to the conservation of the species? By definition, such studies would only be necessary where the species is under threat. In fact, there are a number of other reasons that justify attaching devices to animals. Nonetheless, reducing the animal welfare issue to meaningful numbers can at least go some way to producing a less emotive picture of the effects of proposed research, thereby helping managers to weigh the costs and benefits. When it comes to making decisions on whether research involving the attachment of measuring devices should go ahead or not, the scientists concerned, who are often the most knowledgeable about the animals in question, frequently have the least influence. Companies with a vested interest in the research may have substantial political influence, as do organizations concerned with animal welfare. Many of the problems that arise between the various stakeholders stem from misunderstanding, lack of transparency, and the failure of the researchers to inform the wider community of the purpose of the proposed work. Things are improving, however, and we hope that the data-based approach described above will provide a basis on which to build improved communications. Perhaps measures to quantify the stress of capture and device attachment in wild animals should routinely be included in proposals for field work. The resulting database could prove an invaluable addition to our current understanding of animal behavior. Indeed, it is lack of knowledge that has led to the emergence of the novel field of conservation medicine (Meffe 1999; Osofsky et al. 2000). This new discipline involves the integration of veterinary medicine, conservation biology, and public health in order to advance biodiversity conservation, address issues associated with the inter-relationships between human, animal, and ecosystem health, and study the effects of global environmental change on these inter-relationships. Although this holistic approach is designed to help us understand how animals operate in their natural environment, its integration with numeric assessment of the effects of handling and recording devices on study animals will allow us to assess the extent to which our research results are a product of our own procedures. Ultimately, the combination of conservation medicine with cognizance of research procedure effects should allow us to minimize our impact on wild animals as we study them, as well as to correct for problems that we cause, for the greater wellbeing of animals and humans alike. Acknowledgements We are grateful to N Liebsch and F Quintana for help with fieldwork and related activities. References Andelman SJ and Fagan WF Umbrellas and flagships: efficient conservation surrogates or expensive mistakes? P Nat Sci USA 97: Baker JD and Johanas TC Effects of research handling on the endangered Hawaiian monk seal. Mar Mammal Sci 18: Bishop CA Circulatory variables and the flight performance of birds. J Exp Biol 208: Bradshaw CJA, Hindell MA, Michael KJ, and Sumner MD The optimal spatial scale for the analysis of elephant seal foraging as determined by geo-location in relation to sea surface temperatures. ICES J Mar Biol 59: Burney DA and Flannery TF Fifty millennia of catastrophic extinctions after human contact. Trends Ecol Evol 20: Carignan V and Villard MA Selecting indicator species to monitor ecological integrity: a review. Environ Monit Assess 78: Cooper J, Henley SR, and Klages NTW The diet of the wandering albatross Diomedea exulans at sub-antarctic Marion Island. Polar Biol 12: Culik BM, Putz K, Wilson RP, et al Diving energetics in king penguins (Aptenodytes patagonicus). J Exp Biol 199: Cuthill I Field experiments in animal behaviour: methods and ethics. Anim Behav 42: Dalton R Animal-rights group sues over disturbing work on sea lions. Nature 436: 315. Ehrlich PR and Ehrlich AH The value of biodiversity. AMBIO 21: Farnsworth EJ and Rosovsky J The ethics of ecological field experimentation. Conserv Biol 7: Field IC, Bradshaw CJA, McMahon CR, et al Effects of age, size and condition of elephant seals (Mirounga leonina) on their intravenous anaesthesia with tiletamine and zolazepam. Vet Record 151: Foster C, De Koning JJ, Hettinga F, et al Pattern of energy expenditure during simulated competition. Med Sci Sport Exer 35: Froget G, Gautier-Clerc M, Le Maho Y, and Handrich Y Is penguin banding harmless? Polar Biol 20: Gauthier-Clerc M, Gendner JP, Ribic CA, et al Long-term effects of flipper bands on penguins. P Roy Soc Lond B 271: S Giese M Effects of human activity on Adelie penguin Pygoscelis adelie breeding success. Biol Conserv 75: Hamel NJ, Parrish JK, and Conquest LL Effects of tagging on behavior, provisioning, and reproduction in the Common Murre (Uria aalge), a diving seabird. Auk 121: Hawkins P Bio-logging and animal welfare: practical refinements. Memoirs of the National Institute for Polar Research 58: Heilbrun RD, Silvy NJ, Tewes ME, and Peterson MJ Using automatically triggered cameras to individually identify bobcats. Wildlife Soc Bull 31: Hindell MA, Bradshaw CJA, Sumner MD, et al Dispersal of female southern elephant seals and their prey consumption during the austral summer: relevance to management and oceanographic zones. J Appl Ecol 40: Jabour-Green J and Bradshaw CJA The capacity to reason in conservation biology and policy: the southern elephant seal branding controversy. J Nature Conserv 12: Jackson S and Wilson RP The potential costs of flipperbands to penguins. Funct Ecol 16: Lin HC, Thurmon JC, Bengston GJ, and Tranquilli WJ Telazol a review of its pharmacology and use in veterinary medicine. J Vet Pharmacol Ther 16: MacDiarmid AB, Oliver MD, Stewart RA, and Gopal D Conservation of unique patterns of body markings at ecdysis 153 The Ecological Society of America

59 154 Measuring devices on wild animals enables identification of individual spiny lobster, Jasus edwardsii. New Zeal J Mar Freshwater Res 39: Marsh H and Kenchington R The role of ethics in experimental marine biology and ecology. J Exp Mar Biol Ecol 300: McMahon CR, Burton HR, McLean S, et al Field immobilisation of southern elephant seals with intravenous tiletamine and zolazepam. Vet Rec 146: McMahon CR, van den Hoff J, and Burton HR Repeated handling and invasive research methods in wildlife research: impacts at the population level. AMBIO 34: Meffe GK Conservation medicine. Conserv Biol 13: Minteer BA and Collins JP Why we need an ecological ethics. Front Ecol Environ 3: Osofsky SA, Karesh WB, and Deem SL Conservation medicine: a veterinary perspective. Conserv Biol 14: Phillips RA, Xavier JC, and Croxall JP Effects of satellite transmitters on albatrosses and petrels. Auk 120: Plous S and Herzog H Reliability of protocol reviews for animal research. Science 293: Rickard MD The use of animals in research: counting the costs and the benefits. Aust J Exp Agr 44: Ropert-Coudert Y and Wilson RP Trends and perspectives in animal-attached remote sensing. Front Ecol Environ 3: Sawai A, Ohshige K, and Tochikubo O Development of wristwatch-type heart rate recorder with acceleration-pickup sensor and its application. Clin Exp Hypertens 27: Schwartzkopf-Genswein KS, Stookey JM, and Welford R Behaviour of cattle during hot-iron and freeze branding and the effects on subsequent handling ease. J Anim Sci 75: Stirling I Observations on the behaviour of the New RP Wilson and CR McMahon Zealand fur seal (Arctocephalus forsteri). J Mammal 51: Stonehouse B Animal marking: recognition marking of animals in research. London, UK: Macmillan. Tuyttens FAM, Macdonald DW, Delahay R, et al Differences in trappability of European badgers Meles meles in three populations in England. J Appl Ecol 36: Vincent C, Meynier L, and Ridoux V Photo-identification in grey seals: legibility and stability of natural markings. Mammalia 65: Weimerskirch H, Shaffer SA, Mabille G, et al Heart rate and energy expenditure of incubating wandering albatrosses: basal levels, natural variation, and the effects of human disturbance. J Exp Biol 205: Willis K and Horning M A novel approach to measuring heat flux in swimming animals. J Exp Mar Biol Ecol 315: Wilson RP, Grant WS, and Duffy DC Recording devices on free-ranging marine animals: does measurement affect foraging performance. Ecology 67: Wilson RP, Scolaro JA, Quintana F, et al To the bottom of the heart: cloacal movement as an index of cardiac frequency, respiration and digestive evacuation in penguins. Mar Biol 144: Wilson RP and Wilson MPT A peck activity record for birds fitted with devices. J Field Ornithol 60: Yoda K, Naito Y, Sato K, et al A new technique for monitoring the behaviour of free-ranging Adelie penguins. J Exp Biol 204: Yoda K, Sato K, Niizuma Y, et al Precise monitoring of porpoising behaviour of Adelie penguins determined using acceleration data loggers. J Exp Biol 202: The Ecological Society of America

60 Journal of Experimental Marine Biology and Ecology 360 (2008) Contents lists available at ScienceDirect Journal of Experimental Marine Biology and Ecology journal homepage: Tracking and data logging devices attached to elephant seals do not affect individual mass gain or survival Clive R. McMahon a, Iain C. Field a, Corey J.A. Bradshaw a,b,c,d,, Gary C. White e, Mark A. Hindell d a School for Environmental Research, Institute of Advanced Studies, Charles Darwin University, Darwin, Northern Territory 0909, Australia b Research Institute for Climate Change and Sustainability, School of Earth and Environmental Sciences, University of Adelaide, South Australia 5005, Australia c South Australian Research and Development Institute, PO Box 120, Henley Beach, South Australia 5022, Australia d Antarctic Wildlife research Unit, School of Zoology, University of Tasmania, Private Bag 05, Hobart, Tasmania 7000, Australia e Department of Fish, Wildlife, and Conservation Biology, Colorado State University, Fort Collins, Colorado 80523, USA ARTICLE INFO ABSTRACT Article history: Received 10 March 2008 Received in revised form 22 March 2008 Accepted 25 March 2008 Keywords: Animal welfare Antarctica Climate change Elephant seals Growth Marine Time depth recorders Tracking devices Satellite telemetry Understanding the cryptic lives of wide ranging wild animals such as seals can be challenging, but with the advent of miniaturised telemetry and data logging devices this is now possible and relatively straightforward. However, because marine animals have streamline bodies to reduce drag in their aquatic habitats, attaching external devices to their back or head may affect swimming performance, prey capture efficiency and ultimately, fitness. Given this, and allied welfare concerns, we assessed the short- and long-term consequences of external devices attached to southern elephant seal juveniles and adults under varying environmental conditions. We also assessed the effects of multiple deployments on individuals. There was no evidence for short-term differences in at-sea mass gain (measured as mass on arrival from a foraging trip) or long-term survival rate. The number of times that a seal carried a tracking device (ranging from 1 to 8 times) did not affect mass or estimated survival. Further, there were no tracking device effects in years of contrasting environmental conditions measured as ENSO anomalies. Consequently, we conclude that the current tracking devices available to researchers are valuable conservation tools that do not adversely affect the performance of a large marine mammal in terms of mass gain or survival probability over short (seasonal) or long (years) temporal scales Elsevier B.V. All rights reserved. 1. Introduction Studies of how species respond to variation in their environment require a range of techniques to record pertinent data such as estimates of trends in population size, survival and recruitment, mark recapture and telemetry of individual movements and other behaviours. Combining mechanistic behavioural approaches with population level data is particularly powerful for predicting a species' response to future environmental change (Both et al., 2006; Parmesan and Yohe, 2003; Perry et al., 2005). For wide ranging species, examining foraging dynamics is a particularly important component of these studies because such data summarise information on energy acquisition and expenditure at a variety of spatial and temporal scales. Studies designed to collect such information assume that natural behaviours are not compromised by the experimental procedures themselves. Furthermore, these types of field experiments may raise many ethical issues including the trade-off between individual welfare and information required to conserve threatened species (Minteer and Collins, 2005; Putman, 1995). Corresponding author. Research Institute for Climate Change and Sustainability, School of Earth and Environmental Sciences, University of Adelaide, South Australia 5005, Australia. Documenting the life history of cryptic species can be especially difficult, particularly for marine species that are only rarely observed during brief feeding or breeding events close to or onshore (Bradshaw, 2007). Recent technological advances have provided detailed behavioural information that would be otherwise impossible to collect (Hooker et al., 2007 and references therein). Miniaturisation, long life batteries and large data storage capacity mean that data logging devices can potentially be deployed for years (Hays et al., 2007b). However, it is still necessary that researchers weigh the benefits of these long-term deployments against their potential effects on reproduction, foraging success, energetics and survival of the sampled individuals. Some of the many considerations include the tracking device's (hereafter termed device ) ergonomics, location of attachment, mass relative to body size, additional energetic cost induced by drag, increased agonistic behaviour by conspecifics, and impairment of camouflage and foraging efficiency. Because many studies often require longitudinal information on particular individuals, repeated deployment of devices may also be required (Bradshaw et al., 2004a). It is possible that although a single or short-term deployment may not be harmful to an individual, the cumulative effects of multiple deployments may be ultimately detrimental (as is the case for flipper bands in penguins Gauthier-Clerc et al., 2004). The need for information on the potential effects incurred /$ see front matter 2008 Elsevier B.V. All rights reserved. doi: /j.jembe

61 72 C.R. McMahon et al. / Journal of Experimental Marine Biology and Ecology 360 (2008) Fig. 1. The number of individual deployments of southern elephant seals that carried bio-logging devices from Macquarie Island from 1999 to by multiple deployments of devices is especially important because detrimental impacts may only appear during periods of resource scarcity. For example, a device's effect may be exacerbated in years when prey are scarce (more dispersed or deeper in the water column), thus requiring that the foraging animal expends more energy to catch prey. Despite the importance of these potential detriments to animal performance, there has been little quantification of the effects of the devices especially over multiple deployments (see Wilson and McMahon, 2006 for a recent review). An important Southern Ocean predator that has been the subject of much research in this area is the southern elephant seal (Mirounga leonina). This species is particularly tractable to research because: (a) they are an important Antarctic apex predator that has shown protracted and substantial declines in some regions (McMahon et al., 2005b), (b) there are established demographic links in this species to environmental change (de Little et al., 2007; McMahon and Burton, 2005), (c) they are wide-ranging and incorporate information over broad spatial and temporal scales (Bradshaw et al., 2004a; Bradshaw et al., 2004b; Field et al., 2004; Hindell et al., 2003), (d) they are easily accessible during defined haul-out periods onshore (Hindell, 1991), and (e) their large size means that small devices are less likely to modify behaviour (Ropert-Coudert and Wilson, 2005). Although the effects of marking (McMahon et al., 2006) and handling (Engelhard et al., 2002, 2001; McMahon et al., 2005a) have been examined for this species, the potential effects of data logger deployment on elephant seal performance in terms of energy (mass) gain and survival probability have never before been assessed empirically. Because the potential effects are likely to differ between small and large, and between young and old seals, we calculated age specific survival estimates for seals from a wide range of ages (1 13 years) equipped with devices and those without, as well as assessing the consequences of multiple deployments on individuals. It might be expected that the growth of small and young seals could be compromised by the additional cost of carrying a device, with flow-on effects such as delayed age of primiparity, reduced population growth rate and elevated extinction risk in small populations. The aims of this study were four-fold: (1) To determine if there was any evidence of an energetic cost to seals carrying data loggers by comparing variation in arrival masses between instrumented and noninstrumented elephant seals at Macquarie Island (Pacific sector of the Southern Ocean). We predicted that the attachment of devices may increase the cost of transport (via increase in drag), thus potentially reducing individual fitness. This increased fitness cost, if it compromises survival via poorer foraging performance, may be measureable by either increased time at sea or decreased overall mass gain when compared to animals not carrying devices (Boyd et al., 1997; Ropert-Coudert et al., 2007a,b); (2) We hypothesised that the evidence for any short-term effects of data logger deployment might be masked by subtler long-term effects on average demographic rates. We therefore estimated apparent survival rates of instrumented versus non-instrumented seals relative to the environmental conditions encountered while foraging; (3) To assess the additional influence of multiple deployments on individuals; (4) To assess the influence of inter-annual variability in environmental conditions on mass gain and survival when carrying a device. 2. Materials and methods 2.1. Deployment A large sample (n=12251) of recently weaned southern elephant pups was hot branded between 1993 to 1999 on Macquarie Island Fig. 2. Post-moult arrival weights of adult female elephant seals expressed as a function of (a) age and (b) the number of times the seal had carried a device.

62 C.R. McMahon et al. / Journal of Experimental Marine Biology and Ecology 360 (2008) Table 1 Ranking of the generalised linear mixed effects models (GLMM) relating arrival mass (AM) during the post-moult period at sea to age (age), number of times that an individual carried a tag (num), and individual seal (ID) Models k LL AIC c ΔAIC c waic c %DE AM~age+ (1 ID) E AM~age+ num+(1 ID) E AM~age+ num+age num+(1 ID) E AM~num+(1 ID) E AM~1+(1 ID) E The models are ranked in order of Akaike Information Criterion corrected for small sample size (AIC c ) weights (waic c ). LL: maximum log-likelihood of the model; k: number of estimated parameters; ΔAIC c : difference between the model's AIC c and the minimum AIC c ; %DE: per cent deviance explained by model. Table 2 Ranking of the generalised linear mixed effects models (GLMM) relating arrival mass (AM) during the post-breeding period at sea to age (age), number of times that an individual carried a tag (num), and individual seal (ID) Models k LL AIC c ΔAIC c waic c %DE AM~age+ num+(1 ID) E AM~age+ num+age num+(1 ID) E AM~age+ (1 ID) E AM~num+(1 ID) E AM~1+(1 ID) E The models are ranked in order of Akaike Information Criterion corrected for small sample size (AIC c ) weights (waic c ). LL: maximum log-likelihood of the model; k: number of estimated parameters; ΔAIC c : difference between the model's AIC c and the minimum AIC c ; %DE: per cent deviance explained by model. (54 30 S, E) (McMahon et al., 2006). Of these, 124 (aged between 1 to 9 years at deployment) were equipped with time depth recorders (Mk6, Mk7 and Mk8 Wildlife Computers, Redmond, USA), light loggers (Platypus Engineering, Hobart, Australia) or platform transmitter terminals [PTT] (Sea Mammal Research Unit, St. Andrews, Scotland). Data loggers and transmitters were attached to seals that were captured during one of their two annual haul-outs: (1) at the end of breeding and (2) at the end of the moult between 1999 and 2005 (see Bailleul et al., 2007; Bradshaw et al., 2004a; Field et al., 2004 for attachment procedures, and Field et al., 2002; McMahon et al., 2000 for capture details), and then recaptured upon their subsequent return to the island, representing an average of 70 and 280 days at sea, respectively for post-breeding and post-moult deployments. Time depth recorders were combined with VHF transmitters and weighed b350 g, and platform transmitter terminals weighed 550 g. These represented b1.0% of the departure mass for the smallest seal in the study (78 kg). Because TDRs were combined with VHF transmitters to facilitate retrieval, PTTs and TDRs were of similar size and mass, and both had protruding antenna. We therefore did not distinguish between unit type in the analysis. Daily searches of the isthmus beaches and tussock areas (main study area) and monthly searches of the entire island were made to resight (and recapture) marked seals and to search for seals that were equipped with devices ( ). Seals were caught within 3 days of coming ashore for attachment and retrieval. To assess the short-term effects of devices on seals, we weighed all of the animals to the nearest kilogram in a net sling suspended from an aluminium tripod using an electronic balance precise to 1.0 kg. To ensure that mass measurements were accurate, the scales were tared each day prior to operation with a known-mass gymnasium weight. Measuring and comparing mass changes to quantify the effects of devices in endotherms, like seals, is both a convenient and appropriate way to evaluate fitness because mass changes reflect foraging success during the previous trip to sea (Bradshaw et al., 2004a) and are greater in endotherms than they are in ectotherms (e.g., marine turtles) where the rates of mass change are generally small (Hays, 2008) Arrival mass To test the hypothesis that mass gain varied between seals carrying or not carrying a device, we were obliged to use arrival mass as the response. Although mass gain during the time at sea would likely represent a better index, we only rarely had access to non-instrumented seals when returning to Macquarie Island. Therefore, to examine the evidence for an effect of carrying a device on a seal's subsequent arrival weight, we fitted a series generalized linear mixed effect models (GLMM) to the data using the lmer function implemented in the R Package (R Development Core Team, 2004). The mixed effects structure of the GLMM allows us to partition the variance within and among individuals from that associated with the Fig. 3. Post-breeding arrival weights of adult female elephant seals expressed as a function of (a) age and (b) the number of times the seal had carried a device. We also collected less data from this time only getting 4 years as opposed to 6 years for post-moult.

63 74 C.R. McMahon et al. / Journal of Experimental Marine Biology and Ecology 360 (2008) Table 3 The five most parsimonious models showing the effects on model parsimony of applying the over-dispersion metric ĉ in program MARK ΔQAIC c Model Likelihood k Deviance Model ĉ= [φ(a5 +t+sex+ device) p(a9+t+sex+device)] [φ(a5 +t+sex) p(a9+t+sex)] [φ(a5 +t+sex) p(a8+t+sex)] [φ(a5 +t+sex) p(a7+t+sex)] [φ(a7 +t+sex) p(a7+t+sex)] input files for program MARK (White and Burnham, 1999). Multiple sightings of a seal within a seal year (15 October t 14 October t+1 )were treated as a single sighting in the capture history matrix. Age specific estimates of apparent survival (φ) and recapture or resighting probability (p) were estimated using the Cormack-Jolly-Seber (CJS) model in program Model ĉ= [φ(a5 +t+sex+ device) p(a9+t+sex+device)] [φ(a5 +t+sex) p(a9+t+sex)] [φ(a5 +t+sex) p(a7+t+sex)] [φ(a5 +t+sex) p(a8+t+sex)] [φ(age11^3 +t+sex) p(a9+t+sex)] b Model ĉ= [φ(a5 +t+sex+ device) p(a9+t+sex+device)] [φ(a5 +t+sex) p(a7+t+sex)] [φ(a5 +t+sex) p(a9+t+sex)] [φ(a5 +t+sex) p(a8+t+sex)] [φ(age11^3 +t+sex) p(a9+t+sex)] b While ĉ adjustments did affect model weightings, the top-ranked model remained the most parsimonious in all adjusted cases (ĉ= and ĉ =2.7009) and there was little support for any of the nearest competing models [ΔQAIC c 7 highlighted (Burnham and Anderson, 2001)]. k is the number of parameters included in each of the models. fixed effects of main interest. The five models defined and compared were: (1) AM~num+(1 ID), (2) AM~age+num+(1 ID), (3) AM~age+num+ age num+(1 ID), (4) AM~age+(1 ID) and (5) AM~1+(1 ID), where: AM=arrival mass (kg), num=the number of times a seal had been equipped with a device, age=age in years, ID=seal identity. Note that model 4, containing age alone, was the control model because all seals necessarily aged during the course of the study. All weights were corrected to an estimated arrival weight based on the number of days ashore prior to capture and an estimated constant rate of mass loss calculated from the difference between subsequent weighings during the moulting period (4.03 kg day 1 ). We analysed only adult female data because the arrival masses for sub-adults were confounded by their tendency to make multiple visits to Macquarie Island during winter, making it difficult to compare individuals. We analysed the post-moult and post-breeding datasets separately because the two periods at sea are fundamentally different with respect to duration, function and destination. Arrival mass (log transformed) was set as the response variable, and models included various combinations of seal age and the number of times that a device was carried as fixed effects. Individual seals were coded as a random effect to account for repeated measures. Each model was constructed using a Gaussian error distribution and an identity link function. Model goodness of fit was assessed as the per cent deviance explained (%DE). We used an index of Kullback-Leibler (K-L) information loss to assign relative strengths of evidence to the different competing models (Burnham and Anderson, 2002), Akaike's Information Criterion corrected for small sample sizes (AIC c ). These indices of model parsimony identify those model(s) from a set of candidate models that minimize K L information loss (Burnham and Anderson, 2004). The relative likelihoods of candidate models were calculated using AIC c weights (Burnham and Anderson, 2002), with the weight (waic c ) of any particular model varying from 0 (no support) to 1 (complete support) relative to the entire model set Survival probability Individual capture history matrices were constructed from the resight histories (McMahon and Burton, 2005; McMahon et al., 2003) and used as Fig. 4. (a) Survival estimates (φ) for male seals carrying time depth recorders devices (open squares), and male not carrying devices (open circles), and survival estimates for female seals carrying devices (closed squares) and those without devices (closed circles). Data are only presented to age five because survival for all seals is constant after age five. While no differences between survival estimates were apparent, seals carrying devices (squares) had higher survival estimates than seals that were not equipped with devices. Because this may have been the result of differences in capture probabilities we calculated the recapture probabilities (b). The recapture probability estimates (p) for male seals carrying time depth recorders devices (open squares), and males not carrying devices (open circles), and recapture estimates for female seals carrying devices (closed squares) and those without devices (closed circles). Data are only presented to age nine because survival for all seals is constant after age nine. Recapture probability estimates for seals carrying devices (squares) were higher than for seals that were not equipped with devices. (c) Survival estimates (φ) for elephant seals carrying devices (closed squares), and those not carrying devices (open circles) were identical even when including the effects of environmental stochasticity in the form of the Southern Oscillation Index. Again data are only presented to age five because survival for all seals is constant after age five. In all cases the vertical lines represent the 95% confidence intervals for each estimate.

64 C.R. McMahon et al. / Journal of Experimental Marine Biology and Ecology 360 (2008) MARK (White and Burnham,1999). Southern elephant seals in general, but particularly breeding females, show strong site fidelity to their natal areas (Bradshaw et al., 2004a; McMahon et al., 1999; Nicholls, 1970) and consequently, we do not expect much permanent emigration from Macquarie Island. Capture resight matrices were constructed as follows. For each seal, two individual time variant factors identified when seals were equipped with a device (coded 1) and when they were not (coded 0). An example history of a typical individual was: ; 0; 1; 0; 0; 0; 0; 0; 0; 0; 0; 0; 1; 1; 0; where the first 12 numeric codes define the capture history of the seal ( ,0,1,0,0,0,0,0,0,0,0,0,1,1,0), then its sex (male=1,0; female=0,1) in the next two columns ( ,0,1,0,0,0,0,0,0,0,0, 0,1,1,0), and the last 12 columns, we define whether the animal was equipped with a device or not ( ,0,1,0,0,0,0,0,0,0,0,0,1,1,0). Seals in this study could only carry a device for a maximum of one year because they undergo an obligatory annual moult. Individuals are exposed to a variety of environmental conditions over time, so we included the mean Southern Oscillation Index (SOI, available from from March to September as an extra covariate in the models constructed (de Little et al., 2007; McMahon and Burton, 2005) to reflect environmental variation. Parametric goodness of fit (GOF) tests within MARK were used to test whether the CJS model assumptions were met. To accommodate lack of fit, the amount of over-dispersion (ĉ) was quantified. Program RELEASE (Burnham et al.,1987) run in MARK was used to explain the causes for any the lack of fit (using the saturated model [φ(sex t) p(sex t)], where sex=seal gender and t=time). Consequently, when there was sufficient evidence for over-dispersion, we corrected for the extra-binomial variation in the data by the variance inflation factor ĉ,(lebreton et al., 1992) to adjust the deviance in the calculation of the AIC c (quasi likelihood AIC c =QAIC c ) and parameter standard errors (Lebreton et al., 1992). 3. Results 3.1. Arrival mass We deployed devices on 124 adult seals that were also weighed after their time at sea. Many seals were equipped with a device more than once over the period of investigation (Fig. 1), although most individuals only ever carried a device once, and some carried a device up to eight times. Post-winter arrival weights averaged 550±82 kg, although this varied considerably among individuals (Fig. 2a). The control model (age only, with seal coded as a random factor) was the top-ranked model (waic c =0.66), although it only accounted for 6.8% of the deviance (Table 1). The model including the number of times that a seal carried a device made a negligible addition to the per cent device explained (0.01%), suggesting little evidence for a measureable effect on arrival mass (Fig. 2b). Seals weighed substantially less after their post-breeding trip to sea (mean arrival mass=458±51.5 kg), due partly to the much shorter duration of the post-breeding trip (70 days for the post-breeding trip versus 280 days for the post-moult trip). Post-breeding arrival mass results were similar to the post-moult results (Fig. 3); however, the control model received little support (waic c =0.002). The top-ranked model (waic c =0.58) included age and the number of times an animal carried a device (Table 2). Most importantly, adding the number of times that a seal carried a device added less than 1% to the %DE, again indicating little evidence of an important or measureable effect on arrival mass Survival Our first model set ignored individual time-variant covariates. The most highly ranked of these basic models included sex and time (t) effects: φ (sex t) p(sex t). To this model we added various age, time and sex effects as well as the individual time-variant factor device (i.e., whether or not a seal carried a device over an interval) (Table 3). The top-ranked model was φ(age5+t+sex+device) p(age9+t+sex+device) whereage5=survival probability to 5 years of age, age9=survival to 9 years, and device=times when a device was attached. The goodness of fit simulations indicated some over-dispersion (ĉ bootstrap =2.701, ĉ median =2.122), thus demonstrating moderate violation of the assumption that all individuals were equally catchable or have similar apparent survival probabilities. We accounted for this by applying the ĉ correction factor, but this did not affect model ranking (Table 3). There was little evidence for a survival difference between males and females (Fig. 4). Likewise, there was little evidence for an effect of device on survival at any age (Fig. 4). Survival estimates for seals with devices were consistently higher than those without, although there was considerable overlap in the parameter estimates. One possible explanation may be that the recapture probability for seals that carried devices was higher (Fig. 3). To test whether the presence of a device affected survival or recapture probabilities, we applied the information theoretic evidence ratio (ER) which is the waic c of one model divided by that of a simpler comparison model. Indeed, adding the device factor to the recapture probability models improved model performance (ER=51.7). There was also weak evidence for an effect of environmental stochasticity as measured by the Southern Oscillation Index on survival probability (ER=2. 6), but little evidence for an SOI effect on capture probability (ER=1.2). Importantly, survival of seals that carried devices was unaffected by SOI (Fig. 4). 4. Discussion Assessing the potential effects of research procedures on animal performance is an important component of data interpretation and ethical justification of research (Wilson and McMahon, 2006). However, acquiring such information is not always straightforward, especially for animals that spend much of their lives in remote areas and challenging environments. Few studies have assessed the potential long-term impact of telemetry or data logging devices on animal performance and fitness. One of the key reasons for investigating the potential effects of externally borne devices is concern for the welfare of the animals, but in addition and closely aligned to this, is concern for the integrity of the data being collected. Consequently, establishing that the research procedure does not compromise the animal's performance in the short term (e.g., foraging success during foraging trips) or long term (e.g., survival over many years) is an important step to ensure that the information collected accurately represents the life history of the animal under study. Moreover, quantifying the effects of devices on animal performance can help drive technological advancements in instruments and attachment procedures that reduce potentially negative effects on study animals. For leatherback turtles, satellite tags have traditionally been attached using harnesses (e.g., James et al., 2006); however, it has been shown recently that harness attachments, when compared to direct attachment of devices to the carapace, compromise diving ability and reduce speed of travel (Fossette et al., 2008). Hence, direct attachment is now being adopted as the standard attachment system for leatherback turtles (Doyle et al., 2008). This work also highlights how subtle changes in travel speed and other behaviours may be indicative of negative device impacts, just as longer-term indicators such as mass and survival may change when devices are attached. We found no evidence that devices attached to southern elephant seals used to study behaviour and foraging have any short-term (arrival mass) or long-term (survival) effects on performance. This conclusion held true for all seals of all age classes, even for the smallest seals that may be the most sensitive to research manipulation. Importantly, this was true even during periods when ENSO conditions were below average (i.e., low SOI values) when juvenile seal survival is

65 76 C.R. McMahon et al. / Journal of Experimental Marine Biology and Ecology 360 (2008) more likely to drop below average (McMahon and Burton, 2005). Although the potential effects of data logging devices will depend on the target species and technology employed, our study provides a benchmark that can be used for other species. From our long-term monitoring dataset we addressed each of the three main welfare concerns for ecological research: (1) the stress due to handling and capture, (2) eco-physiological limitations of device attachment and (3) long-term effect of handling and device attachment. While our results are encouraging, using arrival mass as a proxy for changes in body condition may not necessarily reflect changes in body energy content (Coltman et al., 1998), although for elephant seals mass is thought to provide a reasonable index of condition (Biuw et al., 2003). It is such small discrepancies that can pose problems when assessing device impacts and consequently highlights the need for more study on this topic using other indices of performance and fitness. Another component that we did not address specifically was that the additional drag created by attaching an external instrument to an otherwise highly streamlined body form (Wilson et al., 2004) is that its potentially negative effects on performance may be mitigated by altering the instrument's buoyancy. Indeed, for deployments of short duration that inevitably incur high drag, negative effects can be minimised by ensuring neutral buoyancy in the attached device (e.g., Hays et al., 2007a; Williams et al., 2004). Two key findings set this study apart from previous work investing the effects of devices on animals: (1) we could detect no amplification of effects in poor years (i.e., low SOI) when negative effects are hypothesised to be exacerbated due to food shortages either in the form of reduced quantity or quality; and (2) we could find no evidence that multiple deployments reduced fitness (quantified by survival estimates). Together these observations lend powerful support to our main conclusion that devices attached to elephant seals of any size do not compromise fitness by showing that even under some of the more extreme deployment regimes, seal performance is not compromised. This is an important result for elephant seals but more importantly, it has wider applicability for studies focussing on species of similar size and foraging dynamics such as other seals and marine turtles (Hooker and Boyd, 2003; McMahon et al., 2005c). With increasing emphasis on using wide-ranging marine predators as automonous oceanographic samplers (Biuw et al., 2007), establishing that there are no negative effects due to the presence of recording devices is an essential first step. Acknowledgements We thank the many people who aided in the collection of resight data and deployment of bio-logging devices at Macquarie Island from 1993 to We also thank S. de Little for her insightful discussion on survival data analyses. An anonymous reviewer provided helpful comments to improve the manuscript. This study was funded by the Australian Research Council (A and DP ), the Antarctic Science Advisory Committee (1171 and 2794), Sea World Research and Rescue Foundation Inc., and the Australian Antarctic Division. Data were collected with Australian Antarctic Animal Ethics and University of Tasmania Animal Ethics approvals and Tasmanian Parks and Wildlife Service permits.[ses] References Bailleul, F., Charrassin, J.-B., Ezraty, R., Girard-Ardhuin, F., McMahon, C.R., Field, I.C., Guinet, C., Southern elephant seals from Kerguelen Islands confronted by Antarctic Sea ice. Changes in movements and in diving behaviour. Deep Sea Research II 54, Biuw, M., McConnell, B.J., Bradshaw, C.J.A., Burton, H.R., Fedak, M.A., Blubber and buoyancy: monitoring the body condition of free-ranging seals using simple dive characteristics. Journal of Experimental Biology 206, Biuw, M., Boehme, L., Guinet, C., Hindell, M., Costa, D., Charrassin, J.-B., Roquet, F., Bailleul, F., Meredith, M., Thorpe, S., Tremblay, Y., McDonald, B., Park, Y.-H., Rintoul, S.R., Bindoff, N., Goebel, M., Crocker, D., Lovell, P., Nicholson, J., Monks, F., Fedak, M.A., Variations in behavior and condition of a Southern Ocean top predator in relation to in situ oceanographic conditions. Proceedings of the National Academy of Sciences of the USA 104, Both, C., Bouwhuis, S., Lessells, C.M., Visser, M.E., Climate change and population declines in a long-distance migratory bird. Nature 441, Boyd, I.L., McCafferty, D.J., Walker, T.R., Variation in foraging effort by lactating Antarctic fur seals: response to simulated increased foraging costs. Behavioural Ecology and Sociobiology 40, Bradshaw, C.J.A., Swimming in the deep end of the gene pool global population structure of an oceanic giant. Molecular Ecology 16, Bradshaw, C.J.A., Hindell, M.A., Sumner, M.D., Michael, K., 2004a. Loyalty pays: potential life history consequences of fidelity to marine foraging regions by southern elephant seals. Animal Behaviour 68, Bradshaw, C.J.A., Higgins, J., Michael, K.J., Wotherspoon, S.J., Hindell, M.A., 2004b. At-sea distribution of female southern elephant seals relative to variation in ocean surface properties. ICES Journal of Marine Science 61, Burnham, K.P., Anderson, D.R., Kullback-Leibler information as a basis for strong inference in ecological studies. Wildlife Research 28, Burnham, K.P., Anderson, D.R., Model Selection and Multimodal Inference: A Practical Information Theoretic Approach. Springer-Verlag, New York, USA. Burnham, K.P., Anderson, D.R., Multimodel inference - understanding AIC and BIC in model selection. Sociological Methods and Research 33, Burnham, K.P., Anderson, D.R., White, G.C., Brownie, C., Pollock, K.H., Design and analysis methods for fish survival experiments based on release recapture. American Fisheries Society, Monograph 5, Coltman, D.W., Bowen, W.D., Iverson, S.J., Boness, D.J., The energetics of male reproduction in an aquatically mating pinniped, the harbour seal. Physiological Zoology 71, de Little, S.C., Bradshaw, C.J.A., McMahon, C.R., Hindell, M.A., Complex interplay between intrinsic and extrinsic drivers of long-term survival trends in southern elephant seals. BMC Ecology 7, 3, doi: / Doyle, T.K., Houghton, J.D.R., O'Súilleabháin, P.F., Hobson, V.J., Marnell, F., Davenport, J., Hays, G.C., Leatherback turtles satellite tagged in European waters. Endangered Species Research 4, Engelhard, G.H., van den Hoff, J., Broekman, M., Baarspul, A.N.J., Field, I.C., Burton, H.R., Reijnders, P.J.H., Mass of weaned elephant seal pups in areas of low and high human presence. Polar Biology 24, Engelhard, G.H., Baarspul, A.N.J., Broekman, M., Creuwels, J.C.S., Reijnders, P.J.H., Human disturbance, nursing behaviour, and lactational pup growth in a declining southern elephant seal (Mirounga leonina) population. Canadian Journal of Zoology 80, Field, I.C., Bradshaw, C.J.A., McMahon, C.R., Harrington, J., Burton, H.R., Effects of age, size and condition of elephant seals (Mirounga leonina) on their intravenous anaesthesia with tiletamine and zolazepam. Veterinary Record 151, Field, I.C., Bradshaw, C.J.A., Burton, H.R., Hindell, M.A., Seasonal use of oceanographic and fisheries management zones by juvenile southern elephant seals (Mirounga leonina) from Macquarie Island. Polar Biology 27, Fossette, S., Corbel, H., Gaspar, P., Le Maho, Y., Georges, J.-Y., An alternative technique for the long-term satellite tracking of leatherback turtles. Endangered Species Research 4, Gauthier-Clerc, M., Gendner, J.P., Ribic, C.A., Fraser, W.R., Woehler, E.J., Descamps, A., Gilly, C., Le Bohec, C., Le Maho, Y., Long-term effects of flipper bands on penguins. Proceedings of the Royal Society B 271, S423 S426. Hays, G.C., Sea turtles: a review of some key recent discoveries and remaining questions. Journal of Experimental Marine Biology and Ecology 356, 1 7. Hays, G., Marshall, G., Seminoff, J., 2007a. Flipper beat frequency and amplitude changes in diving green turtles, Chelonia mydas. Marine Biology 150, Hays, G.C., Bradshaw, C.J.A., James, M.C., Lovell, P., Sims, D.W., 2007b. Why do Argos satellite tags deployed on marine animals stop transmitting? Journal of Experimental Marine Biology and Ecology 349, Hindell, M.A., Some life history parameters of a declining population of southern elephant seals, Mirounga leonina. Journal of Animal Ecology 60, Hindell, M.A., Bradshaw, C.J.A., Sumner, M.D., Michael, K., Burton, H.R., Dispersal of female southern elephant seals and their prey consumption during the austral summer: relevance to management and oceanographic zones. Journal of Applied Ecology 40, Hooker, S.K., Boyd, I.L., Salinity sensors on seals:use of marine predators to carry CTD data loggers. Deep Sea Research I 50, Hooker, S.K., Biuw, M., McConnell, B.J., Miller, P.J.O., Sparling, C.E., Bio-logging science: Logging and relaying physical and biological data using animal attached tags. Deep Sea Research II 54, James, M.C., Davenport, J., Hays, G.C., Expanded thermal niche for a diving vertebrate: a leatherback turtle diving into near-freezing water. Journal of Experimental Marine Biology and Ecology 335, Lebreton, J.D., Burnham, K.P., Clobert, J., Anderson, D.R., Modeling survival and testing biological hypotheses using marked animals: a unified approach with case studies. Ecological Monographs 62, McMahon, C.R., Burton, H.R., Climate change and seal survival: evidence for environmentally mediated changes in elephant seal, Mirounga leonina, pup survival. Proceedings of the Royal Society B 272, McMahon, C.R., Burton, H.R., Bester, M.N., First-year survival of southern elephant seals, Mirounga leonina, at sub-antarctic Macquarie Island. Polar Biology 21, McMahon, C.R., Burton, H.R., McLean, S., Slip, D., Bester, M.N., Field immobilisation of southern elephant seals with intravenous tiletamine and zolazepam. Veterinary Record 146, McMahon, C.R., Burton, H.R., Bester, M.N., A demographic comparison of two southern elephant seal populations. Journal of Animal Ecology 72,

66 C.R. McMahon et al. / Journal of Experimental Marine Biology and Ecology 360 (2008) McMahon, C., van den Hoff, J., Burton, H., 2005a. Handling intensity and the short- and long-term survival of elephant seals: addressing and quantifying research effects on wild animals. Ambio 34, McMahon, C.R., Bester, M.N., Burton, H.R., Hindell, M.A., Bradshaw, C.J.A., 2005b. Population status, trends and a re-examination of the hypotheses explaining the recent declines of the southern elephant seal Mirounga leonina. Mammal Review 35, McMahon, C.R., Autret, E., Houghton, J.D.R., Lovell, P., Myers, A.E., Hays, G.C., 2005c. Animal borne sensors successfully capture the real-time thermal properties of ocean basins. Limnology and Oceanography Methods 3, McMahon, C.R., Burton, H.R., van den Hoff, J., Woods, R., Bradshaw, C.J.A., Assessing hot-iron and cryo-branding for permanently marking southern elephant seals. Journal of Wildlife Management 70, Minteer, B.A., Collins, J.P., Ecological ethics: Building a new tool kit for ecologists and biodiversity managers. Conservation Biology 19, Nicholls, D.G., Dispersal and dispersion in relation to the birth site of the southern elephant seal, Mirounga leonina, of Macquarie Island. Mammalia 43, Parmesan, C., Yohe, G., A globally coherent fingerprint of climate change impacts across natural systems. Nature 421, Perry, A.L., Low, P.J., Ellis, J.R., Reynolds, J.D., Climate change and distribution shifts in marine fishes. Science 308, Putman, R.J., Ethical considerations and animal welfare in ecological field studies. Biodiversity and Conservation 4, R Development Core Team, R: A language and environment for statistical computing. R Foundation for Statistical Computing, Vienna, Austria. Ropert-Coudert, Y., Wilson, R.P., Trends and perspectives in animal attached remote sensing. Frontiers in Ecology and the Environment 3, Ropert-Coudert, Y., Wilson, R.P., Yoda, K., Kato, A., 2007a. Assessing performance constraints in penguins with externally attached devices. Marine Ecology Progress Series 333, Ropert-Coudert, Y., Knott, N., Chiaradia, A., Kato, A., 2007b. How do different data logger sizes and attachment positions affect the diving behaviour of little penguins? Deep Sea Research II 54, White, G.C., Burnham, K.P., Program MARK: survival estimation from populations of marked animals. Bird Study 46, Williams, T.M., Fuiman, L.A., Horning, M., Davis, R.W., The cost of foraging by a marine predator, the Weddell seal Leptonychotes weddellii: pricing by the stroke. Journal of Experimental Biology 207, Wilson, R.P., McMahon, C.R., Measuring devices on wild animals: what constitutes acceptable practice? Frontiers in Ecology and the Environment 4, Wilson, R.P., Kreye, J.A., Lucke, K., Urquhart, H., Antennae on transmitters on penguins: balancing energy budgets on the high wire. Journal of Experimental Biology 207,

67 The Effects of Handling and Anesthetic Agents on the Stress Response and Carbohydrate Metabolism in Northern Elephant Seals Cory D. Champagne 1 *, Dorian S. Houser 2, Daniel P. Costa 1, Daniel E. Crocker 2 1 Department of Ecology and Evolutionary Biology, University of California, Santa Cruz, California, United States of America, 2 Department of Biology, Sonoma State University, Rohnert Park, California, United States of America Abstract Free-ranging animals often cope with fluctuating environmental conditions such as weather, food availability, predation risk, the requirements of breeding, and the influence of anthropogenic factors. Consequently, researchers are increasingly measuring stress markers, especially glucocorticoids, to understand stress, disturbance, and population health. Studying free-ranging animals, however, comes with numerous difficulties posed by environmental conditions and the particular characteristics of study species. Performing measurements under either physical restraint or chemical sedation may affect the physiological variable under investigation and lead to values that may not reflect the standard functional state of the animal. This study measured the stress response resulting from different handling conditions in northern elephant seals and any ensuing influences on carbohydrate metabolism. Endogenous glucose production (EGP) was measured using [6-3 H]glucose and plasma cortisol concentration was measured from blood samples drawn during three-hour measurement intervals. These measurements were conducted in weanlings and yearlings with and without the use of chemical sedatives under chemical sedation, physical restraint, or unrestrained. We compared these findings with measurements in adult seals sedated in the field. The method of handling had a significant influence on the stress response and carbohydrate metabolism. Physically restrained weanlings and yearlings transported to the lab had increased concentrations of circulating cortisol (F 11, 46 = 25.2, p,0.01) and epinephrine (F 3, 12 = 5.8, p = 0.01). Physical restraint led to increased EGP (t = 3.1, p = 0.04) and elevated plasma glucose levels (t = 8.2, p,0.01). Animals chemically sedated in the field typically did not exhibit a cortisol stress response. The combination of anesthetic agents (Telazol, ketamine, and diazepam) used in this study appeared to alleviate a cortisol stress response due to handling in the field without altering carbohydrate metabolism. Measures of hormone concentrations and metabolism made under these conditions are more likely to reflect basal values. Citation: Champagne CD, Houser DS, Costa DP, Crocker DE (2012) The Effects of Handling and Anesthetic Agents on the Stress Response and Carbohydrate Metabolism in Northern Elephant Seals. PLoS ONE 7(5): e doi: /journal.pone Editor: Juan Fuentes, Centre of Marine Sciences & University of Algarve, Portugal Received December 10, 2011; Accepted May 8, 2012; Published May 31, 2012 Copyright: ß 2012 Champagne et al. This is an open-access article distributed under the terms of the Creative Commons Attribution License, which permits unrestricted use, distribution, and reproduction in any medium, provided the original author and source are credited. Funding: This research was supported by National Science Foundation (NSF) grant # The funders had no role in study design, data collection and analysis, decision to publish, or preparation of the manuscript. Competing Interests: The authors have declared that no competing interests exist. * champagn@biology.ucsc.edu Introduction Free-ranging animals often cope with fluctuating environmental conditions such as weather, food availability, predation risk, the requirements of breeding, and the influence of anthropogenic factors. An animal s response to perturbation is, in large part, mediated by stress hormones (e.g. cortisol & epinephrine) [1]. These hormones have strong impacts on energy balance and metabolism, especially the maintenance of blood glucose levels [2,3]. Thus, glucocorticoid concentrations have been correlated with food availability [4], increased feeding behavior [5], human disturbance [6 8], and survival [9 11]. Consequently, researchers are increasingly attempting to measure stress markers, especially glucocorticoids (e.g. cortisol), to understand stress, disturbance, and health in free-ranging populations [6,12 14]. Studying free-ranging animals comes with numerous difficulties posed by environmental conditions and the particular characteristics of study species. Some tissues may be collected without animal handling (e.g. hair and feces) for glucocorticoid measurement [15,16]. Ideally, however, researchers gather information not only on the indicators of stress but also a measure of physiological state (e.g. energy expenditure or metabolism) across multiple lifehistory stages [17]. These measures of metabolism in free-ranging animals, however, can be challenging. The doubly-labeled water method revolutionized the measurement of metabolic rate in freeranging animals [18,19] while advances in instrument technology now allow for the remote measurement of foraging behaviors [20 22] as well as estimates of metabolic rate by heart rate [23,24] or accelerometry data [25]. For some studies, physiological measurements can be made in free-ranging animals by temporarily implanting probes and attaching recording devices [26]. In each of these cases, some degree of animal handling is required to investigate vital function in free-ranging animals. Usually, samples can only be collected after either physically restraining the animal or using chemical sedatives both have potential confounding effects on the measured parameters. To counteract these stress artifacts researchers typically attempt to minimize any stress response due to capture by re-assessing and adjusting handling protocols [27 29]. For example, corticosterone measurements from blood samples collected within 2 3 minutes of capture PLoS ONE 1 May 2012 Volume 7 Issue 5 e38442

68 Stress Response to Handling in Elephant Seals probably reflect basal conditions in many bird species [30]. Conversely, studying an animal s response to a capture-stress protocol can provide insight into the individual s ability to cope with stressors [31]. For some study objectives and select physiological variables, transient stress responses to handling may not be important sources of measurement artifact. Since the hormones released during a stress response impact metabolic pathways [32,33] investigations of whole-animal metabolism may be particularly sensitive to artifacts of stress responses. For example, acute stress responses result in increased levels of glucocorticoids and catecholamines, which affect the release of glucose into plasma [34]. Thus, studies of fuel metabolism are potentially influenced by stress artifacts from handling. These stress responses may be most quickly observed in carbohydrate metabolism, which is normally tightly regulated [35]. A variety of studies in free-ranging animals, including investigations of carbohydrate metabolism, e.g. [36,37 42], and static measures of metabolites and hormones, e.g. [43,44 47], are potentially impacted by responses to handling and sampling. Few studies, however, have quantitatively examined the impact of handling, chemical immobilization, or stress on glucose metabolism in wildlife. The aim of this study was to measure the stress response from handling and sedation and determine its influence on physiological parameters (e.g. plasma glucose concentration and the rate of glucose production and use). We compared the metabolic responses to handling and restraint using standard metabolic tracer techniques to measure endogenous glucose production (EGP) and radioimmuno assay (RIA) to measure hormone levels in a well-studied species, the northern elephant seal (Mirounga angustirostris). We investigated the variability in metabolic and endocrine responses to capture and handling among four age classes: weaned pups, yearlings, adult females, and adult males. In one year, measurements were conducted under experimentally manipulated handling conditions 1) chemically sedated, 2) physically restrained, and 3) unrestrained seals. The response to handling in these controlled conditions was then compared with measurements conducted in the field under chemical sedation. Methods Study Design & Experimental Groups Animals were studied during natural fasts while hauled-out at Año Nuevo state park (San Mateo county, CA) and included four age classes weaned pups, yearlings, adult females, and adult males; these broad age classes are easily identified by size and pelage coloration. The study design and measurement conditions for each group are summarized in Table 1. There were two separate study groups: the field sedated and handling manipulated groups. Field sedated animals were only investigated while under chemical sedation at the field site and included weaned pups, adult females, and adult males. In the handling manipulated group, measurements were conducted under three experimental conditions 1) chemically sedated, 2) physically restrained, and 3) unrestrained. The handling manipulated group was composed of weanlings and yearlings. Measurements were made in weanlings in the field under chemical sedation and while physically restrained. Yearlings were studied while chemically sedated and while unrestrained but confined within a transport cage (see below for details). Under each experimental condition, EGP was measured over a minute sampling period. Blood samples were drawn periodically for subsequent analysis of cortisol concentration in all study animals and plasma glucose and epinephrine in a subset of study animals. Table 1. Summary of the experimental design and treatment groups used in this study. Age Class Study Year Animal State Restraint Type n Handling Manipulated Group Weanling 2008 mid postweaning chemical sedation 5 fast physical restraint 5 Yearling 2008 late molting chemical sedation 7 unrestrained 6 Field Sedated Group Weaned Pup 2003 early postweaning chemical sedation 5 fast late postweaning chemical sedation 5 fast Adult Female 2003 early lactation chemical sedation 5 late lactation chemical sedation 7 late molting chemical sedation 6 Adult Male 2007 early breeding chemical sedation 5 late breeding chemical sedation 5 late molting chemical sedation 5 total: 66 Measurements were made in four age classes at various times during natural fasts. This study used data from 46 elephant seals and reports cortisol responses for 66 procedures. Samples were collected in three separate years: 2003, 2007, and The handling manipulation measurements were made in 2008 on weanlings and fully molted yearlings both fasting for approximately 3 4 weeks. Using these handling manipulated animals, we tested the effects of restraint in a paired sample design. To make measurements in an unrestrained condition, yearlings were transported to the animal holding facility at Sonoma State University for both chemical sedation and unrestrained measurements. All other procedures were conducted in the field. Field sedated study groups consisted of weaned pups, measured early and late in their post-weaning fast (less than 2 weeks and over 6 weeks after weaning); adult females were measured early (5 days post-partum) and late in lactation (23 days postpartum). Late molting measurements, of both adult males and females, were made in fully molted animals with estimated fasting durations of 3 4 weeks. Breeding season measurements were made in adult males early (fasting less than 3 weeks) and late (fasting over 2 months) in the season. doi: /journal.pone t001 Treatment Procedures A summary and timeline of treatment procedures is shown in Figure 1. All procedures were approved by the Institutional Animal Care and Use Committee of Sonoma State University and conducted in accordance with the Guide for the Care and Use of Laboratory Animals published by the National Research Council ( and the Guidelines for the Treatment of Marine Mammals in Field Research published by the Society for Marine Mammalogy ( Chemical sedation was performed similar to previous studies [36,48 51]. Sedation was achieved using an initial intramuscular injection of Telazol (tiletamine and zolazepam) at a dose of approximately 1.0 mg kg -1. Intravenous access for anesthetic administration and blood sampling was via the extradural vessel using an 18 G, 3.5-inch needle or catheter. Intravenous doses of ketamine ( mg kg -1 ) and diazepam (1 5 mg) were administered as needed to maintain immobilization (all drugs from Fort Dodge Laboratories, Ft. Dodge IA). Sedated procedures were performed under light anesthesia (Plane 1) and animals were eupnic throughout. A summary of the chemical doses used is provided in Table 2. PLoS ONE 2 May 2012 Volume 7 Issue 5 e38442

69 Stress Response to Handling in Elephant Seals Figure 1. Summary of the treatment procedures used in this study. The measurement of endogenous glucose production (EGP) was performed by administering [6-3 H]glucose as a tracer and periodically drawing blood samples over 180 minutes. Animals in the field sedated study group were all measured at the rookery. Weanlings and yearlings in the handling manipulation group were each measured twice over two consecutive days. Weanlings were studied at the rookery while yearlings were transported to an animal holding facility for study. doi: /journal.pone g001 Initial blood sampling occurred min after Telazol administration (5 10 minutes before the onset of the EGP measurement). Following initial blood sample collection, the glucose tracer was administered and blood samples were collected periodically for 3 hours; these serial samples were used for the EGP measurement and assessment of hormone concentrations. Immediately upon returning to the lab, blood samples were centrifuged at 800 g and 4uC, the plasma or serum collected and was stored at -80uC until further analysis. The injection of glucose tracer was defined as time zero and subsequent sample times are reported relative to tracer injection. Field Sedated Group Five weaned pups (4 female and 1 male) were measured early and late in their post-weaning fast in a paired sampling design. Adult females were measured early and late in the lactation period and after the completion of molting in a mixed sampling design. Five adult males were studied early and late in the breeding season and after molting in an unpaired (cross-sectional) sampling design. The duration of the blood sampling period was typically 180 minutes but limited to 150 minutes in adult males. Lactation and fasting durations were determined by marking and monitoring seals daily during the breeding season and throughout postweaning fasts. Molting seals were studied after the completion of molting, determined by pelage coloration. PLoS ONE 3 May 2012 Volume 7 Issue 5 e38442

70 Stress Response to Handling in Elephant Seals Table 2. The anesthetic doses used during selected procedures. Study Group mass induction dose ketamine diazepam kg mg mg/kg mg mg/kg mg mg/kg sedated weanling 106 (22) 58.0 (11.0) 0.55 (0.10) 630 (73) 5.94 (0.69) 4.6 (3.9) 4.36 (3.72) sedated yearling 111 (11) 64.3 (11.0) 0.58 (0.10) 594 (110) 5.35 (0.99) 12.0 (2.9) (2.59) early weaning 114 (17) 50.0 (0.0) 0.44 (0.00) 674 (194) 5.91 (1.70) 2.0 (2.3) 1.75 (1.99) late weaning 94 (14) 45.0 (6.8) 0.48 (0.07) 372 (100) 3.96 (1.06) 0.3 (0.7) 0.32 (0.71) early lactation 536 (33) (22.5) 0.41 (0.04) 2200 (262) 4.10 (0.49) 42.5 (24.1) 7.93 (4.50) late lactation 374 (50) (12.6) 0.48 (0.03) 2286 (507) 6.11 (1.36) 14.3 (10.6) 3.82 (2.83) molting female 311 (21) (32.0) 0.54 (0.10) 1675 (301) 5.39 (0.97) 10.2 (7.1) 3.28 (2.28) Grand Mean (0.062) 5.25 (0.88) 4.61 (3.62) The total drug doses are reported as the mean and (sd). The induction dose was equal parts tiletamine and zolazepam values shown are for each. For each agent the total dose, in mg or mg, and mass-specific doses are reported. The induction dose was administered in a single intramuscular injection; ketamine and diazepam were administered intravenously periodically over hours of sedation. Data of anesthetic doses for adult males are not reported. doi: /journal.pone t002 Handling Manipulated Group Experimentally manipulated weanlings (2 female and 3 male) were studied at the Año Nuevo rookery in two states chemically sedated and physically restrained. These weanlings had been weaned 3 4 weeks prior to measurements, determined by monitoring mother-pup pairs during the breeding season. To minimize diurnal variability these measurements were made midday, between We varied the order of treatment procedures among animals, with three animals receiving sedation on the first day of handling and two sedated on the second day. The following morning weanlings underwent the second sampling procedure. The physical restraint measurement was conducted by placing the subject on a specially designed restraint board with nylon straps to minimize animal movement. Chemical sedation was performed as described above and weanlings were kept in custom-made aluminum transport cages between the two procedures. Yearlings (4 female and 3 male) were captured from the same field location, placed in transport cages, and transported by truck to the animal holding facility at Sonoma State University, Rohnert Park, CA, for study. The cage dimensions were approximately m, large enough for juvenile seals to move freely while minimizing their ability to turn around. Yearlings were captured after the completion of molting, in May June, and had an estimated fasting duration of 3 4 weeks, similar to that of the weanlings. Yearlings were measured in two handling states chemically sedated and unrestrained but confined within a transport cage. EGP was measured under chemical sedation on the same day as capture. At the end of the procedure an indwelling catheter (16 G620 cm, MILA# 1610) was inserted into the extradural vessel and a 600 extension tube filled with saline was attached to the catheter and sealed with a cap. The animal was allowed to recover from sedation overnight and the catheter was maintained patent by a periodic saline flush. The following morning we quietly performed a second measurement in the unrestrained yearling confined within the transport cage with minimal disturbance to the seal. Tracer injection and blood sampling were conducted as before but via the catheter and extension tube. The degree of alertness varied between individual study animals and over time during the measurement. One unrestrained EGP measurement was not made due to loss of catheter patency. Hormone Analyses To assess the stress response to the different animal handling methods, cortisol and epinephrine concentrations were measured from blood samples drawn immediately prior to and approximately every 30 min during the EGP measurement. Epinephrine concentrations were only measured in the handling manipulated seals. Both hormones were measured using commercially available radioimmuno assay (RIA) kits (Siemens cortisol coat-a-count kit TKCO2; and Alpco epinephrine double-antibody kit 17-EPIHU- R50, Salem NH). The cortisol kit has previously been validated in this species [36,44]. The epinephrine kit was validated for this study using serially diluted elephant seal plasma and significant parallelism with the standard curve was observed within the range of concentrations detected in this study. Average CV s for the cortisol and epinephrine assays were 2.9 and 3.1%, respectively. Several samples did not contain detectible levels of epinephrine. These non-detectible values were assigned the detection limit of the kit, 55 pm, for statistical analysis. To assess the total hormone response during the measurement period, we calculated the area under the curve (AUC) over time by summing the areas under the hormone vs time polygons between sampling points and standardized for procedure duration by dividing by the total duration of the sampling period (e.g. 180 min). Plasma Glucose & EGP For all groups a bolus injection technique and non-compartmental model were used to measure the rate of tracer dilution [52]. A description of EGP measurement methods for field sedated animals may be found in Champagne et al [36,53]. The rates of EGP for the field sedated animals have been reported previously: weaned pups [36], adult females [53], and adult males [54]. For the handling manipulated group, each seal was administered 100 mci of [6-3 H]glucose via the extradural vein. After injection, blood samples were serially drawn for 3 hrs. Typically samples were taken over the measurement period, although performing these procedures without the use of immobilizing chemicals dictated some variation in the precise sampling intervals among study animals. The specific activity of [6-3 H]glucose was determined as described in [36,53]. Briefly, plasma samples from each sample time point were thawed and deproteinated using barium hydroxide and zinc sulfate (each 0.3 N, Sigma-Aldrich, St Louis, MO). Deproteinated plasma was then passed through ion PLoS ONE 4 May 2012 Volume 7 Issue 5 e38442

71 Stress Response to Handling in Elephant Seals exchange columns; the eluant was collected, dried, and reconstituted in water. The glucose concentration of each reconstituted sample was measured using a glucose analyzer (YSI 2300, Yellow Springs, Inc, Yellow Springs, OH). Reconstituted samples were then aliquoted in duplicate, scintillation cocktail was added and samples were counted on a liquid scintillation counter (Beckman LSC 6500; Fullerton, CA). [6-3 H]glucose specific activity was calculated for each sample as the disintegrations per minute (dpm) per mole glucose. The rate of glucose production was measured by the dilution of isotopically labeled glucose by unlabeled glucose produced over time and was determined by dividing the dose injected by the area under the clearance curve ð R a ~Dose dpm = yt ðþdt Where R a is the rate of appearance of unlabeled glucose, Dose dpm is the radioactivity of the injected tracer in disintegrations per minute (dpm), and y(t) is the exponential function describing the decay of the tracer specific activity with respect to time [52]. Two exponential functions were fit to the clearance curve by maximizing the r 2 value for each function; curve-fitting and integration were performed using Mathematica (Wolfram Research, Champaign, IL). The typical inflection point occurred 20 minutes post glucose administration. Mean r 2 values were 0.91 for the initial tracer dilution curve and 0.98 for the latter turnover curve (before and after the inflection point, respectively). Representative glucose dilution curves with and without the use of anesthetic agents are shown in Figure 2. The volume of the tracer administered to each study animal was determined by gravimetric calibration of the injection syringe. In this model of glucose kinetics, the rate of tracer dilution, R a, is equal to EGP and to the total uptake by all body tissues. Plasma glucose concentration was measured from blood samples drawn at the onset of the EGP measurement for all study animals and approximately every 30 minutes during the EGP measurement in the field manipulated study animals using a glucose analyzer (YSI 2300, Yellow Springs Instruments). In these study animals, glucose concentrations were averaged across the sampling period as an index of circulating glucose concentration during the procedure. Data Analysis Paired t-tests were used to detect differences between groups of paired individuals. To test for significant differences among groups of unpaired individuals a linear mixed effects model with seal ID as a random effect was used, followed by post-hoc tests to compare between groups. In each instance we tested the full model, including interaction terms; when the interaction terms were not significant they were removed from the model. To investigate changes in hormone concentrations during the sampling period we performed repeated-measures analysis using a linear mixed model with sample time and study group as fixed effects and seal ID as a random effect; when differences among sample times were detected we tested for differences from initial concentration using LSD post-hoc tests. There was no apparent order effect of procedure day between weanlings physically restrained on day one versus day two so this factor was not included in analyses. Statistical tests were performed using R (version , R Development Core Team, and JMP ver 9 (SAS institute, Cary NC). Figure 2. Example of [6-3 H]glucose clearance curves used to calculate endogenous glucose production (EGP). Curves are shown for one weanling and one yearling with and without the use of anesthetic agents using the same tracer dose. The lower specific activity observed in the weanling under physical restraint compared to chemical sedation indicates increased dilution of the label from higher rates of EGP under physical restraint. Equivalent 3 H doses were administered to each seal, 100 mci. dpm disintegrations per minute. doi: /journal.pone g002 Results Cortisol Response The average cortisol concentrations at each sampling point and each study group are shown in Figure 3. Only a few treatment groups showed changes in cortisol concentration with sampling time. Within the chemically sedated weanlings there was no significant change in cortisol concentration with sample time (F 6, 20.7 = 0.9, p = 0.53) but physically restrained weanlings had elevated cortisol concentration during much of the measurement period (F 6, 23 = 5.2, p = 0.002; Figure 3A). There was no significant change in cortisol concentration with sample time in unrestrained yearlings (F 6, 23.0 = 1.4, p = 0.26) whereas it was elevated under chemical sedation (F 6, 27.6 = 3.5, p = 0.01; Figure 3A). Among weaned pups sedated in the field, cortisol concentration did not significantly change with sample time early in the post-weaning fast (F 7, 28.0 = 2.1, p = 0.07) whereas there was a significant change late in the fast (F 7, 28.0 = 4.36, p = 0.002; Figure 3B). Among adult females, cortisol concentration varied by study group (F 2, 29.5 = 227.7, p#0.001) but there was no effect of sample time on cortisol concentration (F 14, 213 = 1.0, p = 0.51; Figure 3C). Within PLoS ONE 5 May 2012 Volume 7 Issue 5 e38442

72 Stress Response to Handling in Elephant Seals Figure 3. Cortisol concentrations during handling. The average cortisol concentration at each sample time within each study group; error bars represent standard errors. Note that the y-axis scales are different between the top and bottom graphs. RM ANOVA followed by pairwise post-hoc t- tests were used to test for significant differences from initial cortisol concentration. A) Handling manipulated group physically restrained weanlings and chemically sedated yearlings showed increased cortisol levels during sampling. + and * indicate significant differences from initial (time = 0) cortisol value for physically restrained weanlings and chemically sedated yearlings, respectively (pairwise post-hoc t-test, p,0.05). B) Weaned pups early and late in post-weaning fast late in the post-weaning fast, pups showed increased cortisol concentrations after 100 minutes of chemical sedation. * indicates significant difference from initial cortisol value (pairwise post-hoc t-test, p,0.05). C) Adult females early and late in lactation and after molting and D) adult males early and late in the breeding season and after molting. There was no significant difference in cortisol concentration with sample time among the adult samples. doi: /journal.pone g003 adult males there was no difference in cortisol concentration among study groups (F 2, 12 = 0.1, p = 0.9) nor did cortisol vary with sample time (F 10, 120 = 0.4, p = 0.95; Figure 3D). Cortisol AUC values varied among study groups (F 11, 46.5 = 25.2, p#0.001; Figure 4). The least cortisol response to handling was observed in animals chemically sedated in the field while the greatest was in physically restrained weanlings and yearlings that were transported to the lab for study. Physical restraint increased cortisol levels in weanlings; both initial and AUC values were greater under physical restraint than chemical immobilization (paired t = ; p = 0.04, 0.008, respectively). In yearlings, cortisol levels were not different between the sedated PLoS ONE 6 May 2012 Volume 7 Issue 5 e38442

73 Stress Response to Handling in Elephant Seals Figure 4. Cortisol AUC value for each study group. The total cortisol present during the sampling period (cortisol AUC) for each study group; symbol and color coding matches that of Figure 3. Groups without overlapping letters were significantly different (p,0.05). See text and Table 1 for additional descriptions of study groups. Central horizontal lines indicate median of each group; whiskers extend to data points within 1.5 times the interquartile range from each box. doi: /journal.pone g004 and unrestrained states for either initial or AUC values (paired t- tests, p.0.1). Epinephrine Response Epinephrine was only measured in the handling manipulated group. The average epinephrine concentrations during these procedures are shown in Figure 5A. Epinephrine concentration varied with sample time (F 5, 88.3 = 2.4, p = 0.04). There was no effect of study group (p = 0.17) but the group-by-sample time interaction was significant (F 15, 88.3 = 2.0, p = 0.02). There was no change in epinephrine concentration with sample time in sedated weanlings (LSD post-hoc tests, p.0.05) but sample time had a significant influence on epinephrine concentration in physically restrained weanlings. Within this group the initial epinephrine concentration was different than that from any other sample time (LSD post-hoc tests, p,0.05). In contrast, there was no significant difference in epinephrine concentration with sample time in sedated yearlings but there was in the unrestrained group, though only the 180 min sample was different from time zero (LSD posthoc test, p,0.05). Epinephrine AUC values varied among study groups (F 3, 12 = 5.8, p = 0.01; Figure 5B). Physical restraint resulted in higher initial epinephrine and AUC values in weanlings (paired t = 3.6, 2.9, p = 0.02, 0.045, respectively) and unrestrained yearlings had higher epinephrine AUC values than during chemical sedation (paired t = 2.9, p = 0.03; Figure 5B) but the yearlings initial epinephrine concentrations were not different between the sedated and unrestrained states (paired t-test, p.0.6). Glucose Metabolism During Handling Plasma glucose concentrations were measured from samples taken periodically during the EGP measurement in handling manipulated animals (Figure 6A). Both the initial plasma glucose concentrations and average levels during the EGP procedure varied by study group (initial concentration: F 3, 11.9 = 6.9, p = 0.006; average levels: F 3, 11 = 15.5, p#0.001). Glucose concentration was higher during physical restraint than during chemical sedation, both the initial concentrations and average levels throughout the sampling period (paired t = 4.7, 8.2; p = 0.018, 0.004, respectively). There was, however, no difference in plasma glucose level between chemically sedated and unrestrained yearlings (paired t-test, p.0.6). EGP was 20% higher in weanlings under physical restraint compared with chemical immobilization (paired t = 3.1, p = 0.04, Figure 6B). There was no difference in EGP between sedated and unrestrained yearlings (paired t-test, p.0.2). Additionally, there was no difference between yearlings and weanlings of any group (F 3, 12 = 2.2, p = 0.14; individual variation accounted for 74% of the variability in EGP). The rates of EGP for field-sedated animals have been reported elsewhere [36,53,54]. There was no relationship between EGP and cortisol AUC when accounting for body mass and study group (F 1, 14.7 = 1.0, p = 0.33; Figure 7). Within physically restrained weanlings alone, PLoS ONE 7 May 2012 Volume 7 Issue 5 e38442

74 Stress Response to Handling in Elephant Seals Figure 5. Epinephrine concentration in handling manipulated seals. A) Epinephrine concentrations were generally stable during procedures except in physically restrained weanlings. These restrained seals had elevated epinephrine concentrations at the beginning of the procedures; + and * indicate significant difference from initial epinephrine concentration in physically restrained weanlings and unrestrained yearlings, respectively (pairwise post-hoc t-test, p,0.05). Error bars are standard errors. B) The lowest epinephrine AUC values occurred while study animals were chemically sedated in both weanlings (paired t = 2.9, p = 0.045) and yearlings (paired t = 2.9, p = 0.03). Central horizontal lines indicate the median of each group; whiskers extend to data points within 1.5 times the interquartile range from each box. doi: /journal.pone g005 however, there was a trend toward increased EGP with cortisol AUC values in a multiple regression analysis of EGP by cortisol AUC and mass (full model: F = 38.5, p = 0.025; effect test for cortisol AUC: F = 71.3, p = 0.014; Figure 7C). Similarly, there was no correlation between EGP and epinephrine AUC values in a mixed-model analysis of EGP with study group, mass, and epinephrine AUC as predictors and seal as a random effect (F 1,16.5 = 0.03, p = 0.8). Discussion The method of handling had a significant influence on cortisol release and metabolism in northern elephant seals. Extended sedation is necessary to conduct metabolic measurements such as the EGP measurements described here, as well as other measurements including glucose tolerance tests [40,55] and measures of lipolysis [38]. This study did not detect a cortisol response during extended sedation in adult northern elephant seals. Physical restraint caused increases in circulating cortisol, PLoS ONE 8 May 2012 Volume 7 Issue 5 e38442

75 Stress Response to Handling in Elephant Seals Figure 6. Glucose response in handling manipulated seals. A) The average glucose levels in physically restrained weanlings during the EGP measurement were significantly higher than the other groups (F 3, 11 = 15.5, p,0.001). Error bars represent standard errors. B) Physical restraint significantly increased EGP (*) in weanlings (paired t = 3.1, p = 0.04) but there was no difference in EGP between chemically sedated and unrestrained yearlings (paired t-test, p.0.05). Central horizontal lines indicate the median of each group; whiskers extend to data points within 1.5 times the interquartile range from each box. doi: /journal.pone g006 epinephrine, and glucose concentrations as well as increased EGP in weanlings. Transport appeared to sensitize seals to further manipulation as chemically sedated yearlings displayed a significant cortisol response, similar to that of physically restrained weanlings. The cortisol response in yearlings, however, was not associated with increased plasma glucose concentration or increased rates of EGP. These findings are similar to reports in other species, where sedation reduced or ameliorated the stress impacts of handling [56 59]. Hormonal Response to Handling Sustained physical restraint led to a marked stress response. Cortisol concentration increased from an initial level of 586 nm to over 1000 nm during physical restraint. During chemical sedation cortisol concentration remained steady throughout procedures in nearly all study groups. In order to study unrestrained seals, yearlings were first captured and transported to the lab before any measurements were made. These yearlings displayed cortisol and epinephrine responses similar to that of physically restrained weanlings. This contrasted with the response of animals sedated in the field. Regrettably, we did not sample yearlings in the field and there are no published data on cortisol concentration in molting northern elephant seal yearlings. However, Kelso [60] conducted a study of 40 yearlings over two years (in 2008 & 2009) during their annual fall haul-out and reported cortisol concentrations of 223 (s.d. 26) and 260 (s.d. 29) nm at the beginning and end of fasting, respectively. These values were significantly lower than cortisol concentrations of yearlings measured in this study (F 3, 53 = 17.0, p#0.001) which had similar fasting durations but were sampled in different seasons. The lower cortisol levels reported in yearlings measured in the field suggest that cortisol concentrations increased during transport and were elevated by the time we collected an initial blood sample. Cortisol concentrations increased further during chemical sedation in yearlings. The similarity in cortisol responses between physically restrained weanlings and chemically sedated yearlings suggests an acute response due to capture and transport. By the following morning, cortisol concentrations in yearlings held at the lab returned to their earlier levels but these were higher than reported values measured from animals sedated in the field. These patterns suggest that transport may be inherently stressful [61], despite the apparent tolerance of northern elephant seals to this type of handling and transport. The timing of the cortisol release was similar in physically restrained weanlings and sedated yearlings (Figure 3A). Peak cortisol levels occurred at min and declined after 90 min in both groups. These findings are similar to those of Engelhard and co-authors [57] who reported increased cortisol levels in southern elephant seal pups, M. leonina, during 45 min of physical restraint. Both the absolute cortisol concentration and the timing of the response to physical restraint were similar between the two studies. Investigations in grey seals, Halichoerus grypus, also detected increased cortisol levels with handling and restraint, in weaned pups [62] and adult males [63]. Grey seal pups had increased cortisol within ten minutes of initial handling whereas, in adult males, cortisol levels began to plateau after 30 minutes of continued restraint. In the present study, the sampling period was prolonged and we identified a peak and subsequent decrease in cortisol level while seals were still under physical restraint. The magnitude of the cortisol response was much greater in northern elephant seals peak levels were over 1000 nm in physically restrained elephant seals, compared with,100 nm in grey seal weanlings and less than,480 nm in adult grey seals [62,63]. Adult elephant seals displayed remarkably stable cortisol levels during hrs of chemical sedation (Figure 3C and D). If the initial handling or anesthetic induction caused a substantial cortisol release, we would expect to find declining cortisol concentrations during the subsequent three hours of sampling under sedation. The stable cortisol concentrations observed suggest that there was not a cortisol release in response to typical sedation procedures and cortisol concentrations measured under these conditions are near baseline levels. Cortisol concentrations closely match those reported for southern elephant seals sedated using similar methods during lactation [57]. Engelhard et al did, however, detect a small but statistically significant increase in cortisol concentration,23 minutes after induction. While not statistically significant, data from early lactation and molted females in this study do show a similar trend (see Figure 3C). However, when sampling for 180 min vs 45 min in Engelhard et al, the parabolic trend in the data during the first 30 minutes appears even less substantial. There is sizeable evidence that the stress response is suppressed during lactation in several species [64 66]. Engelhard and co-authors therefore cautioned that the mild cortisol response observed during lactation in southern PLoS ONE 9 May 2012 Volume 7 Issue 5 e38442