Present address: Point Reyes Bird Observatory, 4990 Shoreline Highway, Stinson Beach, California, 94970, USA.
Maternal and birth colony effects on survival of Weddell seal offspring from McMurdo Sound, Antarctica
Article first published online: 5 FEB 2004
Journal of Animal Ecology
Volume 67, Issue 5, pages 722–740, September 1998
How to Cite
Hastings, K. K. and Testa, J. W. (1998), Maternal and birth colony effects on survival of Weddell seal offspring from McMurdo Sound, Antarctica. Journal of Animal Ecology, 67: 722–740. doi: 10.1046/j.1365-2656.1998.00242.x
- Issue published online: 5 FEB 2004
- Article first published online: 5 FEB 2004
- maternal effects;
- birth colony;
- offspring survival;
- Leptonychotes weddellii;
1. Maternal and birth colony effects on offspring survival to weaning and to reproductive age were examined for Weddell seals using mark–recapture models and over 25 years of mark–resight data from McMurdo Sound, Antarctica.
2. Pre-weaning mortality (proportion of the total pups found dead on the ice surface) varied significantly with maternal age and was significantly higher for primiparous mothers than for multiparous mothers.
3. Offspring survival to reproductive age also increased significantly with maternal age and experience. Survival of offspring from birth to 6 years of age increased significantly with maternal body length only for male offspring.
4. Increased survivorship of offspring with maternal age was only evident for offspring of multiparous (rather than primiparous) mothers. This suggests that the age when mothers have their first pup may not affect survival of offspring, although lack of effect of age on offspring survival for primiparous mothers may be due to insufficient variation in age of first reproduction in our sample for us to detect a trend.
5. Pre-weaning survival of offspring also varied significantly among pupping colonies. In 1983, a year of low reproduction and low first-year survival for the McMurdo population (Testa 1987b; Hastings 1996), first-year survival was lower for offspring born in Outer Erebus Bay (west of 166·7°E) than for offspring born in Inner Erebus Bay (east of 166·7°E).
6. Maternal effects on offspring survival may be related to quality of breeding site. Maternal age and experience varied significantly among pupping colonies. That pup population size was stable at all colonies over the study period suggests non-random age-distribution of mothers may not be an artefact of population growth, but instead result from intraspecific competition for space in good pupping colonies.
7. Offspring survival, pup population size, and mean maternal age at colonies were significantly positively correlated. This suggests that favourable environmental conditions and high maternal or paternal quality may interact to enhance survival of offspring at larger colonies.
Survival of juveniles, particularly to breeding age, is one of the most important determinants of lifetime reproductive success in mammals and birds (Clutton-Brock 1988). Population dynamics and conservation of these species also depend greatly on survival or recruitment of immature animals (Eberhardt & Siniff 1977; York 1994). Although mark–recapture studies provide ideal data for examining juvenile survival, few studies of factors affecting survival of juveniles to reproductive age have been conducted because long-term monitoring is required. Studies of juvenile survival are also often plagued by low return rates of tagged animals or small sample sizes (Testa 1987a). Thus, until the development of new statistical software packages (Lebreton et al. 1992; Smith et al. 1994), survival analyses were limited by the lack of powerful statistical techniques to handle sparse recapture data.
Maternal effects (effects of the mother's phenotype on the phenotype of her offspring) contribute significantly to breeding success and phenotypic variation in a wide diversity of animals (see Bernardo 1996 for review). However, like studies of juvenile survival, fewer studies of long-lived animals have examined maternal effects (particularly effects of maternal age and experience) because of the need for long-term monitoring (Bernardo 1996). Even larger samples sizes are required to account for covariation among maternal characteristics (e.g. maternal age and body size). Nevertheless, increased clutch size, egg size, and offspring size, growth rate and survival to fledging or weaning with maternal age has been demonstrated for many seabirds (see Ryder 1980 for review), pinnipeds (Reiter, Panken & Le Boeuf 1981; Hill 1987; Costa, Trillmich & Croxall 1988; Arnbom et al. 1993; Iverson et al. 1993; Bowen et al. 1994; Lunn, Boyd & Croxall 1994), and other mammals (see Bernardo 1996 for review). Fewer studies have demonstrated effects of maternal age and experience on offspring survival to reproductive age (Perrins & McCleery 1985; Ratcliffe, Rockwell & Cooke 1988), and to our knowledge, maternal effects on offspring survival to reproductive age remain untested for a pinniped species.
Increased breeding success with parental age or experience has often been attributed to the ability of older individuals to secure higher quality breeding sites (Coulson 1968; Pugesek & Diem 1983). However, correlations of breeding site conditions or location with parental characteristics are often not accounted for, and little evidence exists for effect of quality of breeding site after accounting for parental age (Pugesek & Diem 1983). In pinnipeds and other mammals, colonial breeding is common where females are clumped due to limited breeding habitat or food resources, and is often accompanied by polygynous breeding systems (Trivers 1972; Le Boeuf 1991). Maternal effects in colonially breeding species may be confounded with physical conditions at breeding sites (e.g. protection from weather, high surf, predators, or harassment from conspecifics) or density-dependent effects at colonies. Although density-dependent pup mortality has been demonstrated for several pinniped species (Doidge, Croxall & Baker 1984), colonial breeding has also been associated with increased offspring survival to weaning (Campagna et al. 1992). Few studies, however, have examined variation in maternal quality among colonies (but see Reiter et al. 1981), or variation in offspring survival to reproductive age among colonies. If colonies vary in quality, females may compete for access to breeding sites, resulting in non-random age distribution of females among colonies. However, non-random age distribution of females may also be an artefact of colony growth in philopatric species or of variation in timing of parturition or egg-laying with parental age (Pugesek & Diem 1983).
In McMurdo Sound, Antarctica, individual Weddell seals (Leptonychotes weddellii Lesson) and their offspring have been marked and resighted for several decades, providing a large database of known-aged seals. Data from this study provide a unique opportunity to examine maternal effects and the influence of the breeding site on offspring survival to weaning and breeding age. Weddell seals are colonial breeders and from mid-October to late November, female Weddell seals in McMurdo Sound pup in nine to 11 colonies along perennial cracks in the fast ice which result from tidal action and from movement of Erebus Glacier (Stirling 1969). Weddell seals are also polygynous and from October through December, adult males compete for underwater territories beneath these cracks for access to mates (Kaufman, Siniff & Reichle 1975; Siniff et al. 1975).
Weddell seals are ideal subjects for studying both maternal and birth colony effects on offspring survival. First, adult female Weddell seals appear territorial during the pupping season and competition among females for space in pupping colonies in McMurdo Sound may be substantial (Stirling 1974). Intra-sexual competition may therefore lead to segregation of mothers among colonies based on maternal quality. Secondly, breeding colonies in McMurdo Sound may vary in quality of sea ice conditions for pup rearing. Break-up of sea ice in McMurdo Sound occurs by late January or February in most years (Keys 1984), but fast-ice just north and south of Erebus Glacier does not break out in some years (Stirling 1969). Greater ice stability in the Inner Bay colonies may increase offspring survival through longer pup-rearing times and protection from predators late in the pupping season. Thirdly, return rates of offspring after their natal year varied with maternal age and birth colony for this population, but confounding of probability of resighting, survival, and dispersal could not be accounted for (Hill 1987). More data are currently available to examine these questions using mark–recapture models, which allow survival and resighting probabilities to be estimated separately (Seber 1982). Survival, however, cannot be distinguished from dispersal for the McMurdo population because dispersal rates are unknown. Additionally, because new software for modelling mark–recapture data allows testing for effects of individual covariates on survival (Smith et al. 1994), more powerful tests of maternal effects on offspring survival can be conducted using mark–recapture models than were previously possible.
In this study we examined with univariate approaches: (i) effects of maternal characteristics (age, experience, body size) and birth colony on survival of Weddell seal offspring to weaning and breeding age; and (ii) variation in maternal characteristics and population density among colonies. Offspring survival increased with maternal age, experience and size, and was higher at birth colonies where older, experienced mothers were found. Older, more experienced females also pupped in larger colonies. Our study was limited by the small sample size of known-aged seals so that confounding among maternal characteristics and between maternal effects and breeding colony could not be accounted for. Despite these limitations, this is the first study of a pinniped species to demonstrate significant maternal effects on offspring survival to reproductive age using mark–recapture data. Our results also suggest that increased offspring survival with maternal age and experience may be due to the ability of older females to secure higher quality breeding sites or access to superior mates.
Study site and data collection
Most adult female Weddell seals in McMurdo Sound (77°S, 166°E) give birth in colonies, but a few seals pup away from colonies in scattered areas along cracks (Fig. 1). In some years, a colony formed in the multiyear ice at the North base of Erebus Glacier (North Glacier colony; Fig. 1). Weddell seal pups in McMurdo Sound have been tagged since 1969 as part of a long-term marking programme (Siniff et al. 1977; Testa & Siniff 1987). Pups were tagged with one cattle ear tag in the webbing of each rear flipper (double-tagged). Pups found dead on the ice surface were also tagged to provide a permanent record of these pups in the database. At the time of tagging pups, pup sex, location (latitude and longitude), and identity of the mother (if the mother was tagged) were recorded. At the time of tagging and resighting adult females, location, whether females were with a pup, and the identity and sex of the pup were recorded. Tagged seals were resighted during standardized censuses (6–7 censuses conducted per year; Siniff et al. 1977; Testa & Siniff 1987) and incidentally from October to December and occasionally into February each year. The standard body length (tip of the nose to the tip of the tail) of adult females was measured in some years, after restraining animals using the bagging technique of Stirling (1966). Methods of handling animals were approved by the Institutional Animal Care and Use Committee for work done at the University of Alaska Fairbanks.
Statistical analyses and mark–recapture modelling of survival
Maternal effects on offspring survival
When matching records of mothers and their pups, bias from tag-reading errors and misassociations of mothers and pups was minimized by correcting records that had obvious errors (e.g. transposition of two numbers). Only mothers sighted at the time the pup was tagged were included because misassociations were more likely to occur later in the season when females were less protective of pups (Stirling 1969). A few mothers were assigned to more than one pup and were excluded as erroneous because twinning is rare in Weddell seals (Stirling 1969).
We used only data from adult females tagged as pups to determine effect of maternal age and experience (pup number) on offspring survival. ‘Pup number’ was determined from pupping records of adult females in Erebus Bay, and was defined as the number of the pup in a female's lifetime pupping history. Pup number would be biased low if females pupped outside Erebus Bay, but adult female Weddell seals are philopatric (Stirling 1974; Croxall & Hiby 1983). Except for mark–recapture survival analyses, the significance level for all statistical tests was 0·05 unless stated otherwise.
To examine variation in pre-weaning survival with maternal age and experience, we pooled data across cohorts and tallied the number of pups born to females of different ages and experience (pup number). We calculated the proportion of the total pups born that were found dead on the ice surface for each maternal age and pup number, and used χ2 contingency tables to test for significant variation. We also tested if proportion of pups found dead differed between primiparous (seen with a pup for the first time) and multiparous mothers to increase the power of the test. We excluded pups whose mothers were less than 5 years of age (n = 3 of 928 pups with known-aged mothers) as erroneous because these mothers were seen only one time with the pup or were seen with different pups in the same year. We also pooled pups whose mothers were greater than 10 years of age (or pup number greater than 7) because mothers at these ages had almost no dead pups and therefore expected values were less than 1·0.
Survival from birth to 6 years of age
To determine effect of maternal characteristics on offspring survival to 6 years of age, we pooled resighting data across cohorts and generated capture histories from birth to 6 years of age. Survival estimates included pre-weaning and postweaning periods because all tagged pups, including pups tagged when found dead, were included in the analysis. Seals were observed at birth and potentially observed annually from 1 to 6 years of age (seven capture occasions). Capture histories were truncated at 6 years of age for all cohorts to standardize capture histories across cohorts. Approximately 90% of pups were resighted by 6 years of age if they were ever seen again (Hastings 1996), and onset of pupping occurred at 6 years of age for female seals in McMurdo Sound (Testa 1987b). Survival from birth to 6 years of age therefore estimates the probability that female pups will survive to reproductive age. The average age of first breeding for male Weddell seals is not known (Bartsh, Johnston & Siniff 1992; Testa 1997).
The program surph (Smith et al. 1994) was used to estimate effect of maternal age (n = 228), experience (n = 228), and body size (n = 307) on offspring survival to 6 years of age. surph allowed individual covariates such as maternal age, pup number, and body size to be included in the mark–recapture model, such that each capture history was associated with a unique value of the covariate. Variables included in the model were age of offspring (in years) and the effect of the covariate. Effect of the covariate on offspring survival was included in surph models at different offspring ages and then set equal across offspring age (Smith et al. 1994). Model selection and hypothesis testing were conducted using likelihood ratio tests (LRT) with a significance level of 0·10.
We were not able to test the effect of maternal mass, girth, or condition (girth/length) on offspring survival because these measurements decline dramatically over lactation (Hill 1987) and sample sizes of measurements at parturition were too small. To provide a larger sample, we used maternal body length, which does not vary during lactation, to test effect of maternal body size on offspring survival. Pup sex was included in this analysis to determine if the survival of one sex was more dependent on maternal size than the other. Our analysis was limited in that we were not able to account for covariation in effects of maternal age, pup number, and maternal body length on offspring survival. Instead each factor was modelled separately because of small sample sizes.
To examine the interaction between effects of maternal age and experience on offspring survival, we analysed the effect of maternal age on offspring survival of primiparous vs. multiparous mothers (n = 228). Because primiparous mothers ranged from 2 to 11 years of age and multiparous mothers ranged from 3 to 17 years of age in our sample, we repeated this analysis for only pups of mothers 6–11 years of age to allow the same range of ages for both the primiparous and multiparous groups (n = 172). Pups of 2–5-year-old mothers were also excluded because of small sample size (n < 20) and, as described previously, the greater likelihood of error when assigning pups to mothers of these ages.
Unlike the Cormack model (Cormack 1964), mark–recapture models using individual covariates do not assume homogeneity of resighting and survival probabilities among individuals, but instead assume the relationship between the covariate and survival is correctly modelled by the selected link function (Smith et al. 1994). To test this assumption, we plotted the standardized residuals vs. the capture histories and generated normal Quantile-Quantile plots. These diagnostics are described in detail in Smith et al. (1994). Like all mark–recapture models, models using individual covariates cannot distinguish between mortality and permanent emigration unless emigration is also monitored. Emigration rates of juveniles from this population are unknown, and therefore survival is confounded with permanent emigration in our study. We did not correct survival estimates for tag loss, but assumed probability of tag loss did not vary significantly with the characteristics of the mother. However, tag loss does occur (Testa & Rothery 1992) and therefore, uncorrected survival estimates (i.e. response curves) were biased low. Tag loss rates, however, did not differ significantly between the sexes (Testa & Rothery 1992).
Birth colony effects on offspring survival
Pre-weaning survival. To determine if pre-weaning survival varied among birth colonies, we generated capture histories from data of multiple sightings of pups during the pup-rearing period. Six standardized censuses were conducted annually at weekly intervals from November to mid-December (Siniff et al. 1977; Testa & Siniff 1987). Capture histories included a pooled ‘precensus’ occasion (tagged before the first census), to account for mortality that occurred before censuses began, and six census occasions (seven occasions total).
Open population mark–recapture models assume that all individuals within a population have the same probability of survival and resighting (Cormack 1964; Jolly 1965; Seber 1982). If these probabilities differ among individuals, the population is said to be heterogeneous, and underestimation of standard errors may result (Burnham et al. 1987). Heterogeneity in survival and resighting probabilities was introduced in this analysis by pooling data from pups with different time intervals from birth (or tagging) to the first census. Pooling over ages and time intervals in the ‘precensus’ occasion, however, did not result in substantial bias in pre-weaning survival estimates or standard errors in previous studies of the McMurdo population (Schreer, Hastings & Testa 1996). In birds, survival estimates or their standard errors also are not biased significantly by pooling individuals over a lengthy tagging period (Smith & Anderson 1987).
Data also were pooled across cohorts (year of birth) because data were too few to include effects of both birth colony and cohort in the mark–recapture model. To reduce heterogeneity due to pooling across cohorts, only data from 1990 to 1994 cohorts were used. Resighting data from South Base and North Glacier colonies were pooled due to small sample sizes and because ice conditions were similar at these sites. Only pups born before the first census were included in the analysis to reduce heterogeneity due to pooling across pup age (n = 1594 pups). We tested for heterogeneity in survival and resighting probabilities using the program release (Burnham et al. 1987) in which two tests (tests 2 and 3) utilizing chi-square (χ2) statistics were conducted. These tests are explained in detail in Burnham et al. (1987).
If data were not significantly heterogeneous, we used the program surge (Lebreton et al. 1992) to select the best model for the data, to test the hypothesis that pre-weaning survival differed among colonies, and to estimate survival probabilities. Variables included in the model were birth colony and time, where time indicated survival from one census occasion to the next. We tested for significant differences between simpler and more complex models using LRT (Lebreton et al. 1992). We modelled the data by simplifying the most complex model (colony and time effects in probability of resighting and survival) using the following succession of tests for effects of:
1. time, colony, and the interaction of colony and time (colony*time) on probability of resighting;
2. time on survival;
3. interaction of colony and time on survival; and
4. colony on survival.
The simpler model (model with fewer parameters; see Appendix I) was rejected if the significance level of the LRT was less than 0·10. To reduce the number of formal tests conducted with the data, simpler models were also accepted if their Akaike's Information Criterion (AIC; Akaike 1973) was lower than that of the more complex model (Lebreton et al. 1992). LRTs were performed, however, when AIC values differed by less than five and when testing main hypotheses (e.g. effect of birth colony on survival; Lebreton et al. 1992).
If data were significantly heterogeneous, a correction factor (i.e. variance inflation factor) was included in AIC and LRT in surge to account for this heterogeneity. The correction factor (cf) was equal to the releaseχ2 divided by its degrees of freedom (rdf) (Lebreton et al. 1992). AIC was modified by dividing the deviance of the model by cf and then adding the result to two times the number of parameters in the model (Lebreton et al. 1992). LRT was modified by dividing the LRT statistic divided by its degrees of freedom ( df), by cf. Instead of treating the resulting statistic as a χ2, the statistic was treated as an F-test on df and rdf degrees of freedom (Lebreton et al. 1992). Survival estimates were taken from the best model of the data but were not corrected for tag loss because tag loss rate for pups during the nursing period was essentially zero (Testa & Rothery 1992). If average tagging dates differed among colonies, survival estimates for the ‘precensus’ occasion could be biased low in colonies with earlier mean tagging dates (i.e. longer time intervals from tagging to census 1). If survival during the ‘precensus’ period varied significantly among colonies, we tested if survival differences among colonies resulted from differences in time intervals from tagging to census 1 (i.e. ‘precensus’) among colonies by comparing median tagging dates of pups at colonies.
Survival estimates from the surge analysis reflect both seen (deaths on the ice surface) and unseen (deaths in the water) mortality. While deaths in the water may result from ice conditions at colonies, deaths on the ice surface are probably related to maternal condition or behaviour, such as abandonment, starvation, crushing and stillbirths (Schreer et al. 1996). To determine the amount of pre-weaning mortality attributable to deaths on the ice surface, we tallied the total number of pups born at each colony and calculated the proportion of the total pups born that were found dead on the ice surface for each colony. We used a χ2 contingency table to test if the proportion of total pups found dead varied significantly among colonies. We used only data from the 1987–94 cohorts pooled in this analysis because all dead pups found on the ice surface were tagged (and therefore recorded) during these years.
Survival from birth to 6 years of age.
Data were pooled across cohorts because data were too few to analyse colony effects in each cohort. After determining that proportion of total pups born at each colony varied significantly among years (Table 1), only data from the 1984–88 cohorts (n = 2066 pups) were used in this analysis. This restriction allowed us to analyse separately the effects of birth colony and cohort on survival, because juvenile survival did not vary significantly among the 1984–88 cohorts (Hastings 1996). Pooling across the 1984–88 cohorts, however, introduced heterogeneity into the data because probability of sighting varied among these cohorts (Hastings 1996). Data from Inaccessible Island, Little Razorback Island, and Tent Island were pooled into the Delbridge Islands because of small sample sizes. Data from North Base and North Glacier were also pooled because sample size was small for North Glacier and maternal characteristics were similar between the two colonies.
Heterogeneity in survival and resighting probabilities was tested with release as described previously. If the data were not significantly heterogeneous, the program surge was used to select the best model for the data, to test for effect of birth colony on juvenile survival, and to estimate survival probabilities (Lebreton et al. 1992). The variables included in the model were birth colony and age, where age indicates survival from one year to the next. Model selection proceeded as described previously for the analysis of pre-weaning survival. A correction factor was included in AIC and LRT in surge if the data were significantly heterogeneous. Survival estimates were not corrected for tag loss because we assumed tag loss rates did not differ significantly among seals born at different colonies.
Inclusion of cohort in the model.
Variation in juvenile survival among birth colonies may be pronounced in some years. Substantial variation in annual reproductive rate of adult females and juvenile survival among cohorts in McMurdo Sound may indicate substantial environmental variation in the Ross Sea (Testa 1987b; Hastings 1996). Environmental variation may affect some colonies more than others through greater variation among years in ice conditions and maternal condition at some colonies. To test if effect of birth colony on juvenile survival varied among cohorts, resighting data were pooled into two locations, Inner Bay (east of 166·7°E) and Outer Bay (west of 166·7 °E; Fig. 1) because data were too few to include each colony and cohort separately. Colonies were pooled into these two locations because ice stability can vary greatly between Inner and Outer Erebus Bay in some years (J.W. Testa, personal observation). Only data from the 1982–86 cohorts were included because a maximum of 10 groups could be analysed simultaneously using surge (5 cohorts with 2 locations = 10 groups; n = 2009) and survival differed significantly among these cohorts due to particularly low first-year survival of the 1983 cohort (Hastings 1996). The data were modelled in surge, after testing for heterogeneity in release as described previously. Variables included in the model were location (Inner and Outer Bay), cohort, and age.
Variation in maternal characteristics and population size among colonies
After pooling data across cohorts, we used anova to examine variation in the mean age of mothers at colonies. To test if the proportion of primiparous mothers varied significantly among colonies, we tallied the number of primiparous and multiparous mothers that were of known-age for each colony and compared proportions using a χ2 contingency table. We tested for significant annual variation in the proportion of total pups born at each colony using a χ2 contingency table and data only from 1982 to 1994, because only after 1981 were all pups born in the study area tagged. Linear regression was used to test for trends in pup population size and proportion of total pups born at each colony from 1982 to 1994.
We used linear regression to test for significant relationships among (i) number of pups born at colonies, (ii) mean maternal age at colonies, and (iii) pre-weaning and first-year survival of seals born at colonies. We calculated pre-weaning survival at colonies as the product of the six survival intervals (from censusi–1 to censusi) estimated from the surge analysis. First-year survival estimates were taken from the surge model including the effect of birth colony on first-year survival.
maternal effects on offspring survival
The proportion of the total pups born that were found dead on the ice surface varied significantly with maternal age (P = 0·03; Table 2). A higher proportion of dead pups for 8-year-old mothers and lower proportion of dead pups for mothers older than 10 years of age contributed most to the overall χ2 (Table 2). The proportion of total pups found dead did not vary significantly with pup number (P = 0·09; Table 2). When data from multiparous (pup number > 1) mothers was pooled, pre-weaning mortality was significantly higher for pups of primiparous (7·1% mortality) than multiparous (3·2% mortality) mothers (P < 0·01; Table 2).
|Pup number||Proportion dead pups||Maternal age (years)||Proportion dead pups|
|1||0·07† (298)||5||0·00 (26 )|
|2||0·05 (197)||6||0·05 (95 )|
|3||0·01 (143))||7||0·05 (148)|
|4||0·02 (97)||8||0·10* (137)|
|5||0·04 (74)||9||0·03 (126)|
|6||0·04 (47)||10||0·05 (100)|
|7+||0·01 (69)||11+||0·02* (293)|
Survival to 6 years of age
Offspring survival was best modelled as two age-classes in all surph models: (i) survival from birth to 1 years of age, and (ii) equal annual survival after 1 years of age. Model diagnostics in surph indicated adequate fit of models to data because less than 5% of the values of the standardized residuals were outside the range (– 2,2), and significant deviations from the 45° line were not apparent in Quantile-Quantile plots (Smith et al. 1994).
Annual survival of offspring after 1 year of age increased significantly with the age of their mothers (P = 0·02), but first-year survival did not (P = 0·37; Table 3). Effect of maternal age on first-year survival, however, was significantly different from effect of maternal age on annual survival after 1 years of age because effect of the maternal age on survival could not be pooled over the two age-classes (P = 0·02; Table 3). Three pups had mothers less than 5 years of age and, as mentioned previously, these records were most likely erroneous. Results were essentially identical when these pups were removed from the analysis (LRT χ21 = 5·009, P = 0·03; Fig. 2a).
|Model||AIC||np||–Lnlk||Tests between models (LRT)|
|(1) φ(a) p(a)||500·520||11||239·260||Effect of MA on p:|
|(2) φ(a) p(a + MA, 1–5)||504·164||16||236·415||(2) vs. (1), P = 0·337|
|(3) φ(a) p(a + MA, 1 = 5)||502·164||12||239·080||(3) vs. (2), P = 0·255|
|(4) φ(a1) p(a + MA, 1–5)||502·438||12||239·219||Modelling offspring age:|
|(5) φ(a2) p(a + MA, 1–5)||499·846||13||236·923||(5) vs. (4),P = 0·032*|
|(6) φ(a3) p(a + MA, 1–5)||501·552||14||236·778||(6) vs. (5), P = 0·588|
|(7) φ(a2 + MA, 1–5) p(a + MA, 1–5)||499·211||18||231·606||Effect of MA on φ:|
|(8) φ(a2 + MA, 1 = 5) p(a + MA, 1–5)||499·406||14||235·703||(9) vs. (5), P = 0·365|
|(9) φ(a2 + MA, 1) p(a + MA, 1–5)||501·026||14||236·513||(11) vs. (10), P = 0·964|
|(10) φ(a2 + MA, 2–5) p(a + MA, 1–5)||501·800||17||233·900||(11) vs. (5), P = 0·016 *|
|(11) φ(a2 + MA, 2 = 5) p(a + MA, 1–5)||496·079||14||234·039||(12) vs. (11),P = 0·115|
|(12) φ(a2 + MA, 1, 2 = 5) p(a + MA, 1–5)||495·594||15||232·797||(12) vs. (8), P = 0·016 *|
Like effect of maternal age, annual survival of offspring after 1 year of age increased significantly with pup number (P = 0·03), but pup number did not significantly affect survival of offspring from birth to 1 years of age (P = 0·46; Table 4). Effects of pup number on the two survival periods (birth to 1 years of age, and annual survival after 1 year of age) could not be pooled (P = 0·05; Table 4). Results were essentially identical when pups of mothers less than 5 years of age were removed from the analysis (LRT χ21 = 4·569, P = 0·03; Fig. 2b).
|Model||AIC||np||–Lnlk||Tests between models (LRT)|
|(1) φ(a) p(a)||500·520||11||239·260||Effect of PN on p:|
|(2) φ(a) p(a + PN, 1–5)||502·038||16||235·019||(2) vs. (1), P = 0·132|
|(3) φ(a) p(a + PN, 1 = 5)||500·170||12||238·085||(3) vs. (1), P = 0·125|
|(4) φ(a1) p(a + PN, 1–5)||499·668||12||237·834||Modelling offspring age:|
|(5) φ(a2) p(a + PN, 1–5)||497·324||13||235·662||(5) vs. (4), P = 0·037 *|
|(6) φ(a3) p(a + PN, 1–5)||498·916||14||235·458||(6) vs. (5), P = 0·523|
|(7) φ(a2 + PN, 1–5) p(a + PN, 1–5)||499·544||18||231·772||Effect of PN on φ:|
|(8) φ(a2 + PN, 1 = 5) p(a + PN, 1–5)||497·223||14||234·612||(9) vs. (5), P = 0·457|
|(9) φ(a2 + PN, 1) p(a + PN, 1–5)||498·722||14||235·386||(11) vs. (10), P = 0·443|
|(10) φ(a2 + PN, 2–5) p(a + PN, 1–5)||497·763||17||231·881||(11) vs. (5), P = 0·030 *|
|(11) φ(a2 + PN, 2 = 5) p(a + PN, 1–5)||494·625||14||233·313||(12) vs. (11), P = 0·288|
|(12) φ(a2 + PN, 1, 2 = 5) p(a + PN, 1–5)||495·497||15||232·748||(12) vs. (8), P = 0·054 *|
Maternal body size.
Annual survival of offspring to all ages increased significantly with body length of the mother (P < 0·01; Table 5). Unlike effects of maternal age and pup number, maternal size affected survival from birth to 1 year of age (P = 0·01) as well as annual survival after 1 year of age (Table 5). When sex was included in the model, survival increased significantly with maternal size for male offspring (P < 0·01), but not for female offspring (P = 0·44; Table 5; Fig. 3a,b). Survival of male and female offspring differed significantly when the effect of maternal length on survival of male offspring was included in the model (P < 0·01; Table 5). For small mothers, survival was higher for female offspring than for male offspring, whereas for larger mothers, survival was higher for male offspring than for female offspring, for both first-year survival and annual survival after 1 years of age.
|Model||AIC||np||–Lnlk||Tests between models (LRT)|
|(1) φ(a) p(a)||889·867||11||433·934||Effect of L on p:|
|(2) φ(a) p(a + L, 1–5)||894·874||16||431·437||(2) vs. (1), P = 0·417|
|(3) φ(a) p(a + L, 1 = 5)||890·715||12||433·358||(3) vs. (1), P = 0·283|
|(4) φ(a1) p(a + L, 1–5)||904·463||12||440·232||Modelling offspring age:|
|(5) φ(a2) p(a + L, 1–5)||894·954||13||434·477||(6) vs. (5), P = 0·686|
|(6) φ(a3) p(a + L, 1–5)||896·790||14||434·395|
|(7) φ(a2 + L, 1–5) p(a + L, 1–5)||889·810||18||426·905||Effect of L on φ:|
|(8) φ(a2 + L, 1 = 5) p(a + L, 1–5)||888·362||14||430·181||(8) vs. (7), P = 0·162|
|(9) φ(a2 + L, 1) p(a + L, 1–5)||890·236||14||431·118||(8) vs. (5), P = 0·004 *|
|(10) φ(a2 + L, 1, 2 = 5) p(a + L, 1–5)||890·199||15||430·100||(10) vs. (8), P = 0·687 (9) vs. (5), P = 0·010 *|
|Sex differences inφ|
|p(*) = p(a*s + L, 1–5(males) = 1–5(females))||Effect of L on φ(males):|
|(11) φ(a2*s) p(*)||902·690||21||430·345||(12) vs. (11), P = 0·001 *|
|(12) φ(a2*s + L, 1 = 5 (males)) p (*)||893·360||22||424·680||Effect of L on φ(females):|
|(13) φ(a2*s + L, 1 = 5(females)) p(*)||904·097||22||430·048||(13) vs. (11), P = 0·441|
|(14) φ(a2 + L, 1 = 5(males)) p(*)||900·185||20||430·093||Sex effect in φ: (14) vs. (12), P = 0·005 *|
Effect of maternal age on survival of offspring for primiparous vs. multiparous mothers.
Annual survival after 1 year of age increased with maternal age only for offspring of multiparous mothers (P < 0·01; Table 6). Differences in survival between seals born to primiparous vs. multiparous mothers were significant when effect of maternal age on offspring survival of multiparous mothers was included in the model (P = 0·03; Table 6). This may indicate a significant age*pup number interaction. Results were nearly identical when only pups whose mothers were 6–11 years of age were included in the analysis (primiparous mothers: LRT χ21 = 0·04, P = 0·84; multiparous mothers: LRT χ21 = 3·60, P = 0·06; Fig. 4a,b). In younger mothers, offspring survival after 1 year of age was higher for primiparous mothers than for multiparous mothers (Fig. 4a). In older mothers, offspring survival after 1 year of age was lower for primiparous mothers than for multiparous mothers (Fig. 4a). Probability of resighting offspring of primiparous mothers was also lower at older ages than that of offspring of multiparous mothers (P = 0·04; Table 6).
|Model||AIC||np||–Lnlk||Tests between models (LRT)|
|(1) φ(pn*a) p(pn*a)||510·845||22||233·422||Effect of pn on p:|
|(2) φ(pn*a) p(a)||508·688||17||237·344||(2) vs. (1), P = 0·165|
|(3) φ(pn*a) p(a + MA, 1–5 prim, 1–5 mult)||516·676||27||231·338||Effect of MA on p:|
|(4) φ(pn*a) p(a + MA, 1–5 prim = 1–5 mult)||513·453||22||234·726||(4) vs. (3), P = 0·238 (4) vs. (2), P = 0·388|
|p(*) = p(a + MA, 1–5 prim = 1–5 mult)||Modelling offspring age:|
|(5) φ(a1*pn) p(*)||505·487||14||238·743||(6) vs. (5), P = 0·077 *|
|(6) φ(a2*pn) p(*)||504·366||16||236·183||(7) vs. (6), P = 0·300|
|(7) φ(a3*pn) p(*)||505·959||18||234·980|
|(8) φ(a2 *pn + MA prim, 2 = 5) p(*)||506·362||17||236·181||Effect of MA on φ(prim):|
|(9) φ(a2*pn + MA mult, 2 = 5) p(*)||497·920||17||231·950||(8) vs. (6), P = 0·162|
|(10) φ(a2*pn + MA 2–5 prim = 2–5 mult) p(*)||499·833||14||232·917||Effect of MA on φ(mult):|
|(11) φ(a2 + MA mult, 2 = 5) p(*)||500·979||15||235·489||(9) vs. (6), P = 0·004 * Effect of pn on φ: (11) vs. (9), P = 0·029 *|
|Retesting p differences in pn||Effect of pn on p:|
|(12) φ(a2*pn + MA mult, 2 = 5) p(*) = p(a*pn+MA, 1–5 prim, 1–5 mult)||499·093||27||222·546||(12) vs. (9), P = 0·043 *|
|(13) φ(a2 + MA mult, 2 = 5) p(*) = p(a*pn+MA, 1–5 prim, 1–5 mult)||505·641||25||227·821||Effect of pn on φ: (13) vs. (12), P = 0·005*|
Birth colony effect on offspring survival
Data were insufficient to test for heterogeneity by birth colony using release. When all colonies were pooled, the data were significantly heterogeneous (releaseχ211 = 80·25, P < 0·01). An approximate goodness-of-fit test for data separated into birth colonies was calculated by subtracting the releaseχ2 from the surge LRT χ2 between the models φ(c*t) p(c*t) and φ(t) p(t) (surgeχ277 = 192·36, P < 0·01; see Appendix I for model notation; Lebreton et al. 1992). The data grouped by birth colonies were also significantly heterogeneous (χ266 = 112·11, P < 0·01). Therefore, a correction factor was used to modify AIC and LRT in surge as described previously (cf = 112·11/66 = 1·70).
After modelling the data in surge, two groups of colonies with similar survival probabilities were apparent (Table 7). Because these two groups were not chosen a priori but were suggested by results from surge, we adjusted the significance level of the F-test from 0·10 to 0·0125 to test for survival differences between these two groups (0·10 significance level/8 colonies = 0·0125). After grouping colonies, pre-weaning survival of pups was significantly lower at the Delbridge Islands, South Base (and North Glacier), and scattered areas than at other colonies (P = 0·0124; Table 8). To test if differences in survival among colonies resulted from differences in time intervals from tagging to census 1 (i.e. ‘precensus’) among colonies, we pooled data of tagging dates of pups into two groups of colonies: (i) Delbridge Islands, South Base (and North Glacier) and scattered areas; and (ii) other colonies (Big Razorback, Hutton Cliffs, North Base, Turks Head and Turtle Rock). The median tagging date of pups from colonies with lower survival (group 1) was 3 days later than that of pups from colonies with higher survival (group 2), indicating that lower survival in these colonies was not due to a longer precensus time period.
|Colony||Population sizea||Proportion dead pupsb||Total pre-weaning mortalityc||First-year survivald||Mean maternal age (years)e||Proportion multiparousf|
|Delbridge Islands||533§||0·172*||0·156a||0·303 (0·062)||7·9 (0·4)||0·42|
|Scattered Areas||381¶||0·073||0·202a||0·299 (0·060)||8·3 (0·4)||0·56|
|(North and South)||248||66||66|
|South Base||125||0·050||0·194a‡||0·480 (0·160)||8·5 (0·7)||0·58|
|Big Razorback||662||0·050||0·087†||0·292 (0·042)||8·5 (0·3)||0·55|
|Turtle Rock||524||0·056||0·082†||0·427 (0·131)||9·2 (0·3)||0·58|
|North Base||787||0·038||0·088†||0·481(0·102)‡||9·5 (0·2)||0·69|
|Turks Head||1133||0·063||0·102†||0·388 (0·074)||9·6 (0·2)||0·69|
|Hutton Cliffs||1007||0·061||0·084†||0·473 (0·120)||10·0 (0·2)||0·74|
|North Glacier||115||0·078||9·6 (0·6)||0·88|
|Model||Modified AIC||np||Dev||Tests between models|
|(1) φ(c*t) p(c*t)||4397·047||88||7175·780||Colony effect in p:|
|(2) φ(c*t) p(c*T)||4362·959||64||7199·430||(4) vs. (3)|
|(3) φ(c*t) p(c + T)||4353·010||57||7206·317||F7,66 = 6·634, P < 0·0001*|
|(4) φ(c*t) p(T)||4385·446||50||7285·259|
|(5) φ(c*t1) p(c + T)||4361·441||17||7356·650||Colony effect in φ:|
|(6) φ(c*t2) p(c + T)||4314·350||25||7249·395||(9) vs. (8)|
|(7) φ(c + t2) p(c + T)||4308·691||18||7263·575||F2,66 = 4·689,P = 0·0124*|
|(8) φ(c2 + t2) p(c + T)||4298·855||13||7263·854|
|(9) φ(t2) p(c + T)||4304·234||11||7279·797|
The proportion of total pups found dead during the pupping season also varied significantly among pupping colonies (P < 0·01), due to greater proportion of dead pups at the Delbridge Islands and fewer dead pups at North Base (Table 7). When the Delbridge Island group was removed from the analysis, proportion of pups observed dead on the ice surface did not vary significantly among the remaining colonies (χ27 = 6·17, P = 0·52).
Survival from birth to 6 years of age
Five of the eight colonies were heterogeneous with respect to survival or resighting probabilities (releaseχ2109 = 228·24, P < 0·001). We could not determine if heterogeneity was due to pooling across the 1984–88 cohorts because data grouped by pupping colony and cohort were too sparse to test for heterogeneity in release. Instead, data were modelled in surge with a correction factor (cf = 228·24/ 109 = 2·09) included in AIC and LRT to account for heterogeneity. No significant variation among colonies in probability of sighting or survival to any age was detected in the surge analysis for data pooled over the 1984–88 cohorts when the LRT tests or modified F-tests were used to compare models (P > 0·10).
Inclusion of cohort in the model
Data for cohort (1982–86) and location (Inner and Outer Bay) were significantly heterogeneous (releaseχ2101 = 148·81, P = 0·001). Thus, a correction factor was included in AIC and LRT during model selection in surge (cf = 148·81/101 = 1·47). Probability of sighting was lower at all ages and for all cohorts for animals born in the Outer Bay. Probability of first-year survival was significantly lower in pups born in the Outer Bay for the 1983 cohort only (P = 0·06; Table 9; Fig. 5). Lower first-year survival of Outer Bay pups also occurred in 1986, but was not significant (P = 0·21; Table 9; Fig. 5).
|Model||Modified AIC||np||Dev||Tests between models|
|(1) φ(l*c*a) p(l*c*a)||3674·008||116||5070·078||Effect of (c*a) on p:|
|(2) φ(l*c*a) p(l + c + a + (l*c) + (l*a) + (c*a))||3653·661||100||5087·243||(3) vs. (2)|
|(3) φ(l*c*a) p(l + c + a + (l*c) + (l*a))||3712·077||80||5232·210||F20,101 = 4·921, P = 0·001*|
|(4) φ(l*c*a) p(l + c + a + (l*c) + (c*a))||3649·168||95||5095·354|
|(5) φ(l*c*a) p(l + c + a + (l*a) + (c*a))||3646·682||96||5088·746|
|(6) φ(l*c*a) p(l + c + a + (c*a))||3642·370||91||5097·125|
|p(*) = p(l + c + a + (c*a))||Location effect on φ1(1983):|
|(7) φ(l*c*a3) p(*)||3591·825||61||5111·052||(9) vs. (8)|
|(8)φ(l*c*age1,a3) p(*)||3573·632||43||5137·282||F1, 101 = 3·616, P = 0·060*|
|(9) Inner/Outer Bay equal except 1983||3575·248||42||5142·609||Location effect on φ1(1986):|
|(10) Inner/Outer Bay equal except 1986||3573·210||42||5139·607||(10) vs. (8) F1, 101 = 1·578, P = 0·212|
Variation in maternal characteristics and population size among colonies
Females that pupped at the Delbridge Islands, Big Razorback and Scattered areas were significantly younger than females that pupped at Turks Head, North Base, and Hutton Cliffs (Bonferroni comparisons: all P < 0·007; Table 7). A greater proportion of females that pupped at the Delbridge Islands and Big Razorback were primiparous compared with females that pupped at the Hutton Cliffs and North Glacier colonies (all P < 0·001; Table 7).
The proportion of total pups produced in the population that were born at each colony varied significantly among years (P < 0·001; Table 1). From 1982 to 1994, proportion of total pups produced (and pup population size) increased only in the Hutton Cliffs colony and declined only at Turks Head (P < 0·05; Table 1). No trend in pup population size or proportion total pups produced was apparent at other colonies nor for the population as a whole from 1982 to 1994.
All variables, (i) mean maternal age at colonies, (ii) population size at colonies, and (ii) pre-weaning and first-year survival of offspring, were positively correlated (all P < 0·05), with two exceptions. First, while pre-weaning survival of seals born at colonies increased significantly with pup population size at colonies (P = 0·04), first-year survival of seals from colonies did not (P = 0·70). Second, mean maternal age at colonies was not related to pup population size at colonies when the North Glacier colony was included in the analysis (P = 0·22). The North Glacier colony was potentially an outlier because this colony formed only in 4 of 13 years. When this colony was removed from the analysis, mean maternal age at colonies increased significantly with pup population size at colonies (P = 0·03).
Maternal effects on survival of weddell seal offspring
This study has demonstrated significant effects of maternal age, experience and body size on offspring survival to weaning and to reproductive age for Weddell seals. However, our results may also suggest that offspring of younger, smaller females are more likely to permanently emigrate from McMurdo Sound. Although confounding among maternal factors could not be accounted for, the curvilinear relationships between offspring survival and both maternal age and experience suggest maternal body size may be the more influential of the three factors on offspring survival. Body size of Weddell seals increases curvilinearly with age (Stirling 1971; Bryden et al. 1984; Hill 1987), and offspring survival to all ages increased linearly with maternal body length. Why maternal age and experience did not affect first-year survival of offspring is not known. First-year survival included the pre-weaning period, however, and the relationship between maternal age and pre-weaning mortality of offspring appears parabolic, increasing until mothers are 8 years old and declining thereafter (Table 2). This may indicate increased abandonments, stillbirths, and crushing of pups by young multiparous mothers or older primiparous mothers.
Patterns between offspring survival and maternal age vary among studies and species. Many studies of birds or mammals have demonstrated a positive correlation between parental age or body size and offspring survival to independence (Ainley & Schlatter 1972; Reiter et al. 1981; Pugesek & Diem 1983; Ozoga & Verme 1986; Rockwell et al. 1993), through the first winter (Festa-Bianchet 1988), to 6 months of age (Derocher & Stirling 1996), from fledging to 1 year of age (Fitzpatrick & Woolfenden 1988), or to at least 1 year of age (Perrins & McCleery 1985; Clutton-Brock, Albon & Guinness 1987; Hill 1987). Other studies have observed increased offspring survival or recruitment until mothers are middle-aged, followed by declines at older ages (Perrins & Moss 1974; Guinness, Clutton-Brock, & Albon 1978; Ratcliffe et al. 1988). We were unable to test if offspring survival was reduced for the oldest Weddell seal mothers due to small sample sizes.
Offspring survival may increase with maternal age and size for several reasons. Older or larger mothers often have larger eggs (see Ryder 1980 for review), larger offspring (Costa et al. 1988; Arnbom et al. 1993; Iverson et al. 1993), higher offspring growth rates (Iverson et al. 1993), and greater milk energy output (Oftedal 1984). These patterns are also true in Weddell seals with larger and older mothers producing pups with heavier birth and weaning weights, faster growth rates, and greater overall weight gain (Hill 1987; Hastings 1996). Large size therefore primarily confers nutritional benefits to Weddell seal offspring.
Other potential benefits of maternal body size or experience to offspring survival, such as variation in ability of mothers to transfer non-nutritional benefits (or detriments) via the placenta or milk (antibodies, pathogens, hormones, or toxins; see Bernardo 1996 for review) or to teach offspring swimming or foraging skills, remain to be tested. Unlike species where intra- or interspecific predation or aggression leads to significant neonatal mortality (Pugesek & Diem 1983; Ozoga & Verme 1986), the ability to fend off predators is probably not an important determinant of maternal quality for Weddell seals. Killer whales (Orcinus orca Linnaeus) and leopard seals (Hydrurga leptonyx Blainville), the potential predators of Weddell seal pups, are not usually present during the pupping season due to limited breathing holes in the fast-ice. Regardless of the mechanism, maternal factors appear to affect, not only immediate survival of offspring, but also offspring survival to older ages.
Two additional results concerning maternal effects are worth discussing. One, maternal age did not significantly affect survival of offspring of primiparous mothers, suggesting that age of first reproduction may not affect offspring survival. However, a trend in the data of primiparous mothers may have been difficult to detect due to insufficient variation in age of primiparous mothers (most primiparous mothers were 6–7 years of age; Fig. 4b). A cost, in terms of survival of future offspring, for young mothers pupping in consecutive years is suggested by lower offspring survival of young multiparous mothers than of young primiparous mothers. A similar cost of pupping, in terms of future reproduction, was reported by Testa (1987b) for young female Weddell seals.
Secondly, maternal body length affected survival of male but not female offspring, possibly indicating that male survival may be more dependent on maternal investment or condition than female survival. There is no strong evidence for differential maternal investment in male and female Weddell seal pups (Hill 1987; Hastings 1996), and why males would benefit more than females from larger mothers given their similarities in size and growth rates as pups (Hill 1987) is not obvious. This result is consistent, however, with several other recent findings from the McMurdo population: (i) more female than male pups are born to primiparous than multiparous mothers; and (ii) in poor reproductive years (i.e. years when maternal condition may be reduced), more females are born and male survival is particularly reduced (Hastings 1996). Studies of other mammals and birds have also observed that male offspring are more affected by food shortages or generally more susceptible to mortality than female offspring, even in the absence of differential investment or behaviour on the part of the mother (see Clutton-Brock 1991 for review).
COLONY EFFECTS AND COLONY DYNAMICS IN McMURDO SOUND
Maternal effects on offspring survival of Weddell seals may be related to the quality of the breeding colony. Survival of offspring from colonies increased with maternal experience and mean maternal age at colonies. These two factors are indistinguishable because sample sizes are currently insufficient to analyse colony and maternal effects simultaneously. Regardless of whether effects of the mother, breeding colony or both explain the observed variation in offspring survival, non-random distribution of age-classes of mothers in McMurdo Sound suggests that colonies vary in quality and females are competing for access to good pupping colonies. Non-random age-distribution of parents among colonies may result not only from competition for breeding sites, but also from (i) colony growth where new and younger recruits are found at the periphery, or from (ii) earlier arrival, laying, or parturition dates by older females (Pugesek & Diem 1983).
However, there are several lines of evidence that non-random age distribution of females in McMurdo Sound may result from competition for space at colonies. First, pupping colonies in McMurdo Sound have existed since at least the turn of the century (Wilson 1907) and are not growing (except for a 1% increase and decrease at Hutton Cliffs and Turks Head, respectively, since 1982). Adult population size and pup production have also been stable since 1979 (Testa & Siniff 1987). Second, unlike several seabirds (Ryder 1980; Pugesek & Diem 1983) and other pinnipeds (Lunn et al. 1994), parturition date was not related to maternal age in this population (Hastings 1996), and therefore variation in parturition date with maternal age (i.e. older females arrive earlier) cannot account for age-class segregation in McMurdo Sound. Lastly, mothers more likely to abandon their pups or have still-births may give birth at the Delbridge Islands, the farthest north pupping colonies, rather than compete for space at inner colonies. Only at the Delbridge Islands were greater proportion of dead pups (17% of total born) found on the ice surface, whereas greater proportion of dead pups were not observed at several other colonies where younger mothers were found.
Females may compete for space at some colonies to increase their reproductive success by increasing their own survival or fecundity, or their offsprings’ survival. Although prey abundance at colonies may be important to Weddell seal mothers because they may feed opportunistically during lactation (Testa, Hill & Siniff 1989), colony preference is most likely not related to food availability because prey abundance increases with distance from breeding colonies (Testa et al. 1985). Additionally, females pupping in scattered (less dense) areas were generally younger and their offspring suffered greater mortality (this study), despite the likelihood of greater prey abundance at these areas (Siniff et al. 1977).
Instead, colony preference may be influenced by ice conditions at colonies; females may prefer to pup at colonies where the fast-ice remains stable for a longer period of time. Break-up of fast ice does not usually occur until after pups are weaned (mid to late December), but timing of break-up can vary greatly among colonies. Some Inner Bay colonies remain ice-locked for several years, whereas open water was present during the pupping season at Tent and Inaccessible Islands in 1987 and 1988 (J.W. Testa, personal observation). Maternal segregation among colonies may therefore be a long-term adaptation to unusual ice years. Over 45% of Weddell seal pups died during an unusually early break-up of sea ice at Pointe Géologie (Cornet & Jouventin 1980). Greater ice stability in the Inner Bay colonies may increase offspring survival through longer pup-rearing times and protection from predators late in the season. Killer whales are seen in McMurdo Sound from November to March (Testa 1994), although predation rates on Weddell seal pups have not been measured.
Weddell seals may prefer to pup in colonies where ice is stable throughout the pupping season, but not so thick that pups cannot haul out of the water. At South Base, North Glacier and scattered areas greater numbers of dead pups were not found on the ice but pre-weaning survival estimated with mark–recapture models was low. Therefore, more unseen pup deaths must have occurred at these colonies. Unseen deaths in the water account for approximately one-half the pre-weaning mortality in the McMurdo population (Schreer et al. 1996). Young pups have difficulty hauling out of the water (Thomas & Demaster 1983) and particularly thick ice at South Base and North Glacier may increase this difficulty. Smaller colonies where fewer adults are present to maintain breathing and access holes in the sea ice may also increase the risk of drowning to pups.
Maternal or colony effects may be particularly important to offspring survival in some years. Only in 1983 cohort was first-year survival and probability of sighting at all ages lower for pups from the Outer Bay than for pups from the Inner Bay. Reproductive rate of adult females and first-year survival were lower in 1983 than in other years in the McMurdo population (Testa 1987b; Hastings 1996). Results from our study suggest that low juvenile survival of the 1983 cohort was due to poor survival of pups born in the Outer Bay. Greater annual variation in offspring survival from the Outer Bay than that from the Inner Bay may indicate greater annual variation in ice conditions in the Outer Bay or that younger mothers are more affected by environmental variation. Lower survival of seals born in the Outer Bay, however, may reflect a greater tendency of these seals to permanently emigrate, as significantly lower resighting probabilities may indicate greater movement of these seals.
Lastly, colony preference may also be due to variation in mate quality among colonies, and thereby influence offspring survival or fecundity of females. Older females pup at larger pupping colonies and colony size is probably determined by the size of cracks, with the largest cracks occurring just north and south of Erebus Glacier (Stirling 1969). If superior males fight for access to the largest colonies (i.e. the greatest number of females), females may prefer to pup at these colonies for access to superior mates. If this is true, higher survival of male offspring of larger mothers may also be due to the superior fathers of those seals. Therefore, higher survival of seals born at larger colonies may be due to the interaction of several factors beneficial to offspring survival including, greater ice-stability (because larger colonies occur in Inner Erebus Bay) and higher maternal or paternal quality.
This study was funded by grant OPP-9119885 from the National Science Foundation to J. Ward Testa and Mike A. Castellini. We extend our thanks to the many researchers that have assisted with the population study of Weddell seals in McMurdo Sound by collected tagging and resighting data over the years. We particularly thank Prof. Donald B. Siniff for initiating and maintaining the Weddell seal study in McMurdo Sound through grants from the National Science Foundation from 1968 to 1986. The logistic support by Antarctic Support Associates over the years is also greatly appreciated. Jennifer Burns, Mike Castellini, Brian and Janey Fadely, Tom Gelatt, Rob Jensen, Lorrie Rea, and Tania Zenteno-Savin assisted in the field for the 1993 and 1994 field seasons. We are grateful to Tom Gelatt and Donald Siniff for use of resighting data from the 1995 field season. Special thanks are due to Eric Rexstad for statistical assistance and critical review of the manuscript. This manuscript also benefited greatly from critical reviews by Mike Castellini, Sue Hills, Jim Sedinger and anonymous referees.
- 1972 Chick raising ability in adélie penguins. Auk, 89, 559–566. &
- 1973 Information theory and an extension of the maximum likelihood principle. International Symposium on Information Theory, 2nd edn(eds B.N.Petran & F.Csáki), pp. 267–281. Akadémiai Kiadi, Budapest, Hungary.
- 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. Canadian Journal of Zoology, 71, 1772–1781. , , &
- 1992 Territorial behavior and breeding frequency of male Weddell seals (Leptonychotes weddelli) in relation to age, size, and concentrations of serum testosterone and cortisol. Canadian Journal of Zoology, 70, 680–692. , &
- 1996 Maternal effects in animal ecology. American Zoologist, 36, 83–105.
- 1994 The effect of maternal age and other factors on birth mass in the harbour seal. Canadian Journal of Zoology, 72, 8–14. , , &
- 1984 Growth of the Weddell seal, Leptonychotes weddelli (Pinnipedia). Australian Journal of Zoology, 32, 33–41. , , &
- 1987 Design and analysis methods for fish survival experiments based on release–recapture. American Fisheries Society Monograph 5. , , , &
- 1992 Group breeding in sea lions: pups survive better in colonies. Animal Behaviour, 43, 541–548. , , , &
- 1991The Evolution of Parental Care. Princeton University Press, Princeton, New Jersey, USA.
- 1988Reproductive Success: Studies of Individual Variation in Contrasting Breeding Systems. University of Chicago Press, Chicago, Illinois, USA. , ed..
- 1987 Interactions between population density and maternal characteristics affecting fecundity and juvenile survival in red deer. Journal of Animal Ecology, 56, 857–871. , &
- 1964 Estimates of survival from the sighting of marked animals. Biometrika, 51, 429–438.
- 1980 The social plasticity of the Weddell seal (Leptonychotes weddelli L.) at Pointe Géologie. Mammalia, 44, 497–521. &
- 1988 Intraspecific allometry of neonatal size in the Antarctic fur seal (Arctocephalus gazella). Behavioral Ecology and Sociobiology, 22, 361–364. , &
- 1968 Differences in the quality of birds nesting in the centre and on the edges of a colony. Nature, 217, 478–479.
- 1983 Fecundity, survival and site fidelity in Weddell seals, Leptonychotes weddelli. Journal of Applied Ecology, 20, 19–32. &
- 1996 Aspects of survival in juvenile polar bears. Canadian Journal of Zoology, 74, 1246–1252. &
- 1984 Density-dependent pup mortality in the Antarctic fur seal Arctocephalus gazella at South Georgia. Journal of Zoology, London, 202, 449–460. , &
- 1977 Population dynamics and marine mammal management policies. Journal of the Fisheries Research Board of Canada, 34, 183–190. &
- 1988 Age-specific reproduction of bighorn ewes in Alberta, Canada. Journal of Mammalogy, 69, 157–160.
- 1988 Components of lifetime reproductive success in the Florida scrub jay. Reproductive Success: Studies of Individual Variation in Contrasting Breeding Systems (ed. T.H.Clutton-Brock). University of Chicago Press, Chicago, Illinois, USA. &
- 1978 Factors affecting calf mortality in red deer (Cervus elaphus). Journal of Animal Ecology, 47, 817–832. , &
- 1996 Juvenile survival and maternal strategies of Weddell seals in McMurdo Sound, Antarctica. MSc Thesis, University of Alaska Fairbanks, Fairbanks, Alaska, USA.
- 1987 Reproductive ecology of Weddell seals (Leptonychotes weddelli) in McMurdo Sound, Antarctica. PhD Thesis, University of Minnesota, Minneapolis, Minnesota, USA.
- 1993 The effect of maternal size and milk energy output on pup growth in grey seals (Halichoerus grypus). Physiological Zoology, 66, 61–88. , , &
- 1965 Explicit estimates from capture–recapture data with both death and immigration-stochastic model. Biometrika, 52, 225–247.
- 1975 Colony behavior of Weddell seals, Leptonychotes weddelli, at Hutton Cliffs, Antarctica. Rapport et Procès-Verbaux Ré-Unions Du Conseil International Pour L’Exploration de la Mer, 169, 228–246. , &
- 1984 Antarctic Marine Environments and Offshore Oil. Commission for the Environment, Wellington, New Zealand.
- 1991 Pinniped mating systems on land, ice and in the water: emphasis on the Phocidae. The Behaviour of Pinnipeds (ed. D.Renouf). Chapman & Hall, London.
- 1992 Modeling survival and testing biological hypotheses using marked animals: a unified approach with case studies. Ecological Monographs, 62, 67–118. , , &
- 1994 Reproductive performance of female Antarctic fur seals: the influence of age, breeding experience, environmental variation and individual quality. Journal of Animal Ecology, 63, 827–840. , &
- 1984 Body size and reproductive strategy as correlates of milk energy output in lactating mammals. Acta Zoological Fennica, 171, 183–186.
- 1986 Relation of maternal age to fawn-rearing success in white-tailed deer. Journal of Wildlife Management, 50, 480–486. &
- 1985 The effect of age and pair bond on the breeding success of great tits Parus major. Ibis, 127, 306–315. &
- 1974 Survival of young great tits in relation to age of female parent. Ibis, 116, 220–224. &
- 1983 A multivariate study of the relationship of parental age to reproductive success in California gulls. Ecology, 64, 829–839. &
- 1988 Recruitment and maternal age in lesser snow geese Chen Caerulescens Caerulescens. Journal of Animal Ecology, 57, 553–563. , &
- 1981. Female competition and reproductive success in northern elephant seals. Animal Behaviour, 29, 670–687. , &
- 1993 Age and reproductive success in female lesser snow geese: experience, senescence and the cost of philopatry. Journal of Animal Ecology, 62, 323–333. , , &
- 1980 The influence of age on the breeding biology of colonial nesting seabirds. Behavior of Marine Animals, Vol. 4 (eds J.Burger, B.L.Olla & H.E.Winn), pp. 153–168. Plenum Press, New York.
- 1996 Preweaning mortality of Weddell seal pups. Canadian Journal of Zoology, 74, 1775–1778. , &
- 1982The Estimation of Animal Abundance and Related Parameters, 2nd edn. Macmillan, New York.
- 1977 An analysis of the dynamics of a Weddell seal population. Ecological Monographs, 47, 319–335. , , &
- 1975 Movements of Weddell seals in McMurdo Sound, Antarctica, as monitored by telemetry. Rapport et Procès-Verbaux Ré-Unions Du Conseil International Pour L’Exploration de la Mer, 169, 387–393. , , &
- 1987 Effects of lengthy ringing periods on estimators of annual survival. Acta Ornithologica: Ringing Recovery Analytical Methods 23 (ed. P.M.North), pp. 69–76. &
- 1994SURPH.1 Manual: Statistical Survival Analysis of Fish and Wildlife Tagging Studies. Center for Quantitative Science, University of Washington, Seattle, Washington. , , , &
- 1966 A technique for handling live seals. Journal of Mammalogy, 47, 543–544.
- 1969 Ecology of the Weddell seal in McMurdo Sound, Antarctica. Ecology, 50, 573–586.
- 1971 Population dynamics of the Weddell seal (Leptonychotes weddelli) in McMurdo Sound, Antarctica, 1966–68. Antarctic Pinnipedia. Antarctic Research Series 18 (ed. H.Burt), pp. 141–161. American Geophysical Union, Washington, DC, USA.
- 1974 Movements of Weddell seals in McMurdo Sound, Antarctica. Australian Journal of Zoology, 22, 39–43.
- 1987a Juvenile survival and recruitment in a population of Weddell seals (Leptonychotes weddellii) in McMurdo Sound, Antarctica. Canadian Journal of Zoology, 65, 2993–2997.
- 1987b Long-term reproductive patterns and sighting bias in Weddell seals (Leptonychotes weddelli). Canadian Journal of Zoology, 65, 1091–1099.
- 1994 Over-winter movements and diving behavior of female Weddell seals (Leptonychotes weddellii) in the southwestern Ross Sea, Antarctica. Canadian Journal of Zoology, 72, 1700–1710.
- 1997 A comment on the age at onset of breeding of male Weddell seals (Leptonychotes weddellii). Canadian Journal of Zoology, 75, 156–157.
- 1989 Diving behavior and maternal investment in Weddell seals (Leptonychotes weddellii). Marine Mammal Science, 5, 399–405. , &
- 1992 Effectiveness of various cattle ear tags as markers for Weddell seals. Marine Mammal Science, 8, 344–353. &
- 1987 Population dynamics of Weddell seals (Leptonychotes weddelli) in McMurdo Sound, Antarctica. Ecological Monographs, 57, 149–165. &
- 1985 Weddell seal – Antarctic cod interactions in McMurdo Sound, Antarctica. Antarctic Nutrient Cycles and Food Webs. Proceedings of the 4th SCAR Symposium on Antarctic Biology (eds W.R.Siegfried, P.R.Condy & R.M.Laws), pp. 561–565. Springer-Verlag, Berlin, Germany. , , &
- 1983 Parameters affecting survival of Weddell seal pups (Leptonychotes weddelli) to weaning. Canadian Journal of Zoology, 61, 2078–2083. &
- 1972 Parental investment and sexual selection. Sexual Selection and the Descent of Man, 1871–1971 (ed. B.Campbell), pp. 136–179. Aldine, Chicago, Illinois.
- 1907 Mammalia (Seals and Whales). National Antarctic Expedition 1901–04, Natural History II, Zoology, pp. 1–66. British Museum (Natural History), London.
- 1994 The population dynamics of northern sea lions. 1975–85. Marine Mammal Science, 10, 38–51.
Received 14 February 1997;revision received 21 December 1997
Example of model notation for surge models. Variables in survival models for the effect of birth location on offspring survival for the 1982–86 cohorts (Table 9): l = birth location (Inner Bay vs. Outer Bay), c = cohort (year of birth), a = age. Examples show only modelling of survival (φ); identical notation was used for modelling resighting probability (p).
|Model||Variables in model||Explanation of variables|
|Most complex model||φ(l*c*a)||l + c + a + (l*c) + (l*a) + (lc*a) + (l*c*a)||All main effects (location of birth, cohort, and age), two-way and three-way interactions|
|φ(l + c + a + (l*c) + (1*a) + (c*a))||All main effects, two-way interactions only, no three-way interaction|
|φ(1 + c+ a)||All main effects only, no interaction terms|
|φ(l*a)||l + a + (l*a)||Birth location and age main effects, birth location and age interaction|
|φ(l + a)||Birth location and age main effects only, no birth location and age interaction|
|Simplest models||φ(l)||Birth location effect only, survival constant across age and cohorts|
|φ(a)||Age effect only, survival equal across birth locations and cohorts|