© 2003 by ICES/CIEM International Council for the Exploration of the Sea/Conseil International pour l'Exploration de la Mer
Trends in age-at-maturity and growth parameters of female Northeast Atlantic harp seals, Pagophilus groenlandicus (Erxleben, 1777)
a Norwegian College of Fisheries Science, University of Tromsø N-9037 Tromsø, Norway
b SevPINRO 17 Uritsky Street, 163002 Arkhangelsk, Russia
c Greenland Institute of Natural Resources DK–3900 Nuuk, Greenland
d Institute of Marine Research Sykehusveien 21, 9291 Tromsø, Norway
*Correspondence to A. K. Frie, Institute of Marine Research, Sykehusveien 21, 9291 Tromsø, Norway; tel: +47 776 82422; fax: +47 776 29100. e-mail: anne.kristine.frie{at}imr.no; lmm{at}sevpinro.ru; mcsk{at}natur.gl; tore.haug{at}fiskforsk.norut.no.
We analyzed and compared trends in age-at-maturity and body growth in the Greenland Sea and Barents Sea stocks of harp seals, Pagophilus groenlandicus, from the early 1960s to the early 1990s. Mean and median age at sexual maturity (MAMPM and MdAM) were estimated from Richards curves fit to age-specific proportions mature. No long-term trends were found in the Greenland Sea seals, where a common value of MAMPM (5.6 years) and MdAM (4.8 years) could be fit to samples from 1959 through 1990. There were also no significant changes in length-at-age of molting females between 1964 and 1987. For Barents Sea harp seals, MAMPM increased significantly from 5.4 years in the period 1962–1972 to 6.6 years in 1976–1985 and 8.2 years in 1988–1993, concurrently with a decline in body growth rates. Tests on MdAM also showed an increasing trend, but the grouping of samples was slightly different. Estimates of MAMPM for the Barents Sea stock were similar to previously published back-calculated values of MAM, but simulations showed that this method is sensitive to the age distribution of the sample, thus complicating comparisons between samples with different age structures. The high values of MAMPM and low growth rates in the Barents Sea stock in the late 1980s to early 1990s coincided with severe depletion of important prey species in the Barents Sea, reports of mass invasions of harp seals along the Norwegian coast and indications of reduced body condition. All these are consistent with a hypothesis of reduced per-capita resource levels within the distribution area of Barents Sea harp seals at that time, but no cause-and-effect relationship for the long-term trend in age-at-maturity can be established.
Keywords: age-at-maturity, Barents Sea, density dependence, Greenland Sea, growth, harp seals, maturity curves, Northeast Atlantic, Pagophilus groenlandicus
Received 25 February 2002; accepted 17 February 2003.
 |
Introduction |
|---|
 Top Introduction Material and methodsResultsDiscussionReferences |
|---|
Female harp seals, Pagophilus groenlandicus, aggregate in dense "whelping patches" on pack ice in the North Atlantic to give birth to a single pup, which is nursed for about 12 days (Kovacs, 1987). Adult males gather at the periphery of the breeding patches, and females ovulate and mate shortly after their pups are weaned (Sergeant, 1991). As in other pinnipeds, implantation of the embryo is delayed for some months after fertilization (Boyd, 1991). Two management stocks are defined in the Northeast Atlantic based on differences in both timing of whelping and geographical location of whelping and molting grounds (Sergeant, 1991). The Barents Sea stock whelps in February/March in the White Sea and moults in the same area and adjacent pack-ice areas in the southeastern Barents Sea known as the "East Ice" about a month later (see Figure 1). The Greenland Sea stock whelps near the end of March in pack-ice areas off Northeast Greenland known as the "West Ice" and starts molting in approximately the same area in late April.
|
Genetic separation of the two Northeast Atlantic harp seal stocks has not been demonstrated, but both are genetically distinct from a third stock of harp seals, which inhabits the Northwest Atlantic (Perry et al., 2000). Mark–recapture studies suggest a high degree of site fidelity to native areas during breeding and molting. As yet, no mature harp seal has been reported from a non-native breeding patch and only a few animals have been reported in the molting patches of another stock (Øien and Øritsland, 1995).
In the past century, the two Northeast Atlantic harp seal stocks have been commercially exploited and jointly managed by Norway and Russia. The commercial hunt has taken place in both breeding and molting patches. Hunting pressures have varied widely, but decreased markedly for both stocks in the 1960s after a postwar period of heavy exploitation (Ulltang and Øien, 1988). Exploitation levels remained relatively low throughout the 1970s and 1980s (ICES, 2001) and there are suggestions that this has resulted in increasing numbers in both stocks (Ulltang and Øien, 1988; Øien and Øritsland, 1995). However, owing to problems and inconsistencies in the methods used for abundance estimation, no firm conclusions can be drawn about population trajectories. The Greenland Sea is thought to be a less favorable breeding habitat than the White Sea, mainly owing to worse weather and ice conditions, and hence the pup production and total population size of the Greenland Sea stock are generally expected to be considerably lower than for the Barents Sea stock (Khuzin, 1972; Sergeant, 1991). Population estimation techniques for harp seals are based on estimates of pup production, and estimates of female reproductive rates are therefore required to calculate total stock size. In the Northeast Atlantic, reproductive samples have mainly been collected during the commercial hunt in breeding and molting patches, and since pregnancy rates cannot be directly determined from any of these samples, both Norwegian and Russian monitoring efforts have focused on estimating age at first ovulation from analyses of ovaries and teeth (Khuzin, 1972; Kjellqwist et al., 1995).
Mean age-at-maturity (MAM), defined as mean age at first ovulation, has been found to vary over time in many marine mammal populations, including harp seals (e.g. Gambell, 1976; Bengtsson and Siniff, 1981; Kjellqwist et al., 1995; Sjare et al., 2000). This has commonly been explained as an effect of changes in food availability on growth rates, assuming that marine mammals generally attain sexual maturity at a fixed proportion of final body size (cf. Laws, 1956, 1959). Therefore, MAM may be a useful indicator of numbers in relation to available resources (Eberhardt and Siniff, 1977; Bengtsson and Siniff, 1981).
Much attention has been devoted to possible density-dependent changes in female reproductive parameters and other life-history characteristics in exploited populations, because of the need to determine population levels that maximize sustainable yields or accommodate other management goals (e.g. Fowler, 1984 and references therein). In the Northwest Atlantic harp seal population, Bowen et al. (1981) found a significant linear correlation between values of MAM and the 1+ population size lagged 5 years, which suggested a decreasing trend in MAM from about 6.2 years in the early 1950s to 4.5 years in the late 1970s, concurrently with an estimated decrease in 1+ population size from about 3.0 to about 1.1 million. However, because harp seal abundance estimates are calculated from estimated reproductive rates, it is impossible to prove density-dependent relationships. In addition, it is being increasingly recognized that density-independent environmental changes often interfere with density-dependent mechanisms that modulate life-history parameters (Gaillard et al., 2000). In the 1990s, the size of the 1+ component of the Northwest Atlantic harp seal population has consistently been estimated above 3.6 million, but by 1995–1997, the estimated MAM had only increased moderately to 5.6 years. Although this discrepancy may be partly due to problems with the earliest estimates of MAM and population size (Healey and Stenson, 2000; Sjare et al., 2000), it may also suggest a more complex relationship between MAM and abundance than previously indicated.
In the Barents Sea harp seal stock, the estimated MAM increased from 5.5 years in the period 1963–1972 to 8.1 years in the period 1990–1993, suggesting a considerably larger range of MAM than previously seen in harp seals (Kjellqwist et al., 1995). Caution is warranted, however, owing to differences in methods. In the Northwest Atlantic population, MAM was estimated from age-specific proportions mature in the year of sampling by the method of DeMaster (1978), whereas Kjellqwist et al. (1995) back-calculated age-at-maturity (BAM) of individual females from the number of corpora in the ovaries. This method is based on the assumption that corpora remain visible for at least 3 years after formation and that ovulation alternates between ovaries, producing a regularly alternating size sequence of corpora in the two ovaries. To distinguish clearly between the two methods, we will hereafter refer to estimates obtained by the former as MAMPM and the latter as MAMBC. Only mature animals are used for estimating MAMBC, which technically allows the use of breeding ground samples (Øritsland, 1971; Khuzin, 1972), in contrast to MAMPM, which requires representative sampling of mature and immature females. However, the back-calculation method relies on several untested assumptions and is thought to be sensitive to the age distribution of the sample, whereas MAMPM is not (DeMaster, 1984; MacLaren and Smith, 1985).
The importance of MAM in population dynamics of large mammals has been disputed. Simulation studies suggest that changes in age at first reproduction, of which age at first ovulation is a component, have little effect on population growth rates in long-lived iteroparous species compared with the same proportional change in adult mortality rates (Eberhardt, 1977; Eberhardt and Siniff, 1977; Heppell et al., 2000). On the other hand, studies on large herbivores suggest that adult mortality rates are remarkably constant and other more variable life-history parameters such as juvenile survival and fecundity rates of young females may be more significant than indicated by theoretical elasticity studies (Gaillard et al., 2000). In order to increase our understanding of harp seal population dynamics, it is important that reliable and comparable estimates of life-history parameters accumulate in the literature.
In the present study, we estimate and compare MAMPM for the two Northeast Atlantic harp seals stocks based on Richards curves fit to age-specific proportions mature in samples collected in Greenland Sea molting patches in the period 1959–1991, as well as previously published reproductive data on Barents Sea harp seals covering the period 1962–1993. We also estimate median age-at-maturity (MdAM), which is a more robust measure of central tendency than MAMPM, when the shape of maturity curves differs between samples. The different methods for estimating age-at-maturity in harp seals are evaluated and the observed trends are discussed in relation to available information on body growth rates and general ecology of the stocks.
|
Material and methods |
|---|
|
TopIntroductionMaterial and methodsResultsDiscussionReferences |
|---|
Biological sampling
Greenland Sea stock
The data set consists of ovarian data from 693 harp seals sampled in Greenland Sea molting patches by Russian scientists during the period 1959–1991 (see Table 1). Most of this data set (624 females) consists of previously unpublished data sampled and analyzed by scientists from SevPINRO in the period 1964–1991, with a spread in sampling time from 20 April to 11 June. Reproductive data from 69 molting females in the age group 3–9 years sampled in the period 1959–1964 presented in Khuzin (1972) were pooled with available data for 12 females in age group 10–15 years from SevPINRO's 1964 data set.
|
Age was determined by counting dentinal annuli in a lower canine tooth (Khuzin, 1972). Ovaries fixed in 4% formalin were cut into 2 mm thick sections and the number of corpora in each ovary was recorded. Three types of corpora were identified: (1) fresh corpora lutea from a recent ovulation of the new cycle, (2) regressing corpora lutea resulting from pregnancy in the preceding cycle, i.e. new corpora albicantia and (3) corpora albicantia from earlier pregnancies or ovulations.
The presence of any of these types of corpora was considered an evidence for attainment of sexual maturity. Body length was measured to the nearest centimeter in a straight line from snout to tail (belly down). All the data are taken from the original journals written at the time of analysis of teeth and ovaries performed shortly after sampling.
Barents Sea stock
Reproductive data for the Barents Sea stock are taken from Kjellqwist et al. (1995) and Timoshenko (1995). Kjellqwist et al. (1995) presented reproductive data for a total of 582 females pooled over three different sampling periods: 1963–1972, 1976–1985 and 1990–1993. All data from the first two periods were sampled during Norwegian commercial sealing operations in molting patches outside the White Sea in March–April, whereas the 1990–1993 data were collected throughout the year and from different areas in the Barents Sea. Timoshenko (1995) presented age-specific ovulation rates of 706 females sampled by scientists from SevPINRO in molting patches within the White Sea from the end of April to the beginning of May in 1962–1964 and 1988.
As in the Greenland Sea sample, all animals were assigned an age corresponding to the last breeding period, e.g. 50 animals taken in February 1993 were analyzed according to their age and maturity status in 1992. The presence of at least one corpus was used as a criterion for attainment of sexual maturity. For the Norwegian samples, age was determined according to Bowen et al. (1983). Although neither teeth nor ovaries have been systematically cross-read between Russia and Norway, readers have been exchanged, and according to the principal Norwegian reader, there are no significant differences in the technical routines used in the two countries (B. Bergflødt, Institute of Marine Research, Bergen, Norway, pers. comm.). In the data presented in Timoshenko (1995), animals older than 8 years were pooled into a 9+ age class, which was omitted from our analyses for the sake of consistency, since none of the other samples contained pooled age classes. In the present study, we have included age classes 1–15 years in the curve fittings following Bowen et al. (1981).
Statistical methods
Curve fitting
MAMPM and MdAM were estimated based on sigmoidal maturity curves fit to the age-specific proportions mature in each sample. The chosen maturity curve was the Richards function (Richards, 1959) in the parameterization presented by Sugden et al. (1981)
|
| (1) |
is the estimated proportion mature at age x, x the integer age (years) in the most recent breeding season before capture, 
the asymptote set to 1 in this study, assuming that all females will eventually mature, M the age (years) at the point of inflection, k the slope at the point of inflection (maximum rate of maturing; proportion per year) and m is the shape parameter related to at the point of inflection as 
.
Other sigmoidal curves like the logistic curve, Gompertz curve and the von Bertalanffy curve for length that have previously been fit to maturity and growth data in pinnipeds (York, 1983; Hammill et al., 1995; Hårding and Härkönen, 1995; Kingsley and Byers, 1998) are all special cases of the Richards curve with the shape parameter m equal to 2, 1 and 0, respectively (Richards, 1959; Sugden et al., l981). The Richards curves were fit in an Excel® 97 (Microsoft Corp.) spread sheet (using the Solver function) by optimizing the parameters M, k and m to maximize the binomial likelihood function
|
| (2) |
(x) the fit proportion at age x (of a sample size nx) that has ovulated at least once and Px is the observed proportion at age x in the sample population that has ovulated at least once. To facilitate fitting, appropriate artificial values were constructed in parameter regions, where the Richards curve is not defined, and starting values were carefully chosen. In the form used here, parameters have clear graphical meaning, and good starting values could be taken off from curves hand-fit to plots of the data. Two measures of central tendency in age-at-maturity were defined:
- MAMPM was estimated by Equation (1) based on fit Richards curves constrained to have equal shape parameters for all samples within each dataset
where
(3) (x) is the estimated proportion mature at age x and w is the oldest age group in the sample.
If
, this expression is equivalent to DeMaster's (1978) formula for MAMPM. If 
, Equation (1) is based on the assumption that all animals will be mature at age w+1.
- MdAM for each sample was estimated from unconstrained Richards curves by solving Equation (1) for
.
Different models for time-series of maturity curves were specified by constraints imposed on MAMPM or MdAM and compared by likelihood-ratio (LR) tests using the approximation that 
(ln L) is distributed as 2
2. MAMPM and MdAM of adjacent samples were compared in a stepwise procedure retaining the constraint giving the smallest change in overall likelihood, provided that it was not significant at the 5% level. Degrees of freedom (d.f.) for the LR tests in each series of tests were set equal to the maximum number of new constraints in the series to allow for unplanned comparisons. Before comparing MAMPM, we attempted to fit a common shape parameter to all samples within each data set. Comparisons of MdAM were based on unconstrained Richards curves.
Confidence (support) intervals for curve parameters
Maximum likelihood (ML) equivalents of univariate 95% confidence intervals (support intervals) for Richards parameters were estimated as the width of the interval resulting from maximizing and minimizing each parameter in turn under the constraints of the chosen model without decreasing the original log-likelihood (ln L) value by more than 1.92. The offset in ln L of 1.92 was derived from the 2.5% critical level of the standard normal distribution (since a 95% confidence interval has 2.5% outside each side). Since a variate at any critical level in a 2 distribution with 1 d.f. equals the square of the value of the standard normal distribution at half that level, the appropriate 2 value is 3.84, corresponding to a change in ln L of 1.92.
Simulations of the effects of age structure on back-calculated values of MAMBC
To elucidate the effect that different age distributions may have on back-calculated estimates of MAMBC in harp seals, a re-sampling experiment was performed based on the age structures and age-specific probabilities of BAM for the three samples of Barents Sea harp seals analyzed by Kjellqwist et al. (1995). Matrices of age-specific probabilities of BAM were calculated for each sampling period and a re-sampling model for MAMBC with optional age structures was constructed in MATLAB® 5.3 (The Mathworks Inc.). The three probability matrices were sampled by all three sample age vectors and MAMBC was estimated for each pair as the mean value of 1000 simulations.
Comparisons of length-at-age
Gompertz growth curves were fit to length-at-age data for 84 females caught in 1964 (15–26 May) and 175 females caught in 1987 (11–26 May). All animals were caught in Greenland Sea molting patches. The following formulation of the Gompertz curve was used
|
| (4) |
is the asymptotic length. Because only seals sampled during molting were included, the growth curves are likely to describe lean growth and l0 is the estimated length of pups approximately 1 month after birth (beaters). Overall differences between growth curves were tested by F-tests (Hammill et al., 1995). In addition, differences in body length-at-age between the two samples were tested by two-way ANOVA for age groups 1–10 years. Growth curves were fit by least-squares regression using the Solver function in Excel® 97.
|
Results |
|---|
|
TopIntroductionMaterial and methodsResultsDiscussionReferences |
|---|
Age at sexual maturity of Greenland Sea harp seals
In the Greenland Sea data set, females matured from the age of 3–4 years in the samples from 1959–1964 to 1990 and from 6 years in the 1991 sample (see Figure 2). With a few exceptions, mainly in the 1978 sample, all 8+ animals were mature. The Richards maturity curves fit to the Greenland Sea samples had shape parameters (m) ranging from 0.0 (1990) to 2.3 (1959–1964) in unconstrained models (Table 2), but a common m could be fit to all samples (see Table 3) and was retained for subsequent comparisons of MAMPM. Pairwise LR tests between adjacent samples showed that a common MAMPM could be fit to the samples from 1959–1964 to 1990 but not to the 1991 sample (Table 3). A common k could be fit to samples with common MAMPM (Table 3) and the most parsimonious Richards model thus had a MAMPM of 5.6 years for the first four samples (1959–1990) and 6.9 years for the 1991 sample (see Table 2). The common shape parameter was 0.8, i.e. close to a Gompertz curve. MdAM estimated from unconstrained Richards curves ranged from 4.3 years in 1990 to 6.1 years in 1991 (Table 2) and the most parsimonious model had a MdAM of 4.8 years for the first four samples and 6.1 years for the 1991 sample (Table 2).
|
|
|
It seems unlikely that the large and significant increase in age-at-maturity from 1990 to 1991 reflects a real change in age-at-maturity of the whole stock. Rather, it raises concern of unrepresentative sampling. Several authors have found that reproducing females arrive at the molting grounds later than immature animals and adult males (Khuzin, 1972; Roff and Bowen, 1983; Sergeant, 1991), which may give a positive bias in calculations of MAMPM and MdAM from early molting patch samples. The 1991 sample includes 17 females aged 6–10 years caught in early molting patches on 20–22 April, while the rest of the sample was caught during peak molting on 23–29 May. All of the 1978 samples were taken during early molting from late April to early May, and may therefore also be positively biased. Excluding all samples taken before 10 May (see Table 1) decreased MAMPM for the 1991 sample by 0.3 years (results not shown), but a test for overall differences in MAMPM was still significant (p=0.01, d.f.=3). Thus, temporal segregation per se does not seem to fully explain the high value of MAMPM in the 1991 sample.
Growth rates of Greenland Sea harp seals
Length-at-age for the 1964 sample was well described by the fit Gompertz curve as indicated by a high coefficient of determination and a low coefficient of variation (CV) (see Figure 3 and Table 4). The CV for the 1987 sample was also small, but the coefficient of determination was low and the estimated l0 was unrealistically high (122.1 cm) owing to the very few young animals in the sample. No significant differences were found in length-at-age of 1–10 years old females between the two Greenland Sea samples by two-way ANOVA (F1,9=3.73, p=0.085) and a single growth curve could be fit to the two samples (F3,253=0.513, p=0.673) (see Table 4).
|
|
Age at sexual maturity of Barents Sea harp seals
There was an increasing trend in minimum age of mature females from 3 to 4 years in the 1962–1964 and 1963–1972 samples to 5 years in the 1976–1985 and 1988 samples and 6 years in the 1990–1993 sample (see Figure 4). In the first three samples, full maturity was reached at about 7 years of age, although some older immature outliers were found in the 1976–1985 sample. In the 1988 and 1990–1993 samples, only about 50% of the 7-year-olds were mature and full maturity was not reached before the age of 12 years in the last sample. The age distributions of the Barents Sea data set were restricted in comparison with the Greenland Sea data set, especially in the Russian samples from 1962 to 1964 and 1988, which only included animals up to the age of 8 years. Scarcity of older animals in the early samples is of little significance, since full maturity was reached within the age range represented in the data. Even in the later samples, scarcity of older samples per se should not greatly affect the estimated values of MAMPM and MdAM, if the proportions mature of the available data are unbiased and the fit of the Richards curve is reasonable. As a minimum, a good fit of the Richards model requires that age-specific proportions mature generally increase with age and that the data display a single point of inflection, which is the case for the 1988 sample. Of course, maturity curves fit to cross-sectional data represent a weighted average of the maturity curves of several cohorts, and in periods characterized by a strong trend in MAMPM and MdAM, the relative weighting of different cohorts in a sample may have a significant effect on the estimated values and the fit of the Richards model.
|
In the Barents Sea data set, Richards curves fit well to all but the 1976–1985 sample, which contained several outliers and did not display a clear point of inflection. The unconstrained Richards curves for the Barents Sea data set had shape parameters ranging from 0.0 (1976–1985 and 1990–1993) to 3.1 (1963–1972) (see Table 2). A common m could not be fit to all samples (
2=12.38, d.f.=4, p=0.014) owing to conflict between the 1963–1972 and the 1976–1985 samples. Both of these samples are subsampled over many years in periods that seem to be characterized by considerable changes in timing of age-at-maturity, and the extreme values of m could possibly be due to uneven subsampling of age groups over time. This may also have contributed to the presence of outliers and the poor fit to the Richards curve in the 1976–1985 sample. Although the 1963–1972 sample has the most extreme shape parameter, we decided to fit a separate m to the data from 1976 to 1985, because this sampling period appears to be characterized by the largest change in MAMPM and could not be fit well by any model. Based on a model with separate m for the 1976–1985 sample, values of MAMPM increased from 5.3 years in 1962–1964 to 8.4 years in 1990–1993. Pairwise LR tests between adjacent samples showed that a common MAMPM could be fit to the first two samples (1962–1972) and the last two samples (1988–1993) (see Table 3), thus splitting the data set into three periods. A common k could be fit to samples with common MAMPM and a parsimonious Richards model thus had an MAMPM of 5.4 years for the period 1962–1972, 6.6 years for 1976–1985 and 8.2 years for the period 1988–1993 (see Table 2). The common shape parameter of the first and the last period was 1.2 (i.e. close to a Gompertz curve). Fitting a separate m to the 1963–1972 sample instead of the 1976–1985 sample did not change the grouping of samples in the final model and only affected the estimated MAMPM slightly in the last period (8.1 instead of 8.2 years). MdAM estimated from unconstrained Richards models ranged from 4.7 years in 1962–1964 to 7.4 years in 1990–1993 (Table 2). There was a significant difference in MdAM between the first two samples, while the rest of the samples split into two groups (see Table 3). The most parsimonious model had an MdAM of 4.7, 5.4 and 7.1 years for the periods 1962–1964, 1963–1985 and 1988–1993, respectively (Table 2).
As for the Greenland Sea samples, seasonal timing of sampling also varied considerably for the Barents Sea data set, with most of the Norwegian samples taken in March–April and both of the Russian samples taken from late April to early May. Barents Sea harp seals start molting in late March and peak in early May (Khuzin, 1972), and MAMPM and MdAM could therefore be overestimated in the Norwegian samples owing to under-representation of mature females. In addition, the 1990–1993 sample may not be entirely comparable with the other samples, because most of this sample was taken outside the molting period and in different areas of the Barents Sea. Restricting the analysis to the two Russian samples, which were both taken in late molting patches located within the White Sea, the differences in both MAMPM and MdAM were still highly significant (p<0.001, d.f.=1).
To facilitate comparisons between stocks, we performed an additional test on all late molting patch samples from both stocks including the 1962–1964 and 1988 samples from the Barents Sea and the 1959–1964, 1987, 1990 and late 1991 samples from the Greenland Sea (see Table 1). A common m could be fit to all of these samples (p=0.7, d.f.=5). Comparing MAMPM between adjacent sampling periods by a similar stepwise procedure as applied earlier, the most parsimonious Richards model grouped the 1962–1964 sample from the Barents Sea data set together with the 1959–1964 and 1987 samples from the Greenland Sea data set (results not shown). The 1988 Barents Sea sample could not be grouped with any of the adjacent samples, but the 1990 and the late 1991 sample from the Greenland Sea data set were subsequently grouped together. MAMPM was estimated at 5.3 years for both the first and the second group of samples, while MAMPM of the 1988 sample was estimated at 7.7 years. Comparisons of MdAM based on a model with free m grouped the samples similarly with a MdAM of 4.7 years for the first three samples, 7.0 years for the 1988 sample and 4.5 years for the 1990 and 1991 samples. The difference in MdAM between the first and the last group of samples was not significant.
Sensitivity of MAMBC to sample age distributions
Age structures of the three samples of Barents Sea harp seals analyzed by Kjellqwist et al. (1995) are shown in Table 1. The mode of the age distributions of maturing animals (3+) ranged from 3 years in 1963–1972 to 8 and 11 years in 1976–1985 and 1990–1993, respectively. Pairwise Kolmogorov–Smirnov tests for differences in age structure were all significant at the 0.1% level. There was a strong positive correlation between age at capture and BAM with product-moment correlation coefficients of 0.911, 0.902 and 0.960 in samples from 1963–1972, 1976–1985 and 1990–1993, respectively. Re-sampling probability matrices of age-specific BAM by age vectors with increasing dominance of older animals gave rise to increasing estimates of MAMBC (see Figure 5). The bootstrapped confidence intervals of estimates based on the 1963–1972 age vector did not include any of the estimates based on the other two age vectors, thus demonstrating an overall significant effect of age distribution on MAMBC. However, estimates based on the latter two age vectors did not differ significantly from one another. In comparison, exchanging age distributions between any of the three Norwegian samples did not change the estimated values of MAMPM and MdAM measurably and only changed the support intervals slightly (results not shown).
|
|
Discussion |
|---|
|
TopIntroductionMaterial and methodsResultsDiscussionReferences |
|---|
Age at sexual maturity of Greenland Sea harp seals
No significant long-term trend in age at sexual maturity could be detected for the Greenland Sea data set, where a single value of MAMPM (5.6 years) and MdAM (4.8 years) could be fit to all the data from 1959 through 1990. However, small differences between the early and late samples may not have been detected owing to the paucity of early samples. A notable feature of the Greenland Sea data set is the large and significant differences in both MAMPM and MdAM between samples from 1990 and 1991. Explaining this difference as a result of real changes in the age-specific proportions mature requires the invocation of high differential mortality rates of young mature females from one year to another, for which there is no evidence. Temporal segregation of mature and immature animals is a potential problem in molting patch samples, but does not seem to provide an entirely satisfactory explanation for the difference between 1991 sample and the other samples, since the differences in both MAMPM and MdAM were still significant after removing the April subsample from the 1991 sample. Alternatively, the high value of MAMPM in the 1991 sample could be due to spatial segregation of mature and immature animals in the molting patches resulting from non-uniform mixing of juvenile and parous females, as the latter join the molting patches. Øien and Øritsland (1995) report indications of spatial clustering of recaptures of harp seals tagged simultaneously in the same area as pups, which might suggest non-uniform mixing throughout the year. Any tendency of juvenile seals to associate in more or less stable feeding groups tending to haul out together in the molting patches could lead to spatial substructuring in age-specific proportions mature owing to possible effects of differential ecological conditions on timing of first ovulation. The distribution of juvenile Greenland Sea harp seals indicated by mark–recapture experiments ranges from Newfoundland to the southeastern Barents Sea (Øien and Øritsland, 1995) and the potential spatial variation in habitat quality experienced by juvenile Greenland Sea harp seals is thus considerable. Modeling of mark–recapture data from the Greenland Sea further suggests that a variable fraction of the cohorts does not return to moult in their native area for the first one or few years after birth (Øien and Øritsland, 1995). If such temporarily emigrated juveniles tend to return to their native area only as they attain sexual maturity, this could positively bias the age-specific proportions mature estimated from molting patch samples. However, data from the most recent large-scale tagging operation in the Greenland Sea covering the period 1977–1991 and comprising a total of about 17 000 pups (Øien and Øritsland, 1995) only suggest a potentially significant effect of temporary emigration in 2 of the 10 cohorts. We therefore believe that the effect of temporary emigration on estimates of MAMPM based on cross-sectional samples is likely to be rather small.
Another potential bias relates to the possible presence of animals from other stocks in the Greenland Sea molting patches. So far, no recaptures of about 56 000 animals tagged in the Northwest Atlantic over the period 1949–1993 have been reported from the Greenland Sea molting patches (Sergeant, 1991). From 1989 through 1994, about 18 000 pups were tagged in the White Sea breeding patches, and two juvenile Barents Sea harp seals were recaptured in Greenland Sea molting patches in 1994 (Øien and Øritsland, 1995). Tagging in the White Sea continued in 1995–1997, but no more have been reported as recaptured in the Greenland Sea molting patches (ICES, 2001). In contrast, most of the 870 recaptures of Greenland Sea harp seals were juvenile animals caught in the Greenland Sea molting patches (Øien and Øritsland, 1995). However, the recapture rate of Barents Sea harp seals in Greenland Sea molting patches in the late 1990s is unreliable owing to a dramatic decrease in the number of 1+ animals caught in the Greenland Sea in the period 1996–1999 (ICES, 2001).
Growth rates of Greenland Sea harp seals
There were no significant differences in length-at-age for female Greenland Sea harp seals in samples from 1964 and 1987, which is consistent with the lack of trend in MAMPM and MdAM, if sexual maturity is reached at a rather fixed proportion of final body size in marine mammals (Laws, 1956, 1959). In approximately the same period, a significant reduction in length-at-age was seen in young (1–10 years old) female Barents Sea harp seals for samples between 1963–1972 and 1990–1993 (Kjellqwist et al., 1995). The asymptotic length of Greenland Sea harp seals was estimated at 164.8 cm (95% CI: 162.4 cm–167.2 cm), which is somewhat smaller than the asymptotic lengths of female Barents Sea harp seals estimated at 172.5 cm (95% CI: 162.7 cm–182.4 cm) in 1963–1970 and 186.4 cm (95% CI: 162.7 cm–173.9 cm) in 1990–1993. However, the differences are unlikely to be statistically significant owing to the wide confidence interval of the estimates for the Barents Sea stock. Caution should furthermore be exercised in comparisons of absolute lengths between Russian and Norwegian samples, because measurements in the former were done on animals lying on their venter, while Norwegian researchers measured the lengths of animals lying on their backs. The effect of this difference in measuring technique is not known.
Chabot et al. (1996) found that the length of harp seals is positively correlated with body condition and thus comparisons of fat and lean seals may be misleading. Both of the Greenland Sea samples were collected during molting in late May, when feeding activity is likely to be reduced and both samples are thus expected to represent lean growth. The early length-at-age sample from the Barents Sea was collected in molting patches in March–April, when the seals are generally at their leanest, while the later sample contained some animals caught in February, when the seals are often in good condition (Chabot et al., 1996; Nilssen et al., 1997). Nevertheless, length-at-age of young harp seals was significantly smaller in the 1990–1993 sample than in the earlier sample. The Gompertz curve asymptotic length of female Northwest Atlantic harp seals also measured on their backs has been estimated at 164.1 cm in April samples taken in the period 1979–1994 (Chabot et al., 1996), which is within the confidence limits of all the samples from the Northeast Atlantic.
Age at sexual maturity of Barents Sea harp seals
There was a significantly increasing trend in both MAMPM and MdAM in the Barents Sea data set from the 1960s to the late 1980s and early 1990s, but the exact grouping of the first three samples differed between the two methods: comparisons of MAMPM based on curves with a common m for all but the 1976–1985 sample grouped the first two samples together, while comparisons of MdAM based on curves with independent shape parameters grouped the second and third samples together. Because the mean is not insensitive to the distribution of data in the ends of the curves, comparisons of MAMPM are preferably done between samples with equal shape parameters. MdAM is a more robust measure of central tendency, because it is almost exclusively determined by the distribution of data in the center of the curve, where the sample sizes are often largest. MdAM is therefore a useful alternative to MAMPM, when there are significant deviations from a common shape parameter. The results are, however, not easily comparable with most other estimates of age-at-maturity in pinnipeds, which are based on MAMPM (Bowen et al., 1981; Hårding and Härkönen, 1995; Sjare et al., 2000).
The values of MAMPM for the Norwegian samples from the Barents Sea stock closely resembled back-calculated values of MAM estimated by Kjellqwist et al. (1995), but simulations showed that a large part of the differences in MAMBC could be explained by major differences in the age structures of the samples. In contrast to our results, Hårding and Härkönen (1995) found no significant effect of simulated changes in age distribution on MAMBC in crabeater seals, Lobodon carcinophagus. This discrepancy could be due to different procedures for back calculation in the two species. Kjellqwist et al. (1995) included only females with less than four corpora in the calculation of MAMBC to avoid bias due to disappearance of corpora. However, since female harp seals appear to mature from the age of 3–11 years or more, individuals maturing early will tends to be increasingly under-represented in samples aged 6 years and more, leading to a positive correlation between age at capture and BAM. In addition, individual variability in the retention of corpora, which seems to be considerable (Kjellqwist et al., 1995), may lead to erroneous inclusion of older females in the calculation of MAMBC and thus also tend to bias the estimate upwards. Another problem is related to the assumption of alternation between ovaries. Kjellqwist et al. (1995) interpreted a missing alternation between ovaries in the size sequence of corpora as a missing ovulation and corrected for this by subtracting an extra year from the age at capture when estimating age at first ovulation. However, according to Fisher (1954), alternation between ovaries appears to be much less regular in harp seals than in hooded seals (Cystophora cristata) and corrections for missing alternations, that were in fact not due to a missing ovulation, may therefore bias age-at-maturity downwards. Owing to the strong correlation between MAMBC and the age structure of the sample, comparisons of MAMBC between samples with different age structures are difficult to interpret, because changes in age distribution may also be due to changes in mortality and fertility rates and not just to changes in age at first ovulation. Empirical observations suggest that newly matured females have lower birth rates than prime age females in many large mammal species including gray seals (Boyd, 1985; Schwartz and Stobo, 2000), harp seals (Øritsland, 1971; Bowen et al., 1981) and several terrestrial large herbivores (Gaillard et al., 2000). Values of MAMBC based on breeding patch samples may therefore often be biased upward by a deficiency in young females.
The only assumption regarding formation and retention of corpora underlying determination of maturity status for estimation of MAMPM and MdAM is that one corpus remains visible until the next is formed. The validity of this assumption is supported by data presented by Bowen et al. (1981), indicating that ovulation rates are close to 100% in harp seals even at times of low late-term pregnancy rates. However, the validity of the estimated values of MAMPM and MdAM is critically dependent on representative sampling of mature and immature animals, which may be hindered by temporal and spatial segregations of reproductive classes. There were no strong indications of this kind of sampling problem in the Barents Sea data set, but significant variabilities may have been masked by pooling of samples. Another potential source of bias is that the most extreme values of MAMPM and MdAM for the Barents Sea stock were observed in a period characterized by an unusual migration pattern of harp seals. Beginning in 1978, large numbers of harp seals started to appear on the coast of northern Norway in winter and early spring (Haug et al., 1991). From 1986 to 1988, the magnitude and extent of the invasions in time and space increased dramatically and about 100 000 harp seals may have drowned in fishing gear all along the Norwegian coast line (Haug et al., 1991). It cannot be excluded that the composition of mature and immature animals in the traditional molting patches were affected by these changes in migration patterns. During the invasions, 13 harp seals tagged in the Greenland Sea in 1985 and 1987 were recaptured along the Norwegian coast, primarily in northern Norway from December through May. However, according to Øien and Øritsland (1995), Greenland Sea harp seals are unlikely to have accounted for more than a few percent of the total number of harp seals caught or by-caught during the invasions, suggesting that the proportion of Greenland Sea harp seals in the Barents Sea and White Sea molting patches is also likely to be small. Nevertheless, three harp seals tagged in the Greenland Sea in each of the years 1989, 1990 and 1991 were recaptured in molting lairs in the East Ice in 1992, and it cannot be excluded that temporarily emigrated immatures from the Greenland Sea at times may influence the age-specific proportions mature in molting patches of Barents Sea harp seals. On the whole, it is clear that a better understanding of migration patterns and social structure of Northeast Atlantic harp seals is required in order to assess the reliability of different sampling regimes with respect to obtaining representative compositions of mature and immature animals.
Evaluation of different indicators of age-at-maturity in harp seals
Although none of the estimates of age-at-maturity given here are free of potential errors, we find the estimates based on age-specific proportions mature the most useful for harp seals due to their independence of age structures and the technical and theoretical simplicity of the determination of maturity status of individuals. In contrast to determining the exact number and size distributions of corpora in each ovary as needed for calculation of MAMBC, the identification of mature versus immature females based on the presence or absence of corpora is relatively straightforward and little variation is expected between readers. Probably, the most serious potential problem associated with these estimates is segregation of mature and immature animals, which requires further study. The estimates of MAMPM for Barents Sea harp seals in the late 1980s and early 1990s are well beyond the historical range of values seen in the Northwest Atlantic, which according to the latest revised estimates of historical and recent data (Sjare et al., 2000) appear to have ranged from 5.8 years in the late 1950s to 4.6 years in the early 1980s. Although these values may be biased to some extent, the total range of MAMPM seen in all three stocks of harp seals do suggest a range of values to be explored by simulations to identify critical parameters in the population dynamics of harp seals.
Possible ecological implications
Comparisons of the trends in age-at-maturity found by separate analyses within stocks suggest that the two Northeast Atlantic harp seal stocks had similar values of MAMPM and MdAM in the 1960s, but subsequently diverged and followed different trends up to the late 1980s. This was also supported by a more rigorous comparison of late molting samples from both stocks. Testing a Richards model including all the late molting patch samples from the Barents Sea (1962–1964 and 1988) and the Greenland Sea (1959–1964, 1987, 1990 and late 1991 subsamples) revealed that the 1988 Barents Sea sample was significantly different from all the other samples, which did not differ significantly from one another with respect to MAMPM or MdAM. In spite of possible biases, these results do suggest that some factor influencing age-at-maturity differed between the two Northeast Atlantic harp seal stocks at least in the 1980s.
The significant increase in MAMPM and MdAM in Barents Sea harp seals found in the present study is consistent with the results of previous studies of MAMPM by Kjellqwist et al. (1995), who found that the increasing trend in age-at-maturity of Barents Sea harp seals from 1963–1972 to 1990–1993 was likely to be due to both density-dependent factors and large-scale ecosystem changes in the Barents Sea over the same period.
The support for a change in population density of Barents Sea harp seals is mainly based on projections of a 1965 pup production estimate of approximately 100 000 based on the survival index method, giving an expected pup production of 172 000 in 1978 assuming a median age at first whelping of 5 years (Benjaminsen, 1979). Assuming a median age at first whelping of 6 years, which is more in line with estimates of MdAM in the present study, the projected pup production in 1978 was 141 000. However, as the 95% confidence interval of the 1965 pup production estimate ranged from 74 000 to 221 000, these calculations cannot be taken to indicate an increasing trend in pup production. Meanwhile, Soviet aerial surveys also suggested a moderate increase in pup production from the early 1960s to 1980s (Benjaminsen, 1979). While this may be true, some caution is warranted, because surveys of whelping females rely heavily on the accuracy of correction factors for the proportion of females in the water, which may vary considerably with time of day, meteorological conditions and perhaps also local food availability, since female harp seals have been found to feed opportunistically during lactation (Nilssen et al., 1995; Lydersen and Kovacs, 1996). Thus, to our knowledge, there is no hard evidence of an increase in population size of Barents Sea harp seals from the 1960s onwards, although there may be some reason to expect an increase in numbers owing to the implementation of several catch regulations such as full protection of whelping females from 1963, a stop in Soviet catches of 1+ animals and a general decrease in catches due to a new quota system from 1965 (Benjaminsen, 1979).
The observed changes in age-at-maturity and growth rates in Barents Sea harp seals may also have been at least partly due to large-scale changes in the Barents Sea ecosystem. Several diet studies (summarized in Nilssen et al., 2000) conducted in the Barents Sea area indicate that the diet of harp seals is dominated by krill (mainly Thysanoessa sp.) and amphipods (mainly Themisto sp.) in summer and early autumn, when the majority of the stock is thought to be in the northern part of the Barents Sea (Haug et al., 1994), while fish, mainly capelin, Mallotus villosus, herring, Clupea harengus, and Arctic cod, Boreogadus saida, dominate the diet in late autumn to early spring, when most of the seals are thought to approach the southern part of the Barents Sea. Seasonal variation in body condition in harp seals caught in the Barents Sea area suggests a peak in energy intake in July through October. The seals appear to maintain a high body condition until whelping in February March, suggesting high energy intake in winter also (Nilssen et al., 1997).
One of the most conspicuous changes in the Barents Sea ecosystem within the study period is the collapse of the Atlanto-Scandian herring stock in the 1960s, leading to the near absence of this species from the Barents Sea until the early 1990s (Gjøsæter, 1995). The Barents Sea capelin stock exceeded 8 million tonnes in the mid 1970s and then decreased steadily until 1983 (Anon., 2000), and from 1984 to 1986 it decreased dramatically from ca. 3 million tonnes to about 100 000 t and stayed at a very low level until 1990. Large fluctuations have also been observed for the Barents Sea stock of Arctic cod, which declined from an estimated total biomass of more than 2 million tonnes in 1970 to about 100 000 t in 1975. The stock was still below 500 000 t in 1985–1990 (Gjøsæter, 1995), but no estimates exist for the intervening period. It has been suggested that the invasions of harp seals to the Norwegian coast in 1986–1988 were at least partly due to food shortage resulting from simultaneous low levels of capelin, herring and Arctic cod in the southeastern Barents Sea in winter (Haug and Nilssen, 1995). Reduced body condition in seals caught in February 1988 compared with seals caught in February 1993, when capelin was abundant, provided some support for this hypothesis (Nilssen et al., 1997). Thus, although the data are insufficient to establish any causal relationships, the low growth rates and high values of MAMPM and MdAM for Barents Sea harp seals in the late 1980s and early 1990s do seem to fit into a general scenario suggesting reduced per-capita resource levels for harp seals in the Barents Sea area in the preceding years. The long-term increasing trend in age-at-maturity and decrease in growth rates suggest that this situation may have built up gradually since the 1970s or even before.
The significantly lower values of MAMPM and MdAM in Greenland Sea harp seals in the late 1980s compared with Barents Sea harp seals may suggest that Greenland Sea harp seals experienced more plentiful per-capita resource levels than Barents Sea harp seals in the preceding years, perhaps as a result of differences in seasonal distribution pattern.
As a consequence of the different locations of breeding and molting patches, the distribution of mature animals differs between stocks at least from some time before parturition till some time after the molting period. Most of the immature animals also appear to be present in their native areas around the time of molting (Øien and Øritsland, 1995). To have an effect on age-at-maturity, differences in resource levels must obviously act on the immature stages, but food availability scenarios affecting adult females may also have an indirect effect on the weaning mass of the offspring. Harp seal mothers may lose about 30% of postpartum body mass during lactation, and stored body reserves are thought to account for the majority of the energy allocated to lactation (Lydersen and Kovacs, 1996). In other phocids relying mainly on stored body reserves during lactation, maternal postpartum mass has been found to explain 40–60% of size at weaning (e.g. Kovacs and Lavigne, 1992; Iverson et al., 1993; Deutsch et al., 1994). Food availability during the last period of gestation, when the seals are close to the breeding grounds, is likely to be important for the ability of pregnant females to meet the increasing energy demands of the fetus (Stewart et al., 1988) at the same time as maintaining a high level of body reserves for lactation.
Data from both the Northwest Atlantic and the Barents Sea area suggest that the energy intake of both juvenile and adult harp seals is reduced during spring and early summer, while the most intensive accumulation of body reserves appears to occur in late summer and autumn (Chabot et al., 1996; Nilssen et al., 1997). In the Northwest Atlantic, a minor drop in body mass is observed in late autumn, but this appears to be compensated by an increase in late winter (Chabot et al., 1996). There are no comparable data for the Greenland Sea, but it seems likely that energy intake of both juveniles and adults in this area is also reduced in spring at least during molting, because more time is spent hauled out on the ice.
According to Rasmussen and Øritsland (1964), harp seal pups from the West Ice tend to drift passively southwards with prevailing currents during the first couple of months after birth. This is in agreement with the numerous tag returns from northern Iceland of harp seal pups caught 1–2 months after tagging in the Greenland Sea (Øien and Øritsland, 1995). No recaptures of Barents Sea harp seals have been reported from this area. Harp seals are recorded in Icelandic waters from November to July, with a clear peak in May (Hauksson and Bogason, 1997a). Almost all of these animals are pups of the year or yearlings. Haug et al. (unpublished material) observed a large aggregation of harp seals in the Denmark Strait between Iceland and East Greenland in February March 2001, but the stock identity of these animals cannot be stated with certainty.
A relatively large number of tags from Greenland Sea harp seals aged 1–5 years have been reported from southeast Greenland during summer and autumn (Kapel, 1996), while only a few Barents Sea harp seals have been recaptured west of Spitsbergen. By far, most of the tag returns from Barents Sea harp seals are from coastal waters of northern Norway (Øien and Øritsland, 1995), and a significant number of recaptures of Greenland Sea harp seals from the same areas indicate a certain overlap between juveniles of the two stocks along the coast of Norway (Øien and Øritsland, 1995). Based on reports from sealers, Sivertsen (1941) further suggested that the two Northeast Atlantic harp seal stocks share summer feeding grounds around Spitsbergen. A large number of summer sightings in this area were also reported by Haug et al. (1994), but stock identity of these animals is uncertain.
Based on the mark–recapture data, it does seem likely, however, that at least juvenile Greenland Sea harp seals may have a more western distribution than Barents Sea harp seals, also outside the breeding and molting periods, and therefore may have experienced different food availability scenarios during the study period. Unfortunately, information on harp seal diet and availability of suitable prey in the East Greenland–Iceland area is too scarce to provide a useful basis for evaluation of per-capita resource levels in this area (Hauksson and Bogasson, 1997b; Haug et al., 2000; Kapel, 2000; Potelov et al., 2000).
Stock abundance is another factor influencing per-capita resource levels, of which we have very little knowledge for Greenland Sea harp seals. An increase in stock size since the mid-1960s has been expected based on a decreasing trend in the catches (Ulltang and Øien, 1988). However, while catch per unit effort data do suggest a decrease in stock size during the period of high catches from the late 1940s to 1964, no clear trend is evident in the subsequent indices up to the early 1980s, when the CPUE series stops (Ulltang and Øien, 1988). There is also no clear trend in the mark–recapture pup production estimates covering the period 1977–1991, which range between 40 000 and 115 000 pups with large interannual variability and associated confidence intervals (Øien and Øritsland, 1995).
It may be concluded that harp seals appear to display considerable variability in age-at-maturity, which may be related to changes in per-capita resource levels. However, in the absence of reliable data of trends in abundance, seasonal distribution and feeding ecology, monitoring of age-at-maturity adds little to our understanding of harp seal ecology.
|
Acknowledgements |
|---|
We thank Drs Mike Hammill, Garry Stenson, Peter Corkeron, Kjell T. Nilssen and Ulf Lindstrøm for critical review of the manuscript.
|
Footnotes |
|---|
1 tel: +7 818 2 440366; fax: +7 818 2 440376.
2 tel: +299 32 10 95; fax: +299 32 59 57.
|
References |
|---|
|
TopIntroductionMaterial and methodsResultsDiscussionReferences |
|---|
-
Anon. (2000) Havets Ressurser 2000. Fisken og Havet, Særnummer 1/2000. 156 pp.
Bengtsson J.L. and Siniff D.B. (1981) Reproductive aspects of female crabeater seals (Lobodon carcinophagus) along the Antarctic Peninsula. Canadian Journal of Zoology 59:92–102.
Benjaminsen T. (1979) Pup production and sustainable yield of White Sea harp seals. Fiskeridirektoratets Skrifter Serie Havundersøkelser 16:551–559.
Bowen D.W., Capstick C.K., Sergeant D.E. (1981) Temporal changes in the reproductive potential of female harp seals Pagophilus groenlandicus. Canadian Journal of Fisheries and Aquatic Sciences 38:495–503.
Bowen W.D., Sergeant D.E., Øritsland T. (1983) Validation of age estimation in harp seal, Phoca groenlandica, using dentinal annulli. Canadian Journal of Fisheries and Aquatic Sciences 40:1430–1441.
Boyd I.L. (1985) Pregnancy and ovulation rates in grey seals (Halichoerus grypus) on the British coast. Journal of Zoology (London) 205:265–272.
Boyd I.L. (1991) Environmental and physiological factors controlling the reproductive cycles of pinnipeds. Canadian Journal of Zoology 69:1135–1148.
Chabot D., Stenson G.B., Cadigan N.G. (1996) Short- and long-term fluctuations in the size and condition of harp seal Phoca groenlandica in the Northwest Atlantic. Northwest Atlantic, Fisheries Organisation Scientific Council Studies 26:15–32.
DeMaster D.P. (1978) Calculation of the average age of sexual maturity in marine mammals. Journal of the Fisheries Research Board of Canada 35:912–915.[Web of Science]
DeMaster D.P. (1984) Review of techniques used to estimate the average age of sexual maturity in marine mammals. Report of the International Whaling Commission 6:175–179 (Special Issue).
Deutsch C.J., Crocker D.E., Costa D.P., Le Boeuf B.J. (1994) Sex- and age related variation in reproductive effort of northern elephant seals. In Le Boeuf B.J. and Laws R.M. (Eds.). Elephant Seals: Population Ecology, Behavior and Physiology(University of California Press, Berkeley) pp. 169–210.
Eberhardt L.L. (1977) Optimal policies for conservation of large mammals, with special reference to marine ecosystems. Environmental Conservation 4:3205–212.
Eberhardt L.L. and Siniff D.B. (1977) Population dynamics and marine mammal policies. Journal of the Fisheries Research Board of Canada 34:183–190.[Web of Science]
Fisher H. D. (1954) Studies on reproduction in the harp seal Phoca groenlandica Erxleben in the Northwest Atlantic. Fisheries Research Board of Canada. MS Report, 588. 109 pp.
Fowler C.W. (1984) Density dependence in cetacean populations. Report of the International Whaling Commission 6:373–379 (Special Issue).
Gaillard J.M., Festa-Bianchet M., Yoccoz N.G., Loison A., Toïgo C. (2000) Temporal variation in fitness components and population dynamics of large herbivores. Annual Review of Ecology and Systematics 31:367–393.[CrossRef][Web of Science]
Gambell R. (1976) World whale stocks. Mammal Review 6:41–53.
Gjøsæter H. (1995) Pelagic fish and the ecological impact of the modern fishing industry in the Barents Sea. Arctic 48:267–278.[Web of Science]
Hammill M.O., Kingsley M.C.S., Beck G.G., Smith T.S. (1995) Growth and condition in the Northwest Atlantic harp seal. Canadian Journal of Fisheries and Aquatic Sciences 52:478–488.
Hårding K.C. and Härkönen T. (1995) Estimating mean age at sexual maturity in the crabeater seal Lobodon carcinophagus. Canadian Journal of Fisheries and Aquatic Sciences 52:2347–2352.
Haug T. and Nilssen K.T. (1995) Ecological implications of harp seal Phoca groenlandica invasions in northern Norway. In Blix A.S., Walløe L., Ulltang Ø. (Eds.). Whales, Seals, Fish and Man(Elsevier, Amsterdam) pp. 545–556 720 pp.
Haug T., Krøyer A.B., Nilssen K.T., Ugland K.I., Aspholm P.E. (1991) Harp seal (Phoca groenlandica) invasions in Norwegian coastal waters: age composition and feeding habits. ICES Journal of Marine Science 48:363–371.
Haug T., Nilssen K.T., Øien N., Potelov V. (1994) Seasonal distribution of harp seals Phoca groenlandica in the Barents Sea. Polar Research 13:163–172.[Web of Science]
Haug T., Nilssen K.T., Lindblom L. (2000) First independent feeding of harp seal (Phoca groenlandica) and hooded seals (Cystophora cristata) in the Greenland Sea. In Vikingsson G.A. and Kapel F.O. (Eds.). Minke Whales, Harp and Hooded Seals: Major Predators in the North Atlantic Ecosystem(The North Atlantic Marine Mammal Commission, Tromsø) pp. 40–49 NAMMCO Scientific Publications 2 132 pp.
Hauksson E. and Bogason V. (1997) The occurrence of vagrant seals in Iceland, in 1989–94. Journal of Northwest Atlantic Fishery Science 22:47–54.
Hauksson E. and Bogason V. (1997) Comparative feeding of grey Halichoerus grypus and common seals Phoca Vitulina in coastal waters of Iceland, with a note on the diet of hooded Cystophora cristata and harp seals Phoca groenlandica. Journal of Northwest Atlantic Fishery Science 22:125–135.
Healey B. P. and Stenson G. B. (2000) Canadian Stock Assessment Secretariat. Research Document, 2000/081.
Heppell S., Caswell H., Crowder L. (2000) Life histories and elasticity patterns: perturbation analysis for species with minimal demographic data. Ecology 81:654–665.[CrossRef][Web of Science]
ICES. (2001) Report of the Joint ICES/NAFO Working Group on Harp and Hooded Seals ICES CM 2001/ACFM: 8. 40 pp.
Iverson S.J., Bowen W.D., Boness D.J., Oftedal O.T. (1993) The effect of maternal size and milk output on pup growth in grey seals (Halichoerus grypus). Physiological Zoology 66:61–88.
Kapel F.O. (1996) Recoveries in Greenland, 1949–94, of tagged or branded harp and hooded seals. Northwest Atlantic Fisheries Organisation. Scientific Council Studies 26:87–99.
Kapel F.O. (2000) Feeding habits of harp and hooded seals in Greenland waters in the Greenland Sea. In Vikingsson G.A. and Kapel F.O. (Eds.). Minke Whales, Harp and Hooded Seals: Major Predators in the North Atlantic Ecosystem(The North Atlantic Marine Mammal Commission, Tromsø) pp. 40–50 NAMMCO Scientific Publications 2, 132 pp.
Khuzin R. S. (1972) Ekologo-Morfologicheskij analiz razlichii i perspektivy promysla grenlandskogo chulenia Belomorskoi, Ian-Maienskoi i Niufaundlendskoi populjatsii (Ecological and morphological analysis of differences between the White Sea, Jan Mayen and Newfoundland harp seal populations and perspectives for their exploitation). Murmanskoe knizhnoe izdatel'stvo, Murmansk. 174 pp.
Kingsley M.C.S. and Byers T.J. (1998) Failure of reproduction in ringed seals Phoca hispida in Amundsen Gulf, Northwest Territories in 1984–1987. In Heide-Jørgensen M.P. and Lydersen C. (Eds.). Ringed Seals in the North Atlantic(The North Atlantic Marine Mammal Commission, Tromsø) pp. 197–210 NAMMCO Scientific Publications 1 273 pp.
Kjellqwist S.A., Haug T., Øritsland T. (1995) Trends in age-composition, growth and reproductive parameters of Barents Sea harp seals, Phoca groenlandica. ICES Journal of Marine Science 52:197–208.
Kovacs K.M. (1987) Maternal behaviour and early behavioural ontogeny of harp seals Phoca groenlandica. Animal Behavior 35:844–855.[CrossRef][Web of Science]
Kovacs K.M. and Lavigne D.M. (1992) Mass-transfer efficiency between hooded seal (Cystophora cristata) mothers and their pups in the Gulf of St. Lawrence. Canadian Journal of Zoology 70:1315–1320.
Laws R.M. (1956) Growth and sexual maturity in aquatic mammals. Nature (London) 178:193–194.[CrossRef]
Laws R.M. (1959) Accelerated growth in seals with special reference to the Phocidae. Norsk Hvalfangsttidende 9:425–452.
Lydersen C. and Kovacs K.M. (1996) Energetics of lactation in harp seals (Phoca groenlandica) from the Gulf of St. Lawrence, Canada. Journal of Comparative Physiology 166:295–304.
MacLaren I.A. and Smith T.G. (1985) Population ecology of seals: retrospective and prospective views. Marine Mammal Science 1:54–83.[Medline]
Nilssen K.T., Potelov V., Stasenkov V.A., Timoshenko Y.K. (1995) Food habits of harp seals (Phoca groenlandica) during lactation and moult in March–May in the southern Barents Sea and White Sea. ICES Journal of Marine Science 52:33–41.
Nilssen K.T., Haug T., Potelov V. (1997) Seasonal variation in body condition of adult Barents Sea harp Seals (Phoca groenlandica). Journal of Northwest Atlantic Fisheries Science 22:17–25.
Nilssen K.T., Pedersen O.-P., Folkow L.P., Haug T. (2000) Food consumption estimates of Barents Sea harp seals. In Vikingsson G.A. and Kapel F.O. (Eds.). Minke Whales, Harp and Hooded Seals: Major Predators in the North Atlantic Ecosystem(The North Atlantic Marine Mammal Commission, Tromsø) pp. 9–27 NAMMCO Scientific Publications 2 132 pp.
Øien N. and Øritsland T. (1995) Use of mark–recapture experiments to monitor seal populations subject to catching. In Blix A.S., Walløe L., Ulltang Ø. (Eds.). Whales, Seals, Fish and Man(Elsevier, Amsterdam) pp. 35–45 720 pp.
Øritsland T. (1971) The status of Norwegian studies of harp seals at Newfoundland pp. 185–209 ICNAF Redbook 1971, Part III.
Perry E.A., Stenson G.B., Bartlett S.E., Davidson W.S., Carr S.M., Bartlett S.E., Davidson W.S. (2000) DNA sequence analysis identifies genetically distinguishable populations of harp seals (Pagophilus groenlandicus) in the Northwest and Northeast Atlantic. Marine Biology 137:53–58.[CrossRef]
Potelov V., Nilssen K.T., Svetochev V., Haug T. (2000) Feeding habits of harp (Phoca groenlandica) and hooded seals (Cystophora cristata) during late winter, spring and early summer in the Greenland Sea. In Vikingsson G.A. and Kapel F.O. (Eds.). Minke Whales, Harp and Hooded Seals: Major Predators in the North Atlantic Ecosystem(The North Atlantic Marine Mammal Commission, Tromsø) pp. 40–50 NAMMCO Scientific Publications 2, 132 pp.
Rasmussen B. and Øritsland T. (1964) Norwegian tagging of harp and hooded seals in North Atlantic waters. Fiskeridirektoratets Skrifter 13:743–55.
Richards F.J. (1959) A flexible growth curve for empirical use. Journal of Experimental Botany 10:290–300.
Roff D.A. and Bowen D. (1983) Population dynamics of the Northwest Atlantic harp seal (Phoca groenlandica). Canadian Journal of Fisheries and Aquatic Sciences 40:919–932.
Schwartz C.I. and Stobo W.T. (2000) Estimation of juvenile survival, adult survival, and age-specific pupping probabilities for the female grey seal (Halichoerus gryprus) on Sable Island from capture–recapture data. Canadian Journal of Fisheries and Aquatic Sciences 57:1247–253.
Sergeant D.E. (1991) Harp seals, man and ice. Canadian Special Publication of Fisheries and Aquatic Sciences 114:1–153.
Sivertsen E. (1941) On the biology of the harp seal Phoca groenlandica Erxleben. Investigations carried out in the White Sea 1925–37. Hvalrådets Skrifter 26:1–164.
Sjare B., Stenson G. B., Warren W. G. (2000) Recent estimates of reproductive rates for harp seals in the Northwest Atlantic. Canadian Stock Assessment Secretariat Research Document, 2000/077. Fisheries and Oceans Canada, Ottawa (URL: http://www.dfo-smpo.gc.ca/CSAS/CSAS/English/Research_Years/2000/2000_077E.htm).
Stewart R.E.A., Stewart B.E., Lavigne D.M., Miller G.W. (1988) Fetal growth of Northwest Atlantic harp seals, Phoca groenlandica. Canadian Journal of Zoology 67:2147–2157.
Sugden L.G., Driver E.A., Kingsley M.C.S. (1981) Growth and energy consumption of captive mallards. Canadian Journal of Zoology 59:1567–1570.
Timoshenko Y.K. (1995) Harp seals as indicators of the Barents Sea ecosystem. In Blix A.S., Walløe L., Ulltang Ø. (Eds.). Whales, Seals, Fish and Man(Elsevier, Amsterdam) pp. 509–523 720 pp.
Ulltang Ø. and Øien N. (1988) Bestandsutvikling og status for grønlandssel og klappmyss. Stock development and current status of harp and hooded seals. Fiskets Gang 6:8–10.
York A. (1983) Average age at first reproduction of the northern fur seal Callorhinus ursinus. Canadian Journal of Fisheries and Aquatic Sciences 40:121–127.
CiteULike 
Connotea 
Del.icio.us What's this?
This article has been cited by other articles:
|
|
 |
|
  H. J. Skaug, L. Frimannslund, and N. I. Oien Historical population assessment of Barents Sea harp seals (Pagophilus groenlandicus) ICES J. Mar. Sci., October 1, 2007; 64(7): 1356 - 1365. [Abstract] [Full Text] [PDF]
|
|
|
|
|
|
B. A. Krafft, K. M. Kovacs, A. K. Frie, T. Haug, and C. Lydersen Growth and population parameters of ringed seals (Pusa hispida) from Svalbard, Norway, 2002-2004 ICES J. Mar. Sci., January 1, 2006; 63(6): 1136 - 1144. [Abstract] [Full Text] [PDF]
|
|
| ||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||