Spatiotemporal variation in pup abundance and preweaning survival of harbour seals ( Phoca vitulina ) in the St. Lawrence Estuary, Canada

Marine mammal populations worldwide greatly beneﬁtted from conservation measures put in place since the 1970s following overexploitation, and many pinniped populations have recovered. However, threats due to bycatch, interspeciﬁc interactions or climate change remain, and detailed knowledge on vital rates, population dynamics, and their responses to environmental changes is essential for efficient management and conservation of wild populations. In this study, we quantiﬁed pup abundance and survival of individually marked harbour seal ( Phoca vitulina Linnaeus, 1758) pups during the preweaning period at Bic Island and Métis sites in the St. Lawrence Estuary from 1998 to 2019. We used mark-recapture models to evaluate competing hypotheses regarding variation in daily preweaning survival rates and capture probability during the pups’ ﬁrst 30 days of life. Pup abundance increased from 76 [95% CI: 59, 101] to 323 [95% CI: 233, 338] in the past two decades at Bic Island and from 66 [95% CI: 47, 91] to 285 [95% CI: 204, 318] at Métis. Preweaning survival was generally higher at Bic (0.73 [95% CI: 0.58, 0.82]) than at Métis (0.68 [95% CI: 0.52, 0.79]). We hypothesize that differences between habitats and human disturbance contribute to lower preweaning survival at Métis, but behavioural studies are needed to understand the impacts of disturbance on mother–pup interactions during the nursing period.


Introduction
Identifying the factors driving changes in population growth, composition, and size is central to ecology (Krebs and Myers 1974;Krebs 1996Krebs , 2015)).Population growth depends on vital rates, including emigration, immigration, fecundity, and survival rates, which are generally affected by environmental conditions (Gamelon et al. 2014).However, understanding the effects of ecological changes on population processes is challenging because these changes are often complex as they involve density-dependent and independent factors and their interactions (Bonenfant et al. 2009;Volzke et al. 2021).
In long-living species, population growth should be most sensitive to perturbations in adult female [also named primeaged, see Loison et al. (1999)] survival (Gaillard et al. 2000;Saether et al. 2013) because this segment of the population makes the greatest contribution to population growth (Coulson et al. 2006).Thus, several demographic studies have focused on the adult component of the population (Gaillard et al. 1998b(Gaillard et al. , 2000;;Eberhardt 2002).However, due to canalisation of fitness-related traits, adult survival is typically very high and shows little inter-annual fluctuations (Gaillard et al. 2000;Gaillard and Yoccoz 2003).In contrast, juvenile survival rates are typically much lower and show larger inter-annual variability at both temporal and spatial scales (Guinness et al. 1978;Clutton-Brock et al. 1987b;Hanski 1999;Gaillard et al. 2000;Herfindal et al. 2022).This large variability in juvenile survival can have a major influence on population trend (Eberhardt 1977;Clutton-Brock et al. 1985, 1987b;Coulson et al. 1997;Fay et al. 2015).Large interannual variation in juvenile survival can be due to predation (Linnell et al. 1995), competition for resources (Clutton-Brock et al. 1987a), environmental factors such as seasonal drought or rainfall (Owen-Smith 1990), variations in maternal care (Pomeroy et al. 1999), late parturition, and genetic factors (Clutton-Brock et al. 1987b;Gaillard et al. 1998a).As such, the quantification of spatiotemporal variation in juvenile survival can represent an important tool to anticipate population trends and help identify the key variables upon which conservation and management plans should be directed.
Worldwide, pinnipeds (seals, fur seals, walrus, and sea lions) show the highest proportion of recovering populations within the marine mammal group, potentially due to their faster life histories, the protection of their haul-outs and breeding habitats, and protective management measures (Magera et al. 2013).They are long-living mammals that aggregate at terrestrial sites to which they show high fidelity (Boness and Bowen 1996;Härkönen and Harding 2001).Although pinniped species generally share similar life history strategy, females' phocids and otariids' reproductive tactics are generally assumed to be at opposite ends of the continuum between capital and income tactics (Jönsson 1997).Capital breeders, observed among phocids, build energy reserves prior to lactation, then fast while remaining close to their pups, during a shortened lactation period, lasting only days or weeks (Oftedal et al. 1987;Bowen et al. 1992;Costa 1993), and typically have high preweaning survival rates (Hall et al. 2001;McMahon and Bradshaw 2004).Otariids (i.e., eared seals), on the other hand, are income breeders; they forage during lactation, have longer lactation periods, lasting several months and they display slower pup growth (Hammill et al. 1991;Boness et al. 1994;Lydersen and Kovacs 1999).However, many phocids have developed intermediate tactics possibly due to physical or ecological constraints resulting in longer lactation durations than expected for their size (e.g., Weddell seals (Leptonychotes weddellii (Lesson, 1826)), Testa et al. 1989) or because they are unable to store energy due to their small size (e.g., harbour seals (Phoca vitulina Linnaeus, 1758), Boness et al. 1994).Detailed behavioural observations on mother-pup pairs are needed to document their reproductive tactics; however, preweaning survival is also affected by these tactics with capital breeders generally showing high pup survival until weaning (Hall et al. 2001).Thus, quantifying preweaning survival rates among pinnipeds may offer insight into the potential evolutionary advantages of these various breeding tactics.
The harbour seal Phoca vitulina is the most widely distributed pinniped, occupying a wide variety of habitats and climatic zones across the Northern Hemisphere (Liu et al. 2022).Harbour seals, with a relatively small body size, have developed a hybrid tactic of resource use for reproduction, which includes at-sea foraging behaviour, similar to Otariids, and a longer lactation duration than other phocids (Oftedal et al. 1987;Muelbert and Bowen 1993;Boness et al. 1994;Cottrell et al. 2002).Weaned harbour seal pups rely on their stored energy reserves during their transition period to independent feeding following weaning (Bowen et al. 1994).Growth during lactation is thus important to fitness in harbour seals (Bowen et al. 1994;Harding et al. 2005).
Harbour seals in eastern Canada have received less attention compared to larger and more abundant species such as the harp seal (Pagophilous groenlandicus (Erxleben, 1777)) and grey seal (Halichoerus grypus (Fabricius, 1791)) with which harbour seals spatially overlap.Both the commercially harvested harp and grey seals have shown remarkable recovery from harvesting to levels not seen in over a century (Hammill et al. 2015;Rossi et al. 2021).Harbour seal numbers were greatly reduced by hunting between the 1930s and 1970s, and as a precautionary measure, hunting was banned in 1976 (Boulva and McLaren 1979).Since then, the abundance of harbour seals has likely increased overall (Robillard et al. 2005;Waring et al. 2015).However, sharp declines in harbour seal abundance at important haul-out areas such as Sable Island have also been documented (Bowen et al. 2003a).The harbour seal and the endangered beluga (Delphinapterus leucas (Pallas, 1776)) are the only two marine mammals to occur year-round in the St. Lawrence estuary.Under a changing climate this is expected to change as the winter ice cover recedes and other species such as grey seals expand their winter distribution (den Heyer et al. 2021).Therefore, our characterisation of the population dynamics of harbours seals in the estuary is timely.
We conducted a mark-recapture study at two harbour seal haul-outs in the St. Lawrence Estuary from 1998 to 2019 to estimate pup abundance and spatiotemporal variation in preweaning survival.Expecting that harbour seals in this area have benefited from a ban on hunting since 1976, we hypothesize that pup abundance has increased over the last two decades.Previous studies have suggested that female harbour seals have adopted a breeding tactic closer to the one used by otariids (Boness et al. 1994).If this hypothesis is true, one might expect lower preweaning survival in harbour seals compared to other phocids that employ a capital breeding tactic, given higher risks of mother-pup separation when female harbour seals forage at sea.We also hypothesize that preweaning pup survival should be higher at an uninhabited offshore site comprising numerous interconnected islands (Bic Island) than at a smaller coastal site located near summer cottages and potentially of lower quality (Métis).Finally, we tested for density-dependence effects by quantifying the correlation between pup abundance and preweaning survival.

Study site and data collection
Harbour seals were monitored in all years from 1998 to 2019, with the exception of 2004-2007 and 2017-2018, at two colonies in the St. Lawrence estuary (Bic Island and Métis area).The Bic Island colony (hereafter, Bic; 48 • 24 N, 68 • 51 ), which is closer to deeper water (Lesage et al. 2004), is situated 70 km to the southwest of the Métis colony (48 • 41 N, 68 • 01 W) on the south shore of the estuary (Fig. 1).In this area, harbour seals generally remain <11.0 km from shore in waters <50 m deep.The median date of pupping is 25-28 May, lactation lasts for 18 to 32 days (Dubé et al. 2003), and the median date of weaning is ∼25 June to 1 July (Dubé et al. 2003;Van de Walle 2013).Sites were alternately visited using small boats each study year and the number of days of surveys varied among sites and years due to weather (Supplementary Table S1).Each year, pups were captured from the end of May to early July.The captures were conducted by one boat catching a swimming or hauled-out pup using a 1 m diameter dip net attached to a 2 m pole (Dubé et al. 2003).Once captured, pups were sexed and individually marked with a Can.J. Zool.Downloaded from cdnsciencepub.com by UNIVERSITY OF ARIZONA LIBRARY on 07/03/24 coloured and numbered head tag (Seal Hat , Dalton, England; Hall et al. 2001) glued to the fur and a colour-coded and uniquely numbered hind flipper tag (Jumbotag , Dalton, England; Hall et al. 2001).Pups were subsequently recaptured, and their tag numbers recorded as often as possible until weaning, allowing a minimum of 2 days between captures.Each pup was weighed to the nearest 0.5 kg at each capture (Salter Scale, West Bromwich, England).Sightings without capture were rare and were not used in this study.Capture and handling of animals followed the Canadian Council on Animal Care guidelines and were approved by Fisheries and Oceans Canada and Université de Sherbrooke (permits numbers: IML 2019-034, UdeS 2020-2423).

Mark-recapture models
We first used Cormack-Jolly-Seber (CJS) models to estimate "apparent" preweaning survival (i.e., the probability that a pup survived and remained within the study area before weaning) and capture probability between two successive capture occasions (Cormack 1964;Jolly 1965;Seber 1965).We developed a priori models that included two parameters: apparent daily survival rate (ϕ) and capture probability (p), where ϕ was defined as the probability that a pup alive in occasion t remains available for resighting until the next occasion t + 1 (i.e., survives and remains in the study area), and p t was defined as the probability that a pup marked alive and present in the study area was captured at occasion t (Table 1).Given the year effect on both ϕ and p of the best CJS model and computational constraints, we derived pup abundance, N, using the "superpopulation" formulation of the Jolly-Seber model (JS) with the same structure as the CJS for each year separately (Jolly 1965;Seber 1965;Schwarz and Arnason 1996).JS models take into account the probability of capture of each individual and estimate abundance based on the subsequent capture histories (Kéry 2008;Kéry and Schaub 2011).Because they do not condition on first capture (i.e., capture histories include 0s before the first capture), JS models allow the estimation of gains from births and losses (e.g., due to weaning and death in our case) to the population (Kéry and Schaub 2011).

Preweaning survival analysis: Cormack-Jolly-Seber model
Each individual entered the study when it was first captured and marked.As pups were not all captured at the same time, pups entered at different ages.We defined the preweaning period as 0 to 30 days of age (Dubé et al. 2003), and we focused on daily survival until age 30 days.Previous studies report that weaning occurs at ∼30 days of age (range of 18-32 days).Capture probability also declines after 30 days (Dubé et al. 2003).Thus, we allowed capture probability to vary at each sampling occasion in the models until 30 days of age.Difficult sea conditions prevented daily surveys, resulting in unequal intervals between capture occasions among sites and years.We accounted for unequal time intervals between capture occasions by including "dummy" sampling occasions with no observations for any pups and a fixed sighting probability of 0 (Sanz-Aguilar et al. 2019).We used estimates of daily apparent survival probabilities from the most supported model to Can.J. Zool.Downloaded from cdnsciencepub.com by UNIVERSITY OF ARIZONA LIBRARY on 07/03/24 estimate the probability of surviving from age 1 to age 30 days (ϕ 30 ) and we derived the uncertainty of ϕ 30 directly from posterior distributions.After day 30, we fixed the capture probability to be constant across sampling occasions.Ages of some pups (6% of all pups) were known because they were first captured with visible, not yet dried, pink umbilical cords and were likely 1-day old (Dubé et al. 2003).However, ages of most pups were uncertain and instead, their approximate birth date was back-calculated based on pup mass at capture, assuming a linear growth rate for all individuals in a given site-year and a birth mass of 10 kg, and the additional constraint that birth had to occur before the first capture (Dubé et al. 2003).These birth date estimates were then used to approximate each pup age at first capture and therefore allowed us to determine, for each of them, whether they were first captured before or after weaning (i.e., 30 days of age).The sex of unsexed individuals (n = 7) was drawn from a Bernoulli distribution with a probability of 0.5.

Pup abundance: Jolly-Seber models
The JS superpopulation formulation uses entry probabilities, b, which are the probabilities that an individual enters the population at time t, and capture probabilities p to model the observation process (e.g., seen, not seen; Schwarz and Arnason 1996).In JS models, p is the probability of capture of both unmarked and marked animals that are in the population at time t and ϕ refers to the survival probabilities of both marked and unmarked animals between each capture date (Schwarz and Arnason 1996).To estimate pup abundance, we added hypothetical "unmarked" individuals to the data set.This technique, called data augmentation, is used to account for individuals that were never observed during a study and for potential capture heterogeneity that may affect the capture and recapture rates (Royle and Dorazio 2012).The superpopulation formulation formally admits the zero inflation of the augmented data set to model capture probability (Kéry and Schaub 2011).JS models assume that (1) all marked and unmarked individuals present in the population at time t have the same capture probability, (2) all marked animals in the population immediately after time t have the same probability of surviving to time t + 1, (3) marks are not lost or missed, and (4) sampling is instantaneous, relative to the interval between time t and t + 1, and each release is made immediately after sampling (Jolly 1965;Seber 1965).

Model fitting and selection
Sampling occasions were at-sea capture occasions, each corresponding to a date (Supplementary Table S1).We first fitted 13 different CJS models (see Table 1 for full model description).We ran the models on a dataset of pooled years and sites, testing effects of combinations of site, year and sex, and date and their interactions on capture and survival probabilities.We used the Watanabe Akaike Information Criterion (WAIC; Vehtari et al. 2017) and selected the "best" model based on its lowest WAIC to extract preweaning survival and capture probability estimates.Then we used the parameter structure of the selected CJS and extended it into an annual JS model to derive abundance from p, ϕ, and entry probability, b.We used weakly informative prior distributions for all candidate model parameters.For each analysis, 50 000 iterations of three independent chains were used to obtain samples from the joint posterior distribution, with a thinning of 40 and a burn-in of 10 000 iterations, resulting in posterior distributions of parameters comprising 1000 iterations per chain.We report the mean and 95% credible intervals across iterations unless specified otherwise.We assessed model convergence both visually and using the Gelman-Rubin statistics and assumed all parameters converged when the Gelman-Rubin statistic was <1.1 (Gelman and Rubin 1992).Finally, we investigated if preweaning survival in year t was correlated Can.J. Zool.Downloaded from cdnsciencepub.com by UNIVERSITY OF ARIZONA LIBRARY on 07/03/24 with pup abundance the same year or with the derived "population growth rate," calculated as N t+1 /N t , where N was the estimated pup abundance.We extracted mean correlations between survival estimates, pup abundance, and pup growth rate for each of the 1000 iterations per chain.We conducted all analyses using the NIMBLE library [v.0.12.

Results
A total of 1369 pups were captured at Bic and Métis (641 females, 728 males), for a total of 2380 first captures and recaptures over 317 sampling days (Supplementary Table S1).From 6 to 106 pups were marked each year-site.On average, for the two sites combined, ∼42% of pups were recaptured at least once, with this proportion being generally higher at Bic than at Métis (Supplementary Table S1).Individuals were recaptured on average 1.7 times (range: 1-8).Captures occurred between 20 May and 10 July.
All parameters of CJS models converged based on the Gelman-Rubin statistics (Gelman and Rubin 1992).The selected CJS models included fixed year and site effects on capture probability (model 3, Table 1, Supplementary Fig. S1) and random year and fixed site effects on pup preweaning survival (model 9, Table 1).The latter model suggested that mean preweaning survival was slightly higher at Bic (0.73 [95% CI: 0.58, 0.82]) than at Métis (0.68 [0.52, 0.79]), perhaps because of a lower preweaning survival at Metis during the second half of the study (site difference: −0.05 [−0.18, 0.07], Fig. 2).On average, survival rates between occasions were high (>0.93,Fig. 2).Except for 5 years (in 2019 at Bic and 2009Bic and , 2012Bic and , 2013Bic and , 2015 at Métis), mean preweaning survival was >0.70.Uncertainty of estimates was higher in years of low capture effort or when the number of recaptures was low (e.g., 2019 at Bic and 2013 at Métis, Supplementary Table S1).
Overall, pup abundance was higher at Bic than at Métis (Fig. 3).At both sites, pup abundance increased linearly over years by 9.6 [8.3, 11.2] pup/year at Bic and 8.3 [7.0,9.8]pup/year at Métis.Preweaning survival rates were not correlated with either pup abundance at time t or pup production growth rate at Bic and Métis.

Discussion
We quantified temporal and spatial variation in pup abundance and preweaning survival of harbour seals at two locations in the St. Lawrence Estuary using capture-recapture data.We detected an increase in pup abundance over the past two decades at both sites.Substantial inter-annual variability in preweaning survival was found especially at Métis.Preweaning survival at Métis was lower on average and showed higher uncertainty around point estimates than at Bic, but our estimates were within the range of other harbour seal populations (Hastings and Testa 1998;Härkönen et al. 2002).Preweaning survival was not correlated with pup abundance or pup production growth rate in either site or over years, which suggests that covariance between these variables was weak.
Can. J. Zool.Downloaded from cdnsciencepub.com by UNIVERSITY OF ARIZONA LIBRARY on 07/03/24  , 1998-2019, at Bic Island andMétis, Québec, Canada.Estimates are from an annual model with year and site effects on capture probability and preweaning survival.Bars represent 95% credible intervals around posterior means.Data were not available for years 2004-2007 and 2017-2018.The preweaning survival rate at Bic, where abundance has increased, was on average 0.73 [95% CI: 0.58, 0.82], while preweaning survival at Metis averaged 0.68 [95% CI: 0.52, 0.79].These survival estimates should be considered as minimum estimates due to heterogeneity in sampling effort, imperfect detection, and to potential dispersal between sites.In our system, capture probability varied from day to day depending on births and weather.This particularity of the study system has been accounted for by including time variation in entry probability (due to births) and capture probability in the modeling process.Also, in addition to the capturemark-recapture program, other research projects with different goals were conducted in some years at our study sites (e.g., Sauvé et al. 2014Sauvé et al. , 2015)).For these projects, some pups might have been targeted more than others, leading to a permanent trap response in some years, i.e., a different capture probability between the first and subsequent captures, which could not be dealt with using a Cormack-Jolly-Seber model (Kéry and Schaub 2011).We thus acknowledge that uncertainty remains around our survival estimates.
We detected small differences in preweaning survival between the two study sites, with higher survival observed at Bic than at Métis.The Bic site encompasses a large uninhabited island (area = 40 km 2 ), located approximately 9 km offshore from a provincial park and is surrounded by multiple reefs that can serve as potential haul-out sites and by water depths <50 m (Fig. 1), which are preferred by harbour seals in this area (Lesage et al. 2004).The haul-out sites there are also suspected to be closer to local upwelling areas and thus closer to concentrations of prey (Sourisseau et al. 2006), and the presence of multiple haul-out sites provides opportunities for mother-pup pairs to seek shelter from inclement weather conditions or avoid local disturbance.As a result, it is possible that females stay closer to the haul-out site to forage during the overall lactation period, explaining lower risks of abandonment and higher pup survival at Bic.At the Me´tis site, 40 and 60 m isobaths are much closer to shore; seals haul out mainly at two to three reefs that are situated about 200 m offshore in a developed area summer house.This site is also located further from deep and productive waters.Prey density at Me´tis was found to be relatively low (Bérubé and Lambert 1999), but recent studies on fine-scale prey selection by harbour seals in the St. Lawrence Estuary are lacking.Importantly, pups at Me´tis are more frequently disturbed by humans, but have only a few alternative areas to avoid local disturbance or to seek shelter from storms (Henry and Hammill 2001).Females at Métis might need to undergo longer foraging trips (alone or with their pups) to satisfy their energy needs, which might result in higher abandonment risks and lower pup survival at Me´tis compared to Bic.In turn, higher survival at Bic might be due to milder ecological conditions, less difficulties for the pup to maintain contact with the female for feeding, and lower risks of abandonment by the female.Other factors related to mother-pup interactions during the nursing period could also explain site differences in preweaning survival.Field observations also suggest that human activities, including the presence of kayaks, individuals walking their dogs along the beach, and close-range human observation may contribute to an elevated disturbance level at the Métis site (Henry and Hammill 2001).Behavioural Can.J. Zool.Downloaded from cdnsciencepub.com by UNIVERSITY OF ARIZONA LIBRARY on 07/03/24 studies would be needed to gain a greater understanding of the potential consequences of disturbance on mother-pup interactions and help explain site differences in harbour seal pup preweaning survival at the Bic and Métis sites.
Preweaning survival rates from this study, especially at Bic, are comparable to estimates from other stable or increasing harbour seal populations (range of estimates: 0.35-0.9;Härkönen et al. 2002;Hastings et al. 2012).They are also similar to those of other phocids adopting a "typical" capital breeding tactic, with preweaning survival rates often lying between ∼70% and 90% (Twiss et al. 2003).Owing to their small size, harbour seal females are unable to store sufficient energy to support lactation, without supplementing stored reserves with feeding during lactation.This strategy of females feeding at sea during lactation more closely resembles an "income" tactic, associated with Otarids (Costa 1991;Boness et al. 1994;Schreer et al. 1996).It is thus difficult to conclude that reproductive tactic leads to higher preweaning survival in harbour seals than in species using a typical capital breeding tactic.Early in lactation, the pup is either left alone or with other pups at the haul-out site (Boness et al. 1992(Boness et al. , 1994)).Solitary pups may be vulnerable to terrestrial or avian predators.However, after approximately 1 to 2 weeks, the swimming and diving skills of pups improve considerably and they can follow the female as she forages (e.g., Greaves et al. 2004Greaves et al. , 2005;;Lapierre et al. 2004;Clark et al. 2006Clark et al. , 2007;;Prewitt et al. 2010), potentially reducing exposure to predators and mortality.Separation of the pup from the female remains an important source of preweaning mortality (Boulva and McLaren 1979) but may be offset to some extent by pups learning to acquire solid food prior to weaning (Sauvé et al. 2014).We hypothesize that the relatively high preweaning survival estimates reported here confer evolutionary benefits to an intermediate, mobile lactation tactic and explain population growth despite relatively high inter-annual variations in preweaning mortality during the period of maternal care.
We estimated an average annual rate of increase in pup production of ∼15% [95% CI: 12, 18] at Bic and ∼18% [14, 25] at Métis.These rates are generally higher but not significantly greater than the theoretical maximum rate of increase for phocids of 12% (Härkönen et al. 2002) and suggest that the increasing numbers may also include some immigration.If pup production reflects population abundance, our results suggest an increasing population trend over the period 1998-2019.The increase in pup abundance estimated by our models is consistent with field observations from aerial surveys of hauled out animals from Bic and Métis in June 1995June , 1996June , and 2000 during which a total of 59, 63, and 71 individuals (including adults) were reported near Métis and 109, 105, and a total of 128 individuals were reported in the Bic archipelago (Robillard et al. 2005).In 2019, aerial surveys led to the estimation of ∼233 individuals at Métis and ∼372 individuals counted at Bic Island (Mosnier et al. 2023).Overall, our analyses and field observations confirm that the St. Lawrence harbour seal population is increasing.This trend contrasts with the decline in harbour seal pup abundance documented on Sable Island since 1989 and in the northeastern United States between 2012 and 2018 (Bowen et al. 2003a;Waring et al. 2015;Sigourney et al. 2022).There, harbour seal declines were attributed to increased competition from increasing grey seal populations (Pace et al. 2019;Bowen et al. 2003aBowen et al. ,2003b;;den Heyer et al. 2021) and to predation by white sharks at Sable Island (Lucas and Stobo 2000;Bowen et al. 2003a), but evidence of interspecific competition or predation is yet to be shown for the St. Lawrence Estuary during the specific period of maternal care.Lack of data after the weaning period limits our understanding of the contributions of pre-and post-weaning survival to this harbour seal population dynamics.Data on the timing of overlap between grey and harbour seals within seasons are also lacking and would help quantify interspecific competition and predation pressure on unweaned and weaned pups in the St. Lawrence Estuary.However, increased seal pup abundance in this harbour seal population might be explained by the relatively low cooccurrence with grey seals during a substantial part of the year, although the exact timing of overlap is still unknown.
Our findings highlight inter-annual variations in both harbour seal pup abundance and preweaning survival.At the Bic site, pup abundance increased at the beginning and end of the study period and was relatively stable from 2003 to 2013.This not only points to the need for behavioural studies to address disturbance effects, but also suggests that potential environmental changes might be worth investigating.While it is difficult to explain the causes of inter-annual variation in our estimates of preweaning survival in the context of this study, capture effort explains part of the inter-annual variability in preweaning survival estimates.Females could not be captured and maternal attributes were not quantified in this study.Nevertheless, it is possible that inter-annual variations and site differences in preweaning survival are due to changes in maternal attributes (e.g., body mass, breeding experience) that are known to influence milk production and provisioning (Bowen et al. 2001a(Bowen et al. , 2001b) ) or to changes in attributes of pups exposed to different physical conditions each year at each site.Integrating environmental or habitat variables would also deserve future investigation given that substantial changes were recently detected in physical oceanographic conditions in the Gulf of St. Lawrence (Galbraith et al. 2020(Galbraith et al. , 2022)).
This study provides the fundamental knowledge to better inform the conservation and management of harbour seals.Closely monitoring the impact of large and fine-scale environmental fluctuations on harbour seals' demography in the St. Lawrence Estuary is the most important challenge that lies ahead.While predation might be of lower magnitude in the Estuary compared to the Gulf or off the Scotian Shelf as it relates to shark-inflicted mortality occurring during lactation (Lucas and Stobo 2000), Greenland sharks are present in the Estuary (Stokesbury et al. 2005) and have been found to prey on seals, including harbour seal, elsewhere (Lucas and Natanson 2010;Leclerc et al. 2012).The abundance of grey seals, in turn, is closely monitored and is increasing in the southern part of the Estuary and Gulf of St. Lawrence (Swain et al. 2019;Rossi et al. 2021).Given the high probability of trophic niche overlap between harbour and grey seals in the North Atlantic (Lesage et al. 2001;Planque et al. 2021), competition for prey or haul-out sites might impact the long-term population dynamics of harbour seals in the St. Lawrence Estuary.In addi-Can.J. Zool.Downloaded from cdnsciencepub.com by UNIVERSITY OF ARIZONA LIBRARY on 07/03/24 tion, grey seal has been reported to prey on harbour seals (van Neer et al. 2015).Harbour seals can be considered as coastal ecosystem sentinels given their sensitivity to anthropogenic and environmental disturbance [e.g., contaminants (Lake et al. 1995)].Continued monitoring of trends in abundance and demographic parameters should provide insights into longerterm impacts of environmental and anthropogenic changes.

Fig. 2 .
Fig. 2. Preweaning survival of harbour seal (Phoca vitulina) pups in the St. Lawrence Estuary, 1998-2019, at Bic Island and Métis,  Québec, Canada.Estimates are from a model with year and site effects on capture probability and preweaning survival.Bars represent 95% credible intervals around posterior means.Data were not available for years2004-2007 and 2017-2018.
0 (de Valpine et al. 2017, 2022)] and R [v. 4.0.4,(R Core Team 2021)].Codes to run the analyses are available in the Online Supplementary Material.

Fig. 3 .
Fig. 3. Harbour seal (Phoca vitulina) pup abundance in the St. Lawrence Estuary, 1998-2019, at Bic Island and Métis, Québec,  Canada.Estimates are from an annual model with year and site effects on capture probability and preweaning survival.Bars represent 95% credible intervals around posterior means.Data were not available for years2004-2007 and 2017-2018.
WAICModel selection for capture probability, p. p (.) φ (.)A constant capture probability p and survival probability ϕ. 1 21 017 p (yr+site) φ (.) Fixed and additive effects of year and site on p.

(yr * site) φ (.) Fixed and interactive effects year and site effects on p.
Model selection is based on Watanabe Akaike Information Criterion (WAIC) values, and the model with the lowest WAIC was extended into a Jolly-Seber model to model pup abundance.Data were not available for years2004-2007 and 2017-2018.Selected models are in bold.