The annual reproductive cycle of most seal species is characterized by a tight synchrony of births. Typically, timing of birth shows little inter-annual variation. Here, however we show that harbour seals Phoca vitulina from the Wadden Sea (southeast North Sea) have shortened their yearly cycle, moving parturition to earlier dates since the early 1970s. Between 1974 and 2009, the birth date of harbour seals shifted on average by −0.71 d yr⁻¹, three and a half weeks (25 days) earlier, in the Dutch part of the Wadden Sea. Pup counts available for other parts of the Wadden Sea were analysed, showing a similar shift. To elucidate potential mechanism(s) for this shift in pupping phenology, possible changes in population demography, changes in maternal life-history traits and variations in environmental conditions were examined. It was deduced that the most likely mechanism was a shortening of embryonic diapause. We hypothesize that this could have been facilitated by an improved forage base, e.g. increase of small fishes, attributable to overfishing of large predator fishes and size-selective fisheries.

1. Introduction

Harbour seals are seasonal breeders. The female reproductive cycle after parturition consists of lactation followed by oestrus and mating. Placental gestation starts after an obligate period of delayed implantation (Reijnders 1986; Boyd 1991).

The annual reproductive cycle of most seal species is characterized by a tight synchrony of births, ensuring that pups are born at the optimal time of year (Boyd 1991). Typically, timing of birth in most seal species shows little inter-annual variation (Atkinson 1997). Here, we report that harbour seals Phoca vitulina from the Wadden Sea (southeast North Sea) have shortened their yearly cycle, resulting in a birth peak 25 days earlier than in the 1970s. We investigated in which phase of the reproductive cycle these changes might have occurred, and possible underlying mechanisms generating the observed changes.

2. Material and methods

The harbour seal population in the Wadden Sea (southeast North Sea), regarded as a discrete population, is monitored through a series of annual aerial surveys (Reijnders et al. 1997). From 1974 to 2009, the total number of seals and number of pups are counted throughout the whelping period. Data for harbour seals in the Dutch Wadden Sea were used to investigate the timing of pupping. Given the distribution of births over the whelping period (Reijnders et al. 1997), the date where the maximum number of pups is counted each year was chosen as a proxy for the peak timing of birth.

We fitted a model to the pup count data using generalized additive models (Wood 2006). Data collected in years 1988, 1989, 2002 and 2003 were excluded as two virus outbreaks in 1988 and 2002 caused increased variation in the counts. The following model was found to best describe the data:

where PUPS_i ∼ negative binomial, f is a smooth function combining year and julian_day and i indicates the observation.

The fitted model (M1) was used to generate predictions for the number of pups that would have been counted daily in the pupping seasons between 1974 and 2009 and to estimate the date of the maximum pup count each year. To limit the influence of estimated smoothing parameters of M1, we first generated a bootstrap sampling distribution (20 bootstraps) of the parameters to approximate the true function (Wood 2006). Each bootstrap was then used to simulate 50 replicate parameter sets from the posterior distribution of the estimated parameters producing 1000 simulations. The median Julian day at which the maximum number of pups occurred and confidence limits were calculated from the simulations. Note that for these predictions, the years around the virusepizootics were included.

All computations were calculated within the R environment (R Development Core Team 2009), v. 2.10.0, 2009, using package mgcv (Wood 2006).

3. Results

The model outcome is shown in figure 1. The observed annual pup counts enabled the estimation of the most likely date at which the number of pups reached its maximum value. The median Julian day at which maximum number of pups occurred, and the 95% confidence interval, are given in figure 2. (For count dates and variance associated with model fit, see the electronic supplementary material, figure S2.) Between 1974 and 2009, the pupping time of harbour seals in the Dutch part of the Wadden Sea shifted by −0.71 d yr⁻¹ (95% confidence limits (CL): 0.57, 0.83), which equates to a mean pupping date three and a half weeks earlier (25 days, CL 20, 29 days) by 2009. This shift was continual but not constant over the period studied (figure 2), and strongest between 1986 and 1990 (see also electronic supplementary material, figure S1, lower panel). The persistent shift since 1974 indicates that the changed phenology in pupping may have started prior to survey start. However, there are no pup counts to investigate that.

Figure 2. Dates on which maximum number of pups is estimated to have occurred. Estimates were obtained by using a combination of bootstrapping and sampling from the posterior distribution of the GAM coefficients to get 1000 simulations. From each simulation, the yearly date with maximum number of pups was calculated, using a pre-constructed prediction dataset, providing 1000 dates for each year. Horizontal dashes show median date, bars: 95% CL based on the 2.5 and 97.5 quantiles; open circles, day numbers when counts were carried out; black squares, when maximum was counted.

Pup counts available for other parts of the Wadden Sea (Schleswig Holstein and Niedersachsen) for the same period were analysed and similar shifts (−0.68 d yr⁻¹, CL 0.56, 0.82, respectively, −0.59 d yr⁻¹, CL 0.35, 1.0) were found (electronic supplementary material, figures S3 and S4).

4. Discussion

Here, we report on a continual shift in birth date of harbour seals in the Dutch Wadden Sea over decades. Given the similar results found for harbour seals in other parts of the Wadden Sea, we conclude that the factor causing this shift has acted on the entire harbour seal population.

A possible mechanism for this shift in pupping could be a shortening of one or several stages in the reproductive cycle, including lactation, delayed implantation and placental gestation. Most pinnipeds have highly synchronized annual reproduction, however, variation in the mean date of parturition may occur. Photoperiod at the time of implantation is an important factor in the timing of reproduction in seals (Temte 1994). However, there has been no change in day length in the Wadden Sea over the period of study (KNMI Database).

Body condition at the time of implantation may cause changes in the timing of births (Boyd 1984). For instance, reduced prey availability may lead to later birth and implantation, and increase the time from conception to birth (Jemison & Kelly 2001; Bowen et al. 2003). Finally, changes in age structure may also influence pupping phenology, as older females tend to give birth earlier than younger females (Reiter et al. 1981; Boyd 1996).

As a possible explanation for the shift, we examined changes in population age-structure. The two virusepizootics (1988 and 2002), killing about 50 per cent of the population, are considered to have temporarily (5–6 years) affected age structure (Härkönen et al. 1999). Given the episodic character of these events compared with the smoothness of the trend (figure 2), it appears unlikely that changes in age-structure would explain our observations. Except for the epizootics, the relative growth rate remained constant, supporting our conclusion that the change in pupping time is not driven by internal age-structure. We then investigated changes in maternal nutritional condition and related life-history traits as possible explanatory factors, starting with placental gestation. Birth mass is positively correlated with maternal mass in seals, but the duration of active gestation is only marginally influenced by maternal condition (Boyd 1991; Mellish et al. 1999). Apparently heavier mothers produce heavier pups, but in the same time span as lighter mothers. It seems unlikely that a shortened placental gestation would cause the observed shift, as the duration of placental gestation hardly varies among species (Boyd 1991). Consequently, only a shortening of either lactation period and/or period of delayed implantation could explain our observations. Duration of lactation in harbour seals was not affected during the years of reduced prey availability Bowen et al. (2003) and furthermore, in grey seals (Halichoerus grypus) lactation duration was unrelated to postpartum maternal mass (Kovacs & Lavigne 1986). Maternal mass apparently influences weaning mass but not the duration of lactation.

Implantation date in harp seals (Pagophilus groenlandicus) was, however, affected by maternal condition (Stewart et al. 1989). We therefore suggest earlier timing of implantation as the most likely mechanism explaining the observed shift. Possibly, nutritional condition of the mother after lactation and during embryonic diapause is of importance. Females undergo rapid weight loss during lactation and mating, and may need to acquire a fatness threshold prior to implantation (Stewart et al. 1989). However, during the early part of embryonic diapause, animals haul-out frequently for moulting and time for feeding is limited. In other words, the better the food acquisition during and after lactation, the earlier they regain the mass needed for implantation. Improved prey availability, may therefore accelerate the replenishment of used fat reserves, and thus facilitate a shortening of embryonic diapause.

Harbour seals consume a variety of small, generally demersal (bottom dwelling) fish species, and forage in both the Wadden Sea and North Sea (Brasseur et al. 2004). Long-term trawl surveys (1977–2001) in the North Sea revealed that abundance of small fishes including demersal species have increased steadily and significantly from 1977 till 1987 (Daan et al. 2005). Intense fishing of larger fishes in the North Sea has caused both a shift to smaller species and a decrease of large predator fishes, and hence predation on small fishes has decreased (Jennings et al. 2002). In parallel, total biomass in Dutch estuarine and coastal fishes increased from 1970 to mid-1980s (Tulp et al. 2008). It is therefore plausible that over the last three to four decades, as a result of fisheries, the prey available to seals has increased in our study area. This is supported by the increase of avian predators specializing on similar small prey (e.g. red-throated divers Gavia stellata) in our study area since 1972 (Camphuysen 2009). There is a strong correlation (Spearman's rho −0.79) between the shift in pupping and increase in small fishes (less than or equal to 20 cm) in the southeast North Sea (see also the electronic supplementary material, figure S1). We hypothesize that the observed shift in pupping phenology has been facilitated by an improved forage base as a result of overfishing predator fishes and size-selective fisheries. That improved the condition of females in the pre-implantation period and triggered a shortening of embryonic diapause. It is interesting to see whether this shift might be reversed when this exponentially growing population (Reijnders et al. 2009) approaches carrying capacity of the area.

Acknowledgements

We thank Adriaan Rijnsdorp, Amanda Lloyd and Marten Scheffer for valuable comments, Niels Daan for data on changes in fish community in the North Sea. Pup counts in Lower Saxony and Schleswig Holstein were obtained from CWSS, FTZ, Seehundaufzuchtstation Norden-Norddeich.