Field work

PBE (31°42'-31°47'N, 116^o37'-116°40'W) is located in the northwestern part of the Baja California Peninsula, in the southwestern part of Todos Santos Bay (Fig. 1). It is a coastal lagoon environment representative of the Californian ecore-gion and has been designated a Wetland of International Importance (Ramsar Site no. 1604).

Figure 1 Mouth of Punta Banda Estuary, Baja California, Mexico. Sites where the harbor seal (Phoca vitulina richardii) colony is usually found (●). 

The field work was conducted from February to April 2011 and from January to July 2012. Land-based observations were made once a week in 2011 and 3 times a week in 2012, preferably during low tide when typically more individuals are hauled out (^{Jemison and Kelly 2001}). A total of 11 counts were performed in 2011 and 77 in 2012.

Observations were made using binoculars (7 x 50), a telescope (25x), and a digital camera with telephoto zoom lens (200-400 mm).

To determine the timing of pupping, during each count we recorded the total number of individuals on land and in the water at the mouth of PBE (Fig. 1), as well as the number of pups, which are easily distinguished by their new pelage and the small size of their heads. The individuals born during the sampling year and not yet weaned were categorized as pups. Solitary pups that did not interact with an adult during the observation hours were categorized as weaned pups.

The molting process was characterized based on the molt stages described by ^{Daniel et al. (2003)}, in particular the late premolt stage (premolt C), because it was the stage that could be identified unequivocally, and the molt stages, which in the present study were grouped into one single stage. The pelage of animals in premolt C is a uniform tan or brown color and has lost the spots and rings (^{Daniel et al. 2003}). During the molt, the new, shinier hair slowly replaces the old hair until the whole body is covered. The individuals that could not be classified into one of these stages (premolt or molt) were categorized as "undetermined". We also recorded the progression of the molt, that is, the part of the body where the process begins and how the shedding of old hair progresses over the body (^{Daniel et al. 2003}). The most commonly recognized molt pattern (^{Stutz 1967}, ^{Ashwell-Erickson et al. 1986}, ^{Daniel et al. 2003}) is characterized by the appearance of new hair first on the head and flippers and then on the torso. The reverse pattern is characterized by the appearance of new hair first on the torso and lastly on the head and flippers (^{Daniel et al. 2003}).

Data analysis

The model described by ^{Rothery and McCann (1987)} was used to determine the timing of pupping. The model has been successfully used to describe the temporal distribution of elephant seal (Mirounga sp.) females and pups on land (^{Rothery and McCann 1987}, ^{García-Aguilar 2004}). The model assumes that the temporal distribution of pups follows a normal distribution, with a mean pupping date, μ (in days), and standard deviation, σ, and that the nursing period, L, is the same for all the pups. The number of pups at time t, n(t) (days), is given by:

where N _T is the total number of pups born in a single season and p(t, μ_i, σ_i , L) is the proportion of pups expected at time t. For both years studied, t = 1 and corresponds to 1 January. This proportion considers all the pups born before time t minus those that were weaned before t - L; hence, the normal proportion of p is:

where ϕ is the cumulative distribution of the standard normal distribution. This equation describes a symmetric curve around t = μ + L/2, the moment when the maximum number of pups occurs. N _T is the value of the slope obtained after performing linear regression between p(t, μ_i , σ_i , L) and the number of observed pups (^{Rothery and McCann 1987}). The values of μ_i , σ_i , and L were obtained by minimizing the sum of squares of the residual variation between the model and the 2012 data. As there were fewer data for the 2011 analysis, the duration of the nursing period estimated for 2012 was used to estimate μ and σ.

To determine molt timing and progression, the premolt C and molt data collected in 2012 were used. We calculated the relative cumulative frequency of the number of individuals in each of these stages and the data were fit to a nonlinear logistic model. The start and end dates of each stage were defined as when the proportion of individuals was >0.01 and >0.99, respectively. To identify the period with the highest proportion of individuals in the premolt C stage and molt stage, a polynomial regression was done using the Origin 8.5.1 and Curve Expert Professional 1.6.3 programs. To determine the predominant molting pattern and if there were significant differences between the percentage of individuals that presented each pattern, a Mann-Whitney U test (^{Zar 2010}) was performed, using a significance level (α) of 0.05 and STATISTICA 7.1.

To determine whether a greater number of individuals haul out during the pupping or the molting season, the percentage of the colony found on land was calculated for each count. A minimum colony size of 99 individuals was used because it was the maximum count recorded. A time series was generated of the total number of individuals and of adult and immature individuals. Because of the wide fluctuation of the data, a 14-unit moving average was obtained to analyze their trend and be able to determine during which season more than 50% of the colony was on land. The analysis was performed using STATISTICA 7.1.

Timing of pupping of harbor seal

The temporal distribution of the harbor seal pups showed a regular pattern (Fig. 2), which was satisfactorily described by the model (R² = 0.991 in 2011 and R² = 0.887 in 2012). Based on the 2012 data, nursing lasted 22.65 days. In both years (2011 and 2012), the pupping season lasted approximately 9 weeks, from mid-February to mid-April (Table 1, Fig. 2). The mean pupping date occurred in early March and the maximum number of pups occurred in mid-March (Table 1, Fig. 2). The total number of pups estimated was 34 in 2011 and 35 in 2012 (see linear regressions in Fig. 2).

Figure 2 Temporal distribution fit with the ^{Rothery and McCann (1987)} model (-) for counts (●) of the harbor seal, Phoca vitulina richardii, pups observed during the 2011 (a) and 2012 (b) pupping seasons at Punta Banda Estuary. Residuals are shown at the bottom and the regression model is shown on the upper right-hand side. Day 1 = 1 January. 

Table 1 Timing of pupping of the harbor seal, Phoca vitulina richardii, at Punta Banda Estuary in 2011 and 2012, estimated using ^{Rothery and McCann's (1987)} model. Day 1 = 1 January. The pupping season starts when the first newborn pup is observed. The end of pupping season is estimated to be when all pups have been weaned. 

Timing and progression of the molt of harbor seals

The temporal distribution of the cumulative relative frequency of the number of individuals in the premolt C and molt stages showed a satisfactory fit to the model (R ² = 0.995 for premolt C and R ² = 0.998 for the molt stage) (Fig. 3). The premolt C stage lasted 15 weeks, from March to early July (Table 2). The molt stage lasted approximately 11 weeks, from late April to mid-July. The temporal distribution of the proportion of individuals in the premolt C and molt stages was fit with a fifth-degree polynomial and a third-degree polynomial, respectively (R ² = 0.803 for premolt C and R ² = 0.737 for the molt stage). The peak proportion of individuals in premolt C occurred on 5 May (63%) and in the molt stage, on 10 June (48%).

Figure 3 Logistic model for the cummulative frequency of the number of individuals in the late premolt stage (premolt C, continuous grey line) and the molt stage (continuous black line) of the harbor seal, Phoca vitulina richardii, during 2012 at Punta Banda Estuary. The cummulative relative frequencies of the premolt C (dashed grey line) and molt (dashed black line) stages are also shown. Day 1 = 1 January. 

Table 2 Timing of the molt of the harbor seal, Phoca vitulina richardii, at Punta Banda Estuary, estimated using data obtained in 2012. Day 1 = 1 January. The end date is defined as when the proportion of individuals is >0.99. 

Significant differences were found between the 2 molt patterns (U = 37, P < 0.001). Of the individuals that could be classified per day, on average 19% showed that the head and flippers were the first areas to shed and 81% showed the reverse molt pattern, the head and flippers being the last areas to shed.

Temporal variation in the abundance of harbor seals on land

During the pupping and molting seasons, more than 50% of the total number of individuals were hauled out. In the case of adult and immature individuals, only during the molting season were more than 50% observed on land (Fig. 4).

Figure 4 Percentage of total individuals (a) and of immature and adult individuals (b) of the harbor seal, Phoca vitulina richardii, hauled out during the 2012 observation period at Punta Banda Estuary. Moving average (-); percentage of hauled-out individuals during each count (●). Day 1 = 1 January. 

Timing of pupping of harbor seals

The pupping season of P. v. richardii in PBE lasted from mid-February to mid-April. Punctual observations of pups have been made in the study area during this period of time in 1983, 1984, and 1990 (^{Padilla-Villavicencio 1990}, ^{Loya-Salinas et al. 1992}). The mean pupping date estimated for the 2011 and 2012 seasons varied by one day, reflecting the high synchronization of the annual reproductive cycle of harbor seals in specific areas (^{Temte 1991}). ^{Padilla-Villavicencio (1990)}, however, recorded 13 pups in early May 1982 at PBE. These pups may have been weaned pups or the pupping season may have been delayed because of environmental conditions, such as reduced food availability the year before the survey was conducted (^{Bohorquez 2001}, ^{Jemison and Kelly 2001}). The latter hypothesis is less likely because in 1981 the values of the Oceanic Niño Index did not indicate any anomalies (^{CPC 2015}). Two pups were also observed in late January 1984 (^{Padilla-Villavicencio 1990}), but they may have been premature pups.

Compared to more northerly locations, birthing at PBE initiated earlier, in agreement with the latitudinal cline described by ^{Temte et al. (1991)}. Based on the linear regression proposed by ^{Temte et al. (1991)}, in terms of latitude for the colonies from Mexico to Washington, the mean pupping date should be 29 March in PBE; however, it was 5 March, that is, 24 days earlier. Although it falls within the lower limit of the 95% confidence interval (27 February to 28 April), this interval is too wide. If, however, the mean pupping date for PBE (31.77°N) is compared with that for La Jolla, California (32.85°N), which was calculated using data published by ^{La Jolla Friends of the Seals (2012)}, birthing at that site in 2012 occurred 12 days earlier (22 February) than at PBE. In this case the latitudinal gradient is not apparently met, as the mean pupping date should be 4.1 days later for each degree of latitude northward (^{Temte et al. 1991}). Moreover, there seems to be a time lag between the pupping season at PBE and nearby Todos Santos Sur Island, Baja California (31.8°N), since in 2013 the pupping season at PBE lasted at least 11 days longer than on the island (unpublished data). As both sites are located practically at the same latitude and the seals are exposed to the same photoperiod, it would be reasonable to assume that the pupping season would occur at the same time at both locations. The above, therefore, seems to indicate that the PBE colony is reproductively isolated from the island's colony. Such is the case of harbor seals inhabiting Puget Sound, Washington, where differences in the pupping seasons among close colonies have been attributed to genetic differences among populations (^{Huber et al. 2010}). ^{Temte (1994)} observed that each population of harbor seals responds to a specific photoperiod that triggers the reactivation of the blastocyst and controls the moment of implantation. In the Pacific harbor seal, blastocyst implantation occurs 283 days (range: 274-291 days) before birth (^{Temte 1994}). Another factor supporting the hypothesis that the PBE colony may be reproductively isolated is that the critical photoperiod which this colony responds to (between 13.88 and 14.14 h of light per day) does not coincide with the photoperiod reported for the harbor seal from Washington to Mexico (14.3 h of light per day; ^{Temte 1994}).

The pupping season was estimated to last approximately 9 weeks, one week more than the 6-8 weeks reported for the species (^{Bigg 1981}). A total of 35 pups were estimated in 2012, which coincides with our observed data, since the maximum number of pups counted was 33, and 2 had died before the count. The number of pups estimated was similar in both years (34 in 2011 and 35 in 2012), so pup production appears to have been constant at PBE. The nursing period was estimated to last 3 weeks, less than the 4-6 weeks reported for P. v. richardii colonies in Washington (^{Newby 1973}) and southern British Columbia (^{Cottrell et al. 2002}). This can be due to the anthropogenic disturbance experienced by the PBE seals, as observed during the field work. Disturbances can increase the rate of pup abandonment (^{Jemison and Kelly 2001}), which is reflected in a shorter nursing period. Food availability before and after the pupping season may also influence the duration of the nursing period (^{Jemison and Kelly 2001}).

Timing and progression of the molt of harbor seals

The molting period of P. v. richardii at PBE occurred from the end of April to mid-July, earlier than that reported for other more northerly locations, including California (May to June; ^{Stewart and Yochem 1994}), Oregon (July to September; ^{Brown and Mate 1983}), Washington and Canada (August to October; ^{Stutz 1967}), and Alaska (July to September; ^{Daniel et al. 2003}). The molt timing appears to follow a south-to-north latitudinal cline, probably related to a response to photoperiod (^{Mo et al. 2000}) and to the timing of the breeding season, which also shows a latitudinal cline (^{Temte et al. 1991}), since in the case of females and adult males, the molt is inhibited by sexual hormones (^{Thompson and Rothery 1987}, ^{Daniel et al. 2003}). The premolt C stage occurs from late March to early July at PBE, but on Tugidak Island (Alaska) it ends in late August or early September (^{Daniel et al. 2003}), which also indicates a latitudinal cline.

The reverse molt pattern (head and flippers are the last areas to molt) predominated at PBE. At other more northerly locations of the distribution of P. v. richardii, including Canada (^{Stutz 1967}) and Alaska (^{Ashwell-Erickson et al. 1986}, ^{Daniel et al. 2003}), the molt pattern where shedding of hair occurs first on the head and flippers and progresses over the rest of the body is the most commonly recognized. In southwestern Ireland, ^{Cronin et al. (2014)} observed that a large proportion (48%) of individuals of the subspecies P. v. vitulina exhibited the reverse molt pattern. In other species of phocids, the differences in molt progression has been linked to the nutritional status of the individuals (^{Lydersen et al. 2000}). In the present study, the seals exhibiting the reverse molt pattern did not appear to have poor body condition relative to the seals exhibiting the other pattern. A possible explanation for the predominance of the reverse molt pattern at PBE can be differences in ambient temperature. ^{Mauck et al. (2003)} observed, using infrared thermography, that when the seals are hauled out, the temperature in the area of the head, neck, and hind flippers begins to rise when ambient temperatures are low (5-12 °C). Though the same occurs at high ambient temperatures, the areas of heat loss extend towards the torso and, if the individuals remain out of water long enough, the high-temperature areas on the torso exceed those on the head, neck, and hind flippers (32-36 °C; ^{Mauck et al. 2003}). As the epidermis needs high temperatures for follicular regeneration (^{Boily 1995}), the differences in the ambient temperature of the colonies may determine the progression of the molt. That is, at low latitudes, where ambient temperatures are higher, the molt initiates on the torso and progresses to the periphery (head and flippers; reverse molt pattern), but at high latitudes, where ambient temperatures are cooler, the head and flippers are the first areas to molt.

Temporal variation in the abundance of harbor seals on land

The highest numbers of P. v. richardii on land have been recorded during the pupping and molting seasons (^{Brown and Mate 1983}, ^{Jemison and Kelly 2001}). Our findings are in agreement; however, when the number of hauled-out adult and immature individuals (weaned pups, yearlings, and sub-adults) is analyzed, only during the molting season is more than 50% of the colony found on land. This indicates that the high abundance of individuals on land during the pupping season at PBE is due to the pups.

At PBE, as in southern California (^{Stewart and Yochem 1994}) and Alaska (^{Daniel et al. 2003}), the highest numbers of adult and immature P. v. richardii were recorded during the molting season. This is because individuals that are about to molt spend more time on land (^{Daniel et al. 2003}). Individuals likely haul out as a strategy to conserve energy, in view of the amount of energy required during the molt period to maintain the high temperature needed by the epidermis to carry out this process (^{Boily 1995}), and this is better achieved on land (^{Feltz and Fay 1966}). After completing the molt, the individuals once again spend more time in the water feeding (^{Stewart and Yochem 1994}), which would explain the decrease in abundance at the end of the molting season. Moreover, the highest percentage of individuals on land occurs between the peak proportion of individuals in premolt C stage and the peak proportion of individuals in the molt stage. Therefore, it would be advisable to conduct surveys aiming to estimate the abundance of this population during this period of time (between 5 May and 10 June). Counts should preferably be conducted during low tide when more seals tend to be hauled out and, in the case of PBE, when the low tide coincides with the early morning hours and there is less human presence on the beaches.

In summary, at PBE, the pupping season of P. v. richardii begins in mid-February and ends in mid-April, and the molting season begins at the end of April and ends in mid-July. The reverse molt pattern (head and flippers are the last areas to shed) predominated. The number of hauled-out adult and immature individuals was highest during the molt period; hence, the best time to perform counts and estimate the abundance of P. v. richardii at PBE is from early May to early June.