Local recruitment of humpback whales in Glacier Bay and Icy Strait , Alaska , over 30 years

We provide new information on the scale at which fidelity and recruitment underlie observed increases in humpback whale Megaptera novaeangliae populations. We used photoidentification records and DNA profiles from whales in Glacier Bay and Icy Strait (GBIS), southeastern Alaska (SEAK) to investigate 3 sources of population increase over 33 yr (1973−2005): local GBIS recruitment, recruitment from elsewhere in SEAK, and immigration from outside SEAK. We defined 2 temporal strata for these longitudinal records: ‘founder’ individuals identified from 1973 to 1985 (n = 74; n = 46 with DNA profiles) and ‘contemporary’ individuals identified from 2004 to 2005 (n = 171; n = 118 with DNA profiles). To distinguish between local recruitment and recruitment from elsewhere in SEAK, we estimated the proportion of the contemporary stratum that was either a returning founder or descended from a founder female. After excluding 42 contemporary whales without a known mother or genotype to infer maternity, 73.6% of the contemporary stratum was confirmed or inferred through parentage analysis to be either a returning founder or a descendant of a founder mother. Of the 25 females with genotypes in the founder stratum, 24 (96%) were either represented in the contemporary stratum, had at least 1 descendant in the contemporary stratum, or both. We found no significant differences in microsatellite allele or mtDNA frequencies between the strata, suggesting little or no immigration from other feeding grounds. Our results highlight the importance of local habitat protection for a recovering species with culturally inherited migratory destinations.


INTRODUCTION
Humpback whales Megaptera novaeangliae in the North Pacific, first protected by the International Whaling Commission in 1965, were listed under the US Endangered Species Act in 1970 (Johnson & Wolman 1984).Pre-exploitation abundance of North Pacific humpback whales based on catch data has been approximated at about 15 000 prior to 1905 and is thought to have decreased to less than 1000 by 1966 (Rice 1978).This oceanic population has now increased to an estimated abundance of over 21 000 individuals, as estimated using capture−recapture methods with fluke identification photographs collected from 2004 to 2006 (Barlow et al. 2011).
However, this oceanic recovery in the North Pacific humpback whale, and that of other species of exploited whales, is often variable on a local spatial and generational temporal scale (Baker et al. 1993, Best 1993).While some humpback whale populations have shown little sign of recovery since modern whaling (e.g.Moore et al. 1999), others have in creased markedly, as demonstrated in southeastern Alaska (SEAK), a North Pacific feeding ground (Hendrix et al. 2012).
Reasons for spatial and temporal variability in recovery among humpback whale populations are, in most cases, unclear.For humpback whale populations that have overcome extreme depletion from exploitation, recovery can occur through immigration, repopulation from a remnant stock driven by local fidelity and recruitment, or some combination of both (Clapham et al. 2008).
Humpback whales in the North Pacific undertake annual migrations from high-latitude feeding regions along the Pacific Rim to tropical or sub-tropical breeding grounds around Central America, Mexico, Hawaii, and the Ryukyu and Ogasawara Islands (Johnson & Wolman 1984).Fidelity to feeding and breeding grounds has been well documented by photo-identification (photo-ID), using unique mark-ings on the underside of a whale's flukes to identify an individual (Ju rasz & Palmer 1981).Humpback whale fidelity to a feeding ground is maternally directed; a calf will mig rate with its mother to a specific feeding ground and then independently return there in subsequent years as an adult (Martin et al. 1984, Baker et al. 1986, Clapham & Mayo 1987, Straley et al. 1993).Individual fidelity is also reflected in patterns of diversity in the maternally inherited mitochondrial (mt) DNA (Baker et al. 1990(Baker et al. , 2013)).Twentyeight mtDNA haplotypes were resolved in humpback whales throughout the North Pacific in an oceanwide survey of genetic diversity and analysis of population structure in the North Pacific humpback whales (Structure of Populations, Levels of Abundance, and Status of Humpbacks, SPLASH).These 28 mtDNA haplotypes showed marked frequency differences among feeding grounds and among breeding grounds (Baker et al. 2013).Within SEAK, the diversity of mtDNA was notably low, with the large majority of individuals having either haplotype A+ (27%) or A− (66%), suggesting strong fide lity of these 2 maternal lineages to SEAK (Baker et al. 2013).
Site fidelity also influences local habitat use on a fine-scale in SEAK (Hendrix et al. 2012).Using an open-population capture−recapture model based on photo-ID, Hendrix et al. (2012) found that the prob ability of whales returning to their original sighting location year to year (either Glacier Bay/Icy Strait, Sitka Sound, or Frederick Sound/Lynn Canal) was ≥0.75 throughout the study (1994−2008).Movement between feeding regions (both within and between seasons) is relatively infrequent and when it does occur, most transitions take place between adjacent feeding grounds (Witteveen et al. 2011).
In this study, we examined local fidelity, recruitment, and potential immigration on a decadal scale in Glacier Bay and Icy Strait (GBIS), a northern sub-region within the SEAK feeding ground (Fig. 1).Rapid environmental and anthropogenic changes have occurred in GBIS over the past 200 yr, thus providing an opportunity to evaluate recovery from ex ploitation within a changing environment.Glacier Bay was entirely glaciated in 1794 when the English explorer George Vancouver passed through Icy Strait (Bohn 1967).By 1879, naturalist John Muir found that the glacial ice had retreated a distance of 77 km (Muir 1915).Sightings of humpback whales near the mouth of Glacier Bay were first reported in notes from the Harriman Expedition in 1899, and it is believed that humpback whales reached the middle of Glacier Bay as recently as 1938 (Vequist & Baker 1987).Today, the glaciers have retreated more than 90 km from the mouth of Glacier Bay, creating a productive marine ecosystem and suitable summer feeding habitat for humpback whales.Glacier Bay was set aside as a national monument in 1925 and be came a national park and preserve in 1980.To manage the continuing demand for motorized vessel ac cess to the spectacular wilderness, wildlife, and scenic beauty of Glacier Bay, the US National Park Service (NPS) has limited the number of vessel entries and maintained a system of whale approach regulations, vessel speeds, and course restrictions designed to reduce acoustic disturbance and the risk of whale−vessel collisions.These regulations fulfill the mandate of the US Endangered Species Act.In addition, entanglement in fishing gear affects this population in park waters and beyond.When the whales are elsewhere in SEAK or on their wintering grounds in Hawaii, the National Marine Fisheries Service manages human impacts with a 100 yard (91.44 m) minimum approach distance.
Research on humpback whales in SEAK began in the late 1960s, a few years after the cessation of commercial whaling.In one of the first uses of photo-ID, Jurasz & Palmer (1981) began identifying individual humpback whales based on stable black and white pigment patterns on the underside of the tail flukes.The first surveys of humpback whales in Glacier Bay date back to 1973 (Jurasz & Palmer 1981, Vequist & Baker 1987) and produced a catalog of individual whale flukes that allowed tracking of individuals up through the present day.The US NPS began making standardized annual counts of humpback whales in GBIS based on photo-ID surveys in 1985 (e.g.Neilson et al. 2015).This long-term research has been augmented with a substantial increase in photo-ID and genetic sampling efforts throughout SEAK during the collaborative SPLASH program (2004SPLASH program ( −2005)).
Current work by the NPS brings together data from many sources to thoroughly document the GBIS population, making it a unique model for understanding the mechanisms of local population recovery.Longitudinal life history data on humpback whales in GBIS from 1973 to the present has resulted in a large, detailed database of information on the occurrence, distribution, known ages, and mother/offspring rela-tionships of individual whales that have continued to visit GBIS for up to 4 decades (Southeast Alaska Regional Database; Neilson et al. 2013).Based on a robust design, capture−recapture model of the NPS's longitudinal photo-ID database (Saracco et al. 2013), the GBIS population has increased annually by 4.4% in recent years (1985−2009).Genetic sampling has been conducted in GBIS since 1987 (e.g.Baker et al. 1990, Neilson et al. 2014), resulting in DNA profiles for hundreds of whales, many of which are also known from photo-ID (mtDNA control region, microsatellite genotypes, and sex; Baker et al. 2013).
Here, we integrated photo-ID records with DNA profiles to examine changes in the local GBIS population over 33 yr (1973−2005).First, we defined 2 temporal strata of humpback whales in GBIS separated by 20 yr, a period that represents appro ximately 1 generation for humpback whales (Taylor et al. 2007): (1) 'founder' individuals encountered early in the study of humpback whales in GBIS and (2) 'contemporary' individuals encountered during SPLASH.We considered 3 sources of increase: local recruitment, recruitment from elsewhere in SEAK, and immigration.We used photo-ID and DNA profiles to estimate the proportion of the contemporary stratum that was either a returning founder or des cended from a founder female.To evaluate the ex tent of immigration from outside the SEAK feeding ground, we investigated temporal changes in mtDNA haplotype diversity and microsatellite differentiation between the 2 strata.Lastly, we used a paternity analysis to investigate whether founder males contributed through paternity with founder females, as evidence for breeding on the feeding grounds or during migration.

Photo-ID
Photo-ID records of humpback whales have been collected in SEAK since the late 1960s by various research groups, including Sea Search, Ltd. (1968− 1981), J. Straley Investigations and the University of Alaska Southeast (UAS, 1979−present), the University of Hawaii (1980Hawaii ( −1984)), the NPS Humpback Whale Monitoring Program (1985−present), and as part of the international, basin-wide, SPLASH study (2004−2005), which also included tissue sampling (Calambokidis et al. 2008).Rigorous stewardship of these data by the NPS and UAS has resulted in one of the longest-running photo-ID monitoring efforts of a whale species anywhere in the world (e.g.Neilson et al. 2014).This effort has resulted in the SEAK Regional Database, which contains encounter histories for approximately 3500 individual humpback whales in SEAK from 1973 to present.Individual whales that are sighted more than once are assigned a sequential SEAK identification code (SEAK ID).Many maternal relationships and ages inferred for individual whales from photo-ID observations of females with dependent calves are available within this collection (e.g.Gabriele et al. 2007, Neilson et al. 2014).Hereafter, the SEAK Regional Database will be referred to simply as the 'database'.

Genetic samples and DNA profiles
Genetic samples used in this study were extracted from a comprehensive collection of 1026 SEAK tissue samples (692 with an associated SEAK ID), representing 25 yr of genetic sampling.Most samples associated with a SEAK ID were collected using a small biopsy dart (Lambertsen 1987).The majority of tissue samples used in this study were collected in SEAK during the SPLASH project by UAS (J.Straley) from 2004 to 2005.Additional samples were collected by the NPS and UAS pre-SPLASH since 1987, when tissue samples were first collected in association with photo-ID, and post-SPLASH to 2012.DNA profiling (500 bp of the mtDNA control region, 10 microsatellite loci, and sex) followed methods described previously for SPLASH (Baker et al. 2013).

Defining 'founder' and 'contemporary' users of GBIS
The spatial limits of the study area in GBIS are outlined in Fig. 1.For whales encountered within these limits, we defined 2 temporal strata to investigate changes in population structure over time.The 'founder' stratum consisted of all individuals photo-identified in GBIS from 1973 to 1985.This time period was chosen because it starts with the first year that photo-ID was conducted in GBIS and ends with the first year the NPS began systematic annual surveys in GBIS.Whales in the 'contemporary' stratum included all individuals that the NPS or UAS identified within the GBIS study area during the SPLASH program (2004 and 2005), except dependent calves (i.e.unless dependent calves in 2004 were re-sighted in 2005 without their mother).This period coincided with 2 yr of extensive sampling during the SPLASH program.A small number of individuals in the contemporary stratum (n = 8) were identified during 2004 or 2005 by the distinctive shape or markings on their dorsal fins.These 'dorsal fin identifications' (Blackmer et al. 2000) were reconciled with encounters where matching dorsal fin photos and a fluke ID were obtained.

Genetic diversity and temporal differentiation
The DNA profiles provided information on sex, mtDNA haplotypes (as defined by Baker et al. 2013), and 10 microsatellite loci.The program GenAlex v6.5 (Peakall & Smouse 2006) was used to calculate the number of alleles (K ), the observed and expected heterozygosity (H o and H e ), the deviation from Hardy-Weinberg equilibrium (HWE), and the probability of identity (P (ID) − the probability that 2 individuals drawn at random will have the same genotype; and P (ID)SIBS − the probability that 2 siblings drawn at random will have the same genotype) for each microsatellite locus.All pairs of loci were tested for gametic disequilibrium in Genepop v4.2 (Raymond & Rousset 1995).Differences in allele frequencies between strata were performed using Fisher's exact test with Markov chain parameters set to 5000 dememorization steps followed by 500 batches of 2000 iterations per batch in Genepop v4.2.Weir & Cockerham's (1984) F ST was used to assess the difference between founder and contemporary strata using Genepop v4.2 for microsatellite genotypes and Arlequin 3.1 (Excoffier & Lischer 2010) for mtDNA.Sex ratios among strata were compared using an exact binomial test based on the null hypothesis of a 1:1 sex ratio.

Observed and inferred maternity
We searched the database for mother/offspring pairs identified by photo-ID in the summer of the calf's birth.These were referred to as 'observed mater nities.'When DNA profiles were available for both the mother and offspring, the relationship was confirmed by strict exclusion; i.e. an observed mother/ offspring pair was 'confirmed' by a shared allele at all microsatellite loci, allowing for 1 mismatch due to mutation or allelic dropout (e.g.Carroll et al. 2012) and by a matching mtDNA haplotype, when available.
Many dependent calves do not show their flukes and, in these cases, we are not able to link a calf with its mother in future years by observation alone.Thus, in most cases, contemporary whales did not have observed mothers documented by photo-ID.When DNA profiles were available for individuals without a photo-ID confirmed mother, we conducted a likelihood-based parentage analysis in CERVUS v3.0 (Kalinowski et al. 2007) to infer maternity.The set of possible mothers included all confirmed founder females with genotypes (n = 25) and an outgroup of randomly selected non-founder females sampled throughout SEAK (n = 25).We considered an individual to be a likely founder descendant if: (1) a founder female was assigned with the highest, positive logarithm of the odds (LOD) score of all putative parents (i.e. the natural log, ln, of the multilocus likelihood ratio), (2) the mother/offspring pair was not excluded by having different haplotypes (except in cases of hetero plasmy), and (3) the mother/offspring pair had mismatching alleles at no more than 1 microsatellite locus with 8 or more loci compared.Again, 1 mismatch was allowed to account for mutation or allelic dropout (Carroll et al. 2012).

Inferring paternity
We also attempted to infer paternity for 'confirmed' contemporary offspring of founder mothers using CERVUS v3.0 (Kalinowski et al. 2007).For this, we used the 1-parent known option to increase the power of the paternity inference.The set of possible fathers included all founder males with DNA profiles (n = 21) and an outgroup of randomly selected nonfounder males sampled throughout SEAK (n = 21).Photo-ID records for putative fathers were reviewed to determine whether life history information (e.g.age) was consistent or inconsistent with inferred paternity.

Founder and contemporary strata
The founder stratum included 74 whales.Of these 74 individuals, 46 had DNA profiles.The contemporary stratum included 171 whales, of which 118 had DNA profiles.Five individuals were first identified as dependent calves in 2004 by photo-ID but returned in 2005 as yearlings and, thus, were included in the contemporary stratum.
Photo-ID and genotype matching confirmed that 38 of 74 founder whales were represented in the contemporary stratum.Thus, 2 variants of the contemporary stratum were established for all analyses, C 1 : contemporary whales including extant founder whales (n = 171, 118 with DNA profiles), and C 2 : contemporary whales excluding extant founder whales (n = 133, 84 with DNA profiles).In total, 207 individuals were represented in the 2 strata, of which 130 had at least a partial DNA profile, i.e. a haplotype and/or genotype combined with sex information when available (Table 1).
The sex ratio did not differ significantly from 1:1 within the founder stratum (p = 0.659) or within the C 1 stratum (p = 0.108).However, there was a significant male bias within the C 2 stratum (p = 0.031).

mtDNA diversity and temporal differentiation
Mitochondrial DNA haplotypes were available for 46 founder whales, 111 C 1 whales, and 77 C 2 whales (Table 1).There were no significant differences in mtDNA haplotype frequencies between the founder and contemporary strata when sexes were pooled (founder vs. C 1 : p = 0.898; founder vs. C 2 : p = 0.821).However, sex-specific tests revealed a significant difference in haplotype frequencies between founder males and C 2 males (p = 0.012), attributable to an increase in the A+ haplotype (Fig. 2).
There was an increase in the number of haplotypes observed in the C 1 and C 2 strata.The founder stratum included only A− and A+ haplotypes while the C 1 and C 2 strata included 2 additional rare haplotypes (E2 and A8), only 1 of which (E2) has been repor ted previously in the North Pacific (Baker et al. 2013).
Following the discovery of the novel A8 haplotype, electropherograms were reviewed for potential hetero plasmy (the presence of more than 1 mtDNA variant within an organism; e.g.Hoeh et al. 1991).

Microsatellite diversity and temporal differentiation
A total of 46 founders, 114 C 1 , and 80 C 2 individuals were genotyped at 8 or more loci (Table 1).The number of alleles per locus ranged from 6 to 12, the P (ID) for each locus ranged from 0.027 to 0.485, and the combined P (ID) for all 10 loci was 1.1 × 10 −10 .The P (ID) SIBS was 1.8 × 10 −4 .All but 1 locus were in HWE (rw4-10, Table 2).There was no significant gametic dis equilibrium among the loci after Bonferroni cor rection.There was no significant differentiation in microsatellite allele frequencies between the founder and contemporary strata (founder vs. C 1 : p = 0.996; founder vs. C 2 , p = 0.707).

Recruitment and fidelity within the founder stratum
To determine the proportion of founders that were locally recruited (i.e.offspring of founder females), we first searched the database for 'ob served' mother/ offspring relationships within the founder stratum.Eight founder whales were born to founder mothers based on photo-ID records first documenting the offspring as calves (Table 3).In other words, local recruitment was responsible for at least 10.8% of the founder stratum itself between 1973 and 1985.In 3 of   1 because only whales with both mtDNA haplotype and genetic sex are represented here the 8 observed pairs, there were DNA profiles for both the mother and offspring, which, in all cases, confirmed the observed relationship.A total of 38 founders were re-encountered in GBIS from 2004 to 2005 (Fig. 3), thus representing 22.2% of the contemporary (C 1 ) stratum.Of all genetically confirmed females with genotypes in the founder stratum, 24 of 25 (96%) were either represented in the contemporary stratum, had at least 1 descendant in the contemporary stratum, or both (Table 4).

Maternity with photo-ID and DNA profiles
To determine the proportion of the C 2 stratum (contemporary whales excluding extant founder whales) that descended from a founder female, we first searched for mother/offspring relationships using photo-ID and DNA profiles.Thirty-two individuals in the C 2 stratum (18.7% of n = 171) had a founder mother based on photo-ID sighting records (Table 5).Of these 32 pairs, 22 were confirmed by genotypes.The remaining 10 pairs lacked a DNA profile for either the offspring, the mother, or both.
Of the 32 C 2 individuals identified as offspring of founder females, 5 were females observed with a calf in the C 2 stratum (i.e. 5 C 2 whales had a founder grandmother).These mother/offspring relationships were confirmed with a DNA profile when available (Table 6).The 5 C 2 whales with founder grandmothers make up another 2.9% of the total C 1 stratum descended from a founder female (Fig. 3).
Only 5 other C 2 whales had observed mothers based on photo-ID sighting records, none of which were founder females.Two C 2 whales had a mother in the C 2 stratum but did not have a known link to a founder female, i.e. these were internally recruited from the C 2 stratum, but recruitment from the founder stratum was unconfirmed (Fig. 3).

Maternity inference
To investigate the potential for maternity not documented by photo-ID and field observations, we undertook a likelihood analysis of parentage using microsatellite genotypes in CERVUS v3.0.After subtracting the C 2 whales that had observed mothers based on photo-ID, 91 individuals remained without a recorded mother or link to founders.Of these, 49 had a genotype available for maternity inference.Based on a comparison of these 49 individuals to founder females (n = 25) and an outgroup of 25 fe males selected randomly from SEAK, 12 founders were considered likely mothers of 20 of the 49 C 2 whales.These inferred maternity re lationships represented 11.7% of the total C 1 stratum (Table 7).

Paternity inference
To investigate the potential for paternal contribution to local recruitment in GBIS, we used the n = 22 photo-ID and genotypically confirmed C 2 offspring/ founder mother pairs to conduct a '1 parent known' paternity analysis.This analysis provided 1 case of a founder male that matched at all loci with a contemporary offspring and showed a relatively high LOD score when compared to the mother/offspring pair.The individual SEAK ID 117 was assigned as the likely father to ID 1470, a calf born in 1998 whose mother was confirmed to be ID 1460 by photo-ID and DNA profile (trio LOD = 4.19).The probability of non-exclusion was 0.024 and 0.010 for the first and second parent, respectively.A total of 8 loci were compared among the trio with alleles inconsistent with parentage at only 1 locus between the mother and offspring.Because the relationship between the mother and offspring was confirmed with photo-ID, we presume this mismatch is due to mutation or allelic dropout.All 3 whales had the A− haplotype.The individual SEAK ID 117, nicknamed 'White Eyes,' was first documented in 1977 by Sea Search, Ltd (National Marine Mammal Lab, unpublished data).Thus, life history information for this putative father was considered consistent with the assigned paternity.

Haplotype stability suggests lack of immigration
The GBIS area has maintained relatively low haplotype diversity over time (Baker et al. 1998, Fig. 2).Despite the fact that there were 13 different haplotypes on the northern Gulf of Alaska (NGOA) feeding ground when sampled during the SPLASH effort Offspring that were inferred are in italics, offspring that were observed with photo-ID are marked with an asterisk, offspring that were observed with photo-ID and confirmed with a DNA profile are underlined.Only 1 founder female (in bold) was not documented in the contemporary stratum and did not have an offspring in the contemporary stratum.SEAK: southeast Alaska (Baker et al. 2013), only 2 NGOA haplotypes were shared in GBIS (the common A+ and A−).Due to the consistent frequencies of these 2 common haplotypes in GBIS over time, it is unlikely that the population increase within GBIS is due to immigration of whales from other feeding regions (Baker et al. 1998(Baker et al. , 2013)).Instead, the relative stability in haplotype frequencies over time suggests strong regional fidelity of the maternal lineages represented in GBIS between 1973 and 1985 and persisting to the present day.
If immigration had occurred at any appreciable demographic rate in GBIS, we would expect to see additional haplotypes from other feeding grounds appear since 1985.Based on the known distribution of humpback whale haplotype frequencies across the North Pacific (Baker et al. 2013), it seems unlikely that the 2 haplotypes found only in the C 2 stratum (A8 and E2) were introduced into GBIS from outside of the SEAK feeding ground.In fact, A8 seems to have arisen from the fixation of the A−/A8 heteroplasmy detected almost exclusively in SEAK and Hawaii, the prima ry migratory destination for whales in SEAK (Pierszalowski 2014).Based on SPLASH surveys, the only published record of the E2 haplotype on a feeding ground outside of SEAK and NBC is 1 individual sampled in the western Gulf of Alaska (Baker et al. 2013).Therefore, the E2 individual in the contemporary stratum (ID 1489, first documented in GBIS in 1999) is likely to have shifted habitat use to GBIS from another area in SEAK or northern British Columbia.
We detected an increase in the frequency of the A+ haplotype between founder males and C 2 males.However, of the 17 males with the A+ haplotype in the C 2 stratum, at least 8 descended from a founder or C 2 female, indicating that internal recruitment largely drove this increase in males with the A+ haplo type (Rosenbaum et al. 2002

Local fidelity and recruitment
Twenty-four of 25 genetically confirmed females with genotypes in the founder stratum were either represented in the C 1 stratum, had at least 1 descendant in the contemporary stratum, or both.The 1 female without representation in the C 1 stratum (SEAK ID 569) was sighted elsewhere in SEAK during SPLASH and up through 2009.
More than half of the 74 founders were still alive and returning to GBIS 20 yr later, such that 22.2% of the C 1 stratum consisted of individual founder whales returning to GBIS in 2004GBIS in /2005 (n = 38) (n = 38).This finding provides evidence that the increase in population size within GBIS is largely maintained through survival and sustained local fidelity.One-third of the total C 1 stratum (33.3%) was shown to have descended from a founder mother or founder grandmother using a combination of photo-ID, DNA profiles, and maternity inference analysis.This indicates that recruitment is another key driver of population increase in GBIS.
Taken together, a minimum of 55.5% of the total C 1 stratum was found to be either a returning founder whale or a descendant of a founder female.We consider this to be a minimum estimate because the ability to detect all mother/offspring pairs would require photo-ID records of every individual, as well as field notes indicating their relationships to each other or DNA profiles with which to confirm or infer maternity -this was not always the case with the available data.In total, 42 of 171 C 1 whales (24.6%) did not have a photo-ID link to a mother or a genotype with which to infer maternity, i.e. we had no methods of determining whether they were a descendant of a founder female.We can adjust for these unverifiable whales by subtracting them from the C 1 stratum and arrive at a total of 129 C 1 whales for which we were able to investigate maternity.After this adjustment, 73.6% of the C 1 stratum was either a returning founder or a descendant of a founder female.A final bias is the absence of genetic samples for 28 founders, half of which are presumably female.This limitation is more difficult to account for.If any of these founders are sampled in the future, it seems likely they will explain more of the contemporary stratum.
Testing explicitly for recruitment within the greater SEAK feeding ground reaches beyond the scope of this study.However, due to the small impact of immigration, demonstrated by stability in mtDNA haploytpes over time, it is presumable that some proportion of the population increase in GBIS is attributable to recruitment from whales sighted elsewhere in SEAK.

Detecting paternity among GBIS founders
It is generally assumed that most humpback whales from SEAK migrate to Hawaii where males and females mate with individuals from SEAK and other North Pacific feeding grounds (Baker et al. 1986).However, behaviors associated with mating on breeding grounds have also been documented in GBIS, including singing (Gabriele & Frankel 2002) and interactions resembling competitive group behavior (Neilson & Gabriele 2009, Neilson et al. 2012, 2013).Numerous humpback whales are present in SEAK during fall and winter (Straley 1990, 1994, Straley et al. 1993), where some continue to exhibit breeding ground behavior throughout the fall with full song heard in winter (Straley 1990).Researchers have long wondered whether these feeding ground behaviors result in pregnancy or promote preferential mating when whales do return to their breeding ground (Weinrich 1991, Clapham 1993, Ramp et al. 2010).Our finding of only 1 likely paternity among the 22 confirmed C 2 offspring/ founder mother pairs suggests that feeding ground interactions do not often result in breeding.However, the founder males included in the paternity search were selected based primarily on shared feeding habitat with females during summer months.A more comprehensive survey of males from Hawaii and those documented in SEAK during late fall and winter is needed to judge the relative contribution of within-feeding ground paternity (e.g.Garrigue et al. 2004).

Conservation implications
Our results confirm strong maternally directed fidelity on a fine geographical scale.These findings highlight the importance of maternal migratory traditions, which have been key to the recovery of humpback whale populations.Baker et al. (2013) suggested that maternal traditions of migration should be considered as units to conserve.Our study highlights the importance of local habitat protection for species with culturally inherited migratory destinations, given that population growth for this single, wellstudied population is largely attributable to local site fidelity and recruitment.However, we do not intend to suggest that these phenomena are unique to this sub-region of SEAK.Local fidelity and recruitment may have led to population growth in other regions within SEAK and on other humpback whale feeding grounds.
Given the estimated longevity of humpback whales (70+ yr, Gabriele et al. 2010), it is likely that many of the founder whales in this study are survivors of modern whaling and continued to inhabit GBIS despite the overall reductions in humpback whale numbers throughout the North Pacific.Thus, it seems plausible that a remnant stock (i.e. the founder stratum) was sufficient for population recovery in GBIS, driven primarily by local fidelity and re cruitment, versus immigration from other feeding grounds.This population recovery has occurred over the course of 40 yr (1965−2005), demonstrating the slow rate of recovery when it is driven primarily by fidelity and recruitment.Today the whales that re turn to GBIS represent a growing local population.However, the strong fidelity of whales to GBIS makes this population vulnerable to local impacts, and careful management will be needed to ensure the health of this population.Limiting vessel traffic in important habitats, maintaining regulatory vessel approach distances to whales, reducing the risk of fishing gear entanglement, and maintaining stranding networks with the capacity to disentangle whales are all important factors for protecting humpback whales in the North Pacific.However, it is ultimately the health of the marine ecosystem that underlies the success of a population.As the global climate warms and ocean acidification affects the marine food web, marine protected areas like Glacier Bay that control vessel traffic and other anthropogenic impacts will become increasingly important to the continued success of humpback whales around the world.
Acknowledgements.All research was conducted under appropriate permits issued by the US National Marine Fisheries Service, in accordance with the US Marine Mammal Protection Act and the US Endangered Species Act, including nos.325, 571, 767, 933, 0473-1433, 0473-1700, and

Fig. 1 .
Fig. 1.US National Park Service (NPS) Humpback Whale Monitoring Program study area in Glacier Bay and Icy Strait, Alaska (USA), showing the boundary of Glacier Bay National Park and Preserve Fig. 2. mtDNA haplotype frequencies for individual humpback whales Megaptera novaeangliae in the founder (n = 46), C 1 (contemporary whales including extant founders, n = 105), and C 2 (contemporary whales excluding extant founders, n = 71) strata with genetically confirmed sex.Note that totals do not matchTable 1 because only whales with both mtDNA haplotype and genetic sex are represented here

Table 3 .
Founder humpback whales Megaptera novaeangliae born to mothers in the founder stratum (listed by South east Alaska ID code) as documented by photo-ID and, when possible, genotypic confirmation; na: not available Individuals in the contemporary stratum were either a returning founder whale, a descendant of a founder female, or not confirmed to have descended from a founder whale.Shaded boxes highlight individuals that were either a returning founder or a descendant of a founder female (n = 95, 55.6% of the total C 1 stratum [contemporary whales including extant founders], 73.6% of the verifiable C 1 ).Photo-ID observed mother/offspring relationships were confirmed with DNA profiles when available

Table 4 .
Representation and recruitment to the contemporary stratum of 25 founder female humpback whales Megaptera novaeangliae with DNA profiles to confirm or infer offspring.

Table 5 .
).Individual humpback whales Megaptera novaeangliae in the C 2 stratum (contemporary whales excluding extant founders) with a reported founder mother based on photo-ID (n = 32, listed by southeast Alaska ID code of offspring).When available, genotypes were compared to confirm the relationship (n = 22).Hap: haplotype, na: not available, LOD: logarithm of the odds

Table 6 .
Five C 2 (contemporary individuals excluding extant founders) humpback whales Megaptera novaeangliae with mothers in the C 2 stratum and grandmothers in the founder stratum (listed by southeast Alaska ID code).Haplotypes and the first year each founder grandmother was seen in Alaska are reported in parentheses; na: not available

Table 7 .
Twenty C 2 (contemporary individuals excluding extant founders) humpback whales Megaptera novaeangliae with founder mothers assigned using maternity inference (listed by southeast Alaska ID code).Hap: haplotype, LOD: logarithm of the odds, na: not available 14122 issued to J.M.S., nos.945-1499-02 and 473-1700-00 issued to the Glacier Bay National Park, and no.675 issued to C.S.B. Support for this work was provided by a cooperative agreement between Oregon State University and the National Park Service (Pacific West Region Cooperative Ecosystems Study Unit Task Agreement P12AC15004).Additional funding was provided through various Oregon State University awards, including the Mamie Markham Research Award, Joan Crebbin Memorial Fellowship, Lylian Brucefield Reynolds Scholarship, Thomas G. Scott Grant Scholarship, and the Hatfield Marine Science Center Student Organization Travel Grant.Thanks to P. Clapham and S. Mizroch from NOAA's National Marine Mammal Laboratory for instigating this collaboration and for facilitating access to the FlukeFinder Database, respectively.Fluke photography and tissue sampling were conducted under scientific research permits issued by the National Marine Fisheries Service.A special thanks to Charles Jurasz for his insight and foresight in documenting individual whales.