Testosterone levels in hair of free-ranging male northern fur seals ( Callorhinus ursinus ) in relation to sampling month, age class and spermatogenesis

Information about the reproductive status of free-ranging pinnipeds provides useful insight into their population dynamics, which is essential to their management and conservation. To determine the reproductive status of individual animals, blood sampling is often required despite being impractical to collect in open water. Hair as an endocrine marker has been used to less invasively assess the reproductive status of terrestrial animals. However, it is unknown whether pinniped reproductive status can be assessed from hair samples. Here, we examine testosterone levels in hair obtained from 57 male northern fur seals and used it to compare their age class and spermatogenesis during the non-breeding season off Hokkaido. We isolated testosterone from the samples using gas chromatography and measured testosterone levels using time-resolved fluoroimmunoassay. Testosterone levels in hair increased towards the breeding season. In May, testosterone levels were the highestinsealsagedbetween4and7years,followedbythoseovertheageof8yearsandundertheageof4years.Spermatids, thefinalphaseofspermatogenesis,werepresentinthesealssampledbetweenAprilandJune,eventhoughtestosteronelevels were low in April. The seals with spermatids in May showed the highest testosterone levels. Our results demonstrate that seals with higher testosterone levels in May are likely to be mature males ( ≥ 4 years). Since hair can be collected using biopsy darts in the field, it will be possible to less invasively determine testosterone levels of male seals in the future.


Introduction
Understanding population dynamics is essential to the effective management and conservation of threatened species. The reproductive physiology of individual animals is a critical component of population dynamics. Endocrine studies, however, often involve animal captures and require blood sampling, which can be impractical in free-ranging animals (Amaral, 2010). As a result, less-invasive techniques of endocrine monitoring have been sought as a replacement for blood sampling, and these include sampling of urine (e.g. Amaral et al., 2009;Constable et al., 2006), faeces (e.g. Cockrem and Rounce, 1994;Graham and Brown, 1997), saliva (e.g. Amaral et al., 2015), blubber (e.g. Kellar et al., 2009;Mansour et al., 2002) and hair (e.g. Davenport et al., 2006;Koren et al., 2002). Of those, hair as an endocrine marker has been used in various mammals such as humans (Kintz et al., 1999), cats, dogs (Accorsi et al., 2008), bears (Bryan et al., 2013b), chimpanzees (Yamanashi et al., 2013), caribou and roe deer (Capreolus capreolus) (Ashley et al., 2011;Ventrella et al., 2018). The advantage of hair sampling is that hair can be remotely collected in the field using materials such as barbed wire and biopsy darts. Barbed wire and biopsy darts are used to study genetics (Beier et al., 2005;Pagano et al., 2014) in wildlife, and biopsy darts can collect tissue and hair samples to study toxicology (Fossi et al., 1997;Krahn et al., 2007). Remote biopsy darts have been wide spread in studies of marine mammal and can reduce handling stress for animals and improve animal welfare.
Sexually dimorphic northern fur seals (Callorhinus ursinus) breed in summer (Gentry, 1998) and migrate south to open waters without hauling out on land during the nonbreeding season (Bigg, 1990;Kenyon and Wilke, 1953). During this season, sexual segregation of northern fur seals has been observed (Kenyon and Wilke, 1953;Wilke, 1951), and a larger number of males are observed compared to females near the coast of southern Hokkaido in the Sea of Japan (Horimoto et al., 2017;Horimoto et al., 2016). Male northern fur seals have mostly been observed from November to May in the region, and this time period coincides with the non-breeding season of northern fur seals in Hokkaido (Horimoto et al., 2017). However, few studies are conducted to examine the reproductive status of northern fur seals during the non-breeding season even though testosterone levels of male northern fur seals towards the breeding season can be an important indicator for their body condition. The energy gained during the non-breeding, foraging season will likely predict the success of the following breeding season (Beck et al., 2003) since territorial male fur seals at breeding rookeries fast. Moreover, juveniles and mature seals need to increase their body size to survive and reproduce since 30% of mature male northern fur seals are excluded from reproduction for their entire lives (Vladimirov, 1987) and since the survival rate to the age of breeding is only 3% (Lander, 1981-82). Long-term monitoring of demography of northern fur seals can reveal the connection between the non-breeding and breeding seasons. Knowledge of the demography of males can strengthen the understanding of how animal condition can affect reproduction (Beck et al., 2003) and, hence, the growth of the population. This information will be an essential factor for resource management and conservation.
Testosterone provides critical information about the reproductive status and age class of male mammals. Mammals with seasonal breeding cycles show an increase in plasma testosterone levels prior to the onset of the breeding season. For example, pinnipeds such as elephant seals (Mirounga leonina) and hooded seals (Cystophora cristata) also show an increase in plasma testosterone concentrations one month prior to their breeding seasons (Griffiths, 1984;Noonan et al., 1991). Similarly, serum testosterone levels of captive male fur seals start to rise in April towards the breeding season (Kohyama et al., 1999;Otsuki et al., 2020). Male northern fur seals building territories appear in rookeries as early as May, and breeding peaks in July (Gentry, 1998;Vladimirov, 1987). One of the functions of testosterone is spermatogenesis (Roberts and Zirkin, 1991). Along with its influence on sexual dimorphism and reproductive behaviour, testosterone plays a key role in spermatogenesis. Spermatogenesis is enhanced during the breeding season, and gonads are enlarged due to an increase in testosterone levels (Blottner et al., 1996;Holekamp and Talamantes, 1991). The diameter of the seminiferous tubule of a captive northern fur seal was also shown to increase towards the breeding season (Tsubota et al., 2001). Moreover, testosterone levels have been used to differentiate age classes of male mammals. For instance, testosterone levels in hair of free-ranging bears were used to predict male maturity by comparing them with the age obtained from a premolar (Cattet et al., 2018). In northern fur seals, males become sexually mature at around the age of 4 years or older, after which they are capable of reproduction (Gentry, 1998). However, only males older than 8 years typically form harems and successfully sire offspring (Vladimirov, 1987). Characterization of sexual maturity and age class of northern fur seals using hair testosterone levels will allow us to enhance our understanding of fur seal demography in southern Hokkaido.
Hair steroid hormones reflect the integrated free hormone fraction rather than the total concentrations in serum (Russell et al., 2012;Stalder and Kirschbaum, 2012) that are accumulated in hair with a time delay (Ventrella et al., 2018). Hair has been shown to reflect endocrine activities spanning several weeks or months prior to the time of collection (Koren et al., 2002). The interpretation of steroid hormone levels in hair, thus, depends on the timing of hair sampling (Ventrella et al., 2018). Northern fur seals annually moult their pelage after mating in July, and complete moulting in the fall, but the moult in older males starts later in summer compared to younger males (Scheffer and Johnson, 1963). Serum steroid hormones of northern fur seals are possibly integrated in hair spanning after moult, and testosterone in hair of northern fur seals may also increase with a delay after an increase in serum testosterone levels. Although these concerns around the interpretation of hormones in hair still remain, Keogh et al. (2020) demonstrated that testosterone levels in hair of northern fur seal subadults and pups are the same. Since they did not compare testosterone levels across all age classes, it is still unknown whether the sexual maturity of northern fur seals can be determined using testosterone measured from hair. The objective of this study is to compare testosterone levels in hair of male northern fur seals during the non-breeding season off Hokkaido with age class and spermatogenesis.

Study animals
Fifty-seven stranded, bycaught and fatally caught male northern fur seals were collected between 2011 and 2018 along Hokkaido, Japan (S1). All handling and sampling of fur seals were conducted with the permission of the Fisheries Agency under the Sea Otters and Fur Seals Hunting Control Act. Hair samples were collected from the ventral side of seals. They were kept at −20 • C until analysis. The seals were aged by the number of growth layers in tooth sections (Horimoto et al., 2017). The age of seals without teeth due to fracture of jaw bone was determined from the body length based on Trites and Bigg (1996). Based on these ages, seals were classified into three categories: I (<4 years old), II (4-7 years old) and III (≥8 years old) (Gentry, 1998;Horimoto et al., 2017).
The testes were preserved in 10% formaldehyde. Testicular sections were prepared by following the methods in Ishinazaka (2002) and Hirakawa et al. (2021). The central part of the tissue was sliced into sections, which were dehydrated using ethyl alcohol, embedded in paraffin wax. These sections were further sliced into 4-6 μm in thickness using a microtome. These cross sections were stained with haematoxylin and eosin. Spermatogenesis in the cross sections of seminiferous tubules was identified using an optical microscope. Spermatogenesis stages were determined by the presence of spermatogonia, primary spermatocytes, round and elongated spermatids and spermatozoa in the seminiferous tubules of the seals. Samples were classified based on the following observations: spermatogonia only, SG; primary spermatocytes but no spermatids, SC; and round and elongated spermatids and spermatozoa, SP (Hirakawa et al., 2021).

Sample preparation
Hair samples were thawed, and hair (i.e. guard hair and underfur) was cut along the skin. The hair was washed with distilled water until the water became clear. Then hair was washed with 70% isopropanol for 2 min, which was repeated two more times to remove any steroid hormones coating the hair (Yamanashi et al., 2013). The samples were then dried at 40 • C (DKM600, Yamato Ltd, Japan). The hair was pulverized with a ball mill at 1100 rpm for 6 min (Shake Master Auto, BioMedical Science Co. Ltd, Japan). Hair samples (10 mg) were weighed into a test tube, and 99.8% methanol (1 ml) was added. Hair steroid hormones were extracted by sonicating in a water bath at 60 • C for 10 min in standing waves interfered by spherical waves (CPX2800H-J, Branson Ultrasonics, Emerson Japan, Ltd, Japan). After extraction of hormones, the samples were centrifuged at 20400 × g for 15 min (MX-307, Tomy Seiko Co., Ltd, Japan). The resulting supernatant (0.6 ml) was dried under a centrifugal evaporator (CC-181, Tomy Seiko Co., Ltd, Japan) and the dried extract was reconstituted with the same volume (0.6 ml) of Tris-HCl assay buffer. The reconstituted samples were frozen at −20 • C until analysis.

Testosterone measurement
Time-resolved fluoroimmunoassay (DEL-FIA ® , PerkinElmer, Waltham, MA) was used to assay the samples (Yamada et al., 1997). The assay protocols were the same as those described by Otsuki et al. (2020). The coefficient of variation for the intra-assay and inter-assay of control samples was 6.89% (n = 8) and 8.57% (n = 4), respectively. Extraction efficiency was determined by spiking testosterone standards to a sample prior to extraction. The spiked and unspiked samples were assayed to determine the extraction efficiency. Extraction efficiency was 105.7% (n = 4) by spiking with testosterone and compared with non-spiked samples.

Gas chromatography with flame ionization detector
The presence of testosterone in the hair of a seal, one sample, was confirmed by using a gas chromatography-flame ionization detector (GC-FID, GC 390B, GL Sciences, Japan). Hair samples (10 mg) were extracted with methanol as previously described. The extracts (0.6 ml) were dried with N 2 and reconstituted with LC-MS-grade methanol. To make a spiked sample, we added testosterone standard solution (T0027, Tokyo Chemical Industry Co., Ltd, Japan) to the extract. Reconstituted samples (3 μl) were analysed by GC-FID using a 30-m-long column, Ultra ALLOY-5 (MS/HT) with an internal diameter of 0.25 mm and 0.25 μm film thickness. Helium was used as a carrier gas with 2 ml/min constant flow compensation. The injection temperature was 250 • C, and the detector temperature was 350 • C. The oven temperature was set at 120 • C for 5 min, increased at the rate of 7 • C/min to 260 • C for 10 min and then 7 • C/min to 350 • C for 15 min.

Statistical analysis
A parallelism test between serially diluted testosterone standards and hair samples was conducted using an ANCOVA. Testosterone levels of a seal per unit 1-g hair were computed. Testosterone levels in hair were analysed using generalized linear models (GLMs) with gamma error distributions and a log link function. The explanatory variable of each model was sampling month, age class or spermatogenesis. Since all the explanatory variables were categorical, all possible combinations of categorical variables were modelled (Mori  ). The best model was selected using Akaike's information criterion (AIC) score comparisons. For sampling month, only the hair samples collected between March and June were included, since the sample size in each of January, February and November was only one. Samples collected in May were used for modelling age class and spermatogenesis. All of the statistical analyses were conducted in R 3.5.1 (R Core Team, 2019) using a significant level of α = 0.05.

Validations of testosterone extraction and assay
Testosterone extracted from a sample of northern fur seal hair was validated by GC-FID and a parallelism test. The peaks of the sample and those with testosterone standard solution appeared at almost identical retention times of 27.88 and 27.83 min, respectively (Fig. 1). The parallelism test showed no interaction between serially diluted hair samples and standard solutions (F 1,4 = 0.009, P = 0.928; Fig. 2).

Monthly testosterone levels in hair
Monthly testosterone levels in hair of northern fur seals showed seasonal changes (Fig. 3). Means and standard deviations of testosterone levels are shown in Table 1. The GLM model showed that mean testosterone levels of the seals sampled in March and April were the same (Table 2), which was lower than those sampled in May and June (Fig 3). Of all months during the interval between March and June, mean testosterone levels of the seals were the highest in May (Table 2; Fig. 3).

Testosterone levels in hair and age class
Northern fur seals belonging to age classes II (4-7 years old) and III (≥8 years old) that were sampled in March and April showed lower testosterone levels, while those sampled in May and June showed greater levels (Fig. 4). Means and standard deviations of testosterone levels of seals from age classes I, II and III in May were 39.5 ± 10.3, 128.3 ± 69.4 and 94.8 ± 44.3 ng/g, respectively (Table 1). Mean testosterone levels of males from age class III in June was 57.6 ± 52.2 ng/g. The GLM for age class in May showed that mean testosterone levels were different from each other (Table 3). Males from age class II showed the greatest testosterone levels among three age classes in May (Table 3; Fig. 4).

Testosterone levels in hair and spermatogenesis
Seals with SC and SP in March and April showed low testosterone levels (Fig. 5). Testosterone levels for all the seals with SG, SC and SP increased in May compared to those in March and April. In June, testosterone levels in seals with SP decreased compared to those with SP in May. Means and standard deviations of testosterone levels in the seals with SG, SC and SP in May were 38.8 ± 11.8, 42.0 and 109.5 ± 58.1 ng/g, respectively ( Table 1). Testosterone of seals with SG in May ranged from 25.7 to 54.8 ng/g. The GLM model for spermatogenesis in May demonstrated that the seals with SP had higher testosterone levels than those with SG and SC (Table 4; Fig. 5).

Discussion
This study is the first to compare testosterone levels in hair of male northern fur seals to spermatogenesis and age class. Hair as an endocrine marker has been used to examine steroid hormone levels in various terrestrial mammals such as carnivores (Cattet et al., 2017;Terwissen et al., 2014  primates (Kintz et al., 1999), rodents (Carlitz et al., 2019) and ungulates (Ventrella et al., 2018). These studies examined different steroid hormones; however, very few isolated testosterone from hair, possibly due to the relatively small amounts of this hormone in hair. A few studies have examined steroid hormones in the hair of pinnipeds, which measured testosterone in sea lions and fur seals (Keogh et al., 2020;Meise et al., 2016).
Previous studies use various extracting methods such as the combination of incubation and mixing at room temperature (Terwissen et al., 2013;Yamanashi et al., 2013), agitation (Schell et al., 2017), the combination of sonication and rotation at 45 • C (Bryan et al., 2013a) to extract steroid hormones. We sonicated samples in standing waves interfered by spherical waves in a water bath at 60 • C, which is close to the boiling point of extraction solvent, methanol. We first validated the presence of testosterone in hair extracts by GC-FID analysis. Since the retention time for samples and standard solutions was shown to be almost exactly the same, the presence of testosterone in hair was validated. Our extraction technique, thus, was an effective method. In other studies, testosterone in human hair has been detected by gas chromatographymass spectrometry (e.g. Kintz et al., 1999;Scherer et al., 1998). Second, based on the parallelism test, a non-significant interaction of standard and sample solutions indicates that the assay was validated. Testosterone levels in hair of various species have often been measured using enzyme immunoassays (EIAs) (e.g. Bryan et al., 2013a;Terwissen et al., 2014), possibly due to higher sensitivity of EIAs. In this study, we performed a time-resolved fluoroimmunoassay, which is also a highly sensitive assay (Yamada et al., 1997).
Our results showed low testosterone levels in the hair of northern fur seals during the non-breeding season from January to April and higher levels during the pre-breeding season in May. Similar trends were observed in studies of cap-tive northern fur seals that demonstrated low serum testosterone levels during the non-breeding season and high testosterone starting in April, which precedes the breeding season (Kohyama et al., 1999;Otsuki et al., 2020). Higher testosterone levels in hair in May likely resulted from high levels of serum testosterone in April and May. In addition, our results were in line with the study in brown bears (Ursus actos) (Cattet et al., 2017). Testosterone levels in hair of bears increased during the pre-breeding season and decreased in the post-breeding season.
The lower testosterone levels of seals sampled in June compared to May were observed. Breeding males are known to begin returning to their rookeries as early as May to establish their breeding territories (Gentry, 1998). However, seals remaining in open water in June may not participate in breeding, and thus showed lower testosterone levels. Other possible reasons for less hormonal elevation are environmental factors. Wester et al. (2016) demonstrated that ultraviolet radiation in natural sunlight could cause a decrease in glucocorticoid hormones in hair. It might be also possible that marine mammals are exposed to more ultraviolet radiations compared to terrestrial mammals as no shade is available in open water, and thus marine mammals show less steroid hormone levels in hair. Li et al. (2012) showed that longer the immersion of hair in water and hotter the water temperature reduced cortisol concentrations in human hair. In monkey, washing hair also decreased the concentrations of cortisol compared to no-wash control hair (Hamel et al., 2011). Although northern fur seals face much lower temperature at sea, spending most of time in sea may cause the degradation of steroid hormones in hair. This might have reduced hormonal elevations compared to terrestrial mammals.
In relation to the hair growth cycle of northern fur seals, testosterone levels may not be incorporated into hair during the active hair growth. Moult in northern fur seals start in

Research article
Conservation Physiology • Volume 9 2021 June, during which the new generation of hair is still subsurface, and ends in November (Scheffer and Johnson, 1963). In general, steroid hormones are passively incorporated into hair during hair growth (Russell et al., 2012;Stalder and Kirschbaum, 2012). If so, testosterone was incorporated into hair of northern fur seals from June to November during the moulting season. However, in our study testosterone concentrations in hair peaked in May during a resting phase of hair. This may suggest that steroid hormones can be incorporated outside of active hair growth. Further mechanisms of incorporation of hormones need to be clarified. Studying hair samples during the breeding season may help to reveal this mechanism.
One 14-year-old seal in November showed high testosterone levels despite being sampled during the non-breeding season. It may be related to moulting. In older male northern fur seals, the moult starts around August and lasts until November (Scheffer and Johnson, 1963). More than 40% of northern fur seals possess two generations of hair during the early moult of September and October, and a few in November (Scheffer and Johnson, 1963). Furthermore, hair is likely to reflect the integrated free hormone fraction rather than the total concentration in serum (Russell et al., 2012;Stalder and Kirschbaum, 2012), and hair thus reflects endocrine activities integrated over several months (Bryan et al., 2013b;Ventrella et al., 2018). The seal in November might have possibly possessed fur in which testosterone had been incorporated since spring, and thus showed high levels even as late as November.
Age class was another factor for the differences that we observed in testosterone levels. Although testosterone levels in seals from age classes II and III were low in March and April, higher concentrations of testosterone in May likely reflects the reproductive status of these male seals as they approach the breeding season. Northern fur seals become sexually mature at age class II (4-7 years old), and those at age class III (8-10 years old) start to participate in mating (Gentry, 1998). Low testosterone levels in seals from age class I in May likely indicate that these seals were still immature. Similar results were obtained using testosterone from the hair of male brown bears (Ursus arctos) to predict their age class between immature and adult bears (Cattet et al., 2018). This can be explained by the differences in serum testosterone levels between juvenile and mature animals. Juvenile shortfinned pilot whales (Globicephala macrohynchus) tend to show lower serum testosterone levels than adult whales (Kita et al., 1999). These studies agree with our finding that testosterone levels of males in age class I (<4 years old) were lower than levels in seals older than 4 years old. The mean hair testosterone levels in seals from age class I was 39.5 ng/g, which differed from 10 ng/g previously reported by Keogh et al. (2020) for fur seal pups and subadults. This disparity is possibly due to differences in the timing and locations of sampling, as well as the ages of the seals used for each study.
Testosterone levels in the hair of northern fur seals were compared with three stages of spermatogenesis. Even though an increase in serum testosterone levels initiates spermatogenesis (e.g. Brown and Follett, 1977), seals possessing SC and SP in March and April showed low testosterone levels. Monthly changes in testosterone in hair do not follow changes in serum on exactly the same time scale. Although serum testosterone levels in captive northern fur seals started to increase in April (Kohyama et al., 1999;Otsuki et al., 2020), serum testosterone levels in free-ranging northern fur seals may start to increase even earlier than captive males since free-ranging animals tend to have greater testosterone levels than captive animals (Wingfield et al., 1990). If so, low testosterone levels in hair sampled in April along with the presence of spermatocytes and spermatids can be explained. These results were in line with the study of roe deer hair. A temporal delay in accumulation of hormones in hair has been demonstrated. Testosterone levels in plasma and testicular morphometrics increased in the pre-rut group, while testosterone levels in hair were low (Ventrella et al., 2018). Similarly, in male bottlenose dolphins (Tursiops truncatus), the highest sperm density followed the peak of serum testosterone levels (Schroeder and Keller, 1989). The diameter of seminiferous tubules and the testis weight of short finned pilot whales also increased as testosterone levels increased (Kita et al., 1999). Terwissen et al. (2013) demonstrated that the time span of hormone incorporation into hair can be more than one month by injecting adrenocorticotrophin hormone into lynx (Lynx canadensis). In the present study, northern fur seals with testosterone levels greater than 54.8 ng/g in May have spermatids and are sexually mature. In May, northern fur seals with SP showed higher testosterone levels compared to those with SG and SC. Serum testosterone that started to increase towards the breeding season likely integrated in the hair of fur seals in May, while spermatogenesis has already been initiated. Thus, testosterone levels with SP were higher than those with SG and SC.
In summary, we provide the first analysis to compare age class and testis histology of male northern fur seals to testosterone levels from hair. Since the mechanism by which steroid hormones are incorporated into hair is not fully understood, further examination of this process is necessary. However, sexually mature northern fur seals (≥4 years old) showed greater testosterone levels in hair than juvenile seals when hair sampling is conducted in May. In addition, seals with high testosterone in their hair likely possess spermatids, and thus more likely to be mature. Although the sample size is limited other than May, this study provides a basic knowledge of changes in testosterone levels towards the breeding season and gives insight into how testosterone incorporated into   hair of northern fur seals. The knowledge of the connections between breeding and non-breeding seasons can allow us to understand animal conditions. These data on male reproductive status will enhance our understanding of fur seal demography in the future. Understanding the demography in northern fur seals would contribute to expand our knowledge of population dynamics of this species and help resource management and conservation. Less invasive sampling techniques such as biopsy collection of hair can advance fur seal physiology significantly because sampling does not require animals to be captured and sedated. This method can be especially useful for pinnipeds that spend time offshore, since hair can be biopsied when the animals are dispersed in open water, rather than having to wait for them to return to the rookery.

Funding
This work was supported by the Japan Society for the Promotion of Science KAKENHI (grant numbers 15H05709 and 16KT0140). This study was partly supported by the 'Noxious organisms fishery damage preservation program' of the Japanese Fisheries Agency.