Blubber endocrine profiles provide insights into reproductive biology in blue whales from the eastern North Pacific Ocean

Abstract

The goal of the present study was to complement existing data of testosterone and progesterone in blue whale (Balaenoptera musculus) blubber from the eastern North Pacific Ocean to evaluate effects of seasonality and location on these hormones and to better assess reproductive status of individuals. Physiological parameters regarding reproduction are fundamental for describing population dynamics, and hormones can be a valid tool to estimate those for wildlife populations. In this study, blubber tissue was validated for testosterone and progesterone assays. Hormone concentrations were measured in 69 (35 males and 34 females) blubber samples from live (n = 66) and stranded (n = 3) animals collected between 2002 and 2016 from a known winter reproductive ground in the Gulf of California (GoC) and summer feeding areas along the United States West Coast (USWC), specifically off the states of California and Oregon. Results were combined with sighting histories as a tool to determine reproductive status of individual whales. Testosterone concentrations in adult male blue whales were significantly higher (p < 0.05) in blubber biopsies sampled off the USWC between the months of June and November compared to those sampled in the GoC between February and April. Elevated testosterone concentrations likely indicate physiological preparation for reproductive activity while the animals were present off the USWC. Progesterone concentrations were significantly elevated in pregnant females, confirming progesterone as an indicator of pregnancy in blue whales. Probabilities of being pregnant were estimated for adult females with unknown sighting histories based on progesterone concentrations. Testosterone in females was detected and measured only in pregnant whales suggesting its biosynthesis or metabolism is altered during gestation. These results provide updated and new information on the reproductive cycle of blue whales in the eastern North Pacific, posing new milestones to better estimate the timing of the mating season for this endangered population.

1. Introduction

Blue whales (Balaenoptera musculus) occur worldwide and are separated into unique populations by ocean basins (Jefferson et al., 2015). A single subspecies of blue whale (B.m. musculus) is identified in the northern hemisphere (Rice, 1998), and specifically in the North Pacific Ocean, two populations are recognized based on acoustic studies, the western and the eastern North Pacific populations (ENP) (Stafford, 2003). The ENP population is considered to be the most recovered of all blue whales populations (Rankin et al., 2006; Thomas et al., 2016) and many studies have focused on its distribution, abundance, and movements since the late 1980s (Calambokidis et al., 1990; Fiedler et al., 1998; Mate et al., 1999; Reilly and Thayer, 1990; Stafford et al., 1999). However, the taxonomy is still uncertain, with recent identification of two different morphotypes (Gilpatrick and Perryman, 2008; Ortega-Ortiz et al., 2018). Abundance estimates and migration patterns for the ENP blue whales have been investigated since the 1980s using photo-identification (Calambokidis et al., 1990; Gendron and Ugalde de la Cruz, 2012). Currently, the population is estimated at about 2,000 individuals (Calambokidis and Barlow, 2020; Carretta et al., 2020) within a habitat range that extends as far south as the Costa Rica Thermal Dome and as far north as the Gulf of Alaska (Calambokidis et al., 2009; Mate et al., 1999; Reilly and Thayer, 1990). During the summer months, large aggregations of blue whales have been sighted feeding off the west coast of the United States (USWC), particularly off the coast of California. However, data show an increasing occurrence of blue whales off British Columbia and in the Gulf of Alaska, suggesting a potential northward shift in feeding habitat (Calambokidis et al., 2009). The reproductive grounds are less known, where the Gulf of California (GoC) is the only studied reproductive area for this population, with about 300 whales annually present between January and April (Gendron, 2002). Stable isotopes analysis has also shown that blue whales do forage in the GoC, indicating this is a feeding-reproductive ground (Busquets-Vass et al., 2021). Confirmation that these whales are part of the ENP population came from photo ID matches (Calambokidis et al., 1990), satellite tags (Bailey et al., 2010), genetics (Costa-Urrutia et al., 2013), acoustics (Paniagua-Mendoza et al., 2017; Stafford et al., 2001) and stable isotope analyses (Busquets-Vass et al., 2017).

To better understand population dynamics, biomarkers that describe key physiological processes can be useful in determining reproductive parameters such as sexual maturity, pregnancy and parturition rates, and calving intervals. There have been relatively few studies and published articles on reproductive biology for ENP blue whales in the past decade (Atkinson et al., 2020; Sears et al., 2013; Valenzuela-Molina et al., 2018), updating earlier studies on reproductive parameters conducted during commercial whaling (Mackintosh and Wheeler, 1929; Tomlin, 1967). Age of first parturition is calculated between 9 and 10 years old, setting the age of sexual maturity between 8 and 9 years old (Atkinson et al., 2020; Sears et al., 2013; Valenzuela-Molina et al., 2018). This estimate is in line with former studies on this population (Rice, 1963) and other subspecies (e.g., Antarctic blue whales (B. m. brevicauda)) (Ichihara, 1966; Ohsumi, 1979). Atkinson et al. (2020) calculated a pregnancy rate of 0.34 considering a calving interval estimate of about three years; however, those results were based exclusively on animals from the GoC, and therefore might not be reflective of the overall population. For males, population-specific reproductive parameters have not yet been estimated. In Antarctic blue whales, estimates of age of sexual maturity for males, based on counts of laminae of ear plugs (Yochem and Leatherwood, 1985), ranged between 5 and 10 years old (Lockyer, 1981). These estimates have been supported by Trumble et al. (2013) who observed a peak in testosterone concentration, an indicator of sexual maturity, in the ear plug of one ENP blue whale when the animal was about 10 years old.

Hormones have been proven to be effective tools for understanding reproductive physiology in cetaceans (Atkinson and Yoshioka, 2007) and can be measured with commercially available and relatively inexpensive immunoassay kits. Sex steroids are a class of hormones synthesized from cholesterol, mainly in the gonads, and include compounds such as testosterone and progesterone (Norris, 2006). Because of their lipophilic nature, progesterone and testosterone can be detected and have been increasingly measured in blubber tissues of mysticetes (e.g., Atkinson et al., 2020; Carone et al., 2019; Cates et al., 2019; Kellar et al., 2013; Mansour et al., 2002). While not fully described for blue whales, published studies indicate hormone concentrations in blubber to reflect those in blood and be representative of relatively recent (hours to weeks) physiological events (Champagne et al., 2018, 2017; Kellar et al., 2013).

Testosterone is the main circulating androgen in mammals; secreted by the testis and the adrenal glands, it is the main androgen affecting spermatogenesis (Norris, 2006). Its action also influences behavior, development of both primary and secondary sexual characteristics, and the onset of sexual maturity (Atkinson and Yoshioka, 2007). In some mammals, testosterone and androgen concentrations regulate aggressive behavior, territoriality and social ranking in both males and females (Beehner et al., 2006; Bouissou, 1983; Bryan et al., 2014, 2013; Negro et al., 2010). In seasonally breeding cetaceans, testosterone is conventionally thought to show a cyclic trend, peaking before mating, then dropping after breeding has occurred (Shroeder and Keller, 1989). Annual cyclicity in testosterone concentrations has been observed in baleen plates of bowhead whales (Balaena mysticetus), right whales (Eubalaena glacialis) and possibly blue whales, although the sample sizes were limited (Hunt et al., 2018). In male humpback whales (Megaptera novaeangliae) testosterone showed seasonal trends with higher concentrations in blubber during the winter breeding months in Hawaii (Cates et al., 2019). However, Vu et al. (2015) also found unexpected high testosterone levels during fall, which is considered a shoulder season between summer feeding and winter breeding times for humpback whales in the North Pacific, suggesting physiological preparation for reproduction. A similar increase in testosterone concentrations during late summer, as the animals approach their breeding season, was observed in blubber of male fin whales (Balaenoptera physalus) from the GoC (Carone et al., 2019).

Increased androgen (including testosterone) concentrations have been reported in serum of female killer whales (Orcinus orca) (Robeck et al., 2017), bottlenose dolphins (Tursiops truncatus) (Steinman et al., 2016), and in blubber of humpback whales (Dalle Luche et al., 2020) in the late stages of gestation, suggesting these hormones may indicate proximity of the approaching parturition. No studies on profiles of blubber testosterone in blue whales were found in the published literature. Thus, any role of potentially elevated testosterone in female blue whales has yet to be elucidated, but likely would result from the down-regulation of the aromatase enzyme in the biosynthetic pathway from androgens to estrogen production (Norman and Henry, 2015).

In marine mammals, progesterone is secreted by corpora lutea in the ovaries during the estrous cycle and gestation, and is the predominant hormone responsible for sustaining pregnancy (Atkinson et al., 2017; Atkinson and Yoshioka, 2007). For example, progesterone in blubber has been applied as a biomarker for pregnancy in minke whales (Balaenoptera acurostrata) (Mansour et al., 2002), bowhead whales (Kellar et al., 2013), humpback whales (Clark et al., 2016; Pallin et al., 2018), fin whales (Carone et al., 2019) and blue whales (Atkinson et al., 2020).

Profiles of testosterone and progesterone in blubber of blue whales in relation to winter and summer grounds are not fully defined. Only one study has measured hormones in blubber of blue whales from the North Pacific (Atkinson et al., 2020). Specifically, progesterone was validated as a biomarker for pregnancy and preliminary baseline values for cortisol were established in blubber from ENP blue whales (Atkinson et al., 2020). Nonetheless, the individuals used were all sampled in the GoC, limiting the conclusion to a specific migration stage of the ENP population. The present study is meant to complement those data by including individuals from the same population but sampled in multiple seasons, thereby evaluating effects of seasonality on testosterone and progesterone profiles. The specific research questions were:

1. Does testosterone vary across age classes of blue whales, and between the two areas of sampling (GoC and USWC)?

2. Can progesterone concentrations be used to determine the probability of being pregnant for blue whales of unknown status?

3. Do progesterone concentrations in sexually mature non-lactating blue whales differ between the two areas of sampling (GoC and USWC)?

4. Can testosterone be detected and measured in blubber of female blue whales? If so, is there any relationship between testosterone and reproductive status?

2. Methods

2.1. Sample collection and sighting history

Blubber samples (n = 35 males and n = 34 females) were collected from both live (n = 66) and dead (n = 3) blue whales from two main migratory grounds: the GoC and the USWC (Fig. 1A and Fig. 1B). In the GoC (Latitude 24.33°N and 30.84°N; Longitude 114.09°W and 109.51°W; Fig. 1B), 51 blubber biopsies were collected using a 68 kg crossbow and biopsy darts of 7 mm diameter (Costa-Urrutia et al., 2013) in the southwest GoC by Centro Interdisciplinario de Ciencias Marinas (CICIMAR), between 2002 and 2016. All samples were collected between the months of January and May, frozen at sea, stored first in liquid nitrogen (Costa-Urrutia et al., 2013) and then in −80 °C freezer until extraction (Supplementary material, Table S1). Off the USWC (Latitude 32.77°N and 42.54°N; Longitude 124.62°W and 117.40°W; Fig. 1A), 15 biopsies were collected by Cascadia Research Collective (CRC) using crossbow and biopsy darts between 2005 and 2013, in the waters off California. All samples were collected between the months of June and November, frozen at sea, stored first in liquid nitrogen and then in −80 °C freezer at NOAA Southwest Fisheries Science Center (SWFSC) until extraction for hormone analysis (Supplementary material, Table S2). Additionally, blubber samples from three dead whales (1 female, 2 males) were obtained from NOAA SWFSC and Oregon State University and carcass status was classified based on information provided by the Marine Mammal Health and Stranding Response Program (MMHSRP). One male (HMSC15-11-01-Bm) was found stranded in Ophir, OR on November 3, 2015 and classified in advanced decomposition; the second male (LACMNH-DSJ2231) was reported floating in Long Beach Harbor, CA, on September 11, 2007, in severe decomposition (Berman-Kowalewski et al., 2010); the carcass of the female whale (TMMC-C-337-F) was found in an advanced state of decomposition in the San Mateo county, CA and examined on October 2, 2010 by The Marine Mammal Center. Samples from stranded animals were stored at −80 °C until extraction.

Fig. 1.

Study area. Blue whale blubber samples were collected from two known grounds for this population: (A) United States West Coast (USWC) and (B) the Gulf of California (GoC). Sampling locations for biopsies are indicated by circles, while those of stranded animals by triangles.

Since there was no overlap in months of sampling, the area of collection also reflected seasonality. However, to identify possible seasonal trends within each area, seasons were classified into bins of 2–3 months: winter (Jan-Feb), spring (Mar-May), summer (Jun-Aug) and fall (Sept-Nov). Sighting history information including sex, year of birth or first sighting, and reproductive status (i.e., presence of calves) were obtained from CRC and CICIMAR blue whale catalogues (Calambokidis et al., 1990; Gendron and Ugalde de la Cruz, 2012), as well as from stranding reports.

The comprehensive male dataset consisted of 35 individual whales, from biopsies collected in the GoC (n = 27) and the USWC (n = 6), and from stranded individuals (n = 2) (Table 1). Minimum age and age class were estimated for 21 males based on length of sighting history (LSH) or known age (i.e., with known year of birth) for live animals and body length for stranded whales. The youngest animal was sighted as a calf (< 1 year old), while the oldest had a minimum age of 27 years. For the two stranded male blue whales, the age class was assigned based on body length. Length at sexual maturity for blue whale is reported at 22.6–23.8 m (Ichihara, 1966; Mackintosh and Wheeler, 1929; Yochem and Leatherwood, 1985). Specifically, HMSC15-11-01-Bm was 21.3 m long and thus categorized as calf/juvenile (Stranding reports), and LACM-DSJ-2231 was reported to be 24.0 m long and thus categorized in the adult group (Berman-Kowalewski et al., 2010). Age classes were defined as follows: calf/juvenile (n = 3) males with known age of < 8 years (Lockyer, 1981) or body length < 23 m; adult (n = 19) males with known age or with LSH of at least 8 years or body length > 23 m; unknown (n = 13) males that could not be classified as adult or calf/juvenile (Table 1).

Table 1.

Testosterone dataset for male (n = 35) and female (n = 7) blue whales, categorized by sample type (biopsy = b, stranded = s), area (Gulf of California = GoC and United States West Coast = USWC), date of sampling, length of sighting history (LSH), age class (adult, calf/juveniles, unknown), testosterone concentrations (ng/g) and, for females, reproductive state (pregnant or presumed pregnant).

ID	SEX	SAMPLE TYPE	AREA	DATE OF SAMPLING	LSH	AGE CLASS	TESTOSTERONE ng/g	REPRODUCTIVE STATE (for females)
142	male	b	GoC	2/12/07	27	adult	0.09
144	male	b	GoC	3/15/16	20	adult	0.03
198	male	b	GoC	3/8/02	18	adult	0.2
240	male	b	GoC	2/14/02	4	calf/juvenile	0.4
513	male	b	GoC	3/11/09	5	unknown	0.06
817	male	b	GoC	2/13/12	<1	calf/juvenile	0.03
879	male	b	USWC	9/29/05	13	adult	1.02
18	male	b	GoC	2/13/08	19	adult	0.03
249	male	b	GoC	3/14/07	15	adult	0.04
311	male	b	GoC	2/12/08	8	adult	0.24
371	male	b	GoC	2/27/07	23	adult	0.02
374	male	b	GoC	2/5/02	1	unknown	0.22
473	male	b	GoC	4/22/06	3	unknown	0.09
487	male	b	GoC	3/13/05	21	adult	0.15
502	male	b	GoC	4/16/05	18	adult	0.05
547	male	b	GoC	3/25/06	0	unknown	0.11
550	male	b	GoC	3/24/06	12	adult	0.04
587	male	b	GoC	3/16/07	0	unknown	0.12
594	male	b	GoC	4/19/07	0	unknown	0.05
596	male	b	GoC	4/18/07	8	adult	0.08
598	male	b	GoC	4/16/07	0	unknown	0.06
672	male	b	GoC	3/5/08	16	adult	0.02
680	male	b	GoC	2/12/08	7	unknown	0.05
681	male	b	GoC	3/5/08	0	unknown	0.51
684	male	b	GoC	2/21/08	6	unknown	0.02
685	male	b	GoC	2/21/08	13	adult	0.02
829	male	b	GoC	3/22/12	0	unknown	0.02
831	male	b	GoC	3/24/12	5	unknown	0.02
110	male	b	USWC	9/27/05	18	adult	0.18
201	male	b	USWC	7/6/05	18	adult	0.62
210	male	b	USWC	8/8/05	18	adult	2.9
255	male	b	USWC	9/29/05	18	adult	0.89
2214	male	b	USWC	11/18/05	0	unknown	0.33
HMSC	male	s	USWC	11/3/15	NA	calf/juvenile	0.47
LACM	male	s	USWC	9/11/07	NA	adult	2.96
23	female	b	GoC	3/24/02	9	adult	1.06	presumed pregnant
65	female	b	GoC	3/21/07	13	adult	0.15	presumed pregnant
282	female	b	GoC	2/11/08	11	adult	0.12	presumed pregnant
379	female	b	GoC	3/6/02	4	adult	0.2	presumed pregnant
477	female	b	GoC	2/9/06	1	adult	0.92	pregnant
636	female	b	USWC	6/14/08	17	adult	0.26	presumed pregnant
C-337	female	s	USWC	10/2/10	NA	adult	5.4	pregnant

The female dataset included 34 (33 biopsies and 1 stranded) unique individuals, which were classified based on LSH or known age into the following categories: calf/juvenile (n = 6) females seen as calves or of known age < 8 years old (Valenzuela-Molina et al., 2018); adult (n = 28) females with LSH or known age of at least 8 years or females seen with a calf prior to, at the time of, or in the year after sampling. Adult females were further divided into reproductive groups: pregnant (n = 4) if seen accompanied by a calf the year after sampling; lactating (n = 8) if seen with a calf at the time of sampling; unknown (n = 16) if not seen accompanied by a calf in the year of or the year after sampling, or not sighted at all the year after sampling. Based on age of sexual maturity and calving rate (Atkinson et al., 2020; Sears et al., 2013; Valenzuela-Molina et al., 2018), females in the calf/juvenile and lactating groups were considered non-pregnant. The pregnant group included TMMC-C-337-F (C-337), a female that was found stranded with an aborted fetus and estimated to be approximately at 9-month gestational age. Reproductive states were assigned to samples from 34 individual female whales of which 24 were sampled in the GoC and 10 off the USWC (Table 2).

Table 2.

Estimated probability of being pregnant (Pp) for all female and male blue whales based on model-based clustering of progesterone concentrations (ng/g). Samples were collected between 2002 and 2016 in the Gulf of California (GoC) and in the United States West Coast (USWC). For each animal, it is reported sample type (biopsy = b, stranded = s), area and date of sampling, length of sighting history (LSH) and age class, progesterone concentrations (ng/g), reproductive state, probability of a whale to be pregnant (Pp and CI %), and for females, whether previously sighted with a calf and the year sighted after sampling. Whales of unknown status were considered presumed non-pregnant (pres np) if the upper confidence interval around the calculated Pp was < 50% and whales were considered presumed pregnant (pres p) if the lower confidence interval was > 50%. Unknown females that fell outside of these thresholds were categorized as undetermined.

ID	SEX	SAMPLE TYPE	AREA	DATE OF SAMPLING	LSH	AGE CLASS	PROGESTERONE ng/g	REPRODUCTIVE STATE	PROBABILITY OF A WHALE TO BE PREGNANT (Pp and CI %)	PREVIOUSLY SIGHTED WITH CALF?	YEAR OF RESIGHTING (if with calf)
23	female	b	GoC	3/24/02	9	adult	67	pres p	99.3 (97–100)	Yes	2004 (no calf)
39	female	b	USWC	8/6/11	26	adult	17.3	undetermined	8.9 (0.8–99)	No	NA
47	female	b	GoC	2/24/06	12	adult	3.3	pres np	<0.01 (2.5 × 10−13 – 0.9)	No	NA
50	female	b	GoC	2/10/08	14	adult	2.3	lactating	<0.1	Yes	NA
65	female	b	GoC	3/21/07	13	adult	58.1	pres p	98.8 (96–100)	No	2008 (no calf)
111	female	b	GoC	2/18/06	11	adult	1.2	lactating	<0.1	Yes	NA
119	female	b	GoC	3/21/08	17	adult	1.3	pres np	<0.01 (2.6 × 10−23 – 1.7 × 10−2)	Yes	2009 (no calf)
124	female	b	GoC	1/20/07	9	adult	3.3	pres np	<0.01 (2.5 × 10−13 – 0.9)	Yes	2008 (no calf)
134	female	b	GoC	3/28/11	15	adult	2.8	pres np	<0.01 (5.5 × 10−15 – 0.4)	Yes	NA
136	female	b	GoC	4/8/02	6	adult	2.9	lactating	<0.1	Yes	2003 (no calf)
141	female	b	GoC	3/27/08	12	adult	15.5	undetermined	4.1 (0.2–99)	No	2009 (no calf)
162	female	b	USWC	11/18/05	18	adult	5.5	pres np	<0.01 (9.5 × 10−9 – 11.1)	No	2010 (no calf)
232	female	b	USWC	9/9/05	17	adult	8.9	undetermined	0.03 (6.6 × 10−5 – 69)	No	2009 (no calf)
237	female	b	GoC	2/17/07	9	adult	1.9	lactating	<0.1	Yes	2009 (no calf)
266	female	b	GoC	4/18/07	8	adult	2	lactating	<0.1	Yes	2009 (no calf)
282	female	b	GoC	2/11/08	11	adult	45.6	pres p	96.9 (91–99)	Yes	2009 (no calf)
329	female	b	GoC	2/20/08	7	calf/juvenile	4.4	calf/juvenile	<0.1	No	2009 (no calf)
331	female	b	USWC	6/10/08	22	adult	83.1	pregnant	99.61	No	2009 (with calf)
338	female	b	GoC	3/11/12	11	adult	3.2	lactating	<0.1	Yes	NA
379	female	b	GoC	3/6/02	4	adult	184.5	pres p	99.9 (99–100)	Yes	2006 (no calf)
477	female	b	GoC	2/9/06	1	adult	111.8	pregnant	99.8	No	2007 with calf
492	female	b	GoC	3/23/12	17	adult	3	lactating	<0.1	Yes	NA
505	female	b	GoC	4/21/07	2	calf/juvenile	2.5	calf/juvenile	<0.1	No	2008 (no calf)
516	female	b	GoC	3/26/06	1	calf/juvenile	1.3	calf/juvenile	<0.1	No	NA
529	female	b	GoC	5/8/07	1	calf/juvenile	0.9	calf/juvenile	<0.1	No	2013
536	female	b	GoC	3/4/13	7	adult	46.1	pres p	97.0 (91–99)	Yes	NA
544	female	b	USWC	8/1/11	22	adult	11.3	undetermined	0.3 (2.9 × 10−3 – 92)	No	2012 (no calf)
589	female	b	GoC	5/9/07	0	adult	2.7	lactating	<0.1	Yes	NA
636	female	b	USWC	6/14/08	17	adult	54.9	pres p	98.5 (96–100)	No	2010 (no calf)
674	female	b	GoC	2/10/08	<1	calf/juvenile	2.2	calf/juvenile	<0.1	No	2011 (no calf)
1993	female	b	USWC	8/17/10	8	adult	0.6	pres np	<0.001 (5.2 × 10−33 –1.3 × 10−3)	No	2011 (no calf)
2095	female	b	USWC	7/29/11	7	calf/juvenile	0.2	calf/juvenile	<0.1	No	2012 (no calf)
2237	female	b	USWC	7/30/11	15	adult	34.5	pregnant	88.76	No	2012 (with calf)
C-337	female	s	USWC	10/2/10	NA	adult	96.32	pregnant	99.73	NA	NA
142	male	b	GoC	2/12/07	27	adult	1.24	adult	<0.001
144	male	b	GoC	3/15/16	20	adult	1.05	adult	<0.001
198	male	b	GoC	3/8/02	18	adult	1.17	adult	<0.001
240	male	b	GoC	2/14/02	4	calf/juvenile	1.11	calf/juvenile	<0.001
513	male	b	GoC	3/11/09	5	calf/juvenile	1.36	calf/juvenile	<0.001
817	male	b	GoC	2/13/12	<1	calf/juvenile	1.34	calf/juvenile	<0.001
879	male	b	USWC	9/29/05	13	adult	1.9	adult	<0.001

2.2. Steroid hormones extraction and measurement

Blubber samples ranging in weight between 0.02 and 0.30 g (0.10 ± 0.04 g) were placed into 12 × 75 mm borosilicate glass tubes and extracted with a method reported in Atkinson et al. (2020) modified from the methods by Mansour et al. (2002) and Kellar et al. (2006). Each sample was manually macerated in 500 μl ethanol using a Teflon tissue homogenizer and the resultant homogenized tissue in ethanol was centrifuged at 3,000 rpm for 20 min. The supernatant was poured off into a clean tube and this step was repeated. The supernatant from the second extraction was added to the supernatant from the first extraction and dried under forced air. 2 ml of ethanol:acetone (4:1) were added to the residue, vortexed, and centrifuged at 3,000 rpm for 15 min and supernatant dried under forced air. 1 ml diethyl ether was added to this residue, vortexed, centrifuged at 3,000 rpm for 15 min, transferred to clean glass tubes, and dried under forced air. For the final extraction 1 ml of acetonitrile was added to each residue and samples were vortexed before 1 ml of hexane was also added and vortexed. Samples were then centrifuged at 3,000 rpm for 15 min. Acetonitrile and hexane formed two immiscible layers with hexane on top. The bottom layer of acetonitrile was collected, placed in a clean tube and re-extracted with an additional 1 ml hexane. The hexane layer was removed, and acetonitrile was dried under forced air. Extracts were stored frozen at −20 °C until assayed. Before hormone analysis, extracts were rehydrated with a final extraction volume of 1 ml of methanol and aliquoted based on the dilution required for each assay. Each extract aliquot was dried under forced air before being diluted with assay buffer, prepared according to assay manufacturer’s protocols.

Two hormone pools of extracts were used for validating testosterone and progesterone assays for blubber from male and female blue whales using enzyme immunoassays (EIA, Arbor Assay Kit # K025 and K032, Ann Arbor, MI). The analytical validation consisted of testing for parallelism and accuracy using a pool of sample extracts, 6 males and 20 females for testosterone and progesterone, respectively. One pool of extracts from pregnant and one from non-pregnant females were used for validating testosterone in females. Parallelism was used to test if the assay antibody could reliably bind to the targeted hormone and to determine dilution at 50% binding. Accuracy was used to evaluate how well the measured concentrations corresponded to added concentrations of each hormone, thus excluding interference from other compounds in the extracts. Each pool was serially diluted (1:1, 1:2, 1:4, 1:8, 1:16, 1:32) and tested for parallelism with testosterone and progesterone standard curves. For the testosterone assay, a linear model was fitted to standard curve between 80% and 20% binding and to the dilutions of each pool. A Student’s t-test was used to statistically assess the difference in slope and lack of significance was considered evidence of parallelism. For the progesterone assay, parallelism was tested by fitting a four parameters logistic curve (4PLC) to the assay standards and dilutions of each pool using the R package “drc” (Ritz et al., 2015) and then comparing the slope parameter from the standard curve and each curve fitted to the pools of male and female extracts. The difference in slope was again assessed statistically using a Student’s t-test and lack of significance was considered evidence of parallelism. For both assays, the assay standards were spiked with an equal volume of sample pool for the accuracy test. The measured mass was then plotted against the added standard mass and tested for linearity. For testosterone, the dilution binding at 50% was measured at 1:1 for both males and females, while for the progesterone assay, 50% binding occurred at 1:4 and 1:10 for males and females, respectively. Samples for testosterone were run at 1:1 or concentrated and assayed at 2:1, while samples for progesterone were run at 1:10 dilution. Raw data obtained in pg/ml were corrected for dilution factor, extracted blubber mass, and expressed as ng/g.

Each assay was run with standard curve concentrations: for testosterone standards ranged from 40.96 pg/ml to 10,000 pg/ml, with 7 points on a 4PLC; for progesterone, standards ranged from 50 pg/ml to 3,200 pg/ml, with 7 points on a 4PLC. All samples, standards and zerostandards were assayed in duplicate.

For both assays, the intra-assay coefficient of variation was < 10%; any sample with a CV > 10% was re-diluted accordingly and re-assayed. To determine inter-assay variation, the CV (%) was calculated for three internal controls for the testosterone assays (n = 4) and for two internal controls for the progesterone assays (n = 5).

The limit of detection (LOD) was the minimal detectable concentration of the assays, and it usually corresponded to a response (a percent binding in this case) between 90 and 95% (Rodbard, 1974). While there is not a universal method to determine LOD, it can be calculated as the mean blank response plus three standard deviations (Croghan and Egeghy, 2003). In the present study, the LOD was calculated as the mean plus 1 standard deviation of the absorbance of the zero standard and the value interpolated from the standard curve. This choice is justified by an adequately conservative estimate and by consistency with works performed within the same laboratory (i.e., intra-laboratory quality control/quality assurance). The estimated LOD for the testosterone and progesterone assays were 5.5 pg/ml and 25.0 pg/ml, respectively. If samples with concentrations below the LOD had at least 250 μl of extract volume remaining, they were concentrated and re-assayed. If the volume of extracts was not enough, values below the LOD were substituted with a proxy value corresponding to the LOD/√2 (Croghan and Egeghy, 2003). According to Croghan and Egeghy (2003), substituted values should not exceed 25% of the total sample size for each dataset. In this study, the proxy value was applied to 9 out of 42 (21%) values for testosterone and to 5 out of 41 (12%) values for progesterone. The proxy value (LOD/√2) for testosterone was 3.9 pg/ml (0.39 pg/g for mean blubber weight) and for progesterone 17.7 pg/ml (1.7 pg/g for mean blubber weight).

2.3. Statistical analysis

Testosterone concentrations were measured in 35 males (33 biopsies and 2 stranded). Testosterone concentrations were log-transformed and tested for normality using a Shapiro Wilk test (p > 0.05) and for homogeneity of variances using a Bartlett test (p > 0.05). Untransformed concentrations were graphed by day of the year to assess seasonal changes (Fig. 2). However, the effects of season and area were difficult to separate because all animals were sampled in the GoC were sampled between February and April and the USWC samples were collected between June and November. No samples from males used in this study were collected in the months of December, January and May. Therefore, samples from the two locations were analyzed separately using generalized linear models (GLM, R Core Team, 2020) with testosterone concentrations as a response to the following categorical factors: age class (adult, calf/juvenile or unknown), season (winter, spring summer or fall) and sample type (stranded or biopsy) in all possible combinations, using the dredge function (Bartòn, 2019) for automated model selection. This GLM analysis of mean testosterone concentrations was performed with and without the stranded samples, to ensure that the results were not biased by the latter. The most parsimonious model was identified based on the lowest Akaike’s Information Criterion with correction for small sample size (AICc), and the Akaike weights (w_a) were computed to assess the relative weight of evidence for different models. The difference in concentrations between the two migratory grounds, only using adult males and no stranded whale samples, was tested with a Student’s t-test. Untransformed concentrations were graphed by location (Fig. 3). Testosterone concentrations were further measured in 6 live and one stranded female whales (Table 1).

Fig. 2.

Testosterone concentrations (ng/g) in individual male blue whales sampled from February to November by area, Gulf of California (GoC) and United States West Coast (USWC). Shapes of points represent sample type: biopsy (open squares) and blubber from stranded animals (black squares). Season and sampling locations are confounded; hence their effects cannot be differentiated.

Fig. 3.

Testosterone concentrations (ng/g) measured in biopsies of adult male blue whales sampled in the two areas, Gulf of California (GoC n = 13) and United West Coast (USWC n = 5). The mean values are indicated by the stars. Boxplots denote medians (grey bar), lower and upper quartiles (boxes) and whiskers extend to 1.5 times the interquartile range. Outlying points are plotted individually (gray dots). In the GoC, blue whales were sampled in winter/spring (Feb-Apr), and off the USWC blue whales were sampled during summer/fall (Jun-Nov).

Progesterone concentrations were measured in biopsies from 33 females and from 7 males, and in blubber from the one stranded female (C-337) (Table 2), transformed on a natural-logarithmic scale for analyses and then back-transformed for graphical representation (Figs. 4–6). First, only concentrations from whales of known reproductive status (n = 25, including one stranded female) were analyzed using an ANOVA test followed by a post-hoc Tukey test. This analysis was repeated without the stranded female to ensure that the results were not biased by this sample. Second, progesterone concentrations from all females (n = 34, including the stranded female sample), of both known and unknown reproductive status were analyzed to estimate the probability of pregnancy (Pp) as a function of progesterone concentration. A simple logistic regression of the known pregnant (=1) and known non-pregnant (=0) females on progesterone concentration was initially fit to the data. However, the small number of samples and non-overlapping progesterone concentrations between pregnant and non-pregnant females led to convergence issues and confidence bands that ranged from 0 to 1 at all concentrations. Therefore, as an alternative approach, progesterone values from pregnant, non-pregnant, and unknown females were used in the analysis to boost sample sizes and to avoid biases due to including only animals with the lowest and the highest progesterone values. Exploratory analysis showed natural log-transformed progesterone from all females to have an approximate bimodal distribution, likely reflecting two clusters of non-pregnant and pregnant animals, respectively. Therefore, a mixture of two normal distributions was fitted to log-transformed progesterone values using the Expectation-Maximization (EM) algorithm in the R package “mixtools” (Benaglia et al., 2009). The EM algorithm is an iterative method used to find maximum likelihood values of the means and standard deviations of two normal probability density functions that best approximate the bimodal distribution of progesterone concentrations (Benaglia et al., 2009). The Pp at a given progesterone concentration was then calculated as the ratio of the probability density for the presumed pregnant group to the sum of the two probability densities, assuming that the normal distribution with the larger mean corresponds to the distribution of progesterone concentrations for pregnant females. To quantify uncertainty in the estimated probabilities, a modified bootstrap approach, similar to the methodology used in Pallin et al. (2018) was used as follows: first, progesterone concentrations were re-sampled with replacement 10,000 times, then the same approach for estimating probability of pregnancy described above was applied to each bootstrap sample. Some bootstrap samples resulted in biologically unrealistic, dome-shaped probabilities that implied a decrease in the probability of pregnancy at the highest progesterone values. These unrealistic functional forms resulted from long-tailed normal distributions fit to the presumed non-pregnant females, which implied a large probability of high progesterone values in non-pregnant females that were not consistent with the data. Therefore, only bootstrap samples resulting in probabilities that asymptotically approached 1 at high progesterone concentrations (45% of all bootstrap samples) were retained. Finally, a 95% confidence band for the estimated probability of pregnancy was constructed based on the 2.5th and 97.5th percentiles of the bootstrapped probabilities at each progesterone concentration. To assign a reasonable reproductive status to unknown whales, asymmetrical intervals of the confidence bands were used as thresholds and, specifically, whales were considered presumed non-pregnant (pres np) if the upper confidence interval around the calculated Pp was less than 50% and whales were considered presumed pregnant (pres p) if the lower confidence interval was higher than 50%. These thresholds corresponded to progesterone concentrations lower than 8.0 ng/g for pres np and higher than 30.0 ng/g for pres p. Unknown females that fell outside of these thresholds (progesterone concentrations between 8.0 and 30.0 ng/g) were categorized as undetermined. Note that this approach uses weighted probabilities based on the relative proportions of the two groups in the population, assuming that the sampled females were a random subset of all females in the population. A similar approach was used to estimate reproductive status of fin whales (Carone et al., 2019). The model for estimating probability of pregnancy was also reanalyzed excluding the sample from the stranded female. Finally, a t-test was used to compare progesterone concentrations from females sampled in the GoC (n = 24) and off the USWC (n = 10). All statistical and graphical analyses were conducted using the software R v. 3.5.2 (R Core Team, 2020).

Fig. 4.

Progesterone concentrations (ng/g) for female blue whales grouped by reproductive status, and male blue whales grouped by age class. Based on the 95% confidence intervals around the probability of being pregnant (Pp), the group of female whales of unknown reproductive status was divided in presumed non-pregnant (pres np), if the upper confidence interval around the calculated Pp was less than 50% and presumed pregnant (pres p) if the lower confidence interval was higher than 50%. Unknown females that fell outside of these thresholds were categorized as undetermined. Individual whales of known reproductive status/age class are represented by black circles, adult female whales that were not sighted the year after sampling are represented by black triangles, and those sighted unaccompanied by a calf by black squares.

Fig. 6.

Progesterone concentrations (ng/g) in individual female blue whales sampled from January to November by area, Gulf of California (GoC) and United States West Coast (USWC). Shapes of points represent reproductive status assigned using life history data and Pp: calf/juv (black circles), lactating (stars), pres np (diamonds) pregnant (black triangles), pres p (white triangles) and undetermined (crossed squares).

3. Results

3.1. Validation

The testosterone assay validated in males and pregnant females, with serial dilutions of pooled samples exhibiting parallel displacement to the standard curve (p = 0.1; R² = 0.98 and p = 0.9; R² = 0.95, respectively) and the spiked standards showing linear relationships with the measured standards (y = 0.8x-25.9; R² = 0.99 and y = 0.8x+116.5; R² = 0.98, respectively). The progesterone assay validated for both males and females, with serial dilutions of pooled samples displacing parallel to the standard curve (p = 0.2 and p = 0.7, respectively) and the spiked standards showing linear relationships with the measured standards (y = 0.9x-32.4; R² = 0.99 and y = 0.9x-62.8; R² = 0.99, respectively).

For the testosterone assay, the CVs (%) for the three internal controls were 4.7%, 8.2% and 7.5%, whereas for progesterone inter-assay variation, the CVs (%) for the two internal controls were 8.7% and 10.1%, respectively.

3.2. Males

3.2.1. Testosterone

Testosterone concentrations ranged between 0.02 and 2.96 ng/g, with the highest concentration measured in a stranded adult male. Because of the confounding effect of sampling area and season (Fig. 2), samples from USWC and GoC were analyzed separately. Modeling testosterone concentrations within each region did not show significant differences between seasons or between age classes. For the GoC subset, the most parsimonious model identified was the intercept-only model (AICc = −34.2, w_a = 0.63), the second best model included season only (AICc = −31.8, w_a = 0.19), and the third included age class only (AICc = −31.4, w_a = 0.15). However, the models with season and age as explanatory factors were not statistically significant (p = 0.7 and p = 0.1, respectively). For the USWC, the intercept-only model had the lowest AICc with strong support for being the best model (AICc = 29.8, w_a = 0.85). The second best model included season (AICc = 34.6, w_a = 0.08), but this factor was not significant in the model (p = 0.4). Given that the highest testosterone concentration was measured in a stranded animal and could potentially be a result of natural or anthropogenic events that occurred prior to death or decomposition, the GLM model for USWC males was rerun excluding the two stranded samples. Results did not change: the intercept-only model was still the most parsimonious model (AICc = 23.8, w_a = 0.97), followed by the model including season as explanatory factor (AICc = 31.1, w_a = 0.03). Similarly, season was not a significant factor for testosterone concentrations in males sampled off the USWC (p = 0.4).

When mean testosterone concentrations were compared between biopsies of adult males from the two grounds, significantly higher log-transformed testosterone concentrations were found in males sampled off the USWC than those for whales from the GoC (t = −5.5, df = 15, p < 0.001) (Fig. 3).

3.2.2. Progesterone

Progesterone concentrations were measured in biopsies from eight males (calf/juvenile (n = 3) and adult (n = 5)) and ranged between 1.05 and 1.90 ng/g. Differences in hormone concentrations were significant between males of either age class and pregnant females (p < 0.001) but not between males and non-pregnant females, or between male calf/juvenile and adult (p > 0.05) (Fig. 4).

3.3. Females

3.3.1. Testosterone

The testosterone assay validated only for pregnant females and hormone concentrations were measured in seven individuals, two of which were confirmed pregnant and five were presumed pregnant (Table 1). The pregnant group included the stranded female (C-337), who had the highest testosterone concentration of all of the females (Table 1). Testosterone concentrations averaged 1.2 ng/g (range: 0.1 – 5.4 ng/g), and 0.5 ng/g (range: 0.1 – 1.6 ng/g) when the stranded female was excluded. Two animals were sampled off the USWC and had a mean (range) testosterone concentration of 2.8 ng/g (range: 0.3 – 5.4 ng/g), and five in the GoC and had a mean of 0.5 ng/g (range: 0.1 – 1.1 ng/g).

3.3.2. Progesterone

Progesterone concentrations ranged between 0.2 and 184.5 ng/g. Whales in the pregnant group (n = 4) had a mean progesterone concentration of 81.4 ng/g with a range from 34.5 to 111.8 ng/g that did not overlap with any of the non-pregnant groups. Whales in the lactating group (n = 8) averaged 2.4 ng/g (range: 1.2 – 3.2 ng/g) and whales in the calf/juvenile group (n = 6) averaged 1.9 ng/g (range: 0.2 – 4.4 ng/g). Log-transformed progesterone concentrations were significantly different among reproductive status (ANOVA F = 33.3, df = 4, p < 0.001) in whales of known status. Pregnant whales had significantly higher progesterone concentrations than lactating (p < 0.001) and calf/juvenile whales (p < 0.001). Progesterone concentrations were not statistically different between lactating and calf/juvenile whales (p = 0.5). When the pregnant stranded female was removed from the analysis, results did not change: pregnant females had significantly higher progesterone concentrations than calf/juvenile females (p < 0.001), calf/juvenile males (p < 0.001), adult males (p < 0.001), and lactating females (p < 0.001).

There was a clear bimodal distribution of progesterone concentrations best presented by a mixture of two normal distributions. The first mode represented a cluster of animals with lower progesterone concentrations (mean (95% CI): 2.7 (0.4 – 19.3) ng/g) and included all animals known to be non-pregnant. The second distribution represented a cluster with higher progesterone concentrations (mean (95% CI): 68.7 (25.9 – 182.0) ng/g) and included all known pregnant animals. Therefore, it was assumed that these distributions reflected the range of progesterone concentrations of non-pregnant and pregnant females, respectively. There was a little overlap (1.6%) between the two distributions. Based on the models, 50% probability of pregnancy corresponded to a progesterone concentration of 24.2 ng/g. All known pregnant females had Pp higher than 85%, while whales in the calf/juvenile and lactating groups had Pp values lower than 0.1%. The estimated probability of being pregnant for the unknown group ranged from < 0.1% to 99.3% (Table 2). Uncertainty in the estimated Pp was high at intermediate progesterone concentrations between about 8.0 and 30.0 ng/g as indicated by wide bootstrap confidence bands (Fig. 5), and this range included only animals of unknown reproductive status (n = 4). The estimated Pp was well estimated at higher progesterone concentrations. For example, Pp was estimated to be 98.5% (95% CI: 95 – 100%) at a progesterone concentration of 55.0 ng/g.

Fig. 5.

Probability of being pregnant based on progesterone concentrations for all females, with 95% confidence band calculated using a bootstrapping approach. Dashed blue lines indicate the asymmetrical threshold used to assigned reproductive status to unknown (pres p, pres np and undetermined) whales, Shape of points indicate whales confirmed non-pregnant (calf/juvenile and lactating; empty circle), pregnant (black triangle) and unknown (pres p, pres np and undetermined, stars).

Whales of unknown status were classified as pres np (n = 6) if they had progesterone concentrations lower than 8.0 ng/g, pres p (n = 6) if they had progesterone concentrations higher than 30.0 ng/g and undetermined (n = 4) if their progesterone concentrations were between 8.0 and 30.0 ng/g (Table 2). The pres p group included six whales with progesterone concentrations high enough to suggest they could have been pregnant at the time of sampling, but with no evidence to biologically validate their assigned status, since two were sighted the year after sampling, not accompanied by a calf, while three were sighted two or more years later and one was not resighted (Table 2). Similarly, among the six whales in the pres np group, their assigned reproductive status found biological support for three of them that were sighted the year after sampling without calves. The remaining three were not resighted the year after sampling. Of the four whales categorized as undetermined, three were resighted not accompanied by calf, while one was not resighted (Fig. 4).

The thresholds in progesterone concentrations to assign reproductive status to unknown whales did not vary when the stranded whale was removed from the bootstrap analysis: 7.0 ng/g for pres np and 29.9 ng/g for pres p, instead of 8.0 ng/g and 30.0 ng/g, respectively, and 50% probability of pregnancy corresponded to a progesterone concentration of 22.0 ng/g (instead of 24.2 ng/g).

No significant difference in progesterone concentrations was found between females sampled in the GoC (n = 24) and off the USWC (n = 10; t = −0.9; df = 32, p = 0.4; Fig. 6), with or without the stranded female sample (p = 0.6).

4. Discussion

The documented trends in reproductive hormone profiles in male and female blue whales, sampled across habitat and seasons in the eastern North Pacific Ocean, provide preliminary but unique information on the reproductive cycle of the species. EIA techniques were used to measure testosterone and progesterone in blubber of 69 individual blue whales (35 males and 34 females), sampled over 15 years from two ecologically important grounds for this population, the summer grounds along the USWC and winter grounds in the GoC. Increasing testosterone concentrations in adult males sampled in late summer/fall indicate spermatogenesis is occurring and suggest that the onset of breeding in this species may start when the animals are still present off the USWC. This is in agreement with seasonal testosterone trends previously observed in other species of odontocetes (Boggs et al., 2019; Desportes et al., 2003; Richard et al., 2017) and of baleen whales (Carone et al., 2019; Cates et al., 2019; Hunt et al., 2018; Kjeld et al., 2006; Vu et al., 2015). Progesterone was confirmed as an indicator of pregnancy for blue whales (Atkinson et al., 2020; Valenzuela-Molina et al., 2018) and the present study presents the application of hormone concentrations to predict the probability of being pregnant, aiding in definition of reproductive parameters.

The dataset analyzed in the present study included three samples from stranded whales: two males and one female. Blubber of stranded cetaceans is widely used for endocrine studies to better understand physiological stress, reproduction and relative steroid concentrations (Dalle Luche et al., 2019; Kellar et al., 2015; Mello et al., 2017; Mingramm et al., 2020). However, their concentrations may be altered by tissue degradation and decomposition (Mello et al., 2017; Mingramm, 2018; Trana et al., 2015), or natural or anthropogenic events that occurred prior to death (Atkinson et al., 2015); thus, values from these samples should be interpreted with caution. To address this point, all the statistical analyses that included one or more of these stranded animals were also carried out without them. The results with and without samples from stranded animals are discussed for each analysis, but generally there was little or no difference. Likely because of the limited number of stranded animals, the present study did not find any evidence that would suggest tissue decomposition or degradation post-mortem to affect hormone concentrations. Nevertheless, thorough analytical studies are needed to address the feasibility of stranded samples for reproductive endocrine analysis and the robustness of comparison with live animals.

Testosterone concentrations in blubber tissue of male blue whales revealed significant regional differences but no temporal trends or differences among age classes within each region. Significantly higher testosterone concentrations in adult male blue whales sampled between June and November off the USWC compared to those sampled between February and April in the GoC (Fig. 3) were unexpected in this area, since the GoC is considered part of their reproductive grounds (Costa-Urrutia et al., 2013; Gendron, 2002; Valenzuela-Molina et al., 2018). Nevertheless, few courtship events have been previously observed (Gendron, 2002), likely because courtship and probable mating events are of short time duration (Gendron data unpublished). In the present study, elevated testosterone concentrations were measured in adult males sampled between June and November, in their feeding grounds, suggesting the males were physiologically preparing for the upcoming breeding season. Conversely, the low testosterone concentrations in adult males sampled between February and April in the GoC suggest that mating has likely occurred by late winter/early spring. These findings are in agreement with data from whaling in the Southern Ocean, where spermatozoa proliferation was observed to increase in blue whales two months prior the presumed mating season, during the southern winter (Mackintosh and Wheeler, 1929). Physiological preparation for mating has also been documented based on testosterone concentrations in serum for North Atlantic fin whales and North Atlantic minke whales during the feeding season (Kjeld et al., 2006; 2004), and in blubber for humpback whales from the Pacific Ocean (Vu et al., 2015). While male competition for access to breeding females is a known behavior in cetaceans (Connor et al., 2000) and elevated testosterone concentrations have been measured in blubber from the GoC fin whales while a female with high progesterone was present (Carone et al., 2019), no relationship between behavior and hormone profiles has been published for blue whales. Ongoing and future efforts for biopsy collection from blue whales, specifically targeting the monthly interval from November to February would greatly refine testosterone profiles and potentially identify a peak for testosterone concentrations as in other baleen whales (Carone et al., 2019; Cates et al., 2019; Hunt et al., 2018), thus leading to a better delineation of mating grounds for this population.

Additional uncertainty in hormone concentrations may derive from the choice of tissue. The lag-time between the hormone secretion in blood and the accumulation in blubber remains unclear for mysticetes. Kellar et al. (2013) suggested a lag-time of weeks to months in progesterone concentrations measured in serum and blubber of pregnant and non-pregnant bowhead whales. In odontocetes, studies on blubber of bottlenose dolphins suggest a lag time between one to two hours for steroid hormones (Champagne et al., 2018). The rate of perfusion from blood to blubber is unknown for blue whales, and potentially differs from that of odontocetes, therefore lag-time is difficult to estimate for this species. Regardless, hormone concentrations in blubber are likely to be reflective of somewhat recent (days to weeks) physiological events.

No evidence was found that testosterone concentrations varied across age classes in males, but the sample size was limited, as only three individuals were categorized as calf/juvenile based on known ages of those whales. This part of the analysis included two stranded males, one categorized as juvenile and one as adult. When these two samples were excluded, the results of the generalized linear model and of the follow-up tests did not change. The categorization based on age class was very conservative as individuals with LSH < 8 years but without a confirmed year of birth were placed in the unknown group. With the increased use of unoccupied aerial vehicles and aerial photogrammetry, future studies may be able to better determine age class based on body length, thus providing fine-tuned information on any age-related changes in testosterone concentrations.

Male blue whales had progesterone concentrations comparable to non-pregnant females and significantly lower than those measured in pregnant whales (Fig. 4). Furthermore, the lack of a significant difference in progesterone concentrations between adult and calf/juvenile males suggests that the accumulation of this hormone was not affected by age. In gray whales (Eschrichtius robustus), progesterone concentrations have been found to be elevated in calves (animal < 1 year of age) compared to adult males (Melica, 2020), likely as a result of maternal transfer. Due to the small sample size and age of the young whales, evidence of maternal transfer of hormones could not be confirmed in male blue whales.

The present study complemented what was previously published on progesterone concentrations in blue whales (Atkinson et al., 2020; Valenzuela-Molina et al., 2018) with the addition of samples collected from another significant migratory area for this population, the USWC, providing more complete geographic coverage. The analysis presented here further developed a model to calculate the probability of being pregnant based on progesterone concentrations; specifically, a mixture model was used to identify two clusters of progesterone concentrations and to estimate the probability of pregnancy from the measured progesterone concentrations (Fig. 5). The estimated probabilities and the uncertainty around them were used to assign the most probable reproductive status to whales in the unknown group (Table 2), except when the 95% CI included Pp = 50%. The small group of whales in this group were considered of undetermined reproductive state. A similar approach, based on assignment to high (pregnant) or low (non-pregnant) progesterone clusters, was applied to fin whales from the GoC (Carone et al., 2019) and humpback whales (Pallin et al., 2018). For blue whales, a previous study utilized the progesterone concentration from the upper range of lactating females (5.8 ng/g) as a threshold to differentiate non-pregnant versus pregnant whales and calculated pregnancy rate (Atkinson et al., 2020). In the present study, only whales with < 8.0 ng/g progesterone were considered pres np (n = 6), and only whales with a minimum of 30 ng/g progesterone concentrations were considered pres p (n = 6), due to the high uncertainty in the estimated Pp at intermediate progesterone concentrations (Fig. 5). Of these unknown whales with assigned reproductive status, only five were resighted the year after sampling; the low resighting rates for both areas are consistent with previous studies (Atkinson et al., 2020; Valenzuela-Molina et al., 2018). Nevertheless, resighting history supported the model presented in this study for only three unknown adult females, that were considered presumed non-pregnant and were sighted without a calf the year after sampling. When the stranded female was removed from the model to calculate probability of pregnancy, the calculated thresholds (7.0 ng/g and 29.9 ng/g, instead of 8.0 ng/g and 30.0 ng/g) did not alter reproductive status assignment for any of the unknown females (6 presumed pregnant, 6 presumed non-pregnant and 4 undetermined).

The lack of calf in the year after sampling for two whales classified as presumed pregnant (#65 and 282) (Table 2) questioned whether elevated progesterone concentrations may be explained by alternative hypotheses. For instance, fluctuations of progesterone were detected in plasma of mature false killer whales (Pseudorca crassidens) and other odontocetes in correspondence to spontaneous ovarian activity (Atkinson et al., 1999; Atkinson and Yoshioka, 2007). Whales with elevated concentrations of progesterone could potentially be experiencing ovulation. The validation and measurement of 17β-estradiol for this species and tissue may help future studies to ascertain ovulatory events. Alternatively, elevated progesterone might be an indication of pseudo-pregnancy, in which a corpus luteum is retained for a longer than normal time despite the lack of conception (Atkinson and Yoshioka, 2007). Recurring pseudopregnancy has been observed in many odontocetes species and may be related to age, lack of access to males or multiple estrous cycles (Robeck et al., 2018), but there is no evidence of it occurring in baleen whales. An additional explanation for whales with high progesterone but not seen with a calf the following year, is potential reproductive failure, through either the loss of a calf post-parturition or the loss of the fetus, also referred to as spontaneous abortion (Atkinson et al., 1999; West et al., 2000). One female blue whale in the present study (#65) is of particular concern, as she has been seen every two to three years in the GoC since 1994, but was never sighted with a calf. Valenzuela-Molina et al. (2018) reported that progesterone concentrations in feces of the same whale decreased substantially between replicates collected three days apart. This could suggest pregnancy failure for this individual, an event that would have little chance of being detected visually. Finally, it is possible that calves are already weaned by the time they reach the USWC, as Sears et al. (2013) reported females accompanied by calves that were sighted in the GoC, that were seen unaccompanied along the USWC during the same year. Nevertheless, this is an unlikely explanation for the presumed pregnant whales resighted in the GoC, as it would suggest a weaning time of 2 – 4 months, instead of 6 – 8 months known for this species (Sears and Perrin, 2018).

Progesterone concentrations may fluctuate among stages of pregnancy, and this has been documented in many cetacean species. Rapid increases in progesterone concentrations in serum at the beginning of pregnancy have been observed in many cetacean species (Robeck et al., 2018). Similarly, a positive correlation was found between progesterone concentrations in blubber and fetal girth measurements in minke whales killed early in the pregnancy (Mansour et al., 2002), suggesting this hormone increases at the beginning of gestation. In humpback whales, blubber progesterone was found to be stable in the early/middle stages of pregnancy (Clark et al., 2016), followed by a rapid decline during the later stages of gestation prior to parturition (Dalle Luche et al., 2020). Because uncertainty around the estimated Pp was higher at intermediate levels of progesterone, whales with concentration between 8.0 and 30.0 ng/g were considered of undetermined reproductive state. It is possible these whales had recently given birth and potentially lost the calf, or could be in the very early stages of pregnancy or ovulation. Further monitoring efforts and additional biopsy samples are needed to better model reproductive parameters, and pregnancy and parturition rate.

The results and analyses reported in the present study provide benchmark information on progesterone profiles in blue whales, however, it is important to recognize that the sample sizes used in the present study are still relatively small, therefore mistakes in the assignment to reproductive status are possible. Additional samples are necessary to refine the proposed analysis.

This study did not find any temporal trend, nor regional difference in progesterone concentrations, and females with high Pp or confirmed pregnant were sampled both along the USWC and the GoC, possibly indicating stable progesterone concentrations throughout the duration of gestation (Fig. 6). Lack of a temporal pattern in progesterone concentrations has been found also in humpback whales, sampled off California between May and December (Clark et al., 2016). However, the sample size was limited in the present study, and additional samples are necessary to better confirm this lack of a temporal pattern.

Progesterone in blubber has been demonstrated to be a valid and useful biomarker for pregnancy in blue whales, however it appears not to be a useful tool for determining lactation (Fig. 4). Lactating whales had concentrations of progesterone significantly lower than pregnant whales, but not statistically different from whales in the calf/juvenile and presumed non-pregnant groups. These results further confirm previous studies that did not find any difference in progesterone concentrations in blubber and in feces among groups of non-pregnant blue whales (Atkinson et al., 2020; Valenzuela-Molina et al., 2018). In North Atlantic right whales, lactating animals were found to have high fecal estrogen and androgen concentrations compared to resting and juvenile females (Rolland et al., 2005), suggesting these compounds may be better indicators of lactation. Estrogen is yet to be validated in blubber of blue whales, and future studies should focus on validation of both estrogen and testosterone in non-pregnant female blubber, possibly using more sensitive assays, such as radioimmunoassays.

One whale (#134) in the presumed non-pregnant group was sighted the year before sampling with a calf and was potentially resting. Calving intervals for blue whales are estimated to be between 2 and 3 years (Atkinson et al., 2020; Sears et al., 2013). However, whales analyzed in the present study were previosly sighted with calf at intervals ranging from two to ten years. This may indicate a large variability in calving intervals among individuals (Gendron, unpublished data) or the preference of other birthing grounds outside the survey area or season. Most of the females in the unknown group of blue whales had a LSH longer than 10 years, with two longer than 20 years, however, no evidence of senescence has been reported for blue whales.

The present study is the first to publish validations for androgens in the blubber of female blue whales. Testosterone concentrations were measurable only in pregnant whales, while the lack of parallel displacement of the serial dilutions of the pool of non-pregnant whales in the testosterone assay indicated that the concentrations of this hormone were not high enough to be reliably detected by the assay in that group of whales. These results suggest that testosterone biosynthesis or metabolism is likely altered during gestation. However, the limited sample size did not allow proper or confident detection of changes over time in pregnant or presumed pregnant whales. The highest testosterone concentration was measured in a female blue whale that stranded off the coast of Oregon at the beginning of October and was found with a male fetus of 5.3 m. Considering that the size of a blue whale at birth is estimated to be between 7.0 and 8.0 m (Mizroch et al., 1984; Reidenberg and Laitman, 2008; Roston et al., 2013), and using the equation developed by Roston et al. (2013), the fetus was about 37 weeks or about 9 months prenatal age. Assuming a gestation period of 11 to 12 months (Lockyer, 1981), the fetus would have been born around December or January. This coincides with the estimated birth period from late fall to early winter (Brueggeman et al., 1985; Tomlin, 1967), indicating that this sample likely reflected a mid to late stage of pregnancy. The other confirmed pregnant whale (#477) was sampled in the GoC in the month of February and resighted with a calf the following year. Testosterone concentrations in this whale were five times lower than the above, potentially reflecting early gestation, as observed in other species (Robeck et al., 2017; Steinman et al., 2016). In female odontocetes, testosterone was shown to be an indicator of different gestation stages and potentially to be related to fetal health. Specifically, testosterone has been measured in serum of bottlenose dolphins and killer whales and exhibited increased concentrations from mid to late pregnancy (Robeck et al., 2017; Steinman et al., 2016). In baleen whales, androgens were found to be significantly higher in fecal samples from pregnant than from non-pregnant North Atlantic right whales, but the longitudinal profile across pregnancy is unknown (Rolland et al., 2005). Dalle Luche et al. (2020) suggested that androstenedione in blubber may be a better biomarker for near-term pregnancy in humpback whales. Androstenedione is a precursor of testosterone and given that in the present study the highest testosterone concentration was measured in the stranded female, it is possible these levels are the results of steroid metabolism. Future studies should measure multiple androgens in both non-pregnant and pregnant females and in the latter, in animals at different gestational stages.

Elevated steroid concentrations in a stranded animal could also be an effect of decomposition, or may simply reflect natural or anthropogenic events or stressors that occurred prior to death (Atkinson et al., 2015). While elevated progesterone concentrations in the stranded female (C-337) are justified by the obvious reproductive status of this whale (i.e., presence of an aborted fetus), elevated testosterone concentrations in the adult stranded female could be the results of yet to be described metabolic effects from blubber metabolism prior to death or post mortem decomposition. Mello et al. (2017) reported increasing testosterone concentrations in response to days from death in one humpback whale, however, the difference was not statistically significant. Because of this data point and the overall sample size, results on testosterone in pregnant females should be considered preliminary.

Due to the complexity in sampling this species, some blubber samples used in the present study weighed less than 0.15 g, the minimum size recommended by Kershaw et al. (2017). While multiple published studies have used blubber weights of less than 0.15 g of tissue (Atkinson et al., 2020; Carone et al., 2019; Cates et al., 2019; Mingramm et al., 2020), results for such samples could be inaccurate. To further assess the suitability of small samples for endocrine analysis, there is a need for more analytical studies (species and hormone specific) tailored to understand the influence of extracted blubber weight on hormone concentrations.

Concurrent analysis of progesterone and testosterone with a bigger sample size may help to better understand the longitudinal profile of sex steroids during pregnancy and aid the development of more accurate reproductive parameters. Reproductive physiology is very challenging to study in free-ranging mammals the size of blue whales. Collection of blubber is a minimally invasive technique and analyses on this tissue have been proven valid and valuable to provide information on reproductive physiology in many cetacean species (e.g., Atkinson et al., 2020; Carone et al., 2019; Cates et al., 2019; Clark et al., 2016; Kellar et al., 2013; Mansour et al., 2002; Pallin et al., 2018; Trego et al., 2013; Vu et al., 2015). The present study represents an important milestone in understanding reproductive physiology of blue whales, as well as providing insights on their reproductive timing. In addition to confirming progesterone as an indicator of pregnancy, this work provides important insights on the temporal and spatial patterns of the reproductive cycle of this population, posing new questions regarding the mating season and habitat for this endangered population. This knowledge is fundamental to developing more accurate reproductive parameters, as well as to predicting how change in environmental conditions may affect the population dynamics and the ecology of this species.