Tooth collection and dentine sampling

Teeth were taken from H. ampullatus killed by Norwegian whalers in the waters off northern Iceland in 1967 and northern Labrador in 1971 [27] (Fig 1). Northern bottlenose whales are usually found in groups of one to four, and whalers would take all the whales they encountered, regardless of sex or age class, so we assume our dataset has low demographic capture bias [21]. Individuals included in this analysis ranged from 4-27 years old (median age = 14). The teeth of two H. ampullatus that stranded in northeast Newfoundland in 2004 were also analyzed.

The jaws of whaled specimens were originally boiled for two hours to facilitate tooth extraction [27]. Teeth were sectioned along the longitudinal midline and stored unpreserved at room temperature in individual sachets for over 40 years prior to this study. Genetic analysis of gum-tissue from the teeth used in this study confirmed the sex documented in the whaling records [28,29]. The teeth from Newfoundland animals were extracted from decomposed specimens, air dried and stored whole until being sectioned for this study. Similar to other odontocetes [30-32], H. ampullatus dentine is laminated, with one clear and one opaque layer defining each annual GLG within the cone of the tooth [23] (Fig 2). Only teeth with a clear neo-natal line and defined GLG structure across the first five years were retained for isotope analysis, reducing our sample size to 50 individuals (N = 6 from Iceland, N = 42 from Labrador, N = 2 from Newfoundland).  To improve GLG definition, tooth sections were initially polished using 30μm aluminum oxide lapping film [16] and then acid-etched using 10% formic acid

[33]. GLGs were counted and aged assuming annual deposition, starting at the line that divides prenatal and postnatal dentine [16,23]. Using a single section of each tooth, GLGs 1-5 were sampled individually at a depth of 250-μm with a 300-μm-diameter drill bit, using a high-resolution micro-mill (New Wave Research, Freemont, California). When sufficient prenatal dentine was present it was sampled at a depth of 150 μm.  For mature individuals (> 9 years old) [27], we also collected samples from older GLGs to estimate the range in δ¹⁵N in adult diet (N = 29).  However, as whales age their GLGs become compressed and are not wide enough to sample individually. Instead we collected samples representative of the mature age class by drilling across GLGs 8 -12 as a group with a 1 mm-diameter drill bit using a Dremel hand tool.

Stable isotope analysis (δ¹⁵N / δ¹³C)

Powdered dentine from each sampled GLG was weighed (~1 mg) into tin cups for isotopic analysis on a Vario EL Cube elemental analyzer (Elementar, Germany) connected to a DELTA Advantage isotope ratio mass spectrometer (Thermo, Germany). Isotope ratios are reported in Delta notation (δ) as per mil (‰) deviation from isotope ratios of atmospheric N₂ for nitrogen and Vienna Pee-Dee Belemnite (V-PDB) limestone for carbon. δ¹⁵N or δ¹³C are defined as δ = (R_sample – R_standard)/R_standard), where R is the ratio of the abundance of the heavy to the light isotope. Values are normalized to internal standards nicotinamide, ammonium sulfate + sucrose, caffeine, and glutamic acid, whose isotopic compositions cover the natural range of samples (δ¹⁵N -16.61 to 16.58‰, δ¹³C -34.46 to ‒11.94‰) and are calibrated to international standards IAEA-N1(+0.4‰), IAEA-N2(+20.3‰), USGS-40(-4.52‰) and USGS-41(47.57‰) for δ¹⁵N, and IAEA-CH-6(-10.4‰), NBS-22(-29.91‰), USGS-40(-26.24‰) and USGS-41(37.76‰) for δ¹³C. Analytical precision based on repeated measures of laboratory reference materials not used in calibrations was ~0.1‰ for both δ¹⁵N and δ¹³C within multiple laboratory runs. Variation between duplicate measures of ~10% of samples had an absolute mean of 0.26 ‰ for δ¹⁵N and 0.21 ‰ for δ¹³C.

The small size of some GLGs meant it was sometimes necessary to collect amounts less than 1 mg. A linearity study showed samples <0.5 mg appeared to have a positive bias in d¹⁵N but not d¹³C, and further analysis was restricted to samples weighing >0.5mg, reducing the number of GLG samples available for some individuals. Additionally, we omitted the smallest duplicate sample, so that only a single sample from an individual GLG was included in further analysis.

Data Analysis

Following the screening for duplicates and sample weight described above, 50 individuals were included in summary statistics regardless of how many GLGs were available. However ontogenetic trend analysis was restricted to those individuals which had stable isotope data available from at least GLGs 1-3 (N = 37). Data structure, variables, and sample sizes are identified in Table 1 and variable inclusion rationale and data sources are further described in S2 Table.

Table 1. Data structure, variables and sample sizes.

For comparison with other published values and ecological studies, carbon isotope values were adjusted for the oceanic Suess effect, applying a factor of 0.0019‰ yr ^-1 to δ ¹³C measured in GLGs; δ ¹³C_cor values are approximately relative to the year 2000 [16,34,35]. The isotope values sampled from a cross section of mature GLGs (age 8-12) were assumed to represent the average isotopic profile of adult whales, and used as a benchmark for assessing when the weaning associated δ¹⁵N decline ended.