The krill surplus hypothesis of unlimited prey resources available for Antarctic predators due to commercial whaling in the 20th century has remained largely untested since the 1970s. Rapid warming of the Western Antarctic Peninsula (WAP) over the past 50 years has resulted in decreased seasonal ice cover and a reduction of krill. The latter is being exacerbated by a commercial krill fishery in the region. Despite this, humpback whale populations have increased but may be at a threshold for growth based on these human-induced changes. Understanding how climate-mediated variation in prey availability influences humpback whale population dynamics is critical for focused management and conservation actions. Using an eight-year dataset (2013–2020), we show that inter-annual humpback whale pregnancy rates, as determined from skin-blubber biopsy samples (n = 616), are positively correlated with krill availability and fluctuations in ice cover in the previous year. Pregnancy rates showed significant inter-annual variability, between 29% and 86%. Our results indicate that krill availability is in fact limiting and affecting reproductive rates, in contrast to the krill surplus hypothesis. This suggests that this population of humpback whales may be at a threshold for population growth due to prey limitations. As a result, continued warming and increased fishing along the WAP, which continue to reduce krill stocks, will likely impact this humpback whale population and other krill predators in the region. Humpback whales are sentinel species of ecosystem health, and changes in pregnancy rates can provide quantifiable signals of the impact of environmental change at the population level. Our findings must be considered paramount in developing new and more restrictive conservation and management plans for the Antarctic marine ecosystem and minimizing the negative impacts of human activities in the region.

Biopsy collection

We collected skin and blubber samples from female humpback whales during the 2013–2020 austral summers (December-March). This was done in the nearshore waters of the Western Antarctic Peninsula (WAP) using standard biopsy techniques (Fig. 1). We used a crossbow to project modified bolts and 40mm stainless steel biopsy tips (CetaDart) to obtain samples from a distance of 10–30 meters, targeting the area of the body below the dorsal fin.   Samples were collected opportunistically when whales were encountered during prey or visual surveys conducted within ~10 nautical miles of scientific research stations (i.e., Palmer Station, Anvers Island, USA, or Akademik Vernadsky Station, Galindez Island, Ukraine). Dedicated research cruises or platforms of opportunity, including ecotour vessels, were also used. Dependent calves were not sampled during seasons 2013–2019, but all age and sex classes of humpback whales were sampled during 2020. Because of this change in protocol, samples from calves were not included in any analysis. However, the presence of a calf was recorded and identified, as evident by its smaller size (less than half of the presumed mother's length) and close association with an adult, presumed to be the mother.  Supplementary data (including location and group size) were recorded at every biopsy event. Samples were stored frozen whole at -20°C until used for analysis.

DNA profiling

A standard DNA profile, including sex-specific markers and microsatellite genotypes, was used to identify individual whales. DNA was extracted from the skin-blubber interface using a commercially available kit (DNeasy 96 Blood & Tissue Kit, Qiagen, Hilden, Germany). The sex of each sampled whale was determined by amplification of sex-specific markers following the protocols of Gilson et al. Results were compared to controls for a known male and female using gel electrophoresis.

Samples were genotyped using 10 previously published microsatellite loci to resolve the individual identity of each sampled whale and remove potential duplicates (Table S1). Alleles were sized and binned using the software program Genemapper v3.7 (Applied Biosystems). The total number of amplified loci for a given sample was considered as an added quality control threshold, with samples amplifying for less than 7 loci considered poor quality and repeated or removed from final dataset. Given the estimated probability of identity for these loci from previous studies, we assumed that samples matching at a minimum of seven loci to be recaptures of the same individual. Recaptures of the same individual were removed from the analysis. The expected probability of identity (P_ID; the probability that two individuals drawn at random from a population will have the same genotype by chance) for each locus was calculated in GenAlEx v6.5. Cervus 3.0.7 was used to compute the number of alleles (K), observed and expected heterozygosity (HO and HE), and the probability of identity for all individual matches.

Hormone extraction and quantification

We extracted steroid hormones from the blubber portion of the biopsy samples following standard methods. Briefly, to quantify hormone biomarkers (i.e., progesterone), we sub-sectioned a cross-sectional sub-sample (~0.15g) spanning from the epidermis-blubber interface to the most internal layer of the biopsy. These sub-samples were then homogenized multiple times using an automated bead mill homogenizer (Bead Ruptor Elite, Omni International). Following the completion of the homogenization process, we isolated progesterone using a series of chemical washes, evaporations, and separations. The final hormone residue was stored at -20°C until analysis.

We quantified the amount of hormone in each extract using a commercially available enzyme immunoassay used extensively in similar studies. Our progesterone EIA kit (EIA kit 900-011, ENZO Life Sciences, Farmingdale, NY) had a 100% reactivity with progesterone and an assay detection limit between 15 and 500 pg/mL. Two additional standard dilutions were added to allow for a lower detection limit of the standard curve to 3.81 pg/mL. We determined extraction efficiency by spiking subsamples of blubber from a dead, stranded animal of known pregnancy status, with 150ng of progesterone and including these with every extraction. We calculated the percentage of progesterone recovered after each extraction and adjusted each sample concentration to this efficiency prior to statistical analyses. An extraction efficiency greater than 60% was adequate and is based on the reported range of efficiencies seen using these methods.  If the efficiency of an extraction set was less than 60%, the sample extracts were discarded, and the blubber samples were re-extracted and re-analyzed. Each assay was evaluated for color development using a Biotek plate reader Epoch (Gen5^TM software [Biotek, USA]) with reading and correction wavelengths of 405 nm and 630 nm. Blubber hormone concentrations were then transformed into nanograms of progesterone per gram of blubber (wet weight).

Pregnancy classification

We assigned pregnancy of female humpback whales following previously published methods. Biopsy samples (n = 29) were collected from individuals of a known life-history stage from the Gulf of Maine feeding aggregation by the Center for Coastal Studies in Provincetown, MA. Using these control samples from the Gulf of Maine, the pregnancy state relative to blubber progesterone concentrations was modeled using a standard logistic regression model. Each WAP humpback sample of unknown pregnancy status was entered into the model, and the model returned a probability of being pregnant for each female sampled. If the probability of being pregnant was greater than 99.9%, that female was given an assignment of pregnant. If the probability of being pregnant was less than 0.1%, that female was assigned as not pregnant. If a biopsied female's probability of being pregnant was between those two bounds, that female was set as undetermined pregnancy. 

Please see the README document ("README-HumpbackPregnancyEnvironmentalChange.rmd") and the accompanying published article: L.J. Pallin, N.M. Kellar, D. Steel, N. Botero-Acosta, C.S. Baker, J. A. Conroy, D.P. Costa,  C.M. Johnson, D.W. Johnston, R. C. Nichols, D.P. Nowacek, A.J. Read, O. Savenko11,12, O.M. Schofield, S.E. Stammerjohn, D.K. Steinberg, A.S. Friedlaender. A surplus no more? Variation in krill availability impacts reproductive rates of Antarctic baleen whales. Global Change Biology.