Environmental conditions and marine heatwaves influence blue whale foraging and reproductive effort

Abstract

Animal behavior is motivated by the fundamental need to feed and reproduce, and these behaviors can be inferred from spatiotemporal variations in biological signals such as vocalizations. Yet, linking foraging and reproductive effort to environmental drivers can be challenging for wide-ranging predator species. Blue whales are acoustically active marine predators that produce two distinct vocalizations: song and D calls. We examined environmental correlates of these vocalizations using continuous recordings from five hydrophones in the South Taranaki Bight region of Aotearoa New Zealand to investigate call behavior relative to ocean conditions and infer life history patterns. D calls were strongly correlated with oceanographic drivers of upwelling in spring and summer, indicating associations with foraging effort. In contrast, song displayed a highly seasonal pattern with peak intensity in fall, which aligned with the timing of conception inferred from whaling records. Finally, during a marine heatwave, reduced foraging (inferred from D calls) was followed by lower reproductive effort (inferred from song intensity).

Keywords: acoustics; behavior; blue whale; boosted regression tree models; life history; marine heatwave.

FIGURE 1

Study area map and blue whale call spectrograms. Left panel: map of the study area in the South Taranaki Bight region, with hydrophone (marine autonomous recording unit; MARU) locations denoted by the stars. Gray lines show bathymetry contours at 50 m depth increments, from 0 to 500 m. Location of the study area within New Zealand is indicated by the inset map. Right panels: example spectrograms of the two blue whale call types examined: the New Zealand song recorded on 31 May 2016 (top) and D calls recorded 20 September 2016 (bottom). Spectrograms are configured with a 3072 point fast Fourier transform, Hann window, 50% overlap.

FIGURE 2

Occurrence patterns of song and D calls. Left panels: number of hours per day with song detected (light blue bars) and daily song intensity index (dark blue lines) for each hydrophone (MARU 1–5) over the entire recording period. Right panels: number of hours per day with D calls detected (light green bars) and number of D call detections per day (dark green lines) for each hydrophone over the entire recording period. The hydrophone is listed at the top of each panel, corresponding to map locations in Figure 1. Periods with light gray shading indicate gaps in recording due to hydrophone refurbishment.

FIGURE 3

Annual cycle of calling activity. Average annual cycle in the song intensity index (dark blue) and D calls per day of the year, computed across all hydrophone locations and the entire recording period.

FIGURE 4

Functional response curves for environmental correlates of calling activity. Partial dependence plots derived from the boosted regression tree fitted for each within‐season peak in calls, showing the smoothed functional relationships between calling (either song intensity index or D call detections) and each predictor variable while fixing all other variables at their mean value. Color is indicative of the hydrophone, with locations shown on the map in upper right panel. Rug plots show distribution of the values for each predictor.

FIGURE 5

Annual song intensity and the breeding cycle. Top panel: average yearly cycle in song intensity index, computed across the five hydrophone locations and the entire recording period; dark blue line represents a loess smoothed fit. Bottom panel: fetal length measurements from whaling catch records for Antarctic blue whales (gray, measurements rounded to the nearest foot), pygmy blue whales in the southern hemisphere (blue, measurements rounded to the nearest centimeter). Measurements from blue whales caught within the established range of the New Zealand population are denoted by the dark red triangles. Calving presumably takes place around or shortly after fetal lengths are at their maximum (April–May), which implies that mating likely occurs around May–June, coincident with the peak song intensity. Note: Many small fetuses were missed in the Antarctic blue whale data, whereas sampling was more thorough for pygmy blue whale data.

FIGURE 6

Environmental conditions and D calls in summer months. Mean sea surface temperature (top), net primary productivity (middle), and daily D call detections (bottom) for the period of 1 January through 28 February in each of the three years with recording coverage, with each of the five hydrophone recording locations indicated by colors. The number of recording days for each hydrophone in each year are indicated in the bottom panel. January and February were characterized by regional marine heatwaves in 2016 and 2018, while 2017 exhibited more typical summer upwelling conditions. Pairwise comparisons resulting from a linear mixed model for each variable by year (accounting for differences among hydrophone locations) are indicated above each set of boxplots. In all cases, the linear mixed models were significant, with the two marine heatwave years characterized by comparable SST, log(NPP), and D call values that were significantly different from those measured in the more typical upwelling year.

FIGURE 7

Relationship between summer D call detections and fall song intensity. Relationship between the mean number of D call detections between 1 January through 28 February and the mean song intensity index in the subsequent fall peak between 1 April and 31 June. Points are symbolized by year and hydrophone recording location.

FIGURE A1

Correlation between the daily number of D calls identified from the manual validation vs. D calls identified by the automatic classifier, for the period 23 January 2016–31 March 2017.

FIGURE A2

Spatiotemporal detection patterns of blue whale D call occurrence and intensity visualized using D call detections that were manually validated. Hours with D calls detected per day (light green bars) and the daily number of D call detections (dark green lines) are shown between 23 January 2016 and 31 March 2016, at each hydrophone recording location.

FIGURE A3

Spatiotemporal detection patterns of blue whale D call occurrence and intensity visualized using D call detections that were automatically classified using the random forest model. Hours with D calls detected per day (light green bars) and the daily number of D call detections (dark green lines) are shown between 23 January 2016 and 31 March 2016, at each hydrophone recording location. Note no noticeable difference in occurrence patterns from Figure A2.

FIGURE A4

Partial dependence plots derived from the boosted regression tree models for song, showing the smoothed functional relationships between song intensity and each predictor variable while fixing all other variables at their mean value. Color is indicative of the hydrophone. Rug plots show distribution of the values for each predictor, and color corresponds to the hydrophone.

FIGURE A5

Partial dependence plots derived from the boosted regression tree models for D calls, showing the smoothed functional relationships between D call detections and each predictor variable while fixing all other variables at their mean value. Color is indicative of the hydrophone. Rug plots show distribution of the values for each predictor, and color corresponds to the hydrophone.

FIGURE A6

Whaling catches of Antarctic blue whales (blue) and pygmy blue whales (red) in the southern hemisphere, with assumed boundaries in black between each. Acronyms denote each of the currently assumed blue whale populations: central Indian ocean (CIO, Sri Lanka), south‐west Indian Ocean (SWIO, Madagascar), south‐east Indian Ocean (SEIO, Australia/Indonesia), south‐west Pacific Ocean (SWPO, New Zealand). Key land stations in and near the pygmy blue whale region are labeled.

FIGURE A7

Average annual cycle in song intensity index across all five hydrophone locations by year. Dark solid lines represent a loess smoothed fit. Note the alignment of the fall peak in song intensity in both years, but the lower average song intensity in 2016 than in 2017.