Male humpback whales switch to singing in the presence of seismic air guns

The study investigates how seismic air gun noise influences mating tactics of male humpback whales during migration. Anthropogenic ocean noise, particularly from seismic exploration, overlaps in frequency with baleen whale communication and has been linked to altered movement, feeding, and physiology. However, direct links to breeding behaviours are less clear. Male humpback whales exhibit a conditional mating strategy, alternating between singing (advertising/mediating competition) and non-singing "seeker" tactics (actively joining females and forming competitive groups). The authors tested two hypotheses during controlled one-hour exposures to air guns: (1) males would be less likely to sing in the presence of air guns, and (2) males would experience fewer joining interactions (a proxy for mating-related interactions) during air gun activity. The purpose was to quantify whether air gun noise shifts the balance between singing and seeking tactics and to evaluate the apparent payoff (likelihood of joining a female without attracting additional competing males), thereby assessing potential implications for breeding behaviour.

Prior work documents wide-ranging effects of seismic and other anthropogenic noises on baleen whales, but breeding responses are variable and species- and context-dependent. Opportunistic studies reported fin whales ceasing singing and reduced humpback singing during distant seismic surveys, whereas blue and bowhead whales sometimes increased calling in response to impulsive sources such as sparkers or air guns. Earlier work on migrating humpbacks showed reductions in joining interactions during air gun activity, suggesting a potential reduction in breeding-related behaviours. Humpback mating systems involve both singing and intense male–male competition around females; singing may function in male–male mediation and/or female attraction, with singers often ceasing when approached by other whales. Population resilience and recovery post-whaling have been associated with flexible mating strategies, implying that tactic plasticity may buffer disturbances. Collectively, the literature indicates that vocal and social responses to seismic noise are not uniform and may depend on proximity, received levels, duration of exposure, and local social context, motivating controlled assessments of tactic choice and payoff.

Study area and period: Field data were collected off Peregian Beach, Queensland, Australia, during the southward migration (September–October) in 2010, 2011, and 2015 (core analyses focus on 2011, 2014, and 2015 per results description). Observations occurred 07:00–17:00 on days with good visibility. Migrating groups passed within ~6–10 km from shore. Visual tracking: From an elevated land station (Emu Mountain, 73 m), groups (including lone whales) were tracked with a theodolite linked to data acquisition software. Scan sampling continuously catalogued group IDs, positions at ~10 min intervals, composition (adult/calf presence), behaviours (singing, joining, competitive groups), and movements. Joins were defined as directed approaches culminating in surfacing within 100 m and associating. Acoustic monitoring: A moored T-shaped hydrophone array (3–5 High Tech HTI-96-MN sensors with preamps; 22 kHz sampling) recorded vocalizations. Singer locations were estimated by time-of-arrival differences among hydrophones using Ishmael software, producing ~10 min position fixes. Air gun exposures: Experiments used towed single-gun or small-array air guns on set transects across or along the migration stream. Guns fired impulsive signals approximately every 15 s during one-hour exposure periods; control transits (guns off) were also conducted but not analyzed as exposure. Vessel speeds ~7–8 km/h. Exposure ranges included singers within ~10 km of the source. Data structuring: The day was partitioned into three 3-hour periods (08:00–11:00, 11:00–14:00, 14:00–17:00). Each 3-hour block was labeled as baseline (no active guns, excluding any post-exposure periods that day) or air gun (any gun activity within the block; if spanning blocks, the block with greatest exposure proportion was labeled air gun). Baseline periods were distributed across day and season. Whale counts per 3-hour block included numbers of likely males, likely females, and singers. Sex/role inference: Singing adults were assumed male. Adult–calf pairs were assigned as one adult female; adult escorts of female–calf pairs were counted as males. Multi-adult groups were assumed to include one female with remaining adults as males, based on prior biopsy-based sex ratios. Lone non-singers and adult pairs without calves were assigned probabilistic sex (70% male, 30% female) following prior local biopsy results. Ageing was not possible. Tactics and payoff metrics: Males were categorized as singers or seekers (non-singing presumed males). Joining interactions were counted per 3-hour block and attributed to tactic: singer joins (singer actively altered course to join an adult or female–calf group) and seeker joins (non-singing male joined a female/female–calf group). Each observed join contributed a payoff of 1 to the relevant tactic’s total for the block. If an additionally joining male entered a singer-joined group (creating competition), the singer payoff for that join was reduced to 0.5 and the seeker payoff increased by 0.5, reflecting increased competition. Additional joins to seeker-joined groups (up to three additional presumed males) were tallied, reflecting elevated competition risk for seekers. Statistical analyses: Counts were modeled with negative binomial distributions. Baseline-only models (GLM) tested relationships between number of singers and social environment (likely males, likely females). Full-dataset models (GAM, negative binomial) added the covariate air gun presence (baseline vs air gun) and smooth terms for male and female counts to assess whether baseline relationships held during exposure. Payoff analyses proceeded in two stages: (1) baseline GLMs relating tactic-specific total payoff to numbers of males and females; (2) full-dataset GAMs modeling tactic payoff as a function of air gun presence and male numbers (given male density effects at baseline). Individual singer payoff was computed as total singer payoff divided by number of singers per block and analyzed similarly. Significance threshold p < 0.05; model diagnostics included residual checks, overdispersion, and outlier assessments. Sample sizes and ranges: Baseline blocks for singer-count models: n = 67 (males per block 1–20, mean 8; females 1–21, mean 7). Air gun blocks for singer-count models: n = 52; full dataset N = 119 blocks. Payoff models used baseline n = 67 (seekers) and n = 35 (singers), and air gun n ≈ 51 (seekers) and n = 43 (singers) when at least one relevant participant (male or singer) was present.

- Contrary to hypothesis 1, the number of singing males increased during air gun activity relative to baseline: • Full dataset GAM: significantly higher number of singing males during air gun periods vs baseline (N = 119; Z = 3.136; P = 0.0017). • During air guns, the number of singers showed no significant relationship with total males (χ² = 2.903, P = 0.2380) or females (χ² = 0.293, P = 0.5880), in contrast to baseline patterns. - Baseline determinants of singing (baseline GLM, n = 67): • Number of singers increased with number of likely males (Z = 2.395, P < 0.0001). • Number of singers decreased with number of likely females (Z = -2.095, P = 0.036). - Seeker (non-singing) tactic payoff decreased during air gun activity, indicating fewer seeker joins: • Descriptive: average seeker payoff per 3 h decreased from 0.49 (range 0–3.5) at baseline to 0.26 (0–1) during air guns; observed joins and additional joins were also lower during air guns (Table 1). - Singer tactic payoff under baseline vs air gun conditions: • Baseline: singer total payoff did not significantly vary with numbers of males (Z = -0.710, P = 0.4747) or females (Z = 0.659, P = 0.5097). • Air gun effect overall: no significant change in total singer payoff when comparing baseline to air gun periods (GAM, N = 78; z = -0.468; P = 0.634). • However, during air gun periods, total singer payoff increased significantly with increasing male numbers (GAM, N = 78; z = 7.468; P = 0.006), whereas this relationship was absent at baseline (χ² = 0.116; P = 0.7338). • Individual singer payoff similarly rose with male density during air guns (GAM, N = 78; z = 4.352; P = 0.037) but not at baseline (z = 0.162; P = 0.825). - Net behavioural shift: During air gun activity, more males adopted the singing tactic, singers were more often observed joining females, and singer groups did not show an increased risk of attracting additional competing males; conversely, seeker joining interactions declined.

Findings indicate that short-term exposure (one hour) to seismic air guns altered male humpback whale mating tactics. Rather than suppress singing, air gun activity coincided with a higher prevalence of singers, even when overall male or female numbers were low. This undermines the initial hypothesis that males would avoid singing under noise exposure and suggests a context-dependent shift towards an acoustic display tactic. Concurrently, seeker (non-singing) joining declined, implying a redistribution of mating effort from active seeking to singing. Importantly, singers more frequently joined females under air guns without a corresponding increase in additional male joiners, reducing the typical competitive costs associated with singing escorts. The payoff patterns show that during air guns, higher male densities benefited singers (increasing total and individual singer payoff), in contrast to baseline where singer payoff did not scale with male density. These results align with a flexible mating strategy that can buffer transient disturbances. Potential mechanisms include a stimulatory effect of impulsive noise resembling salient social cues (e.g., breaching) that may trigger singing, rather than pure masking. The absence of increased additional joins in singer groups during air guns may reflect the observed decline in seeker activity, reducing competitive pressure. Although overall breeding activity did not appear reduced (given increased singer engagement), the behavioural system shifted in composition. The ecological relevance hinges on female choice and long-term reproductive success, which were not measured. The short exposure duration also limits inference about sustained survey conditions, where responses could differ (e.g., eventual cessation of song as reported elsewhere).

This study demonstrates a tactical shift in male humpback whale mating behaviour during brief exposure to seismic air guns: more males sang, singers were more often observed joining females, and seeker joining declined, with singer payoffs increasing with male density under exposure. These findings suggest that humpback whales can flexibly adjust mating tactics under anthropogenic noise, potentially mitigating some immediate competitive costs. However, the implications for mating success and population dynamics remain unresolved. Future research should: (1) quantify dose–response relationships across distances, received levels, and longer exposure durations typical of full seismic surveys; (2) directly measure female choice and mating success under noise exposure; (3) refine individual-level tracking of tactic switching and outcomes; (4) assess generality across seasons and populations; and (5) evaluate cumulative and repeated exposure effects on breeding behaviour and vital rates.

- Exposure duration was short (1 hour), not representative of multi-day commercial surveys; responses may change over prolonged exposure (e.g., potential eventual song cessation). - Mating success and female choice were not measured; joining is only a proxy for breeding interactions, limiting inference about reproductive outcomes. - Sex assignments for non-singing adults and pairs were probabilistic (70% male assumption) based on prior biopsies; misclassification could affect male/female counts and derived relationships. - Age structure was unknown; potential age-related tactic variation could not be evaluated. - Potential detection and tracking biases due to visibility, group confusion at higher densities, and acoustic detectability could influence counts of singers and joins, though density thresholds were imposed to mitigate this. - The study design emphasizes population-level patterns within the observation area rather than controlled individual-level responses; independence among groups across time blocks cannot be fully guaranteed during migration. - Received sound levels and exact distances varied across years and trials; while ranges included singers within ~10 km, dose metrics were not the central analytical covariates, limiting mechanistic attribution. - Potential displacement effects were considered but not fully excluded; however, comparable male/female numbers across conditions argue against large-scale redistribution driving results.