The future of baleen whales: Recoveries, environmental constraints, and climate change

Abstract

Most baleen whales were severely overexploited during the past century, but many populations have received near-complete protection from exploitation for more than a half-century. Some of these populations have made remarkable recoveries and are now approaching pre-exploitation levels of abundance. Contrary to expectations of baleen whales making minor oscillations around equilibrium abundances, several populations that have made the strongest recoveries have experienced major mortality events. We review examples from the literature showing increasing demographic variability in recovering populations of baleen whales and present a simulation study on the expected response of recovered versus depleted whale population to environmental variability and climate impacts. We propose that baleen whales are more sensitive to environmental variability than previously recognized; that major demographic fluctuations will become the norm as baleen whales recover; and that climate-driven disruptions to whale population dynamics will be most dramatic in populations with the lowest rates of anthropogenic mortality.

Fig. 1.. Variability in canonical baleen whale feeding habitats.

(A to C) krill; (D to F): forage fishes; (G and H): benthic crustaceans and copepods. Note that scales and units are different for each example; the figures are intended to demonstrate interannual variability rather than compare absolute biomass across species or regions. Colors in (B) refer to different krill species, Euphausia pacific and Thysanoessa spinifera. Colors in (D) refer to different study sites, as listed in the caption (SOG, Strait of Georgia; WCVI, West Coast Vancouver Island). Data sources are as follows: (A) (50), (B) (51), (C) (52), (D) (60), (E) (59), (F) (57), (G) (15), and (H) (61).

Fig. 2.. Conceptual figure of population dynamics simulations.

We simulate a baleen whale population that has been depleted by commercial whaling. In one scenario, the depleted population is free from additional anthropogenic mortality (e.g., vessel strikes, bycatch, or continued whaling) and therefore recovers to pre-whaling abundances. In the second scenario, the population continues to experience anthropogenic mortality and therefore remains depleted. These two populations then experience natural variability in environmental conditions and prey availability, applied either through density-dependent (effects applied to K) or density-independent (applied to r) pathways (see Fig. 3). In a second set of simulations, the two populations experience both environmental stochasticity and declining productivity caused by climate warming, applied either through density-dependent or density-independent pathways (see Fig. 4). Simulations are described in detail in the “Simulating Population Responses to Environmental Stochasticity” section and in Materials and Methods.

Fig. 3.. Simulated population responses to environmental stochasticity.

Environmental stochasticity is applied via density-dependent (A to E) versus density-independent (F to J) pathways. In Recovered scenarios (blue), the population had no additional anthropogenic mortality applied and was therefore able to recover quickly to dynamic equilibrium abundance. In Depleted scenarios (yellow), the population had an additional 0.05 anthropogenic mortality hazard rate applied, keeping it far below dynamic equilibrium abundance. Variation in annual growth rates is intended to represent volatility in population dynamics, and the maximum proportional decline is intended to represent the magnitude of mortality events. Density distributions presented for these two metrics represent the summarized values from 1000 unique simulation instances incorporating random stochasticity in environmental conditions. In the top and bottom rows [(A), (D), (E), (F), (I), and (J)], each line represents a unique simulation instance, and the thick line in each panel represents one randomly selected instance with the same environmental variability applied across both density-dependent and -independent simulations, and Depleted and Recovered scenarios. In the density-dependent population trajectory, the gray and black lines indicate realized annual carrying capacity (dynamic equilibrium abundance), whereas environmental effects are not shown in the density-independent population trajectory as they were applied directly to vital rates. Instead, the constant carrying capacity from density-independent simulations is shown as a dashed line. In (E) and (J), only natural survival rates are shown (i.e., anthropogenic hazards are not subtracted from survival rates in the depleted scenarios).

Fig. 4.. Simulated population responses to a climate-induced decline in marine productivity and prey availability.

Prey variability and declines are applied via density-dependent (A to C) versus density-independent (D to F) pathways. In Recovered scenarios (blue), the population had no additional anthropogenic mortality applied and was therefore able to recover quickly to dynamic equilibrium abundance. In Depleted scenarios (yellow), the population had an additional 0.05 anthropogenic mortality rate applied, keeping it far below dynamic equilibrium abundance. Years between climate inflection and peak abundance [(B) and (E)] is intended to represent the responsiveness of a population to climate impacts on prey availability, and the proportional decline from peak abundance [(C) and (F)] is intended to represent the total climate impact on a population within each simulation. Density distributions presented for these two metrics represent the summarized values from 1000 unique simulation instances incorporating random stochasticity in environmental conditions and the extent of climate-driven decreases to environmental productivity. In (A) and (D), each line represents a unique simulation instance, and the thick line in each panel represents one randomly selected instance with the same environmental variability applied across both density-dependent and -independent simulations, and Depleted and Recovered scenarios. In the density-dependent population trajectory (A), the gray and black lines indicate realized annual carrying capacity (dynamic equilibrium abundance), whereas environmental effects are not shown in the density-independent population trajectory (D) as they were applied directly to vital rates. Instead, the constant carrying capacity from density-independent simulations is shown as a dashed line.

Fig. 5.. Demographic variability in recovered and recovering populations of baleen whales.

(A to C) Abundance, birth rate, and total mortality rate of Eastern North Pacific gray whales, from (15). Black lines and shaded polygons represent median and 95% confidence intervals (CIs) of estimates from a population dynamics model that applies environmental stochasticity to carrying capacity. (A) Black dots and error bars represent survey estimates of abundance with SEs. (D) Basin-wide annual mark-recapture abundance estimates for the North Pacific metapopulation of humpback whales, from (14). Points and lines represent annual estimates of abundance and the shaded polygon represents SEs. (E) Annual estimated ratio of calves per female at Glacier Bay National Park, for one subpopulation of North Pacific humpbacks, from (77). Points and lines represent model estimated calf ratio, and the shaded polygon represents the 95% CI of estimates. (F) Annual humpback whale stranding counts from the US west coast, including California, Oregon, Washington, and Alaska, obtained from the NOAA National Strandings Database. (G) Estimated abundance trend for Western South Atlantic humpback whales, from (11). Points with error bars represent survey estimates of abundance, and the line and shaded polygon represent the estimated abundance trajectory (and 95% CI) from a deterministic population dynamics model that assumes a stable carrying capacity. (H) Annual humpback whale stranding counts from the entire coast of Brazil from 2002 TO 2023 (upper line) from (81), and from the coast of the Brazilian state of Rio de Janeiro from 1991 TO 2011 (lower line) from (82). (I and J) Southern right whale abundance (all life stages) and the probability of a resting female becoming pregnant at a breeding area in South Africa, from (86). Lines and shaded polygons represent the maximum likelihood estimates and SEs of estimates, respectively. (K) Annual southern right whale stranding counts from the coast of South Africa, 1979 TO 2019, from (145). (L) Estimated abundance of Western South Atlantic right whales, from (12). The point with error bars represents a single survey-based estimate of abundance from 2010, and the line and shaded polygon represent the estimated abundance trajectory (and 95% CI) from a deterministic population dynamics model that assumes a stable carrying capacity. (M) Detrended annual southern right whale calf counts from Southern Brazil, reported in (91). These values were calculated by subtracting observed calf counts from a mean trend line regressed to annual count data and are not comparable to absolute metrics of fecundity such as pregnancy probabilities or calving rates presented for other populations. (N) Annual female southern right whale mortality probability from Peninsula Valdez, Argentina, estimated using mark-recapture models [calculated as 1 − survival probability, the metric reported in (89), for comparability with (C)].