There is growing evidence indicating that variability and extremes in conditions in the marine environment are as (or more) important as changes in the mean for determining threats to biodiversity, impacts on ecosystem services, and consequences for human systems (1⇓⇓–4). With respect to ocean temperature, long-term persistent warming has been accompanied by an increased frequency of discrete periods of extreme regional ocean warming (marine heatwaves) (5). This poses a threat to biodiversity and ecosystem services, including impacts on foundation species (corals, seagrasses, and kelps) (1, 4). The potential of human and natural systems to adapt to such changes remains unclear. In PNAS, Pershing et al. (6) show that an increasing frequency of extreme heat events—or “surprises”—is challenging autonomous modes of adaptation that rely on historical experience. The authors contrast reactive adaptation that is guided by experiences of past events with proactive adaptation based on forward-looking decision making. They use ocean ecosystems as a case study and, based on mathematical models, consider how temperature trends and the frequency of surprise (high) temperature events could impact natural and human communities under different adaptation strategies.

Pershing et al. (6) define a temperature surprise as an annual mean temperature that is 2 SDs above the mean, where the mean and SD are determined by the prior 30 y of temperature records (a rolling mean), at the scale of 65 large marine ecosystems (LMEs) (7). “Surprising surprises” are those events that are in excess of the number expected, based on the probability of a surprise determined from the rolling mean approach. Importantly, the authors found that the frequency of such surprises is increasing faster than expected and will continue to rise. Indeed, the cumulative increase may be even higher than estimated by Pershing et al. (6). They show that the increase is particularly high in the Arctic but exclude the Antarctic LME (due to missing data), which encompasses one of the fastest-warming regions globally (8).

Environmental conditions that fall outside the typical range of experience have a high potential to drive change in socioecological systems. Pershing et al. (6) use simple models to consider what an increasing frequency of surprises might mean for humans and for ecosystems, under different strategies for adapting to such change. These models suggest that, for humans, there is a higher payoff for strategies that are forward-looking than those that are based on a backward-looking approach (Fig. 1). For ecosystems, an increasing frequency of surprises may lead to a homogenization of the species mix, with specialists being replaced by generalists. The success of generalist species under future climate conditions has also been demonstrated separately in mesocosm experiments (9). Interestingly, based on Pershing et al.’s models (6), the occurrence of surprises appears to be more important for human systems, while the trend in the mean is more important for ecosystem responses to environmental change.

• Download figure
• Open in new tab
• Download powerpoint

Fig. 1.

Summary of implications for human communities and ocean ecosystems of adaptation approaches that are based on information from the past (left-hand side: fewer surprises) versus those that look forward and consider temperature trends (right-hand side: more surprises), based on findings from Pershing et al. (6). For human systems, there is a higher return for strategies that are responsive to temperature trends, while for ecosystems, increasing ocean temperatures and more surprises may lead to the replacement of specialist species with generalists, and a consequent decrease in biodiversity. Image courtesy of Stacey McCormack (University of Tasmania, Hobart, Australia).

What might these findings mean in real-world terms for the characteristics of marine socioecological systems? Pershing et al. (6) give the example of potential increases in the abundance and dominance of fast-reproducing gelatinous zooplankton (including jellyfish), with consequences for marine food webs more generally. For human systems, it might mean greater investment in gears that can target new species, or management actions that factor in likely future change such that they are more robust in supporting conservation outcomes under future conditions. The question remains whether human systems can make such shifts given the inertia that is often present in institutions (and what does it mean if they cannot?).

Pershing et al. (6) point to seasonal and multiyear forecasts of marine species and systems becoming more recognized, but also that the time scale at which these are reliable depends on the region. In a world of surprises where purely statistical predictions based on historical values may have low skill, it is important to include knowledge of mechanistic linkages between climate and marine resource responses (10⇓–12). For instance, skillful 7-y predictions of the size of the commercially and ecologically valuable Barents Sea cod stock are now available (13). These are based on variability in the volume and temperature of Atlantic water masses flowing in from further south and knowledge of temperature–cod production impacts (13). Novel methods from the field of robotics have also been used to enhance the decision-making process for setting catch quotas for commercially exploited fish stocks at the global scale (14). This approach is predicted to provide greater robustness to environmental surprises and higher recovery rate of global stocks by midcentury as compared to present methods (14).

The paper by Pershing et al. (6) is significant in that it provides a theoretical framework for considering how temperature trends and extreme events such as marine heatwaves will impact human and natural communities, as well as formalizing the payoff from different strategies for responding to these changes. This work also opens a series of questions including 1) how to account for surprises from other kinds of climate-driven changes in the marine environment (e.g., storm events, hypoxia, and acidification) and their cumulative effects (15), 2) how to account for surprises that occur at shorter (subannual) timescales but that may have equally important impacts (4), and 3) how adaptation to temperature change via movement (shifts in the habitat range of single species or whole communities) might interact with the strategies described to influence system-level outcomes (16⇓–18).