Effective management of populations threatened by anthropogenic impacts is particularly challenging in marine environments, which are highly dynamic and difficult to observe (Maxwell et al. 2015; Hobday et al. 2016). In both marine and terrestrial systems, spatial management targets regions of high risk for anthropogenic impacts (Maxwell et al. 2015). Static management approaches have proven problematic or ineffective in marine environments, particularly for highly migratory species (Lascelles et al. 2014; Dunn et al. 2016) or for species undergoing distributional shifts due to climate change (Lascelles et al. 2014). Dynamic management can improve management outcomes by adjusting the spatial and/or temporal extent of an area of concern (Dunn et al. 2016). However, implementation of dynamic management where managers make decisions about the future requires anticipating when and where conservation risks are likely to arise (Clark et al. 2001; Lascelles et al. 2014; Dietze et al. 2018). Ecological forecasting tools that use environmental data are one way to anticipate future conservation risks.

Environmental data have been previously used to inform dynamic management by integrating recent environmental conditions into distributional models of marine species (Becker et al. 2016; Dunn et al. 2016; Hazen et al. 2016). In these applications, the most recent available satellite-derived measurements of environmental conditions, also known as “near real-time data”, are used to predict distributions of species of concern in the immediate future (Hazen et al. 2016). Although this approach is useful for making predictions over short time scales, it cannot inform future conditions, which would allow managers to better anticipate conservation risks. However, an alternative approach, commonly referred to as “ecological forecasting”, can inform future risk by incorporating species distribution models into forecasts that predict future environmental variables (Clark et al. 2001; Dietze et al. 2018).

To date, ecological forecasts using environmental data have primarily focused on seasonal time scales (Kaplan et al. 2016). While this work is key to understanding trends on the order of weeks to months, subseasonal forecasts – those that generate predictions over one to several weeks – may be needed to inform certain management decisions (Hobday et al. 2016; Dietze et al. 2018; Jacox et al. 2020). Accurate subseasonal forecasts have been historically difficult to produce, but recent developments have bridged the gap between short-term weather forecasts and monthly (or longer) climate projections (Mariotti et al. 2020). The Subseasonal Experiment (SubX) is a National Oceanic and Atmospheric Administration (NOAA) Climate Test Bed project that provides novel subseasonal global forecasting products for multiple global models by forecasting atmospheric and ocean variables at weekly-to-monthly time scales, which are typically difficult to resolve. SubX is unique compared to traditional weather forecasting models because it combines (1) more frequent model initialization, defined as how often a new model is generated; and (2) longer forecast lead times, which describes the amount of time forecasted into the future from the model initialization date (Pegion et al. 2019). SubX models are publicly available and the forecast is provided in near real-time, which allows for immediate integration with ecological models, and could yield a powerful forecasting tool for dynamic management by anticipating times and places of conservation concern over a weekly timeframe.

Here, we assess the potential for subseasonal forecasts to inform and improve the management of marine mammal populations. Marine mammals are highly mobile and many are impacted by anthropogenic activities such as fisheries bycatch, vessel strikes, and entanglement in fishing gear (Avila et al. 2018). Dynamic factors such as temperature and variability in prey distribution can drive changes in the distribution of marine mammal species (Davies et al. 2019), which can render static management approaches ineffective (Lascelles et al. 2014). Using ecological forecasting to predict future spatial distributions of marine mammals over subseasonal timeframes could improve management efforts.

We use humpback whales (Megaptera novaeangliae) in the Northeast US (NEUS) as a case study to assess the utility of subseasonal forecasts for dynamic management of marine mammal populations. The NEUS is heavily impacted by commercial and recreational fishing, major shipping ports, and recent offshore wind developments. Humpback whales typically undergo seasonal migrations between low-latitude winter breeding grounds and high-latitude foraging grounds including the NEUS (Stevick et al. 2006). Currently, management concerns are focused on anthropogenic mortality of multiple populations of large whales that forage in the NEUS, such as minke whales (Balaenoptera acutorostrata), North Atlantic right whales (Eubalaena glacialis), and humpback whales (Avila et al. 2018). Current mitigation efforts include fishing gear modifications or restrictions and vessel slowdown zones: large vessels are required to reduce speed in seasonal static management areas (SMAs) or are recommended to reduce speed in triggered-closure dynamic management areas (DMAs) when North Atlantic right whales are present (NOAA 2014). We assess the potential for forecasting humpback whale arrival into NEUS foraging grounds by (1) modeling historical variability in the timing of arrival in the NEUS and SMAs using a distribution model; (2) forecasting arrival by integrating the distribution model with SubX forecasts; and (3) assessing the performance of humpback whale density forecasts relative to the performance of density predicted using traditional satellite-derived sea-surface temperature (SST) measurements to assess if forecasts can maintain a high level of model performance.

Download the study: Subseasonal forecasts provide a powerful tool for dynamic marine mammal management