Summary

Marine mammals face a large array of stressors, including loss of habitat, chemical and noise pollution, and bycatch in fishing, which alone kills hundreds of thousands of marine mammals per year globally. To discern the factors contributing to population trends, scientists must consider the full complement of threats faced by marine mammals. Once populations or ecosystems are found to be at risk of adverse impacts, it is critical to decide which combination of stressors to reduce to bring the population or ecosystem into a more favorable state. Assessing all stressors facing a marine mammal population also provides the environmental context for evaluating whether an additional activity could threaten it. Under the National Environmental Policy Act of 1969 (NEPA), federal agencies are directed to assess the environmental impacts of their actions, considering direct, indirect, and cumulative effects. Cumulative effects are defined by the U.S. Council on Environmental Quality as “the incremental impact of the action when added to the other past, present and reasonably foreseeable actions” that might interact with a proposed action. Although significant progress has been made in understanding the responses of marine mammals to specific stressors such as noise and toxins, it is not yet possible to provide quantitative estimates of the impact of repeated exposure to a stressor or to predict how different stressors will interact to affect individuals and populations of marine mammals.

The Office of Naval Research, the National Marine Fisheries Service, the Bureau of Ocean Energy Management, and the U.S. Marine Mammal Commission funded the present study in order to review the understanding of cumulative effects of anthropogenic stressors, including sound, on marine mammals and to identify new approaches that may improve the ability to estimate cumulative effects. The statement of task is detailed in Box S.1.

CUMULATIVE EFFECTS

The definition of cumulative effects under the implementing regulations for NEPA focuses on the incremental effect of a proposed human action when added to those of other human actions. In contrast, most biologists view cumulative effects similarly to the U.S. Environmental Protection Agency’s view of cumulative risk, which focuses on the individual animal or population, with effects accumulating when animals are repeatedly exposed to the same or different stressors. In this ecotoxicology-type approach, a noise source would be considered one of a number of stressors experienced by marine mammals and one component of an overall aggregate exposure to noise. Cumulative risk would derive from the combination of noise and other anthropogenic stressors, such as chemical pollution, marine debris, introduced pathogens, fishing, and warming or lower pH induced by carbon dioxide emissions, as well as natural stressors, such as increased presence of predators, pathogens, parasites, or reduced availability of prey due to natural ecological interactions.

In this report aggregate exposure is defined as the combined exposure to one stressor from multiple sources or pathways and cumulative risk as the combined risk from exposures to multiple stressors integrated over a defined relevant period: a day, season, year, or lifetime.

Cumulative risk from exposure to multiple stressors cannot be predicted based on existing scientific theory and data for individual marine mammals or their populations. The Committee developed a Population Consequences of Multiple Stressors (PCoMS) model to provide a conceptual framework for the challenging task of assessing the risks associated with aggregate exposures to one kind of stressor, such as sound, and the cumulative exposure associated with sound and other stressors. To broaden the analysis of cumula-

tive effects to include multiple species and ecosystems, the concept of interaction webs was introduced.

The report distinguishes between two kinds of stressors: an intrinsic stressor (e.g., fasting), which is an internal factor or stimulus that results in a significant change to an animal’s homeostatic set points,¹ and an extrinsic stressor (e.g., noise or a pathogen), which is a factor in an animal’s external environment that creates stress in an animal. It also distinguishes between stressors, defined by how they influence an individual animal, and ecological drivers, which affect levels of organization from populations to ecosystems. An ecological driver is defined as a biotic or abiotic feature of the environment that affects multiple components of an ecosystem directly and/or indirectly by changing exposure to a suite of extrinsic stressors. Ecological drivers for marine mammals include loss of keystone or foundational species, variations in ocean climate (such as El Niño events), and climate change.

Effects of Sound

In this study, the committee was asked to place sound in the context of other stressors to which marine mammals may be exposed. The National Research Council (NRC) report Marine Mammal Populations and Ocean Noise (NRC, 2005) noted that “[n]o scientific studies have conclusively demonstrated a link between exposure to sound and adverse effects on a marine mammal population.” That statement is still true, largely because these impacts are so difficult to demonstrate, but the intervening decade has seen an increasing number of studies showing the effects of ocean noise on individual marine mammals. Under the U.S. Marine Mammal Protection Act (MMPA), regulation of the effects of human activities on marine mammals requires determining the number of individual animals expected to be “taken”² lethally, by injury or by harassment. One current method is to set an all-or-nothing threshold at the sound pressure level corresponding with an estimated probability of response of 50% from the dose–response function. However, the radiation of sound from point source emissions typically exposes many more animals at sound levels below this threshold compared with the number exposed to higher sound levels. Hence, using this threshold leads to potentially significant underestimates of the total number of animals taken. An “effective received level” can be calculated that gives a more realistic take estimate. Still, the effects of sound on marine mammals cannot reliably be condensed into a single estimate of the number of animals affected by a given exposure. Changes in transmission patterns of sound in the ocean, distribution of animals, variable responsiveness of individual animals, and temporal, spatial, and social determinants of response all create uncertainty in the number of animals that will respond behaviorally or physiologically to any defined sound stimulus. Including measures of uncertainty, such as confidence intervals for estimates of predicted take, would be more consistent with the state of knowledge than providing a single number for the MMPA take estimates.

Estimating the effect of sound on marine mammals requires understanding the relationship between acoustic dosage and the probability of behavioral or physiological

___________________

¹ Homeostasis is a characteristic of a system that regulates its internal environment and tends to maintain a stable, relatively constant condition of properties. The normal value of a physiological variable is called its set point.

² A marine mammal “take” is the act of hunting, killing, capture, and/or harassment of any marine mammal, or the attempt at such.

responses of varying degrees of severity. The criterion used under the MMPA for injury induced by sound is noise-induced hearing loss. The distribution of sound exposures that cause permanent hearing loss is estimated from studies of noise levels that cause the onset of temporary shifts in the hearing threshold (temporary threshold shift [TTS] onset) followed by the increase in the amount of TTS with increasing levels of noise. Currently, data on this relationship exist for one species of fur seal, two species of true seals, two species of mid-frequency dolphins, and two species of high-frequency porpoises. Only a few individuals (one to five) of each species have been tested, and within hearing groups there is wide variation in TTS onset and growth with increasing levels of noise. This variation indicates that the physiological effects of sound cannot be generalized based on testing of a few species of marine mammals but will require studies in more individuals of more species. Understanding how the physiological effects of sound become permanent hearing loss requires audiogrametric measurements. Because there are no audiograms available for baleen whales, physiological sound impacts are estimated based on indirect evidence, such as modeling how sound interacts with tissues in the head, estimated historical ocean noise thresholds, and data from other cetacean hearing groups.

For the recommendations that follow, the chapter number is given where supporting text for a particular recommendation can be found.

Recommendation: Uncertainties about animal densities, sound propagation, and effects should be translated into uncertainty on take estimates, for example, through stochastic simulation. (Chapter 2)

Recommendation: Additional research will be necessary to establish the probabilistic relationships between exposure to sound, contextual factors, and severity of response. (Chapter 2)

Significant progress has been made in developing experiments that can estimate acoustic dose–behavioral response relationships in marine mammals. The response criteria selected for dose–response studies have typically had low severity so as not to harm the subjects, but high enough to act as indicators of harassment under the MMPA. However, in the course of these studies some high-severity responses have been observed for signals that were barely audible. The severity levels were established based on assumed effects on individual fitness, and thus severe responses to low sound levels raise concerns regarding population consequences. This will require research to establish (1) the relationship between levels of exposure and the severity of response, (2) the role of behavioral context in determining the dose–response relationship and the response severity, and (3) the most appropriate acoustic dosage measures for sound exposure.

EFFECTS OF MULTIPLE STRESSORS

There is considerable evidence for single-factor stressor effects on marine mammals. Most of these involve physiological and behavioral responses. Dose–response functions have been estimated for a limited number of single stressors. Particular progress has been made in understanding the effects of anthropogenic sound on behavior. Experiments on a few species have estimated dose–response functions, and, once responses have been characterized in this way, monitoring can be used to estimate the scale of effects from sound-producing activities. Studies of effects of pollutants on marine mammal health and reproduction have also estimated dose–response functions, but there are fewer data on dose–response relationships for other stressors.

While the relationship between the dose of a single stressor and the response of an individual animal is relatively straightforward to predict given sufficient data, the addition of a second stressor can add considerable complexity due to the potential for interaction between the stressors or their effects. Stressors may interact in a synergistic or antagonistic manner, where the resulting response is larger or smaller, respectively, than the sum of the individual stressor responses.

Insight about cumulative effects in the individual can be gained by considering mechanisms at the molecular, cellular, and organ system levels. When stressors act through a common pathway, this provides a high potential for interaction because the stressors may provoke physiological perturbations within the same organ or neuroendocrine system. One common assumption of ecotoxicologists is that, if two or more stressors act through a common molecular mechanism, then their doses can be summed to provide a cumulative dose that can then be used with a single dose–response function (dose addition). Many dose–response functions are sigmoidal in shape or are otherwise nonlinear, and in these cases the sum of two doses may produce a response that is greater or less than the added responses to each stressor alone (response addition). A simple example to illustrate the complexity introduced when a dose–response function is nonlinear is discussed below.

Consider two stressors that act through a common molecular mechanism and are therefore considered eligible for dose addition. After correcting for different strengths (e.g., a toxicity factor for chemical stressors), the doses of the two stressors can be added to give a combined dosage and compared to a dose–response function (see Figure S.1). Stressor A has an effect of 0.10 given a dose of 40 units (see Figure S.1a), and stressor B has an effect of 0.20 given a dose of 60 units (see Figure S.1b). If the responses were additive (response addition), then the response to stressor A and B combined is expected to be 0.30. However, due to the sigmoidal shape of the dose–response function, the added doses of the two stressors (100 units) produces an effect of 1.0, more than threefold higher than the sum of the

individual responses (see Figure S.1c). Therefore, although these stressors are considered additive in terms of dosage (dose addition), they produce a synergistic response. Note that this same phenomenon could also occur with aggregate exposure to a single stressor. Even for this simple situation, a prediction cannot be made of the effects of most stressors unless the dosages, the relative strengths of the stressors, and the dose–response functions are known.

The interaction of stressors that act through different mechanisms but still involve a common adverse outcome pathway may be more difficult to predict due to the complexities of signaling pathways and the existence of feedback loops. For example, stressors such as noise, prey limitation, and some chemical pollutants can induce responses involving the neuroendocrine system known as the hypothalamic-pituitary-adrenal (HPA) axis that controls reactions to stress and regulates many body processes, albeit potentially through differing mechanisms. Chronic activation or perturbation of the HPA axis may be an important mechanism through which cumulative effects arise, and the nature of these effects will be difficult to predict. In cases such as this where there are common adverse outcome pathways but potentially differing mechanisms, the form of interaction between two stressors could be estimated by determining the dose–response relationships for one stressor at different dosages of the second stressor. However, this type of study would be extremely difficult if not impossible to conduct, particularly when more than two stressors are involved, and mechanistic models may be a more appropriate approach to elucidate potential effects. Unfortunately, mechanistic models generally require a detailed understanding of the biochemical and physiological systems, and this is often lacking for marine mammals.

A review of the literature revealed that many stressors

whose effects are mediated through common adverse outcome pathways are therefore more likely to interact. The examination of common adverse outcome pathways underscores the importance of understanding and detecting changes at lower levels of biological organization, such as at the cellular or organ response level, before they exert potentially irreversible effects at individual or population levels. However, it is also imperative to collect information to understand the linkages and processes by which such lower-level responses eventually translate into individual or population-level impacts.

The influences of multiple stressors on marine mammals might be inferred from studies of other species, such as nonmammalian marine species or terrestrial mammals. However, this can be problematic because marine mammals have evolved unique morphologies, behaviors, and physiologies as adaptations for life at sea.

Most existing research on interactions between effects of stressors on marine systems involves factorial experiments with species or systems in settings where treatments can be replicated and controlled. Factorial experiments are useful for detecting the presence of interactions but, because such systems are usually only exposed to one level of each stressor, they rarely provide sufficient information to predict responses at varying levels of stressors present in nature. Meta-analyses of results from studies of multiple stressors on various marine species have been conducted, but no general pattern has emerged for predicting how the effects of stressors will interact. Findings from each specific study were categorized as additive (i.e., noninteractive), synergistic, or antagonistic. One review paper reported that synergy is more common when more than two stressors are added to a system; another study found no evidence of antagonistic interactions between physiological responses. Beyond these generalities, the committee found no information to help predict the influences of multiple stressors on marine mammals. Given the difficulty in predicting interactions, cumulative effects assessments often assume that stressor effects are additive. However, work on other species indicates that this assumption is often wrong.

A rigorous approach for testing interactive effects of multiple stressors involves factorial experiments using a range of levels of each stressor coupled with some tests of mixtures of stressors. But for both practical and ethical reasons, such experimental approaches are often not possible for marine mammals, in which case inferences must be based on quasi-experiments: patterns associated with stressor variation in space or time. Although such data are subject to confounding and thus multiple interpretations, reasonably strong inferences are often possible from time-series analyses and weight of evidence approaches.

One type of single-stressor experimental study design could select subjects from the wild population to sample the cumulative effects of exposure to sound along with the combination of stressors currently found in that population. If this type of study adds one stressor to subjects in the wild whose exposure to other stressors can be documented, the cumulative effects of the single stressor then can be evaluated in the context of the full complement of environmental stressors. The interpretation of these single-stressor experiments in terms of cumulative effects is difficult because the exposures to preexisting stressors are difficult to quantify. Also experimental addition of a stressor is limited for ethical reasons to stressors such as sound, where the added stressor can be controlled in terms of both intensity and duration of exposure. In situations where the current pattern of exposure to stressors is expected to change in the future beyond the levels currently experienced, such as those caused by changes in ocean climate, this approach for studying cumulative effects will be inadequate.

The exposure of marine mammals to stressors has been estimated by mapping stressors in both space and time. However, in order to understand cumulative effects, mapping of stressors needs to be accompanied by mapping the distribution of marine mammal species of concern, because stressors must overlap with the species to exert an effect. Another approach, which is common for chemical stressors, is to sample tissue from a marine mammal to characterize its dosage of the stressor. Biopsies are now a standard remote sampling method for marine mammals that cannot be handled. The development of new methods for remote sampling of blood and other tissues for estimating dosage of stressors from marine mammals at sea are included in a recommendation later in this summary. On-animal dosimeters could also provide a time series of stressor exposure measurements for individual animals.

A MODEL FOR HEALTH AND POPULATION CONSEQUENCES OF MULTIPLE STRESSORS

The PCoMS model (see Figure S.2) developed in this report provides a framework for exploring pathways from stressor exposure to effects on health to effects on populations. Following the general structure of the Population Consequences of Acoustic Disturbance model developed in NRC (2005), PCoMS documents the pathways from exposures to stressors through their effects on physiology, behavior, and health to their effects on vital rates and population dynamics. A key component of this framework is an assessment of the health of individuals. A variety of health indices, including allostatic load, energy stores, immune status, organ status, stress levels, contaminant burden, and parasite load, are discussed. Appropriate health indices integrate the potential effects of physiological and behavioral responses to multiple stressors on fitness over a time scale that is longer than the duration of the responses themselves but shorter than the response time of vital rates. Such indices can provide early indicators of risk of reduced survival and reproduction before an actual alteration in these rates and can increase

understanding of the mechanisms by which these stressors affect fitness.

The committee developed a number of research recommendations that are designed to address the PCoMS model and measures of stressors and health:

Recommendation: Future research initiatives should include efforts to develop case studies that apply the PCoMS framework to actual marine mammal populations. (Chapter 5)

These studies will need to estimate exposure to multiple stressors, predict changes in behavior and physiology from those stressors, assess health, and measure vital rates in order to parameterize the functional relationships between these components of the framework. Where possible, the data on changes in demography, population size, and the health of individuals collected in these studies should be used to improve estimates of the parameters of the PCoMS model and reduce uncertainty.

Recommendation: Future research initiatives should support evaluation of the range of emerging technologies for sampling and assessing individual health in marine mammals, and identification of a suite of health indices that can be measured for diverse taxa and that best serves to predict future changes in vital rates. (Chapter 8)

Potentially relevant measures include hormones, immune function, body condition, oxidative damage, and indicators of organ status, as well as contaminant burden and parasite load. New technology for remotely obtaining respiratory, blood, and other tissue samples and for remote assessment (e.g., visual assessment of body condition) should also be pursued.

Comprehensive health assessments are not only a critical component of the PCoMS framework, but they can also be used to serve as early warning indicators of risk before the consequences have population-level effects. There are some populations of marine mammals where periodic health assessments can include a sufficient sample of individuals to assess population health. To optimize usefulness for management, there is a need to develop databases of stressors and effects measured using established standards. For species that cannot be handled, methods are not currently available to obtain the samples used to assess health.

Establishing baseline values of health indices and their associations across life history stages in marine mammal species will provide critical information for assessing individual and population health. Cross-sectional sampling and repeated sampling from the same individuals of blood or other tissues during critical life-history phases can help to document exposure to and health effects of extrinsic stressors within the context of annual cycles and life cycles of intrinsic stressors. Long-term studies of known individuals are required for longitudinal studies.

Recommendation: Agencies charged with monitoring and managing the effects of human activities on marine mammals should identify baselines and document exposures to stressors for high-priority populations. (Chapter 8)

High-priority populations should be selected to include those likely to experience extremes (both high and low) of stressor exposure in order to increase the probability of detecting relationships. This will require stable, long-term funding to maintain a record of exposures and responses that could inform future management decisions. Information on baselines and contextual variables is critically important to interpreting responses.

Recommendation: A real-time, nationally centralized system for reporting marine mammal health data should be established. (Chapter 7)

Recommendation: Standards for measurement of stressors should be developed along with national or international databases on exposure of marine mammals to high-priority stressors and associated health measures that are accessible to the research community. (Chapter 8)

Recommendation: Techniques should be developed that will allow historical trajectories of stress responses to be constructed based on the chemical composition of the large number of baleen whale earplugs and baleen samples in museums or similar natural matrices in other species. Artificial matrices should be studied for their potential to absorb materials (hormones or chemical stressors) and thereby provide a record of exposures and responses to stressors. (Chapter 8)

Recent work on baleen whales has shown that some tissues that lay down layers with time, such as baleen or a waxy earplug, can provide a record of stress, reproductive hormones, and some contaminants for up to the entire lifespan. Large archival collections of such tissues could be analyzed to provide time series of data that could yield critical information on the relationships between contaminants, stress, and reproductive intervals in baleen whales. Other materials that lay down semiannual layers, such as teeth, could be assessed for their potential to record stressor and life-history information over long periods of time. In addition, artificial materials could be tested for their capacity to store chemical stressors and hormones over long enough time periods to test the relationship between exposure to the stressors and response in terms of health or vital rates.

ECOSYSTEM-LEVEL EFFECTS

The committee broadened its review from cumulative effects of stressors on marine mammals to consider how interactions among stressors may affect entire ecosystems. The distribution and abundance of species in an ecosystem are determined by the interactions among and between species and abiotic environmental elements, which together define an interaction web (see Figure S.3).

In an interaction web, species or abiotic elements that affect the distribution and abundance of a selected species are called drivers of the recipient species. When a driver affects the recipient directly, for example, when gill nets entangle and kill marine mammals, this is called a direct effect. When a driver affects a second driver that in turn affects the recipient, this is called an indirect effect. For example, human fisheries might reduce the population of a fish species that feeds on the same prey as a marine mammal. If this reduction in the competitor species increased the abundance of prey for the marine mammal species, it might have an indirect positive effect on the recipient species. Known or suspected drivers for marine mammals include ocean climate, prey limitation, predators, fishing bycatch, toxins, and pathogens. Interaction webs can help identify the suite of factors that need to be considered in evaluating cumulative effects on populations and ecosystems. As with the PCoMS model, interaction webs do not provide an algorithm for predicting cumulative effects; they serve primarily to identify the most important components of any comprehensive model of cumulative effects, including indirect effects. Interaction webs and the PCoMS model would need to include mathematical functions that describe the relationships between the different compartments before they could be used to predict those effects. Estimating these functions will be extremely challenging.

MANAGEMENT OF CUMULATIVE EFFECTS

The critical question for predicting risk of cumulative effects asks what combinations of stressors dosages elevate the cumulative effect enough to pose a risk to populations and ecosystems. The committee’s review indicates that the strength of effects cannot currently be predicted based on specific levels of exposure to multiple stressors for marine mammals. Once populations or ecosystems are found to be at risk of adverse impacts, the critical issue for selecting management actions is to decide what combination of stressors to reduce in order to bring the population or ecosystem into a more favorable state. The committee concluded that current scientific knowledge is not up to the task of predicting cumulative effects of different combinations of stressors on marine mammal populations. Even though exposure to multiple stressors is an unquestioned reality for marine mammals, the best current approach for management and conservation is to identify which stressor combinations cause the greatest risk. The committee developed a decision tree that can be used to identify situations where a detailed study of potential cumulative effects should be given a high priority (see Figure S.4). The decision tree was applied to three case studies demonstrating its utility.

Recommendation: Situations where studies of cumulative effects should be prioritized can be identified using tools such as the decision tree developed by the committee

and by testing for whether pathways for adverse health outcomes are shared across stressors. (Chapter 4)

Given that it is problematic to predict when stressors may interact to produce strong effects, there is a critical need for early indicators of risk. However, it is not possible to detect even substantial declines in the size of many marine mammal populations, because precision on population estimates is generally low. Although new survey technologies and analysis methods are improving precision somewhat, it is doubtful that the financial resources and scientific methods are sufficient for adequate population assessments.

Despite the uncertainty, regulators must make decisions on whether and where to allow potentially harmful anthropogenic activities to take place. The concept of adaptive (resource) management offers a framework for making such decisions. In this approach, hypotheses are developed based on current understanding; the optimal action is determined taking into account not just this understanding but also what may be learned as a result of each management action. Adaptive management is also used to identify the optimal data collection strategy to reduce uncertainty.

Recommendation: Responsible agencies should develop relatively inexpensive surveillance systems that can provide early detection of major changes in population status. (Chapter 7)

Surveillance systems should be developed first for populations that currently lack adequate stock assessments. To be most effective in providing an early warning, the variables monitored will depend on the species and situation, and may change over time with development of new technology and increasing ecological knowledge. Indices of population health, such as mother-to-calf ratios and body condition, are potentially sensitive measures. Abundance indices, such as calibrated acoustic detection rates, may also be appropriate in some circumstances. All measures considered should be evaluated in the context of their ability to inform alternative hypotheses about the mechanisms underlying population changes so that, if a negative change is detected, an early start on evaluating the possible cause could be made. For example, declines in population health indices may indicate increases in exposure to anthropogenic stressors, but they may alternatively be caused by an increase in population size approaching carrying capacity.

Recommendation: Adaptive management should be used to identify which combinations of stressors pose risks to marine mammal populations, and to select which stressors to reduce once a risk is identified. (Chapter 6)

Once a population of marine mammals has been found to be at risk, managers need to identify a stressor or suite of stressors whose reduction can reduce this risk. It may not be possible to reduce some stressors or ecological drivers that contribute to risk. For example, it simply may not be possible to remove persistent toxicants or reverse warming in the ocean due to climate change. This leaves those stressors that in practice can be mitigated within a time period consistent with the population’s rate of decline or recovery. Among these remaining stressors, or combination of stressors, it will be important to next identify those whose reduction would be most effective at decreasing the risk. These considerations can be used to establish research priorities for estimating dose–response functions. This approach suggests a new form of effect study—experiments that remove or reduce one or more stressors to study effect of reduction. This experimental design may be more appropriate for adaptive management than the more traditional experiments that add stressors to the current baseline.

The committee recognizes that the state of the science of cumulative effects has low predictive power compared to regulatory demands to assess these effects. The most important goals for managing cumulative effects are (1) identifying when the cumulative effects of stressors risk transitioning a population or ecosystem to an adverse state and (2) identifying practical reductions in stressors to reduce this risk.