In aquatic ecosystems like the Chesapeake Bay, when inputs or conditions change, such as a decrease in nutrient loads, threshold-type responses may occur. Examples in the Chesapeake region include the tidal fresh region of the Potomac River proper and a tributary to it (Gunston Cove), the Patuxent River mesohaline region, and the Back River oligohaline region. These examples, which are discussed in greater detail in the Case Studies section of the report, include evidence of linear responses with and without time lags, threshold responses, and hysteretic responses where recovery trajectories differ markedly from degradation pathways. Examples of ecosystem feedback mechanisms observed in the Chesapeake include: (1) a positive feedback relationship between the growth of submerged aquatic vegetation (SAV) and a decrease in total suspended solids (TSS), leading to an improvement in water clarity, followed by abrupt SAV recovery, (2) improved benthic filtration by a clam in the Potomac (Corbicula fluminea), followed by improved water clarity and SAV recovery, and (3) hypoxia in the Bay reinforcing eutrophic conditions (positive feedback) through enhanced nutrient recycling efficiency despite stable or decreasing nutrient loads. Complex interactions between top-down (food web-driven) and bottom-up (nutrient-driven) controls make these kinds of ecological feedbacks difficult to document and analyze. The take-home points here are that we need to: (1) mine existing monitoring and historical data for signs of ecological thresholds and hysteresis, (2) undertake additional research that focuses specifically on quantifying nonlinear feedback mechanisms, and (3) incorporate feedback processes into management efforts in an adaptive and iterative manner.

The Neuse River-Pamlico Sound estuarine system in North Carolina followed particular pathways to reach the eutrophied state that characterizes it today. Its future will be affected by a number of factors, including the prediction that more intense and frequent storms will affect this region with global climate warming.

The Neuse River has a relatively long water residence time that has exacerbated an increase in nitrogen and phosphorus loading by some 30 percent over 40 years of agricultural, urban, and industrial expansion. The region has also felt the impact of several major hurricanes, including Floyd, Dennis, and Irene (1999) and Fran (1996).

Understanding the historical pathways that led to eutrophication will help inform future management strategies. The ModMon and FerryMon monitoring systems help to provide key data by sampling the system at high-frequency temporal and spatial scales. Long-term monitoring data show that phytoplankton chlorophyll levels have decreased upstream over time, but increased downstream. This suggests that the management decision to decrease loading rates for phosphorus only, without parallel decreases in nitrogen loading, led to this regional disparity in chlorophyll abundance and the associated downstream progression of eutrophication effects. To change course, a nitrogen-input threshold must be established. Chlorophyll a levels could serve as the measured response indicator used to evaluate whether management actions are working.

Extreme storm events, like hurricanes, can overwhelm the impact of effective nutrient management strategies. Hurricane Floyd, for example, reduced the residence time of Pamlico Sound from one year to one week. Storms can also change the abundance and community composition of phytoplankton, which can have feedback in other parts of the food web. Establishing nutrient loading thresholds during a period of potentially elevated hurricanes will pose a clear challenge.

Efforts in the Florida Everglades and Grand Canyon provide two case studies where adaptive management has been used to mitigate undesirable shifts.

To pursue effective adaptive management, we need to understand six key concepts related to nonlinear feedback systems: (1) such feedback systems are ubiquitous and can occur in terrestrial, freshwater, or marine systems; (2) variables that drive nonlinear responses occur at different spatial and temporal scales; (3) in most instances, only a “handful” of key ecological variables (3 to 6) are largely responsible for driving state changes; (4) thresholds are dynamic and difficult to predict; (5) resilience can be lost as the result of overcapitalization (increased nutrients and biomass), hyper-connectivity in space, and loss of functional diversity (trophic cascade); and (6) ecosystem structure and function are coupled to human institutions and preferences.

The relationship between the ecological system and human institutions can be characterized by the “Pathology of Command and Control.” This feedback loop links ecosystem state → ecosystem services → human preferences → action, which then in turn feeds back to affect ecosystem state. Ecological changes and accompanying management actions can be described for the Florida Everglades and Grand Canyon examples.

Adaptive management and governance provides a set of tools for accommodating uncertainty in future ecosystem responses. Social response to ecological crisis often plays a key role in initiating management actions. Since regime shifts can be either reversible or irreversible, knowing when to adapt to change or invest in a reverse transformation is key. Ecological resilience plays an important role in this consideration, as it can provide a buffer for experimentation. For adaptive management to succeed, institutional frameworks should be learning-based and open to change.

Biogeochemical feedback mechanisms play an important role in threshold responses. In Europe, the THRESHOLDS of sustainability project serves as an ongoing effort to develop operational tools to identify thresholds, threshold behavior, and point-of-no-return values for coastal systems. Through this effort, scientists hope to use these tools to set policy targets in nutrient and contaminant inputs (learn more at http://www.thresholds-eu.org/).

It is essential to employ appropriate methods for identifying and testing the significance of thresholds in the environment and useful to know how these approaches have been applied to various case studies, including specific statistical tools and models. Case studies presented here focused on the relationship between biogeochemical cycles and hypoxia in Danish estuaries and in the Chesapeake Bay. As these cases show, hypoxia tends to be linked to threshold behaviors because of its effect on changes in benthic communities, with the loss of deep-dwelling organisms that oxidize the sediments and cause dramatic changes in biogeochemical processes. With the sediment’s oxidation capacity diminished, sediment metabolism switches to less efficient anaerobosis, with different pathways for organic matter remineralization.

Both nitrogen and phosphorus cycles are affected by hypoxia, leading to an increase in recycling of ammonium and dissolved inorganic phosphorus, which tends to promote further algal growth. Data from both Chesapeake and Danish waters support the idea that hypoxia may cause a “stuck-in-rut” effect that inhibits a return to a less eutrophic ecological state. Global climate warming will likely make matters worse. Projected temperature increases may mean lower oxygen saturation, higher rates of respiration, and a resulting increase in system heterotrophy, where rates of respiration exceed rates of primary production.

Regime shifts in various lakes in Denmark and elsewhere highlight the importance of bottom-up and topdown controls and trophic dynamics as mechanisms underpinning threshold responses. In many instances, response trajectories in lakes differed depending on whether nutrient loads are increasing or decreasing. Temperate lakes also tend to respond differently from subtropical lakes and brackish tend to respond differently from freshwater lakes.

As nutrient gradients (phosphorus in this case) increase, the trophic organization in the middle of the food web changes — especially among zooplankton and fish. The relative importance of zooplankton-feeding fish (planktivores) increases while that of fish-eating-fish (piscivores) decreases. In eutrophic lakes phytoplankton production dominates, while benthic photosynthesis tends to be more important in oligotrophic lakes. Experimentally increasing the abundance of submerged aquatic macrophytes in lake systems can sometimes reverse this change in food web dynamics. In shallow freshwater lakes, macrophytes remove nutrients for growth and provide refuges for zooplankton (Cladocerans) which control phytoplankton and water clarity. In shallow brackish lakes, macrophytes also assimilate nutrients, enhance denitrification, and stabilize sediment — all of which helps to favor a clear water, oligotrophic state. However, in brackish and saline systems, copepods, which are less efficient in regulating phytoplankton, dominate the zooplankton.

Warm lakes tend to have many fish and few zooplankton, while colder lakes have fewer fish, but many zooplankton. Food web dynamics differ accordingly — in subtrobical lakes, fish feed directly on periphyton. Salinity makes a difference too. Copepods tend to be more abundant at higher salinities. Low salinity tends to correlate with lower fish density and lower chlorophyll.

Chemical and biological resistance might cause time delays in the response of water bodies to decreased nutrient loading. Lakes take an average of 10 to 15 years to respond to changes in nutrient loading. The delay is caused by both biogeochemical factors and lag times in concomitant changes in the organization of the food web (especially birds and fish). In many cases, the delay is also caused by a combination of relatively long residence time for water volumes and by large pools of phosphorus accumulated in lake sediments.

Experiments with biomanipulation (adding/removing fish and plants from the system) have been conducted in multiple European lakes. While such efforts can have an impact, they do not prove a substitute for a decrease in nutrient loading. Treatments have to be repeated in order to have a sustained effect and, therefore, may be more useful as a management tool to maintain a state against natural odds, rather than as a restoration tool.

Reducing nutrient loading is the lynchpin to regime shifts. Take-home messages for Chesapeake Bay are: (1) reduce nutrient loading as much as possible; (2) demonstrate to the public that this works by putting the most concentrated effort toward the upper arms (shallow, freshwater) where the response is likely to be the most dramatic; (3) conduct comparative studies within and between bays and within and between years to help make decisions about where to allocate greatest effort and to set target loadings; and (4) undertake more large-scale experiments (using both exclosures and enclosures).

Dramatic changes have occurred in San Francisco Bay over the past two decades. These demonstrate that the connection between changes in phytoplankton biomass and nutrient loading can be shaped by factors suchas food resources, transport processes, and change in biological community structure.

Like Chesapeake Bay, San Francisco Bay has experienced human-driven nutrient loading, with increasing levels of nitrogen and phosphorus over recent decades. Although similar in scale to Chesapeake Bay, it differs in its ratio of watershed to estuary area, residence time, tidal currents, turbidity, and macrophyte abundance. Additionally, the north and south basins of San Francisco Bay differ greatly from each other — the north bay is river-driven and has low salinity, while the south bay is a marine lagoon.

In the north bay, the invasive clam Corbula amurensis, which first appeared in 1987, dramatically altered spring bloom dynamics and organization of the food web. Primary production decreased and populations of the zooplankton Eurytemora and mysid shrimp declined as a result. Numbers of juvenile striped bass have also fallen since 1987. The south bay did not experience the same changes as the result of the Corbula invasion, continuing to experience spring blooms with a regular pattern, although with varying intensity and duration. In 1999, the spring bloom pattern in the south bay changed, with a secondary bloom appearing that surpassed the spring bloom in magnitude.

Several hypotheses may explain the appearance of a secondary bloom at a time when nutrient loading has been declining as the result of management interventions. These involve turbidity, contaminants, physical transport, and changes in food web structure as possible culprits. Ultimately, both transport processes and trophic dynamics may be interacting in a complex way. A primary hypothesis: Upwelling intensity in the coastal Pacific Ocean has increased in recent years due to climatically driven factors, causing high phytoplankton biomass in the Pacific Ocean. When the wind relaxes or reverses, offshore biomass may be transported into San Francisco Bay. Meanwhile, the abundance of Corbula clams has been decreasing since the late 1990s likely due to increased predation by fish, thus diminishing their impact in limiting phytoplankton abundance. This combination of both climatically-driven and food web-driven factors can explain the change in bloom dynamics, underscoring the importance of using a multi-faceted approach when considering the interaction between phytoplankton abundance and nutrient loading.