Identification and testing of leading indicators of state changes in marine ecosystems

Can early warning indicators be identified and used across marine ecosystems?

Big Picture

When disturbed, marine systems can undergo abrupt state changes that surprise managers and disrupt ecosystem services. Recent studies have proposed a suite of statistical methods for identifying leading indicators (or early warning indicators) of such state changes, and experiments in model systems suggest the potential efficacy of these new tools for managers in the real world. Our goal is to apply these new methods to evaluate leading indicators of state changes across a number of different marine ecosystems. We will pay particular attention to the propensity of alternative indicators to provide accurate information about the occurrence, timing, and extent of state shifts.

Why we are doing it

Identifying leading indicators of ecosystem state changes can offer critical lead-time for managers to make preventative adjustments, or, at a minimum, allow them to prepare.

How we are doing it

We will:

  1. Assemble data from several marine ecosystems where documented state shifts have occurred.
  2. Visually identify temporal patterns in drivers and response variables.
  3. Locate and statistically test one or more breakpoints in time-series of response variables.
  4. Calculate several leading indicators (increased variability, decreased responsiveness, flickering) that have been proposed in theoretical and non-marine systems.
  5. Determine whether leading indicator provides a true/false positive/negative indication of approaching thresholds.
  6. Quantify the accuracy and amount of lead time provided by alternative leading indicators across marine ecosystems, and how these factors are influenced by the extent of ecosystem state shifts.

What we are discovering

We expect to find that:

  1. Leading indicator analyses are data hungry.
  2. Each focal system is likely to require a unique indicator or set of indicators, best suited to the dynamics of the system.
  3. A mechanistic understanding of the underlying drivers of state shifts can help to avoid use of indicators that provide false positive or false negative information. 

What’s next

We are gathering metadata about marine ecosystems in which state shifts have already been documented.

Outcomes

Our evaluation of candidate leading indicators across multiple marine ecosystems will suggest whether generic leading indicators of ocean tipping points exist, and which leading indicators tend to be more risk-averse (higher probability of providing false positive information about an approaching ecosystem threshold) or risk-prone (higher probability of providing false negative information about an approaching ecosystem threshold).

Leads

Mary Hunsicker, Kendra Karr, and Mary Donovan

Anticipated Completion

Fall 2016


Learn more about our other Research Activities with the Ocean Tipping Points project.