Early warning signals, nonlinearity, and signs of hysteresis in real ecosystems

Abstract

Abstract. Early warning signals (EWS) might dramatically improve our ability to manage nonlinear ecological change. However, the degree to which theoretical EWS predictions are supported in empirical systems remains unclear. The goal of this study is to make recommendations for identifying the types of ecological transitions that are expected to show EWS. We conducted a review and meta-analysis of published studies and comparative analysis of eight northeast Pacific Ocean time series to illustrate the importance of testing for nonlinearity in empirical EWS studies. We found that published studies demonstrating nonlinearity in ecosystem dynamics are more likely to support EWS predictions than studies with linear or undetermined dynamics. The northeast Pacific time series in our analysis were often too short for formal tests of nonlinearity, a common problem in empirical studies. To assess the evidence for nonlinear dynamics in these data, we tested for state-dependent driver-response relationships consistent with hysteresis, a central feature of nonlinear ecological models. This analysis supported the results of the literature metaanalysis. Four time series with driver-response relationships consistent with hysteresis generally supported theoretical EWS predictions, while four without evidence of hysteresis failed to support EWS predictions. Theoretical support for EWS is largely generated from nonlinear models, and we conclude that tests for either nonlinear dynamics or hysteresis are needed before employing EWS.

Defining ecosystem thresholds for human activities and environmental pressures in the California Current

“…For example, Bestelmeyer et al. () provided evidence for a linear relationship between euphausiid abundance and the PDO in the Southern California Current, and this conclusion has been complemented more recently by analyses showing that copepod species and coho salmon in the Northern California Current linearly track shifts in the PDO (Litzow and Hunsicker ). In contrast, our analyses imply that as many as five of the nine ecosystem states we evaluated exhibited threshold increases in response to negative PDO s values (copepods, scavengers, groundfish, and marine mammals; Fig.…”

“…For instance, the concept of rising variance, where the variability in an ecosystem state(s) can provide advance warning of an ecosystem shift, has led to new insights about both terrestrial and marine systems (Carpenter and Brock , Sydeman et al. , Litzow and Hunsicker ). Many analytical techniques for identifying ecosystem thresholds based on state–pressure relationships have also been proposed (Samhouri et al.…”

“…Determining where ecosystem thresholds occur, and how large a shift may be induced by crossing them, is key for informing management that prepares for and is designed to avoid abrupt and undesired changes (Doak et al 2008). For instance, the concept of rising variance, where the variability in an ecosystem state(s) can provide advance warning of an ecosystem shift, has led to new insights about both terrestrial and marine systems (Carpenter and Brock 2006, Litzow and Hunsicker 2016. Many analytical techniques for identifying ecosystem thresholds based on state-pressure relationships have also been proposed (Samhouri et al 2010, Large et al 2013, 2015a, b, Hunsicker et al 2016, leaving an open question: Which is most appropriate, and under what conditions?…”

Network-based metrics of resilience and ecological memory in lake ecosystems

“…However, current metrics for evaluating resilience 46 loss (e.g. time-series metrics of critical slowing down) have mixed reliability when applied as 47 early warning signals (EWS) in freshwater [17] and other ecosystems [18][19][20][21]. 48 Methodological issues include the choice of system variable for analysis, false or missed 49 alarms, and subjective parameter choices, while palaeorecord applications presents an 50 additional challenge of variable temporal resolution due to compaction or changing 51 accumulation rates [7,[22][23][24][25].…”

“…However, current metrics for evaluating resilience 48 loss (e.g. time-series metrics of critical slowing down) have mixed reliability when applied as 49 early warning signals (EWS) in freshwater [17] and other ecosystems [18][19][20][21]. 50…”