Characterizing driver-response relationships in marine pelagic ecosystems for improved ocean management

Abstract

Scientists and resources managers often use methods and tools that assume ecosystem components respond linearly to environmental drivers and human stressor. However, a growing body of literature demonstrates that many relationships are non-linear, where small changes in a driver prompt a disproportionately large ecological response. Here we aim to provide a comprehensive assessment of the relationships between drivers and ecosystem components to identify where and when non-linearities are likely to occur. We focus our analyses on one of the best-studied marine systems, pelagic ecosystems, which allowed us to apply robust statistical techniques on a large pool of previously published studies. In this synthesis, we (1) conduct a wide literature review on single driver-response relationships in pelagic systems, (2) use statistical models to identify the degree of non-linearity in these relationships, and (3) assess whether general patterns exist in the strengths and shapes of non-linear relationships across drivers. Overall we found that non-linearities are common in pelagic ecosystems, comprising at least 52% of all driver-response relationships. This is likely an underestimate, as papers with higher quality data and analytical approaches reported non-linear relationships at a higher frequency -on average 11% more. Consequently, in the absence of evidence for a linear relationship, it is safer to assume a relationship is non-linear. Strong non-linearities can lead to greater ecological and socio-economic consequences if they are unknown (and/or unanticipated), but if known they may provide clear thresholds to inform management targets. In pelagic systems, strongly non-linear relationships are often driven by climate and trophodynamic variables, but are also associated with local stressors such as overfishing and pollution that can be more easily controlled by managers. Even when marine resource managers cannot influence 3 ecosystem change, they can use information about threshold responses to guide how other stressors are managed and to adapt to new ocean conditions. As methods to detect and reduce uncertainty around threshold values improve, managers will be able to better understand and account for ubiquitous non-linear relationships.

The Baltic Health Index (BHI): Assessing the social–ecological status of the Baltic Sea

“…The assessment of the likely future status rewards the presence of regulatory and management measures which we estimated using the (Giakoumi et al, 2015;Hunsicker et al, 2015). In general, reference points should be science informed, but optimally would be to develop these reference points in a co-design process with diverse stakeholders and scientists in order to define goals of restorative and active intervention and implement appropriate management measures (Franke et al, 2020).…”

“…This shortcoming is exacerbated by the common practice of setting reference levels based on single sectors and single impacts. However, clearly impacts such as eutrophication and fisheries are linked, and additive and/or synergistic effects of cumulative drivers affect their status which should be reflected in the reference level (Giakoumi et al, 2015;Hunsicker et al, 2015). In general, reference points should be science informed, but optimally would be to develop these reference points in a co-design process with diverse stakeholders and scientists in order to define goals of restorative and active intervention and implement appropriate management measures (Franke et al, 2020).…”

Practical implementation of cumulative‐effects management of marine ecosystems in western North America

“…2016; Hunsicker et al. 2016). Simulation models show that management actions based on the assumption of additive effects can be more damaging than no action when the underlying multiple stressor interactions are actually antagonistic (e.g., Brown et al.…”

Lack of definition of mathematical terms in ecology: The case of the sigmoid class of functions in macro‐ecology

“…This definition also excludes some functions that were pre- Thus, a more flexible use of the sigmoid function in these logistic regressions can prove to be of great use (Godeau and Gosselin, Eide et al, 2012;In prep.). Along the same lines, sigmoid and logistic functions are sometimes confused with each other, whereas the latter is nothing more than a particular type of sigmoid (e.g., Hunsicker et al, 2015). Such confusion may prevent researchers from considering other families of functions that fall into the sigmoid class without being logistic.…”