New diagnostics for multiply stressed marine and freshwater ecosystems: integrating models, ecoinformatics and big data

It is now a truism to state that the Earth is under threat from pervasive human activity, and that we have entered a new era of accelerating planetary change: The Anthropocene (Steffen et al. 2011). The convergent development of earth-observation systems, coupled with high-performance computing platforms linked through the internet have revealed the fragile state of the biosphere at a critical tipping point in terms of the preservation of biodiversity and its associated ecosystem services (Barnosky et al. 2012). The diverse habitats present in aquatic ecosystems – the Earth's oceans, coasts, lakes and rivers – are under imminent threat from a variety of human-induced, interconnecting stressors arising from a changing climate, associated sea-level rise, alterations to the global water cycle and the consequences of globalised trade (Liu et al. 2015). There is now general recognition that environmental problems require complex solutions, and that observing and understanding ecological change in aquatic ecosystems cannot ignore the issue of multiple causality arising from complex stress regimes (e.g. Jackson et al. 2016). Against this background, a workshop was held from 10 to 12 September in 2014 at the Sydney Institute of Marine Sciences in Sydney, NSW. Working groups of scientists from a diversity of backgrounds and disciplines explored three key aspects of multiple-stressor impacts on aquatic ecosystems: (1) defining stressors at ecosystem scale using knowledge of the temporal and spatial scaling of stressor effects and exploring the role of geospatial data in stressor mapping; (2) new technologies for the acquisition of effects data and the development of diagnostics in complex ecosystem-stressor scenarios, including genomics, modelling and statistical tools; and (3) how new technologies and approaches for large-scale ecosystem assessment can be employed in the regulatory arena, to link assessments from headwaters to open oceans in ecological risk assessment (ERA).

The Driver–Pressure–State–Impact–Response framework developed by the European Environment Agency (Smeets and Weterings 1999) was chosen as a suitable heuristic model for multiple-stressor impacts at ecosystem scale, providing a focus for the resulting workshop outputs. Fig. 1 illustrates the linkages and pathways by which environmental ‘drivers’, resulting from a combination of anthropogenic activity and natural variability, result in the action of ‘pressures’ (here, and in the workshop papers, pressures is synonymous with stressors) on ecosystem ‘state’ (e.g. the habitat template, biodiversity). This in turn causes ‘impacts’ which can be observed as ecosystem trends, leading to degradation and, ultimately, ruin (sensu Taleb et al. 2014). Ecosystem ‘responses’ are observed in terms of their capacities to return to a previous state (resilience or antifragility) or their tendency to collapse (fragility) sensu Taleb (2014).

Fig. 1.  The Driver–State–Pressure–Impact–Response (DPSIR) framework adapted to represent the impacts on natural ecosystems caused by multiple stressors, arising from anthropogenic activity (after Smeets and Weterings 1999).

The first paper in the workshop series by Dafforn et al. (2016) focuses on two critical issues in multiple-stressor ecology: (i) developing a conceptual framework for study, which takes into account issues of spatial and temporal scaling in relation to the DPSIR framework; and, (ii) the availability of suitable data, including ‘big data’ sources, which can provide information on Driver–Pressure–State relationships. Their paper includes a brief critique of existing multiple-stressor approaches, which are largely ad hoc and expert knowledge-driven, and thus suffer from a lack of formal rigour, and provide a poor model for a multiple-stressor paradigm, which should be extendable from ecosystem to planetary scale. The second paper, by Chariton et al. (2016), outlines a formal structure to align existing and emerging technologies in ecosystem observation and stress ecology to develop new diagnostics for marine- and freshwater-ecosystem monitoring and assessment. In particular, several emerging technologies, including environmental genomics and new statistical approaches are explored in detail, and their integration in a formal ecological-risk-assessment framework is illustrated. Chariton et al. (2016) highlight the use of new statistical approaches for causal inference, including Bayesian networks, concluding that a variety of modelling approaches are needed to allow flexibility for interpretation of complex causality chains within natural ecosystems under stress. In the third paper, Van den Brink et al. (2016) confront the challenge of integrating new, multi-scale and big-data approaches within a regulatory ERA framework. They view this opportunity as the basis to develop a new paradigm for ERA, which allows a global approach to be taken, yet permits the development of scaleable regulatory guidelines, driven by availability of localised models and data.

Together, these papers offer a start for a new program of research, focusing on the integrated assessment of the Earth's aquatic ecosystems. This program explicitly recognises that marine and freshwater ecosystems are interconnected, and that no matter what spatial or temporal level of impact is of concern, drivers and pressures operating at all levels of organisation must be considered when determining cause. This will not be a simple task, but it is now clear that we possess the necessary tools to achieve it. It is our hope that the papers described above can provide inspiration for new research projects and the basis of a path forward for regulators, and other stakeholders, to implement change across their environmental management and monitoring programs.