Work Package 3: Past and Present Response & Informed Proxies
Work package number: 1
Work package coordination: 5
Work package title: Past and Present Response & Informed Proxies
Participant id: 1, 2, 3, 4, 5, 6
In this workpackage, observed and modelled time series data will be jointly investigated to (i) validate the different biological models used, (ii) understand the effects of multiple drivers (i.e. climate, eutrophication and fisheries) influencing changes in trophodynamics and functional biodiversity, and (iii) develop informed proxies for key species as a basis for scenario predictions within WP4. The specific objectives of this WP are to:
- Reconstruct past ecosystem changes of different parts of the North Sea, including the identification of structural changes such as Regime shifts, Trophic Cascades and Oscillating controls;
- Develop informed proxies for major ecosystem key species to be used in scenario predictions (WP4).
Exploring single time-series of different ecosystem components can provide only an insufficient impression of the changes in ecosystem structure and function. Furthermore, present-day ecosystem and foodweb models are unable to represent the full biodiversity and multiple drivers influencing complex ecosystems such as the North Sea. To provide a holistic impression of past ecosystem changes the aggregated approach of Integrated Ecosystem Assessments (IEAs) is being increasingly employed. IEAs are essentially multivariate statistical analyses (e.g. Principal Component Analyses, Canonical Correlation Analyses) of large data sets integrating knowledge on spatial and temporal trends of all important ecosystem components and driving forces. Examples exist for the northwest Atlantic ecosystems of Georges Bank US and the Scotian Shelf as well as the North Sea and Baltic Sea (Möllmann et al. 2008) ecosystems.
- a possibility to visualize ecosystem changes using the "traffic light approach" used in fisheries management,
- aggregated indicators of ecosystem change which can be used to investigate structural ecosystem changes such as "regime shifts", "trophic cascades" and "oscillating controls" (Frank et al. 2005),
- to identify the major drivers of change (Möllmann et al. 2008), and iv) to derive indications on the functional relationships between the most important ecosystem players as well as biotic and abiotic drivers.
The latter can be used to validate the different biological models and to develop "informed proxies" for predicting the potential future of key ecosystem components.
Description of work
Task 3.1 Identification of ecosystem structural changes
This Task will apply a number of multivariate statistical analyses (e.g., PCA, CCA, RDA) on large abiotic and biotic datasets generated from WP1 and WP2 integrating all ecosystem components and driving forces. Ecosystem changes of different subareas of the North Sea will be visualized using the "traffic light framework" and structural changes will be investigated using methods such as Chronological Clustering (Legendre et al. 1985) and Sequential Regime Shift Analysis (STARS). By doing these analyses sub-areas of the North Sea and on the complete data sets (all variables), as well as separately for drivers (abiotic and anthropogenic variables) and biotic response variables, we will identify if and how distinct changes in pressure variables precede changes in food web structure.
Task 3.2 Analyses of pressures and processes causing food web structural change
To identify the main pressures causing and/or maintaining potential changes in food web structure in each sub-ecosystem direct ordination analyses (e.g., redundancy analyses) will be applied to identify which pressure variables explain most of the overall pattern in the biotic response variables. Furthermore, Generalised linear models (GLM) and for non-linear relationships generalised additive models (GAM) will be developed with the prinicpal components of PCAs (Task 3.1) as response variables and pressure variables as predictors. Finally, to assess whether driving forces differ between regimes, we will develop the GLMs/GAMs described above also separately for each potential 'regime' identified in Task 3.1.
To identify processes (species interactions) that maintain a food web in an altered state following a compositional change temporal trends in trophic control will be analysed and compared between potential regimes. For this, we will analyse temporal trends and discontinuities in predator-prey relationships (Frank et al. 2005). By comparing the timing of potential breakpoints in predator-prey correlations between different species, key inter-specific processes from which further food web changes may cascade can be identified. In addition, thresholds leading to breakpoints will be determined by for example applying threshold-GAMs.
To understand the relative importance of different processes and pressures in causing changes in food web structure, the results from Tasks 3.1-2 will be synthesised in comparative analyses across different North Sea sub-areas. Time series of functional groups as well as scores derived from biotic variables through ordination methods (in Task 3.1) will be investigated for common trends across sub-ecosystems. The overall effect of large-scale pressures, such as climate indices, temperature, salinity, eutrophication, and fishing, on these trends will be analysed using two Generalised Models (including e.g., co-variance analyses and mixed models) to disentangle common trends and area (sub-ecosystem) effects. Additionally, time series analyses will be performed directly incorporating explanatory variables (e.g. dynamic factor analysis), or by subsequently correlating extracted common trends with pressure variables (e.g., min-max autocorrelation factor analysis.
Task 3.3 Development of informed proxies for future ecosystem change
This Task will develop functional models (informed proxies) describing the abundance and/or productivity of key species and/or groups in relation to abioitc and biotic variables. The work of this tasked will be based upon results of the time-series analyses of key species in relation to climate as well as the identified effects of abiotic and biotic factors on key species (WP1). Further background on relationships of key species and their environment including their dependence on ecosystem configuration will be derived based on the multivariate analyses of Tasks 3.1 and 3.2.
These functional models will be used to predict abundance and/or productivity based upon the strength of abiotic (ecophysiological) and biotic (trophodynamic) associations. A suite of statistical approaches (e.g. GLMs and GAMs) will be used to generate the functional models to be used in predictions of future development using different scenarios of anthropogenic and climatic impact in WP4.