Landscape structure and management alter the outcome of a pesticide ERA: Evaluating impacts of endocrine disruption using the ALMaSS European Brown Hare model

• Published in: The Science of the total environment 2016-01, Vol.541, p.1477-1488
• Main Authors: Topping, Chris J., Dalby, Lars, Skov, Flemming
• Format: Article
• Language: English
• Subjects: Abundance occupancy ratio; Agricultural system simulation; Animals; Biodiversity; Conservation of Natural Resources - methods; Endocrine Disruptors - analysis; Environmental Exposure - statistics & numerical data; Environmental Policy; Environmental Pollutants - analysis; Environmental Pollution - statistics & numerical data; Environmental risk assessment; Europe; Hares; Humans; Landscape generation; Models, Theoretical; Pesticide regulation; Pesticides - analysis; Risk Assessment - methods
• ISSN: 0048-9697; 1879-1026
• DOI: 10.1016/j.scitotenv.2015.10.042

There is a gradual change towards explicitly considering landscapes in regulatory risk assessment. To realise the objective of developing representative scenarios for risk assessment it is necessary to know how detailed a landscape representation is needed to generate a realistic risk assessment, and indeed how to generate such landscapes. This paper evaluates the contribution of landscape and farming components to a model based risk assessment of a fictitious endocrine disruptor on hares. In addition, we present methods and code examples for generation of landscape structures and farming simulation from data collected primarily for EU agricultural subsidy support and GIS map data. Ten different Danish landscapes were generated and the ERA carried out for each landscape using two different assumed toxicities. The results showed negative impacts in all cases, but the extent and form in terms of impacts on abundance or occupancy differed greatly between landscapes. A meta-model was created, predicting impact from landscape and farming characteristics. Scenarios based on all combinations of farming and landscape for five landscapes representing extreme and middle impacts were created. The meta-models developed from the 10 real landscapes failed to predict impacts for these 25 scenarios. Landscape, farming, and the emergent density of hares all influenced the results of the risk assessment considerably. The study indicates that prediction of a reasonable worst case scenario is difficult from structural, farming or population metrics; rather the emergent properties generated from interactions between landscape, management and ecology are needed. Meta-modelling may also fail to predict impacts, even when restricting inputs to combinations of those used to create the model. Future ERA may therefore need to make use of multiple scenarios representing a wide range of conditions to avoid locally unacceptable risks. This approach could now be feasible Europe wide given the landscape generation methods presented. [Display omitted] •Uses landscape-scale and agent-based modelling for mammalian ERA.•Prediction of a worst case scenario is difficult from system metrics.•Meta-modelling may not yield acceptable predictive results for novel landscapes.•Realistic combinations of farming, management and ecology are needed for ERA.•Methods for rapid generation of realistic landscapes are presented.