Predicting the response of farmland bird populations to changing food supplies

Summary 1. Agricultural developments over recent decades have had significant negative effects on wildlife. Modern European agriculture faces many further developments in technology, finances and climate. A framework for predicting the consequences of these developments and for informing management that mitigates previous depredation of wildlife is essential. 2. Field trials, monitoring and experimental approaches are all invaluable tools for predicting the consequences of proposed agricultural practices. However, such trials are time-consuming and expensive, with limited spatial and temporal scope. Increasingly, it is recognized that predictive modelling using computer simulations is essential for assessing the likely impacts of novel management practices. 3. Many agricultural developments are likely to have impacts on the availability of food for farmland wildlife. In this paper, we review modelling techniques that are available for predicting the consequences of changing food supplies for farmland birds, a group already shown to be sensitive to agricultural management. 4. We focus on two broad types of simulation modelling: phenomenological approaches (using aggregative responses or full population models), and behavioural depletion models (including daily ration and functional response approaches). We assess the strengths and weaknesses of each, particularly with regard to available data. 5. We conclude that given the constraints on current information, daily ration models probably represent the most useful form of simulation modelling for farmland birds. Few phenomenological population models link demography explicitly to food supply and collecting data on these links should be a priority of autecological studies. More empirical data on aggregative and functional responses is needed to identify the generality of these functions, and to confront uncertainties in their use. 6. Synthesis and applications. Modelling techniques available for predicting the consequences of changing food supply for farmland birds have largely outstripped the availability of data required to parameterize them. We recommend the best available techniques given current constraints on data. Predictive models of biodiversity response should be more widely used when considering proposals for agricultural developments or agri-environment schemes. Models will provide a basis for selecting between competing proposals, whilst their widespread usage will both permit more frequent model validation and aid in model development.