Pasar al contenido principal


    Solicitar acceso a documento de reunión
    Número de documento:
    S. Hill, J. Forcada, P. Trathan and C. Waluda (United Kingdom)

    The CCAMLR ecosystem monitoring programme (CEMP) primarily indicates the short term response of air breathing predators to localised environmental conditions. CEMP data do not directly indicate population size, which is the parameter that many conservation objectives attempt to control. Furthermore, CEMP data cannot be used in a standard environmental impact assessment framework as they lack control sites. Identifying how these data could be used in an ecosystem management strategy is therefore an important challenge. Potential strategies are likely to consist of a method for inferring impacts and a schedule of tactical interventions in response to these impacts (e.g. restrictions on fishing activity). We discuss a range of inference methods which are tractable using CEMP data. These methods either assess the expected probability of an observed value in an unimpacted system, or they assess the frequency of values below a fixed reference point. The former approach allows inference criteria based on changes in this frequency rather than by reference to a critical probability. For example, this approach would have provided a timely and sustained indication of a non-fisheries impact on fur seal pup production at Bird Island from the early 1990s whereas critical probability methods would not have detected the impact until almost a decade had elapsed. Shorter reference periods over which the frequency is assessed increase the risk of Type II error (failing to detect a real impact, which is a risk to the ecosystem). Longer reference periods increase the risk of Type I error (falsely detecting an impact, which is a risk to the fishery) and of detrimental delays in the management response. Higher frequencies of low values required to infer an impact decrease Type I risk while increasing Type II risk. An example in which low values occur according to the binomial distribution illustrates the trade-offs between these risks. No inference method can eliminate these risks, but characterising the trade-offs allows managers to choose inference criteria which match their management approach. This could include minor interventions based on subtle indications of an impact. Our example impact was characterised by an increase in the frequency of very anomalous observations with no detectable change in the frequency of moderately anomalous observations. We therefore recommend that ecosystem managers should compare the state of indicators with several (moderate and extreme) reference points and that the response to an impact should be determined by the dynamics of the system.