Environmental managers are interested in detecting change in order to manage
natural resources. In an applied context, managers want to know i) how much of a resource there is and ii) the quality or value of the resource. I assayed UK
conservation practitioners' views of the effectiveness of current vegetation monitoring methods. Concern was expressed that vegetation mapping and condition assessment (used to assess quantity and quality of a resource respectively) techniques involve too much subjectivity with consequent uncertainty and devaluation of information on which to base management decisions.
These perceptions were explored by quantifying the amount of between-observer variation in maps prepared using the National Vegetation Classification (NVC) and in assessments produced using approaches to Common Standards Monitoring (CSM), both commonly used systems in the UK. NVC mapping produced unacceptably high levels of between-observer variation, with average spatial agreement between seven surveyors of only 34.2%. CSM approaches were shown to produce inconsistent results, with major implications for reporting against national condition targets.
Quantitative methods utilising systematically located plots were also investigated and demonstrated that variation is unpredictable through spatial scales and that sampling intensity and choice of metric both influence the effort required to detect a specified change. Furthermore, quantitative and qualitative approaches require unfeasible levels of effort to detect meaningful sizes of change with sufficient power.
This leads to the following recommendations:
a) Vegetation mapping is unsuitable for monitoring change and should only be used to assess state with full acknowledgement of its subjectivity and uncertainty. Condition assessment could provide widespread snapshots of condition, but should not be used for monitoring temporal change without some quantifiable controls.
b) Quantitative methods must include i) a priori definition of the minimum (or
threshold) change that should be detected, ii) relative importance of type I and II error rates and iii) definition of sample population and sampling frame which specifically includes consideration of spatial heterogeneity.
Finally, this study suggests that research should focus on the development of a multi-scale and nested quantitative sampling design which could address monitoring and surveillance questions in the UK. Since resources do not permit universal application of this approach, quantitative data from a small number of sites could be used via diagnostic test methodology (as used in medicine to assess the accuracy of screening tests) to validate widespread condition assessment.