Inter-observer variation in habitat survey data: investigating the consequences for professional practice

Cherrill, A. (2015) Inter-observer variation in habitat survey data: investigating the consequences for professional practice. Journal of Environmental Planning and Management.

CHERRILL interobservational 5 April a.pdf - Accepted Version

Download (522kB) | Preview


Knowledge of the extent and distribution of vegetation types is essential to underpin conservation assessments, land-use planning and management of wildlife populations (Hill et al. 2005; IEEM 2006; Morris and Thrivel 2009). Despite improvements in remote sensing of land cover, field survey remains an essential method for collection of data on the distribution of habitats and their floristic composition (IEEM 2006). Surveying of vegetation is recognised as a key skill required by ecologists and environmental managers (IEEM 2007, 2011), but studies of variability between surveyors have often revealed significant levels of disagreement in terms of the plant species and habitats recorded (e.g. Scott and Hallam 2002; Milberg et al. 2008; Stevens et al. 2004; Hearn et al. 2011). For example, a study using the National Vegetation Classification (NVC) in the UK found that pairwise spatial agreement between seven surveyors mapping vegetation at the same site averaged only 34% at the community level (Hearn et al. 2011). Comparisons between plant species lists drawn up by different surveyors working in the same plots typically show agreement in the range of 50%–70% for a variety of habitats (Scott and Hallam 2002). Professionals working in the environmental and conservation sectors are therefore aware of the potential for inter-observer variation and its impact on data quality, but there is a dearth of information on the extent to which it is perceived to be an impediment to good decision-making in practice (Cherrill 2013a). If inter-observer variation causes few problems, then the issue may be largely irrelevant in day-to-day practice. However, if inter-observer variation in interpretation of habitat types is a cause of disagreement and poor decision-making there may be a mandate to change training and/or survey methods. The focus of the present study is inter-observer variation in habitat mapping using two of the standard classifications in the United Kingdom, namely the Phase 1 habitat classification (JNCC 1993) and the NVC (Rodwell 2006). Studies focussing on these methods have revealed spatial agreement between surveyors using the same method at the same site in the range of 25%–70% (Cherrill 2013a). These studies, however, were conducted either as bespoke academic research projects designed to directly assess observer variation (Cherrill and McClean 1995, 1999a, 1999b, 2000, 2001; Hearn et al. 2011) or as part of Quality Assurance procedures within a large-scale national monitoring programme designed to detect landscape change (Stevens et al. 2004). The extent to which these results are representative of inter-observer variation in professional practice involving environmental assessment and site management planning is therefore unknown (Cherrill 2013a). None the less, it can be hypothesised that errors made in identifying vegetation types in these spheres of activity may be frequent and that there may be consequences for conservation assessments, site management and planning decisions. The present paper uses a questionnaire survey of members of the Chartered Institute of Ecology and Environmental Management (CIEEM) in the United Kingdom to address two main questions. First, how frequently are errors in data detected in reports describing the results of vegetation surveys? Second, what are the practical consequences of these errors? CIEEM has approximately five thousand members in the UK. They are ideally placed to respond to these questions being employed primarily in environmental consultancy, planning authorities, governmental environmental agencies, and non-governmental conservation organisations. The Phase 1 and NVC survey methods are used only in the UK, but similar approaches are used elsewhere (Alexander and Millington 2000). The wider applicability of the study is, therefore, to illustrate the need to extend academic studies of inter-observer variation to investigate their relevance to the day-to-day experiences of environmental professionals. The implications for further research and development of professional practice are discussed.

Item Type: Article
Keywords: observer bias, data error, vegetation mapping, environmental assessment
Divisions: Crop and Environment Sciences
Depositing User: Ms Kath Osborn
Date Deposited: 19 May 2016 18:37
Last Modified: 23 Jan 2018 12:03

Actions (login required)

Edit Item Edit Item