Headline writers want only black and white, but research results come in shades of grey.
In the past couple of days, the results of Countryside Survey 2000 will have hit (or missed) the newspaper headlines. Depending on the angle taken, there will be rejoicing at the cessation of earlier trends of hedgerow loss or concern about the rise of those types of vegetation associated with higher levels of nitrogen in the British countryside that tend to thrive at the expense of species diversity. At a time when scientific evidence is under closer scrutiny than ever, just how reliable are results from surveys charting the state of the rural environment?
In 1993, when reporting the previous survey, one broadsheet newspaper screamed: "Farms blamed for 14 per cent loss of wild flowers." This was not entirely correct. For one thing, these surveys estimate change, but they do not attribute causation. While the term "wild flowers" is a reasonable metaphor for the recorded range of higher plants, it is also emotive. The baldly stated figure was perhaps the most unfortunate representation. The survey included an estimate of 14 per cent loss in certain types of vegetation, but this figure was accompanied by a statistical error indicating its reliability. But, perhaps inevitably, it is the headline, the soundbite, that is wanted by journalists, politicians and citizens - ifs and buts are annoying peripherals.
Scientists and policy-makers need to have information on habitats, landscape features and vegetation to understand the biophysical processes that take place and to ensure that agri-environment and other countryside policies are producing the desired effect.
To gather information on this enormous resource requires a sampling approach. It is not possible to visit every field or hedgerow. So some 40 teams of surveyors visited sample 1km squares throughout Great Britain in the summer of 1998 to map and record for all they were worth. As soon as sampling enters the equation, results will be estimates. This is why taking notice of the statistical error term is so important. The chances of a separate sample producing a different result are reflected in this term - the smaller the error, the more reliable the result.
So the "good news, bad news" results from Countryside Survey 2000 - the increase in plant diversity in arable field boundaries; the continuing decline in plant species numbers in meadows; the evidence of increasing nutrient levels in road verges with bigger, fast-growing plants dominating and reducing diversity; the increase in broadleaved woodland; the loss of calcareous and acid grassland; the slight increase in the number of lowland ponds; the improvement in the quality of streams and small rivers; the more overgrown riverbanks; and the increase in rush-dominated fens and marshes - need to be recognised as estimates and to be seen alongside their corresponding statistical error terms.
The uncertainty of the results also involves issues of consistency of feature definition (how many gaps do you need before a hedgerow becomes a line of shrubs?), quality of recording and data entry, appropriate forms of analysis and so on.
Countryside Survey has addressed the issues. The approach is scientifically rigorous and the work is of a high standard. The survey is unique in terms of the breadth and depth of its database. All that is needed now is more effort from scientists to assist the delivery and interpretation of the results.
Colin Barr is project leader with the Centre for Ecology and Hydrology at the Merlewood Research Station, Cumbria. Countryside Survey was sponsored by the Department of the Environment, Transport and the Regions, the Natural Environment Research Council and other government departments.