Statistical power analyses suggest that a minimum of 400 sampling points will be needed to detect changes in amphibian populations in Britain, but more would obviously be beneficial. To ensure that the ponds surveyed are a representative sample, they will be chosen through a process of randomly selecting 1-km squares.  The pond nearest to the southwest corner of each square will be surveyed. 

We have stratified the selection of survey squares by Vice County, to select squares in proportion to the size of each Vice County. The total land area of Britain (England, Scotland, Wales), including islands, is 229,800 sq km.  Britain comprises 112 Vice Counties, with a mean area of 2052 sq km each. With a minimum of 400 ponds needed for Britain, each Vice County receives a mean of 3.57 sample sites (range 1-8). Given that Britain, covers 229,800 sq km, each one of the 400 sample points is being used to represent 575 sq km. To visualise this in terms of 10 km dot maps, this is approximately equivalent to one pond surveyed for every six 10 km squares in Britain.

A long-list of 100 squares has been randomly selected in each Vice County and numbered sequentially from 1 to 100. This provides us with more squares per Vice County than required to meet the target of 400 ponds nationally.  If volunteer demand for survey sites exceeds this quota, then additional squares will be allocated in sequence. The minimum number allocated will be commensurate with the size of the Vice County; the maximum will depend on volunteer take-up.

Northern Ireland covers 13,880 sq km, and comprises six Vice Counties (belonging to the separate Irish Vice County system). As it is biogeographically distinct, the sample selection will be carried out independently from the British sample. Ponds will be selected randomly in the same way, and at an equivalent density as Britain. The sample size has not been calculated statistically as for Britain (it would be unrealistically large in a small area), but having the same density of sample sites as Britain maintains some consistency. Therefore, using the British density of one pond per 575 sq km, this gives 24 ponds for Northern Ireland; a mean of four ponds per Vice County.

Squares with >50% urban and suburban cover have been excluded from this survey.  Landscapes unsuitable to amphibians (e.g. sea/estuary, inland water) have also been excluded. We considered stratifying according to landscape or land cover type as well (to ensure proportionate representation of each), but this adds a level of complexity we wished to avoid at this stage, particularly until we have more experience of volunteer take-up.

Within each selected square surveyors will need to identify the pond themselves. Maps and aerial photographs can be used, but confirmation will need to take place through field walking, as not all ponds are shown on the map. The current Ordnance Survey data capture specification (0.1 ha, roughly 30 x 30m) is too coarse to pick up many smaller ponds. Secondly, the periodicity and reliability of map revisions is not compatible with our needs. The only reliable method of identifying a target pond is therefore on foot, starting at the southwest corner of each selected one-km square.

A contingency is in place in case volunteers become discouraged if their ponds have no amphibians. To increase the chances of all volunteers encountering amphibians, we will have a back-up set of ponds expected to be of a higher quality. A sample of up to 400 nature reserves and high quality habitat sites will be available as additional squares for surveyors who are already surveying a randomly selected square. (The intrinsic bias means that these must be treated as a separate sample from the randomly selected sample). The Wildlife Trusts, Defence Estates, National Trust, Forestry Commission and all local authorities approached so far have agreed to participate in principle. Subject to agreement regarding protocol, access arrangements and permissions, a commensurate number of sites will be selected in all areas. This will be arranged centrally by the HCT as much as possible, but volunteers will need to liaise with landowners locally and make access arrangements too.