The relationship between relative and absolute counts may not be linear, so the
fact that one number is twice the size of another does not necessarily mean a
twofold difference in absolute abundance. Such counts cannot be relied on for
comparative purposes across species. Noisy or conspicuously perching or flying
species will appear to outnumber those that are quiet, and otherwise cryptic. Nor
can relative counts be fully relied on within one species across habitats. Birds are
easier to find in open than closed habitats because they are easier to see at greater
ranges and can be detected even if silent. Most detections in closed forests have
to be made by ear. Thus even if absolute numbers are the same; an open country
survey is likely to return higher relative counts.
In many circumstances, relative results are good enough for the purpose of
a study. They are an inevitable consequence of seeking results applicable over
many species and large areas at low cost. They might be especially appropriate
in many tropical studies because of generally poorer knowledge combined
with larger differences between habitats in species complement than are found
elsewhere.
1.2 Designing the fieldwork
Survey fieldwork consists of going to selected places and following a recording
protocol at each. Choosing where to go is a critical element of survey design.
Such sampling is not always properly acknowledged, but it can influence results
as much as the choice of survey or census method. Many studies do not even
mention how the field locations were chosen.
Any well-designed survey has a boundary; the area to which the conclusions
will apply. This should be explicitly identified. In some cases, the whole area
within the boundary will need to be sampled. In other cases, the target might be
just one habitat such as forests. In this case, the location of forests will first need
to be identified from maps or satellite images.
One way to locate study plots is to select them by the generation of randomly
distributed coordinates within the study boundary. This will give the best
unbiased estimates for the whole study area. It will sample individual features of
habitat in proportion to their overall abundance. For many purposes, this might
not be appropriate. If the intention is to study the variation of bird communities
with a feature such as habitat type or human impact, then you will want
a sample that well covers the range of variation. This will mean more intensive
sampling of the rare features and less of the common ones than a random sample
would deliver. The way to do this is by stratification. All possible sample
locations (grid cells or habitat blocks) are allocated to a particular stratum. Strata
Designing the fieldwork| 3