I wrote up some ideas. It proved more difficult than I thought and I am not sure I captured the issue at the core of the discussion -
Currently the vast majority of records in the GBIF archive contain
an observation of one or more individuals of a single taxon. Many additional fields
in the record, based on the Darwin Core, provide context for the observation including
the observation type, the time and place of observation, the observer, etc. At the present time, however, the metadata do not
provide context about whether or not an observation is part of a systematic set
of observations, herein called a survey. A survey is an approach based on the
idea of statistical sampling, whereby an observer is unable to measure an entire
population but instead focuses on a subset of the population to make inferences
about the entire population.
The added scientific value of the survey framework, over
just a collection of unrelate observations indicating present only, is that a
scientist can make inferences about how common or rare a taxon might be (its status)
and over time, measure trends. The basic idea of a survey is intuitive: the
more you look, the more you will find. In the fisheries literature this idea is
called “catch per unit effort”.
The goal of the Humboldt extension is to accurately describe
a survey and its often hierarchical nature. Whereas an observation record is
characterized by general sense of the observation approach (Basis of Record), a
time and a location, a survey has a much more detailed description of the observation
technique(s), and also includes the number of sampling units employed, a time
or time interval (start time and duration), and a measure of the spatial extent
over which the survey was conducted.
A sampling unit, the finest measurement resolution of a
survey, encompasses a variety of ways of looking for species. It might include:
A physical sample such as a leaf or a water sample containing
molecules of DNA
One or several sweeps of a net containing a collection of
insects
Camera trap – collection of images of mammals
Quadrat – estimating
the percentage of space or numbers of space occupying organisms such as plants
or clams
Bird checklist – list of species of birds observed from a
fixed-point
Special considerations arise at the sampling unit level.
First a measurement can detect no individuals or space occupied. In these cases
the data need to reflect this fact with recording 0 for the observation. Second a measurement
might be more than just a number or percentage.
It might a be compound structure that includes the flowering stage of
each plant in a quadrat or the length and body mass of each insect in the sweep
net sample or the location of each bird along a transect.
