When reasonably stop sampling? How to estimate the gain in newly recorded species according to the degree of supplementary sampling effort.
Résumé
Aim: Register new species gradually becomes more difficult as sampling of a community
progresses, addressing increasingly rarer species. Thus, although biodiversity assessments would
ideally require complete samplings, only partial samplings are ordinarily achieved when species
abundances distributions are highly heterogeneous within communities, which is often the case.
Then, in the frequent context of partial samplings, answering knowingly whether to continue or stop
an ongoing inventory require to tentatively assess the “profitability” of the extra sampling effort. That
is, trying to estimate the number of species expected to be newly recorded thanks to a given further
increase of the sampling size.
Methods: Such estimate may be conveniently derived on the basis of the recorded numbers f1, f2,
f3, fx, of species already recorded once, twice, three, … x-times within the ongoing sampling. The
derivation involves a Taylor expansion of the species accumulation curve, with the successive
derivatives of the species accumulation curve being respectively expressed in terms of the
successive recorded values of fx.
Results: A simple nonparametric estimator of the expected number of newly recorded species is
derived as a function of the foreseen additive sampling effort. Depending only upon the directly
recorded values of the fx within the ongoing sampling, this estimator is easy-to-implement and, in
particular, does not require recording explicitly the species accumulation curve.
Conclusion: The practical interest of this estimator is to offer a convenient way to gauge the
additional sampling effort required for a given increase in sample completeness, thus providing
quantitative elements to determine whether further continuing an ongoing sampling looks
appropriate or not, in the context of both limited available time expenditure and possible other
competing priorities.