Counting What You Can't See

All statistics is about extrapolating from the known to the unknown, estimating what cannot be counted.  Demographers may survey a sub-sample of a human population, and use modelling to draw inferences about the population as a whole.  But that is just one species - understanding biodiversity, such as in the context of multiple plant and animal species, is much more difficult.

Ecologists therefore sample environments to estimate species diversity and abundance.  But when environments are surveyed to assess biodiversity, the only certainty is that not all the individuals and species in the population will have been counted.  This is not only because plants and animals can’t fill out forms, but also because some species will be harder to observe than others and different methods of observing will work differently between species.

The challenge is therefore that the sample is often heavily constrained and the sampling mechanism complex.  The extrapolation is often not just for the number of individuals, but also for the number of groups, species and even taxa.  To sample these, there are thus a range of techniques used by ecologists to survey species abundance and diversity.  These include transect surveys, timed area surveys, capture and release trapping programs, monitoring of plants in designated plots and fauna detection using camera traps.

Understanding the impact of a mine or whether an environment is in decline are other common ecological questions.  The challenge is to ensure that these studies are well designed and that the right tools are applied to the right tasks, to produce meaningful inferences and conclusions.  While this is true of any industry, in situations where resources are limited and there is often only a single opportunity to collect the data, applying the right statistical techniques, at both the design stage and the analysis stage, is critical.

Species richness and abundance estimators can not only be used to assess the completeness of surveys retrospectively.  They can also be used to determine the sample sizes needed to adequately survey.  This improves statistical power, confidence in the outcome of statistical testing and saves time, money and effort.

Perhaps more importantly, these statistical methods provide ecologists with the luxury of being able to count what can’t be seen. 

For expert advice on planning, designing and analysing statistically sound environmental studies contact Data Analysis Australia daa@daa.com.au, or phone 9468 2533.

October 2017