eCite Digital Repository

Imperfect observations in ecological studies


Shimadzu, H and Foster, SD and Darnell, R, Imperfect observations in ecological studies, Environmental and Ecological Statistics, 23, (3) pp. 337-358. ISSN 1352-8505 (2016) [Refereed Article]


Copyright Statement

© The Author(s) 2016. Licensed under Creative Commons Attribution 4.0 International License (

DOI: doi:10.1007/s10651-016-0342-2


Every ecological data set is the result of sampling the biota at sampling locations. Such samples are rarely a census of the biota at the sampling locations and so will inherently contain biases. It is crucial to account for the bias induced by sampling if valid inference on biodiversity quantities is to be drawn from the observed data. The literature on accounting for sampling effects is large, but most are dedicated to the specific type of inference required, the type of analysis performed and the type of survey undertaken. There is no general and systematic approach to sampling. Here, we explore the unification of modelling approaches to account for sampling. We focus on individuals in ecological communities as the fundamental sampling element, and show that methods for accounting for sampling at the species level can be equated to individual sampling effects. Particular emphasis is given to the case where the probability of observing an individual, when it is present at the site sampled, is less than one. We call these situations ‘imperfect observations’. The proposed framework is easily implemented in standard software packages. We highlight some practical benefits of this formal framework: the ability of predicting the true number of individuals using an expectation that conditions on the observed data, and designing appropriate survey plans accounting for uncertainty due to sampling. The principles and methods are illustrated with marine survey data from tropical northern Australia.

Item Details

Item Type:Refereed Article
Keywords:compound distributions, detection probability, ecological modelling, marine surveys, sampling, species distribution models, SDMs
Research Division:Mathematical Sciences
Research Group:Statistics
Research Field:Applied statistics
Objective Division:Environmental Management
Objective Group:Terrestrial systems and management
Objective Field:Assessment and management of terrestrial ecosystems
UTAS Author:Foster, SD (Dr Scott Foster)
ID Code:118359
Year Published:2016
Web of Science® Times Cited:3
Deposited By:Directorate
Deposited On:2017-07-11
Last Modified:2018-05-09
Downloads:96 View Download Statistics

Repository Staff Only: item control page