8 – Monitoring threatened ecosystems and ecological communities^105data, and design f laws are a common reason why monitoring fails (Lindenmayer
and Likens 2010). Many monitoring design principles are the same for species and
ecological communities. These include:
● (^) using conceptual models to inform design by identifying key elements,
processes, covariables, management actions, interactions and external linkages
that potentially inf luence change (e.g. Rumpff et al. 2 011)
● (^) stratified coverage across the spatial and thematic domains of interest
● (^) replicating and randomising locations of samples (crucial to statistical power
and generalisation beyond the samples)
● (^) selecting practical, proximal and sensitive response variables (Lindenmayer
and Likens 2010)
● (^) sampling regularly in relation to time or ecological processes of interest
● (^) managing data systematically and using appropriate methods of statistical
inference.
Despite these common principles, surveillance monitoring for reporting and
diagnostic monitoring for management may require somewhat different designs.
Surveillance monitoring is focused on early-warning detection of detrimental
change. Sampling therefore needs to be extensive and randomised across the
domain of interest, relatively frequent and capable of detecting surprises (Duncan
and Wintle 2008). Data collection is therefore focused on end-of-chain variables
that may be responsive to a wide range of causal factors. Good examples of
surveillance design for ecological communities and ecosystems include remote
sensing time series of tree cover in Queensland (DSITI 2015) and time series of
aerial surveys for waterbird communities in eastern Australia (Kingsford et al.
2013), as well as many photopoint studies.
In contrast, diagnostic monitoring is focused more strongly on determining
the causal mechanisms of change and comparing the outcomes of alternative
management strategies. Sampling therefore needs to: (i) be structured for
comparisons; (ii) be stratified, replicated and randomised across management
treatments; and (iii) have unmanaged controls and references determined by the
management questions of interest. Use of ‘appropriate local indigenous reference
ecosystems’ to monitor progress towards goals is a key principle in the new
Australian Standard for Ecological Restoration (McDonald et al. 2016). Even for
relatively simple projects such as photopoint monitoring, the reliability of
outcomes depends upon decisions about design including replication, sampling
frequency and the features, treatments, controls and references to be photographed.
Because the focus is on causality, data typically need to be collected for multiple
variables along the response chain, not only for indicators at its end. Good
examples of diagnostic monitoring include adaptive monitoring of Victorian tall
eucalypt forests to evaluate logging and fire management strategies (Wood et al.