310 Monitoring Threatened Species and Ecological Communities
often only partially fulfilled (creating ‘disillusionment’), before reaching a mature
stage where the pros and cons of a technology are well understood.
Threatened species monitoring relies on being able to detect species. A key
component of how much a new technology can improve a monitoring program is
related to how efficient it is at detecting the species of interest. Two main aspects
determine this: (1) the practical constraints that may limit the field deployment of
the technology; and (2) the cost of getting those detections. First, new technologies
often come with ‘small print’ and challenges to their field deployment. These are
not always obvious, particularly in the early stages of adoption. It is worth
considering these potential caveats very carefully before committing to a new type
of monitoring. Field trials are absolutely essential before deciding to invest time
and money in a new technology: they may reveal unexpected issues that have to be
considered (e.g. raptors attacking drones, or field contamination of eDNA
samples). Second, limited conservation budgets mean that monitoring must be
economically efficient. When a new technology is proposed for monitoring, a
thorough investigation of its sensitivity and cost-efficiency compared with
available and well-tested traditional sampling methods is necessary. Such a
comparison may well be species- and habitat-specific, but general
recommendations will eventually emerge as the technology is evaluated in different
contexts. Evaluating cost-efficiency is best done within the specific decision
context in which the monitoring data will be used (e.g. detecting the presence of a
rare species v. detecting a population trend). Performance evaluation will become
more important than ever as more technologies lure conservationists to use them.
It is key to evaluate these new approaches and devices, testing their efficiency at
improving conservation decisions and management through improved monitoring.
Performance evaluation requires pilot data and a robust statistical framework
of analysis that matches the type of data and the object of inference. This should be
kept in mind from the beginning, so that field trials and pilot data collection can
be best used to compare performance with traditional methods. In some cases, new
types of data may require developing new statistical tools (e.g. multistate
occupancy-detection models that account for false positives to analyse two-stage
sampling for eDNA data; Guillera-Arroita et al. 2017). New statistical
methodologies continue to appear, often coupling hierarchical models with the
Bayesian framework of inference – a very powerful combination (Kéry and Royle
2015). However, despite the promise of new statistical methods, it is best to collect
the highest quality and quantity of data that one can afford, instead of relying on
statistics to clean up your mess.
The discipline of study design provides a set of tools (often specific to the type
of data and statistical analysis) that help structure a monitoring program and get
the most out of a new monitoring method. Power analysis is one such tool, which
provides an estimate of the statistical power to detect a change in a state-variable