Measuring, Reporting, and Verification of Forest Restoration

Restoration management has been hamstrung by a lack of appropriate data on which to act and reward good practice; and there are now a wide array of cost-effective tools to support robust monitoring. Given biodiversity has continued to decline in European landscapes, even within protected areas, implies we cannot rely on assumptions (Pressey et al. 2021). To inform adaptative management and identify issues early enough that decisions can address deficits in a timely fashion, monitoring must be periodically repeated, and indicators chosen to aid the tracking of progress. Corrective actions may be required due to deficiencies in baseline assessments of site capability, but it is also expected to be increasingly important so that external factors such as natural disturbances and climate variability are also captured.

A well-designed monitoring system helps ensure that restoration is measurable, effective, and accountable; key to adaptive management strategy that ensures success and in turn attracting long-term support and investment. A critical part of developing the monitoring design is determining the number of sites, and plots within sites, that need to be monitored to have the adequate power to detect changes in each attribute (Guillera-Arroita and Lahoz-Monfort 2012). The sampling strategy (e.g., transect versus quadrat) and intensity of replication have a large effect on the value of the estimate, and the optimal choices will vary with strategy and ecosystem attribute. For this reason, it is typically worthwhile collecting pilot data during baseline surveys to determine required sample sizes. Although it may seem daunting to estimate sample sizes, with just a small amount of pilot data (or data from past projects at similar sites), but suitable sample sizes can be done quickly using simple mathematical calculations (Edlin et al 2021; Maslen et al 2023).

Figure 3 illustrates how we can anticipate the number of samples required to detect a shift of a specific magnitude and with the given degree of confidence. As the magnitude of changes increase, the number of samples required to achieve a suitably high confidence decreases. Equally, even with large sample sizes, some effects (<0.5) may be too subtle to reliably detect. Optimising the sampling design and replication of indicators is crucial to ensuring results successfully describe changes over the long term and make monitoring cost-effective.



Figure 3. Example of power analysis to inform sampling intensity. Sample size could be the number of hours of acoustic recording, or the number of soil samples. Power is the probability of correctly detecting a restoration effect, and the dotted horizontal line at 0.8 represents a typical threshold for such studies. The effect size is a standardised measure of change.