A new website, http://www.restorationevidence.org, is gathering and summarizing scientific evidence for the effectiveness of interventions to restore degraded habitats. A subset of https://www.conservationevidence.com dealing only with restoration, the site is a collaboration between the Endangered Landscapes Programme and Conservation Evidence, making evidence freely available to all. This will help ensure funds are spent on interventions that are most likely to be effective in achieving a desired outcome, from restoring upland grazing pastures into blanket bog to converting ranches into tropical forests.

So far, the site covers the effectiveness of actions to restore peatland vegetation, shrublands and heathlands, and forests, and evidence for the impacts of habitat restoration on several animal taxa. Work is underway to add the impacts of restoration techniques for grasslands, wetlands and benthic marine habitats, and for more animal taxa. The goal is eventually to cover all interventions that could be used to restore all habitats globally, and to update these regularly to include the latest evidence.

Restoration Evidence makes the evidence for the effects of restoration interventions accessible and digestible, so that planners and managers can rapidly obtain an overview of what has and hasn't worked. For each intervention, such as ‘rewet peat’ or ‘use fences to exclude livestock from shrublands’, each article or publicly available report that has tested the intervention is summarized in a paragraph written in a standardized manner, making the methods and results as clear as possible. The overall findings from these studies are summarized as a set of key messages, giving a rapid overview of the effects of a given intervention. Experts score each intervention in terms of how effective the intervention seems to be (based on available evidence), the level of certainty in the current evidence, and potential harm that might arise to the target taxa or habitat from this intervention. The methods used for finding and synthesizing the evidence can be found on the Conservation Evidence website.

Habitat restoration is increasingly recognized as having an important role to play in conserving biodiversity, mitigating climate change and improving well-being, and this is reflected in the creation of international targets for restoring habitats. The Convention on Biological Diversity aims to restore 15% of degraded ecosystems by 2020, and the Bonn Challenge sets out to restore 350 million ha of deforested and degraded lands by 2030. Regional initiatives contributing to the Bonn Challenge have also been established. Initiative 20×20 aims to bring 20 million ha in Latin America and the Caribbean into restoration by 2020, and AFR100 aims to start restoration on 100 million ha across Africa by 2030. The UK's 25 year Environment Strategy has a target to create or restore 500,000 ha of wildlife-rich habitat outside protected sites. Initiatives such as these will have maximum impact if they use evidence on which interventions are most likely to be most effective. Having the answers to questions such as ‘how should we remove invasives’ and ‘when should we plant trees and when should we sow seeds instead’ is critical to the effective use of resources.

So what does work in restoration? For each ecosystem there are interventions that seem to be effective, although the exact effects are likely to vary with local conditions. In peatlands, rewetting peat and scattering mosses on the peat surface were effective in many studies. For shrublands and heathlands, reducing livestock numbers or density seems to be important. For forests, preparing the ground before planting trees and then thinning the planted trees was effective.

One of the most important findings from the creation of RestorationEvidence.org has been that for each habitat there are relatively few high quality studies testing interventions, meaning that many actions are classified as ‘unknown effectiveness’. If every restoration project included just one experimental component, in which the effects of an intervention were tested, monitored and published, we would know a lot more about the best ways to restore ecosystems.