By Tim Mc Clanahan
Prioritizing protected areas in the ocean is challenged by the difficulties of surveying the diversity of species beneath the sea. A new approach using high resolution satellite and shipboard information has mapped underwater conservation priorities in 11 nations of the western Indian Ocean. Satellite data is more abundant, more resolved, and covers more areas than spotty underwater surveys.
Using both data sources and an AI algorithms helped to identify 119 priority areas among 7039 possible coral reef locations in the western Indian Ocean. These priority areas overlapped in coverage with only 34% of the 208 previously designated marine protected areas in these nations. Consequently, the AI approach identified overlooked potential high priority areas. Past reports had a bias in that they were frequently based on reports of more visible species, such as large fish and sharks, marine mammals, and seabirds.
AI predictions were made using more than 30 variables derived from satellite and shipboard databases. These data cover the ocean at the modelled diversity scale of 6.25km2. The model first evaluated 70 ocean and human impact variables before settling on metrics that made the best predictions based on field samples. Using environmental proxies to make predictions allowed mapping diversity in all 7039 coral reef mapped locations. 119 of these locations with the highest diversity for geographic delineations, such as nations, were selected as conservation priorities.
New methods such as these can assist finding priority area in areas that are difficult or expensive to sample. Moreover, predictions are at a smaller scale than past efforts and therefore useful for locally managed areas, which are often preferred in poorer African nations that cannot afford the cost of ending fishing in large marine areas.
Remoteness or distance from people has been an important criteria of past selection methods but the AI method controls for human effects. Therefore, already impacted locations with potentially high diversity can be identified by the AI methods. This allowed finding impacted areas with high potential to recover with renewed management and conservation efforts. Identified protected area can therefore have high value for local people, in contrast to large remote locations frequented by tourists.
Key findings
- Satellite and shipboard environmental data were able to predict number of marine species in 7039 locations in 11 nations
- Models predicted many new high species diversity conservation priorities areas
- AI models could account for the overriding effect of fishing, which allowed selections of conservation priorities in human impacted sites.
- Poor overlap was found between AI model predictions and currently designated marine protected areas and past consultancy-based selections
- The model allowed predictions for small, protected areas that are increasing with local community-based management projects
Read the paper here: https://conbio.onlinelibrary.wiley.com/doi/10.1111/cobi.14256