Open dialogues with our main partner, the Government of Seychelles during every stage of the project ensured that changes and amendments could be easily made with their input. For example, field locations where easily amended as a result of bad weather, ensuring no time was wasted while at sea. Furthermore, a clear expectation of sample collection and updates during the expedition meant that Convention on International Trade in Endangered Species of Wild Fauna and Flora (CITES) inspections could be made on arrival back into port, expediting the export permits, and thereby helping to facilitate the Government to fulfil their regulatory requirements.

It is not straight-forward to build trust. It can take time, skill, and resources, particularly finance and personnel. Nekton ensured that early engagement with the Government of Seychelles and Seychellois stakeholders began a year prior to the actual start of the Seychelles-Nekton field expedition. This allowed ample time to start building connections and relationships with the stakeholders and partners based in Seychelles. The Government of Seychelles gathered other locally based partners and stakeholders to build and frame a joint agenda of needs that would inform the research during the Seychelles-Nekton Expedition which took place in 2019. Co-production of the expedition included organising workshops to identify research locations, defining the pertinent research questions along with determining the stakeholders’ interest in leading specific projects.

AI-based crack gauge for rockfall is a device that monitors the occurrence of rockfall and the crack displacements in real-time by installing an observation sensor in a rockfall risky area located along the trail. Since 2013, automatic and manual crack gauges have been installed on steep slopes with a high risk of collapse, and 525 units are currently in operation at 174 locations. The rockfall measuring device is divided into risk levels of 'interest, caution, alert, and serious'. In the interest stage, regular and frequent inspections are carried out. In the caution stage where cracks are less than 5 mm and less than 2°, monitoring is strengthened. In the alert stages, precise investigation and action plans for the disaster are prepared. In the serious stage, the adjacent trails are controlled and emergency measures such as rockfall removal are implemented.

AI-based Intelligent CCTV is a scientific safety management system that uses deep learning technology to control emergencies in real-time image analysis. By recognizing and analyzing abnormal behavior patterns, such as intrusion, screaming, wandering, etc., a warning broadcast is immediately sent to the site and delivered to the control system, following the emergency responses.
In addition, in the case of marine/coastal national parks with a high risk of safety accidents due to tides and tides, the broadcast of tide times is automatically issued to the site. Intelligent CCTV was installed in 2020 and is currently being operated in 89 places in 15 national parks.

To enable automatic and instant biodiversity impact assessment enquiry, the BiA tool has been developed to facilitate enquiry services for land planners and other interested parties via Azure platform. The BiA tool works by overlaying the enquiry site or region (or existing construction projects) with multiple geographic layers including species distribution and protected area range to investigate if the site or region is within certain distance (e.g., 3 km, 5 km) from and may cause impact on endangered species habitat and/or protected areas. The assessment reports illustrate ecological and environmental risks of construction projects for decision-makers and could hopefully promotes them to take biodiversity into consideration.

A brief timeline of the BiA tool:

• Apr-Jun 2020: team formation, requirement communication, system development plan
• Jul-Sept 2020: tool development
• Oct 2020: trial test, application and dissemination
• (in preparation) Apr-Sept 2022: system upgrade

During nature watch campaigns, citizen scientists are invited to observe and record wildlife timely, which not only strengthens the connection between citizens and nature but also serves as a promising species distribution data source. Species record data collected by citizen scientists via online questionnaire automatically flows into the visualization platform database (after data cleaning and manually periodically check) and turns into intuitive and attractive visualized charts and maps (two types: spatial, spatial and temporal) via Power BI. The platform, with both web and mobile version, provides real-time feedback to citizen scientists’ nature watch efforts, boosting their sense of accomplishment and motivating their future participation in nature watch activities. Moreover, since the platform integrates multiple nature watch campaigns with links to web articles about specific analysis of each campaign, it offers a broad range of biodiversity knowledge and enables “virtual nature watch” for citizens to get to know wildlife in other regions.

A brief timeline of the platform:

• Jan-Feb 2021: form team, analyze analysis, make blueprint
• Mar-Jun 2021: develop database and platform
• Jul-Aug 2021: trial test
• Sept 2021: go live and promotion

To accelerate camera trap data workflows, an online data management system along with app-based tools and AI image recognition is being developed supported by technical partners, which consists of:

• Community-based camera trap monitoring assistant app: the app allows local monitors to automatically record the time and GPS location of camera trap setup/pickup, saving the cumbersome process of collecting data from local monitors and manual data entry. (blueprint: Jun 2019, development: Oct 2019-Feb 2020, trial and use: Mar-Oct 2020)
• AI image recognition models: AI models help detect animals and identify species in camera trap photos, which greatly reduce the number of photos that need human identification and enhance data processing efficiency.
  • A series of AI models has been trained and/or tested with technical partners, including PU & PKU ResNet18 model (2018), MegaDetector (test only, 2020), MindSpore YOLOv3 model (2021).
• Online data management platform: camera trap information collected via the app along with photos are upload to a structured cloud database. The data management platform not only supports species identification via AI and human, but also enables global data search and statistics reports. (blueprint: Apr-Aug 2021, development: Sept 2021-Jun 2022, trial and use: Jul 2022)

Training of staff is important to ensure the effective implementation and long-term success of the solution. Prioritise training during the designing and deployment phase, as well as after the deployment to ensure continued use of the solution.

Thanks to the Restor.eco platform, we analyze the restoration potential of our reserve, monitoring changes over time with satellite images and geospatial data, thus knowing the local biodiversity and its characteristics, current and potential soil carbon, as well as other variables such as patterns of land cover, soil acidity, or annual precipitation, using machine learning, artificial intelligence, and scientific units of measure.

The use of mobile Apps such as eBird, iNaturalist, Merlin Bird ID, provocated a positive impact for us on monitoring ecosystem and biodiversity.