Jue Wang
Associate Professor & Distinguished Research Fellow of Management Analytics
- Adapted from: “Optimal Learning and Management of Threatened Species”
- Based on Research by: Jue Wang, Xueze Song, Roozbeh Yousefi, Zhigang Jiang
- Journal: Management Science
Key Takeaways
- Scientists have developed a new model to optimize conservation efforts for threatened species, balancing information gathering and protection actions while accounting for uncertainties about the species' presence, extinction risk and detectability.
- The model uses real-time data to adaptively learn and adjust conservation strategies, potentially improving the efficiency of limited conservation resources in protecting biodiversity.
- In some cases, the model suggests that areas where a species has never been found might be more likely to contain the species than areas where it was previously detected, challenging conventional assumptions in conservation planning.
Species loss is taking place at an unprecedented rate with conservation efforts facing the twin challenges of limited resources and information.
Particularly with species whose continued existence now relies on very few survivors, efficient data collection is incredibly and increasingly difficult. Jue Wang and his co-authors developed a partially observable Markov decision process model with unknown parameters to address challenges in data collection in order to protect these threatened species.
The group focused on the Hainan gibbon, the rarest primate, for their case study designed to optimize the spatiotemporal allocation of limited conservation resources.
In 2018, there were only 29 gibbons recorded living in a 15-kilometer square part of the Bawangling conservation reserve. New gibbons have been found in areas well outside the reserve since, and unverified sightings have been reported in forested regions. These unidentified smaller groups are at an elevated risk of extinction and, therefore, increase the need for detection. Following the Hainan Tropical Rainforest National Park’s creation, more resources have been devoted to systematic surveys of heretofore unidentified gibbons. But such efforts require a degree of luck: surveys must be conducted with the right effort in the right place and at the right moment.
This study presents a model to improve the conservation of threatened species by combining learning, monitoring and protection, even when detection is imperfect and dynamics are uncertain. It challenges the current approach of prioritizing areas where species were previously found, noting that regions without records may still have high chances of species presence. The model stresses the need to consider the timing of past detections to better assess extinction risks. Future work should explore applications to multiple species, survey methods, spatial scales and species migration. Leveraging data from citizen scientists will also be essential for more effective conservation efforts.