Mr Jeongu Lee1, Mr Seungwoo Son1, Mr Sanghyuk Lee1, Mr Gilsang Lee1
1Korea Environment Institute, Sichungdaero, South Korea
Biography:
Jeongu Lee is a researcher at the Korea Environment Institute, where his work addresses ecological restoration policy and biodiversity assessment through geospatial analysis. Trained as a geographer, he develops data-driven frameworks that translate national spatial datasets into restoration decisions, including the sequential screening pipeline used to identify candidate degraded lands across the Republic of Korea and the decision-tree typology through which those candidates are classified for restoration. This work supports national implementation of Target 2 of the Kunming-Montreal Global Biodiversity Framework under the Natural Environment Conservation Act. His current interest lies in integrating molecular evidence into spatially derived restoration decisions.
Abstract:
Target 2 of the Kunming-Montreal Global Biodiversity Framework requires nations to place degraded ecosystems under effective restoration. In the Republic of Korea, a nationwide inventory of candidate degraded sites has been generated through sequential geospatial screening, and each candidate is assigned to one of seven restoration typologies by a hierarchical decision tree. One branch directs sites towards habitat restoration for threatened species, yet is resolved by a national rarity grade derived from mapped ecological value rather than observed occurrence.
eDNA is proposed as the evidence this node requires. Soil is sampled across each candidate unit, with species-specific assays applied to designated terrestrial targets and metabarcoding of vertebrate and invertebrate markers screening the same extracts for unanticipated occurrences. Replication is set so that detection probability is estimable under site-occupancy models, and detection thereby enters the tree as a probability with explicit false-negative risk rather than a binary flag.
Sampling exploits the screening pipeline as a national frame. Candidates are stratified by typology and urbanisation grade to define the population represented. Because the tree converts continuous conditions into binary branches, sites close to a decision threshold are sampled preferentially: adjacent parcels separated by a narrow margin receive opposing assignments, and there cartographic inference is least reliable. Parcels excluded at the land-cover stage provide reference assemblages, while restoration projects of differing implementation year indicate how long detection takes to emerge.
Typology assignment thus becomes verifiable rather than inferred, strengthening confidence in the national candidate list on which policy decisions rest.