Evidence-based conservation planning in Taiwan’s agroecosystems

Abstract

Urbanization continues to reduce farmland in land-limited Taiwan, although agricultural landscapes remain important habitats for resident breeding birds and provide carbon-storage functions. Because biodiversity conservation and climate actions are often planned separately, opportunities for integrated management are limited. This study developed a spatial planning framework that combines bird habitat conservation with carbon storage. First, Taiwan Breeding Bird Survey data and generalized linear mixed models were used to identify 33 farmland-associated bird species, classified into four functional groups: omnivores, insectivores, granivores, and carnivores. Second, habitat suitability was estimated using ensemble species distribution models, carbon storage was assessed with InVEST, and spatial overlap analysis was applied to identify co-benefits. Third, 211,585 ha of conservation-interest zones were designated, and Dyna-CLUE was used to simulate land-use change to 2044 under a baseline scenario (S0) and a conservation scenario (S1). Results showed that 51% of farmland had high habitat suitability and 66% had high carbon storage, with positive associations between the two indicators. Compared with S0, S1 increased highly suitable habitat by 12.02% to 38.70%, indicating that conservation zoning can help safeguard farmland habitats under development pressure. The framework provides a replicable decision-support tool for identifying priority agricultural areas and supports farmland OECMs, the Taiwan Ecological Network, and restoration and monitoring planning.

Keywords: Agroecosystems; farmland birds; habitat suitability; carbon storage; conservation planning; other effective area-based conservation measures

Methodology and Implementation Steps

This study used Taiwan’s farmland as the study area and integrated bird habitats, carbon storage, and land-use change into a three-stage spatial conservation planning framework.

In the first stage, bird data were compiled and analyzed to identify farmland-associated species and assess land-use effects. Taiwan Breeding Bird Survey data were combined with international habitat information to determine which resident bird species were significantly associated with agricultural landscapes. Generalized linear mixed models were used to examine the relationship between bird occurrence and farmland coverage. A total of 33 farmland-associated bird species were identified and classified into four functional groups: omnivores, insectivores, granivores, and carnivores. Partial least squares regression was then used to evaluate how the intensity of farmland, abandoned farmland, buildings, roads, pastures, and freshwater areas affected each functional group.

In the second stage, habitat suitability and carbon storage were estimated and spatially integrated. Ensemble species distribution models were constructed using classification tree analysis, generalized boosting models, and random forests to predict habitat suitability for the four bird groups. Carbon storage was estimated using the InVEST model, including aboveground biomass carbon, belowground biomass carbon, and soil organic carbon. Habitat suitability and carbon storage values were normalized and spatially overlaid to identify two types of conservation priority areas: areas with high habitat suitability and high carbon storage for protection, and areas with high habitat suitability but low carbon storage for restoration.

In the third stage, future land-use scenarios were simulated. The priority areas were designated as conservation-interest zones, and the Dyna-CLUE model was applied to simulate land-use change from 2015 to 2044. Two scenarios were compared: a baseline scenario without conservation restrictions (S0) and a conservation scenario in which land conversion was restricted within conservation-interest zones (S1). Changes in farmland extent and highly suitable bird habitat were then evaluated under both scenarios.

Innovation and Cross-Disciplinary Collaboration

This study moves beyond the conventional separation of biodiversity conservation and carbon-storage planning by integrating bird monitoring, land use, habitat suitability, carbon storage, and future scenario simulation within a single spatial decision framework. It combines ecology, landscape planning, geographic information systems, statistical analysis, and ecosystem-service assessment. Ensemble species distribution models, InVEST, and Dyna-CLUE were jointly applied to identify priority areas for protection and restoration that provide benefits for both bird habitats and carbon storage. This cross-disciplinary approach translates scientific evidence into practical farmland management and conservation strategies while supporting collaboration among researchers, government agencies, agricultural sectors, and local communities. The overall workflow integrates bird monitoring data, statistical analysis, spatial modelling, carbon assessment, conservation zoning, and future scenario simulation into an evidence-based decision-support framework for farmland conservation planning.

Expected Results and Contributions

This study establishes an integrated spatial decision-support framework for bird habitat conservation and carbon storage. It identifies 211,585 ha of conservation-interest zones and shows that the conservation scenario could increase highly suitable bird habitat by 12.02% to 38.70% by 2044. The results can help government agencies identify priority areas for farmland protection and restoration, while supporting farmland OECMs, the Taiwan Ecological Network, and net-zero policies. The framework also provides a practical basis for long-term monitoring, biodiversity-friendly farming, and land-use management. By linking habitat conservation with ecosystem-service provision, the study supports more balanced decisions that consider biodiversity, agricultural production, and carbon-storage benefits.