Taitung Moves “Ecological Salary” Upstream with GIS Screening: A Predictive-Service Model for Taiwan’s 55 Indigenous Areas

On September 7, Taimali Township Office posted what could look like a routine forwarded notice, but it marks a useful policy step for Taiwan's 55 Indigenous areas. Taitung County is preparing its 2027 proposal under the ecosystem-service payment program for endangered species and important habitats, and it is first asking where potentially suitable land exists and which farmers are willing to participate. An earlier Donghe Township notice explains the method in greater detail: ecological GIS layers are used to identify possible areas, and those results are then checked against current paddy use, parcel information and farmer willingness.

The distinction matters. This is a preliminary inventory, not approval for a payment. A parcel appearing inside a GIS layer does not establish eligibility. The policy signal is instead that local administration is moving the service upstream. Rather than waiting for farmers to discover a notice on their own, the government can first identify areas that may fit the program, ask local offices to verify current conditions, contact farmers, record willingness, and only then move into formal proposal and review procedures.

What Donghe's notice reveals: GIS is only the first screening layer

Donghe Township Office's September 1 notice says Taitung County is carrying out an advance inventory of potential areas and farmer willingness for its 2027 ecosystem-service payment proposal. The county compiled GIS information supplied by the Taitung Branch of the Forestry and Nature Conservation Agency, including eBird waterbird hotspots, buffer areas around threatened plants on the Red List, and important habitats for threatened plants. Local staff are then asked to verify actual land use, current paddy area, parcel identifiers and farmer willingness. The notice explicitly says GIS is an initial screen and that actual implementation still depends on program rules and field conditions. The original local source is Donghe Township Office, “Preliminary inventory for the 2027 ecosystem-service payment program,” published September 1, 2026; its precise page remains verifiable through ordinary browsing and search, but this repository's current CI link checker cannot reliably reach that single page, so it is intentionally retained here as a source label and date rather than a clickable URL.

That is more mature than simply publishing a “subsidy map.” Ecological layers help staff decide where to ask first, but they do not make the farmer's decision. Parcel data can help direct outreach, but it cannot replace field verification. Even a farmer's decision not to participate is useful information because it reveals the gap between theoretical habitat potential and the area that is realistically implementable. This “data finds clues, people confirm them” division of labor is also a good boundary for AI in public administration.

Taimali's September 7 notice adds a concrete deadline

Taimali Township Office turned the same policy process into a more immediate service action on September 7: interested farmers were asked to provide relevant land information by September 15. See Taimali Township Office | Preliminary ecosystem-service payment inventory notice.

For a farmer, a policy is useful only if several practical questions can be answered: Could this apply to me? Who should I contact? What information do I need? When is the deadline? What happens after I respond? Read together, Donghe's explanation of the screening logic and Taimali's explicit response deadline expose both the opportunity and the remaining service gap. A local digital service should not merely display the same document in another interface; it should help people understand why they may have been contacted and what still requires human confirmation.

“Ecological salary” means paying for public ecological services, not rewarding non-production

The Forestry and Nature Conservation Agency's ecosystem-service payment policy is based on a straightforward idea. When farmers, communities or land managers carry additional responsibilities that maintain important species and habitats, public policy can compensate those ecological services rather than rely only on restrictions or moral appeals. The agency has used the idea of an “ecological salary” to explain this approach. See Forestry and Nature Conservation Agency | Ecosystem-service payment program and Hualien Branch | Important-habitat ecosystem-service payments.

This is why advance screening matters. A passive public notice can miss the very people who are most likely to qualify. Others may see the notice but be uncertain whether their land is relevant, while local staff receive many inquiries that are difficult to sort quickly. A careful GIS screen can narrow the search, and township staff and farmers can then verify the facts together. The system should reduce search costs without removing the right to ask questions, correct records or decline participation.

The transferable lesson for the 55 Indigenous areas is not a map but a predictive-service workflow

The same logic can be adapted to other services in Indigenous areas, including agricultural-disaster relief, environmentally friendly farming, farm-equipment support, youth return-to-community training, tribal-industry equipment, remote transportation and disaster resilience. Government already holds some geographic, agricultural and risk information. People should not always have to start from zero and search through dozens of notices.

A low-risk proactive public service can use four layers. The first is an eligibility clue: lawful and appropriate GIS, crop, hazard or administrative data identifies areas that may deserve outreach. The second is human confirmation: staff or the person concerned verifies land, identity, current conditions and willingness. The third is document guidance: the system explains missing fields, deadlines and the official channel but does not make the final eligibility decision. The fourth is outcome feedback: the service records which outreach was useful, where the screen was wrong and which conditions repeatedly caused confusion so the next round can improve.

AI is best used here for matching records, summarizing official notices, deadline reminders, duplicate detection, missing-field prompts and staff search. It should not use an opaque score to decide who “deserves” a payment. Threatened-species locations, personal data and fine-grained cadastral records are sensitive. A public-facing service should expose only what is necessary, while internal access should be role-based and logged so better service does not become an accidental disclosure mechanism.

Moving from notice-centered to service-centered administration

Taitung's current process remains a preliminary inventory for a future-year proposal. The final scale, implementation areas and funding will still depend on central rules and later approval. Donghe's notice is careful on this point: the GIS layers are a preliminary screen, not an eligibility determination. That limitation is a strength because it keeps the digital tool in the role of identifying possibilities rather than taking over legal or administrative review.

If the 55 Indigenous areas institutionalize this approach, each program could have a compact service record: responsible authority, source notice, publication date, target group, possible eligibility clues, fields requiring human confirmation, deadline, staff contact, last-update date and data-sensitivity level. When a new program appears, AI can compare the service record with lawful data and flag “who may need to know.” A person or authorized official then decides whether contact is appropriate and what happens next. That is more useful than building another portal whose main function is to accumulate notices.

Conclusion: good AI administration helps offices find people earlier; it does not decide for them

The most important policy signal in Donghe and Taimali's ecosystem-service inventory is that several steps have moved earlier. GIS brings forward the search for possible areas. Township offices bring forward outreach. Farmer willingness brings forward a reality check before a proposal is finalized. If AI is added, it should follow the same principle: find likely omissions earlier, organize information faster, and remind people more accurately, while eligibility, obligations and sensitive-data decisions remain governed by formal rules and accountable people.

For Indigenous areas, digital governance improves when public service no longer means simply giving residents one more website to search. The stronger model is a lawful, transparent and correctable service that can recognize who may need help and deliver the right official information before a deadline passes.

Yuan Media AI | Continue asking from your role

• Indigenous township agriculture or industry staff: How can GIS potential layers, farmer records and application conditions become a maintainable service workflow?
• Farmers and community producers: When you receive an inventory notice, what information can be provided now, and what still requires formal review?
• Ecology, GIS and data-governance specialists: How can sensitive species locations and farm data be protected while still supporting public service?
• Digital public-service and AI designers: Which tasks are appropriate for AI assistance, and which decisions must remain with authorized officials?

Main official sources

• Donghe Township Office | Preliminary inventory for the 2027 ecosystem-service payment program | 2026-09-01 (precise official page retained as a source label because the current GitHub CI link checker cannot reliably reach that single page)
• Taimali Township Office | Preliminary ecosystem-service payment inventory notice
• Forestry and Nature Conservation Agency | Ecosystem-service payment program
• Chiayi Branch, Forestry and Nature Conservation Agency | 2025 program revision
• Hualien Branch, Forestry and Nature Conservation Agency | Important-habitat ecosystem-service payments