Why it matters
Without AI surveillance, hidden infections and pest incursions can go undetected, delaying containment, risking wild-origin material, and compromising safety near visitors and staff.
From design to safety
Design
AI-assisted surveillance design
Design data streams from ash dieback, Phytophthora ramorum, oak processionary moth, Xylella risk models, and survey protocols to define detection thresholds, sampling plans, and alert criteria.
- Ash dieback sentinel design
- Phytophthora ramorum sentinel nursery
- Oak processionary moth perimeter plan
Build
Data infrastructure and sensors laid
Install sensor nets, AI-enabled cameras, automatic weather stations, and periphery traps in sentinel plots; integrate observations with accession records and plant health files for real-time verification.
- Ash dieback sentinel plot network
- Phytophthora ramorum nursery block sensors
Operation
Continuous AI surveillance in operation
AI reads imagery, hyperspectral data, leaf moisture indices, temperature, wind patterns; flags anomalies, cross-validates with plant health notes, and publishes alerts to the plant health officer for inspection.
- Integrated plant health surveillance hub
- Comprehensive arboretum surveys data logger
Maintenance
Ongoing calibration and data curation
Regular sensor calibration, phenology alignment with Index Seminum timelines, and reconciliation with accession records to maintain data integrity for risk assessments across arboretum blocks.
- Ash dieback sentinel design
- Oak processionary moth perimeter plan
Safety
Responsible risk governance and sign-off
Tree risk assessments, biosecurity checks, and safety reviews require commissioning by the garden director and plant health officer before any field action.
- Xylella containment perimeter fence sensors
- Integrated plant health surveillance hub
Part by part
- 1
Ash dieback sentinel plot network
- Watch for
- early signs of dieback or lesion spread in ash species or hybrids within monitored plots
- AI action
- AI detects pattern deviations and triggers field inspection orders and updated treatment plans
- 2
Phytophthora ramorum sentinel nursery block sensors
- Watch for
- unexpected wet-dry cycle anomalies and canopy probe readings indicating oomycete stress
- AI action
- Anomaly signals prompt containment halt and sampling for verification
- 3
Oak processionary moth perimeter monitoring log
- Watch for
- clusters of caterpillars or nest buildup near public access routes
- AI action
- Trail checks scheduled and pheromone trap data escalated to pest management team
- 4
Xylella containment perimeter fence sensors
- Watch for
- abnormal sap oozing or vector presence near host plants
- AI action
- Fence sensors trigger boundary patrols and quarantine actions
- 5
Integrated plant health surveillance hub
- Watch for
- multisensor convergence indicating stress patterns across taxa
- AI action
- Cross-correlate with accession records and alert plant health officer
- 6
Comprehensive arboretum surveys data logger
- Watch for
- inconsistent accession provenance or mislabeling during surveys
- AI action
- Survey anomalies trigger verification workflow and record corrections
Worked example
Early detection of ash dieback in a mixed-species stand
Illustrative scenario with sample data.
An AI system analyzes canopy images, leaf reflectance, and soil moisture in a stand of ash and maple, flagging a pattern consistent with dieback progression.
Signal in
sensor readings: canopyIndex=0.72, leafNDVI=0.41, soilMoisture=28%, date=2026-04-12; accession=Q-ASH-047, provenance=Europe, plantingYear=1998, location=Stand A; samplingTime=08:15; weather=overcast; soilTemp=16.2C; leafTemp=22.5C; humidity=68%; sensorID=SN-0834; plotID=P-12
System out
alertID=0521-StandA; riskLevel=high; action=inspect and collect leaf, bark, and soil samples; reportDue=2026-04-12; notes=partial canopy decline; decisionTrail=pending field validation; followUp=soilprint, herbarium check;
Human sign-off: Head of plant health reviews the alert, signs off on sampling protocol, and coordinates containment actions with the garden director.
How to roll it out
Phase 1
Weeks 1 to 4: Baseline
Install sensors, calibrate equipment, integrate with plant records, and establish alert thresholds with key staff.
Phase 2
Weeks 5 to 8: Validation
Test anomaly detection against historical data, refine cross-checks with accession inventories and phenology calendars.
Phase 3
Weeks 9 onwards: Operational
Go-live for routine monitoring, automated reporting, and field response workflows with trained sign-off paths.
Where people stay in charge
- Decisions on interventions and pesticide use must be signed off by the plant health officer and garden director.
- All data handling, quarantine actions, and boundary protections must comply with biosecurity policies and legal requirements.
- Sensor data interpretations must be corroborated by human observers before any deployment or public messaging official.
Questions owners and crews ask
How does AI surveillance integrate with existing plant health records?
AI surveillance reads sensor data, imagery, and phenology notes from plant records to generate risk signals; staff verify and annotate final decisions within the living collections records.
What happens when AI flags a potential outbreak near public areas?
The plant health officer coordinates a rapid assessment, informs the garden director, and implements containment measures while logging actions in the biosecurity module.
How are new pathogens added to the surveillance model?
Staff validate new threat signatures using controlled samples, update the verification rules, and retrain the model with curated reference data in the plant-health system.
Who approves changes to the surveillance thresholds and response plans?
The garden director and plant health officer jointly approve threshold adjustments and response protocols after risk review and policy alignment.