Research Project Overview
The University of Portland's Franz River Campus sits on the banks of the Portland Harbor Superfund Site - a 10-mile stretch of the lower Willamette River designated by the EPA in 2000 as one of the most contaminated waterways in the Pacific Northwest. Following a decade of active remediation, this project asks a fundamental restoration ecology question: have native pollinators returned to this recovering riparian landscape, and are they functionally connected to the broader campus green space network?
Using a custom-built 8-station AprilTag detection network, we individually identify Bombus vosnesenskii (yellow-faced bumblebee) workers at both the Franz River Campus riparian corridor and the student-run SLUG garden on upper campus, tracking whether bees move between sites, stay in their capture location, or establish patterns that reveal the ecological connectivity - or isolation - of this recovering habitat.
Site context & remediation history
The Portland Harbor Superfund Site encompasses a 10-mile stretch of the lower Willamette River contaminated by over a century of industrial activity - including shipyards, manufacturing, and chronic urban stormwater runoff. Sediments and soils along this corridor are heavily tainted with polychlorinated biphenyls (PCBs), dichlorodiphenyltrichloroethane (DDT), heavy metals, and other persistent organic pollutants that accumulated from the late 19th century through the 20th century.
The University of Portland's river-facing property sits directly on this contaminated corridor. In 2006, the university entered into a formal agreement with the EPA to assess and plan remediation of legacy soil and groundwater pollution along the riverbank. Major active remediation - including contaminated soil excavation, heavy metals and PCBs removal, and riverbank capping to prevent further runoff - took place between 2012 and 2014.
Following remediation, the university developed the Franz River Campus, now home to the E.L. Wiegand Environmental Lab, the university boathouse, and restored riparian plantings along the Willamette shoreline. By 2026, the site has had approximately 12 years to begin ecological recovery since active remediation.
Research question
This question is addressed through individual bee tracking at both sites simultaneously. By tagging bees captured at the Franz River Campus (lower campus, river's edge) and at the SLUG community garden (upper campus), we can determine:
A secondary question addresses whether the Franz River Campus riparian corridor is supporting a self-sustaining B. vosnesenskii population 12 years after active remediation, or whether bee presence is primarily driven by movement from established upper campus habitats.
Study sites
The two sites are separated by approximately 0.25 miles and 180 feet of elevation change - a topographic gradient that adds ecological significance to any confirmed inter-site movement, as bees navigating this corridor must traverse both built campus infrastructure and significant elevation.
Bees are captured by hand net at both sites during peak foraging hours. Each bee is briefly chilled in a freezer then fitted with a unique AprilTag identifier attached to the dorsal thorax with cyanoacrylate gel adhesive.
Each station runs continuous AprilTag detection during preset hours. When a tagged bee enters the camera field, the system automatically:
Detection CSVs from all 8 stations sync automatically to a shared Google Sheet each evening, providing real-time data access for the full research team.
Documenting native pollinator recovery at a Superfund remediation site addresses a significant gap in restoration ecology literature. While vegetation recovery following contaminated site remediation is well-studied, invertebrate and pollinator recolonization timelines are far less documented - particularly in urban riparian environments.
The Franz River Campus setting is ecologically unusual: a remediated industrial shoreline embedded within an active university campus, adjacent to both the Willamette River and a productive student-managed food garden. This configuration allows direct comparison of pollinator activity between a recovering contaminated site and an established urban agricultural habitat within the same landscape.
Findings from this study have direct relevance for:
This project is led by Dr. Tara Prestholdt of the University of Portland Biology Department and the Summer Undergraduate Research Experience. Student researchers manage daily station deployment, bee capture and tagging, data collection, and station maintenance under faculty supervision.
Student involvement spans hardware assembly, ecological field protocols, data management, and statistical analysis - providing genuine hands-on experience in restoration ecology research methods, embedded systems, and scientific computing. The project connects Biology, Environmental Science, and the university's sustainability mission through a real-world research problem with publication-quality outcomes.
Technical appendix
| Compute | Raspberry Pi 5 1GB · BCM2712 quad-core 2.4GHz |
| Camera | Waveshare IMX219 · 8MP Sony sensor · 120° FOV · MIPI-CSI |
| Resolution | 3280 × 2464 full sensor · continuous video capture |
| Storage | Samsung PRO Endurance 64GB MicroSD · endurance-rated |
| Power | Anker PowerCore 10K · ~6hr field runtime · USB LED keep-alive |
| Enclosure | IP65 hinged weatherproof box · PATIKIL 15.7" gooseneck arm |
| Mount | PVC pipe stake · 1/4"-20 brass heat-set insert in box lid |
| Network | UPIoT MAC-registered · Pi Connect remote terminal access |
| OS | Raspberry Pi OS Lite 64-bit · Bookworm · headless |
| Detection | OpenCV-contrib 4.13.0 · DICT_APRILTAG_36h11 · ArUco3 |
| Camera | picamera2 · libcamera v0.7.1 |
| Data sync | gspread + google-auth · Google Sheets API v4 |
| Auto-start | crontab @reboot · 20s delay · daily sync 0 20 * * * |
| Source code | github.com/bedell-up/beetracking |
| Tag family | AprilTag 36h11 · 587 unique IDs · strong error correction |
| Printed size | 10–12mm · minimum reliable size for B. vosnesenskii thorax |
| Detection height | 10–15cm · tag appears ~60px in 3280px-wide frame |
| Detection mode | ArUco3 · minSideLengthCanonicalImg = 32 |
| Error correction | errorCorrectionRate = 1.0 · maximum tolerance |
| Cooldown | 10 seconds per tag ID · prevents duplicate logging |
Couvillon, M.J., et al. (2012). Effect of weight on flight performance of bumblebees. Journal of Apicultural Research.
Olson, E., et al. (2011). AprilTag: A robust and flexible visual fiducial system. Proceedings of ICRA 2011.
U.S. Environmental Protection Agency. (2000). Portland Harbor Superfund Site designation. EPA Region 10.
University of Portland Facilities Planning. Franz River Campus remediation documentation. ww1.up.edu/facilitiesplanning/river-campus.html