**ABSTRACT NOT FOR CITATION WITHOUT AUTHOR PERMISSION. The title, authors, and abstract for this completion report are provided below.  For a copy of the full completion report, please contact the author via e-mail at njohnson@usgs.gov. Questions? Contact the GLFC via email at frp@glfc.org or via telephone at 734-662-3209.**

 

Proof of Concept: Use of DIDSON Cameras to Estimate Adult Sea Lamprey Abundance in Streams

 

1Nicholas Johnson, 2Peter Hrodey, 3Erin McCann, 4Jesse Eickholt, and 3Kevin Pangle

 

1U.S. Geological Survey, Great Lakes Science Center, Hammond Bay Biological Station, 11188 Ray Road, Millersburg, MI 49759

 

2U.S. Fish and Wildlife Service, Marquette Biological Station, 3090 Wright Street, Marquette, MI 49855

 

3Department of Biology, Central Michigan University, Bioscience Building 2406, Mt. Pleasant, MI 48859

 

4Department of Computer Science, Central Michigan University, Mt. Pleasant, MI 48859

 

December 2017

 

ABSTRACT:

 

Effective assessment and control of invasive sea lamprey in the Laurentian Great Lakes relies on knowledge of adult sea lamprey migration timing. We investigated adult sea lamprey migration timing by mining historical trap catch data and deploying Dual-frequency Identification Sonar (DIDSON) at the mouth of a Great Lakes tributary. Over the Great Lakes basin during the past 30 years, trap catch generally peaked at 15 C and was highly correlated with stream temperatures. Furthermore, several streams are now reaching 15 C earlier in the spring and are experiencing peak trap catch up to 30 days sooner than in the 1980s. Our analysis of historical trap catchdata did not reveal when sea lamprey enter spawning streams because most assessment traps are located many kilometers upstream of the river mouth. Therefore, the timing of sea lamprey stream entry was assessed in a Lake Huron tributary (Ocqueoc River) during two years using DIDSON. Sea lamprey entered the stream in low densities when temperatures first reached 4C, which was up to 6 weeks earlier than they were first captured in traps located upstream. Stream entry timing peaked when stream temperatures rose to 12C and discharge was high. Examination at a finer temporal resolution (i.e., minutes) indicated that sea lamprey did not exhibit social behavior (i.e., shoaling) during stream entry. Manual processing of all the DIDSON data was not practical, so we were unable to estimate the number of sea lamprey entering the stream. As a first step toward automated processing of DIDSON and other video data, a distributed pipeline was constructed using the Hadoop ecosystem. The pipeline is capable of ingesting raw DIDSON data, transforming the acoustic data to images, filtering the images, motion detection and extraction, and feature generation for machine learning and classification. Future applications of the pipeline could include monitoring migration times, determining the presence of a particular species, and estimating abundance.