Date: 05 Feb, 2025


Dataset Title: CHANGES Project - Fish Collection Curated Data


Dataset Creators: King, Katelyn; Schell, Justin; Alofs, Karen; Thomer, Andrea; Wehrly, Kevin; Lenard, Michael; and Lopez-Fernandez, Hernan


Dataset Contact: Katelyn King kingk42@michigan.gov


Funding: Michigan Institute For Data & AI In Society (MIDAS) Propelling Original Data Science Grant 


Research Overview:
Archives at the Institute for Fisheries Research (IFR) hold records of thousands of lake surveys from the University of Michigan and Michigan Department of Natural Resources. 

Fish collection card types include targeted and non-targeted fisheries surveys by the Department of Natural Resources and this information was transcribed and curated into a csv file (fishc_data.csv). These records include information on the gear types used, the area surveyed and the length and mesh size of nets fished. The number and common name of fish species caught were recorded as well and included in a species table (fishc_species_table.csv). A description of all data fields can be found in the fishc_datadictionary.csv. 

Methodology:
Michigan Department of Natural Resources historically collected lake survey data on index cards. We used the Zooniverse crowdsourcing platform for volunteer transcription of these records using various workflows that captured different data. To be included in the dataset, each card was transcribed by three or more volunteers. Zooniverse transcriptions require significant cleaning and curation before the data is in a usable format. We used code to aggregate the transcribed data from each person in order to provide a consensus-based “final answer” and confidence score for each data field, based on how well entries from the different volunteers matched. We then standardized data using techniques such as changing all text to lowercase, trimming excess whitespace, and converting fractions to decimals. We separated numeric and alphabetic values into different data columns. Finally, we standardized units for each variable into a single unit, and when applicable, transformed to metric units (e.g. inches to millimeters). We checked data numeric values by plotting, identifying outliers, and reviewing the original document. In order to combine multiple sampling events for one lake or connect the transcribed data to more contemporary survey data from the MDNR, we matched the records with the corresponding MDNR unique lake identifiers. The transcribed data included each lake’s name, county, and in some instances geographic reference data in the form of Township, Range, and Section from the United States Public Land Survey System (TRS). We joined data entries on lake names, counties, and TRS when available. Remaining lakes that were unmatched due to issues like lakes crossing county lines or changing names over time, were manually matched to data using experts from the research team. Finally, we were unable to match some of the historical data due to insufficient geographic information.

Instrument and/or Software specifications: NA


Files contained here:
fishc_data.csv
fishc_species_table.csv
fishc_datadictionary.csv

We tried to use a standard naming convention for all of our data fields except for identifiers, dates, and comments. The naming convention is as follows:  [variable name]_[min or max]_[unit]. 

Related publication(s):
King, K.B.S., Schell, J, Wehrly, K.E., Lenard, M., Singer, R., López-Fernández, H., Thomer, A.K., & Alofs, K.M. Community science helps digitize 78 years of fish and habitat data for thousands of lakes in Michigan, USA. under review  


Use and Access: 
This data set is made available under a Creative Commons Public Domain license (CC0 1.0).


To Cite Data: 
King, K.B.S., K.M. Alofs, J. Schell,  A. Thomer, K. Wehrly, M. Lenard, & H. Lopez-Fernandez (2025). CHANGES Project - Fish Collection Curated Data [Data set]. University of Michigan - Deep Blue.