So Much Noise! Extracting Tag Signals from Radio-Telemetry Data Using a Custom Python Script

Many fish species are either rare, cryptic, and/or occur over large spatial and/or temporal scales. Given this, their life histories are rather difficult to quantify. Radio-telemetry has become a valuable tool for the derivation of their life-history data, and a primary mechanism for evaluating such fishes in their natural environment. We employed radio-telemetry to quantify the migratory patterns of a big-river cyprinid (Golden Mahseer, Tor pitutora) in the Himalayan rivers of Bhutan. Large adult fish (N=64) were captured by angling and surgically implanted with radio-transmitters that emitted signals at 3-10s intervals. Data [i.e., frequencies (individual tag) and power (strength of signal)] were subsequently recorded by solar-powered data-loggers located at bankside stations (N=17) along major rivers. All frequencies were recorded, to include those from radio-transmitters as well as non-target sources (=noise). To separate data from noise, we developed a customizable pipeline in PYTHON that filtered valid frequencies from the millions of data points recorded. This was done by diagnosing, then extracting consistent data patterns from the signals, as well as their variability. Such techniques are routinely applied to parse large genomic data sets but, to our knowledge, have not been developed in an automated, customizable manner for radio-telemetric data.