AMMonitor: Remote monitoring of biodiversity in an adaptive framework. Version 2.1

Amid climate change and rapidly shifting land uses, effective methods for monitoring wildlife are critical to support scientifically-informed resource management decisions. As such, monitoring is motivated by ecological hypotheses or natural resource management objectives. The practice of using Autonomous Monitoring Units (AMUs) such as trail cameras or audio recorders to monitor wildlife species has grown immensely in the past decade, with monitoring projects spanning species from birds, to bats, amphibians, insects, terrestrial mammals, and marine mammals.
AMUs can be deployed for long periods of time to collect massive amounts of audio and photographic data. However, the data management requirements can be immense, leaving researchers buried under terabytes of data. A monitoring program is a collection of people, equipment, monitoring locations, location characteristics, research objectives, and data files, with multiple moving parts to manage. Without a comprehensive framework for efficiently moving from raw data collection to results and analysis, monitoring programs are limited in their capacity to characterize ecological processes and inform management decisions in a timely manner.
AMMonitor stands for "monitoring for adaptive management". It is an analysis ecosystem (R + SQLite + media storage) that seamlessly incorporates wildlife monitoring data, species distribution modeling, and decision tools. The analysis ecosystem is characterized by:

Continuous stream of data collection on a variety of wildlife taxa. Autonomous monitoring units (AMUs) such as trail cameras or audio recorders allow the remote capture of digital files that form the backbone of wildlife monitoring. AMUs can be deployed at small or large scales, over short or long-time frames, and are inexpensive relative to human-based surveys.

Standardized yet flexible data and metadata infrastructure. Because many terabytes of monitoring data are collected by AMUs, the data management requirements of an AMU-based monitoring effort are immense. A standardized data infrastructure is critical to store files with strict metadata standards that can be easily incorporated into open access repositories.

Rapid analysis. Vast quantities of AMU data may be collected in a short amount of time, rendering manual inspection of files for target wildlife species impractical. Accordingly, machine learning (ML) algorithms can be used to automatically detect wildlife species in recordings and photos, avoiding the time-consuming task of manual annotation. ML outputs can be seamlessly incorporated into species distribution models and updated as new data are collected via continuous monitoring.