The bird community serves as a very good “barometer” of the effect of environmental change (climate change, wildfires, drought) on biodiversity in our remote mountain ecosystems. The Institute for Biodiversity and the Environment (IBE) (biodiversity-institute.org) has as one of its core research projects the Pacific Crest Trail Biodiversity Megatransect (PCTBM), which is an ongoing survey of the montane bird diversity patterns and breeding phenology (seasonal timing of breeding behaviors) along the Pacific Cordillera – the mountain chains of California, Oregon, and Washington.

Central to this research is the use of novel, noninvasive methods (survey methods that do not require the capture or handling of animals); this includes the use of automated recording units (ARUs) that record bird song to continuously inventory and monitor bird diversity across the wilderness regions of the entire length of the PCT (from Mexico to Canada, 2,650 miles). The PCTBM combines onsite wildlife surveys by field biologists with the use of lightweight ARUs that can be packed in on foot and deployed at remote survey sites along the trail. Since 2010, PCTBM has generated many thousands of 5- and 7-min. recordings of the singing bird community across 3,578 surveys sites along 1,700 miles of the PCT in California, from Mexico to Oregon.

Interpreting these recordings with human observers that listen to all of the recordings to vocalizing birds can be costly, requiring hundreds of interpreter hours. This approach, however, is needed because it provides the necessary level of replication in the dataset that improves statistical models in the analysis. We employ hierarchical statistical models, such as multi-species occupancy models (MSOMs), to estimate species richness, species distributions, population status of individual species, and the breeding phenology in response to underlying climatic variables.

However, due to this high labor cost of using human observers to interpret hundreds of hours of recordings from the ARUs, we are exploring the use of machine learning and artificial intelligence (AI) to identify vocalizing bird species on the audio files. In addition to AI, we employ other types of computer algorithms that make calculations and that help to summarize and synthesize information on the recordings, such as bioacoustic indices, which can serve as metrics of species richness (the number of species detected at each survey site).

In the past, we have implemented surveys along the PCT across 3,578 survey sites, including the entire length of California (1,700 mi) in 2006 and several sections (hundreds of miles long) in California in subsequent years (2010, 2015, 2016, 2017, and 2019).

As of the date of this writing, the PCTBM, thus far, has produced 6 peer-reviewed articles, with 1 more currently in review, on the climatic and environmental drivers of montane avian diversity and breeding phenology.

We feel that the scientific theory, analysis approach, and survey methods has matured to the level that the PCTBM is primed to be implemented at a larger spatial-temporal scales across the entire length of the PCT from Mexico to Canada (2,650 mi).

The overarching goal of the PCTBM is to become established as a regular monitoring program that informs the decision-making response addressing climate change and other sources of environmental change. The PCTBM spans multiple administrative units (federal, state, and private lands) and promises to help inform decision-makers across these boundaries on their response climate change, human-land use change, or other sources of environmental change.

For more details go here.