A project of


Mountain Birdwatch predominantly uses hierarchical Bayesian binomial mixture models to analyze the 'mountain' of data collected by our cadre of field-savvy community scientists.

Mountain Birdwatch Past and Present

Past: 2000–2010

Initiated by the Vermont Center for Ecostudies in 2000, Mountain Birdwatch has always relied on community scientists to conduct annual counts along hiking trails at fixed locations throughout the mountains of northern New England and New York. Originally, Mountain Birdwatch was heavily focused on Bicknell’s Thrush and four other species: Winter Wren, Swainson’s Thrush, White-throated Sparrow, and Blackpoll Warbler. Community scientists used an intensive minute-by-minute procedure to count Bicknell’s Thrush and a more simple counting method for the other species. As such, it is difficult to effectively compare species’ counts from the past (2000-2010) Mountain Birdwatch dataset to our current (2011-onward) data collection. In 2010, we permanently retired the routes and counting methods used in the first decade of Mountain Birdwatch.


Present: 2011–onward

In 2011, we relaunched a more robust version of Mountain Birdwatch which included a new, generalized random tessellation stratified sampling (GRTS) selection of routes across the northeastern United States and a revised survey protocol to allow for more stringent and intensive statistical analyses (for an example from our recent research, read this paper). Since 2011, Mountain Birdwatch has consisted of approximately 750 sampling stations located on 129 routes spread across four states: New York, Vermont, New Hampshire, and Maine. In any given year, approximately 80-90% of the routes are surveyed. Each sampling station is surveyed annually in June by one of the 100+ community-scientists who participate in Mountain Birdwatch. The revised version of Mountain Birdwatch also adopted a common counting method for all species, and expanded the number of monitored species to include Yellow-bellied Flycatcher, Boreal Chickadee, Black-capped Chickadee, Hermit Thrush, and Fox Sparrow, in addition to the original group of five Mountain Birdwatch species.


Routes, Sampling Stations, and the Community Scientists

Mountain Birdwatch is focused on the breeding bird community within the high-elevation spruce-fir forests of the northeastern United States. As a proxy for the extent of this montane boreal habitat, we used a model of the Bicknell’s Thrush breeding range based on Lambert et al. (2005; for an interactive map of that range, click here). As such, most of the sampling stations occur solidly in the spruce-fir zone, but some stations occur in the hardwood-boreal transition or at treeline. Sampling stations occur between 573 and 1500 meters (mean = 1009 meters) in elevation, but 95% of the sampling stations occur between 759 and 1333 meters. Each Mountain Birdwatch route consists of 3-6 sampling stations located on hiking trails (or occasionally retired logging roads). The minimum number of sampling stations on a route is determined by the extent of the trail network and the size of the spruce-fir zone on that mountain.

Each year, community scientists (who self-identify as experienced birders) adopt one or more routes to survey on any day in June with fair weather. Mountain Birdwatch intentionally features a short list of species to survey, and extensive training materials for participants, so just about any birder can contribute. Most observers survey the same route year after year. We encourage observers to bring a companion with them for safety purposes, but that companion does not assist the observer in detecting or counting birds. Due to the remote sampling locations and early morning survey start time, most observers backcountry camp overnight prior to the survey near their first sampling station.

All 750 Mountain Birdwatch sampling stations

All 750 Mountain Birdwatch sampling stations.

Species and Survey Methods

Only a few dozen bird species regularly breed in the spruce-fir zone in our region. From this group, we selected ten bird species for targeted monitoring based on level of conservation concern, degree of habitat specialization, probability of upslope range expansion, and range restriction. We also monitor the abundance of red squirrels to understand how this common nest predator’s cyclical population dynamics interact with those of our monitored bird species. [March 2020 update: we have a paper in review at Ecology Letters describing these red squirrel population dynamics.]

We designed the Mountain Birdwatch surveys to achieve three main goals, to:

  1. estimate on an annual basis the abundance of target species within the mountains of our region,
  2. measure changes in the abundance of target species over time, and
  3. relate trends in abundance to biotic and abiotic variables that may affect the target species.

The Mountain Birdwatch protocol consists of four consecutive 5-minute counts at each sampling station, for a total sampling period of 20 minutes per station. Observers conduct repeated counts for all focal species during each 5-minute period, noting whether individuals were detected within or beyond 50 meters of the station. [Note: as of 2019, observers can optionally record counts for any bird species (not just the focal species) and indicate if their 5-minute counts are complete checklists.] This “repeated measures sampling format” is one of the greatest strengths of our sampling protocol. By conducting multiple back-to-back counts, we can statistically account for the number of  individuals missed by observers during their counts. Surveys are conducted during the month of June, which is the collective period of greatest singing activity for the target species. In order to increase the likelihood of detecting Bicknell’s Thrush, which is most vocal during the pre-dawn period, observers begin surveys 45 minutes before sunrise.

Here are some examples of how detection probability varies with the time (left panel) and date (right panel) of the surveys; Winter Wrens are most likely to be detected during point counts in the early morning (left panel) and beginning and end of June (right panel). The dark lines represent the mean response, and the thin colored lines represent the 95% Bayesian credible interval (which are less-likely forms of those relationships). We account for covariates that affect detection probability of the birds to improve our estimates of density for those species. Otherwise, the variation attributable to un-modeled detection processes would surface as increased uncertainty in our density estimates.

Inclement weather can greatly reduce an observer’s ability to detect birds in the field, so each survey is conducted when temperatures are above 35ºF and when precipitation and wind conditions are favorable. Occasional drizzle or a brief shower is acceptable, but surveys are not conducted during periods of steady drizzle or prolonged rain. Likewise, a light wind with occasional gusts is considered acceptable weather for surveying birds, but a stiff breeze of >18 miles per hour is not. Observers also report background noise levels at each sampling station; this information is used to help model the detection process within our models.

Mountain Birdwatch observers record the background noise (e.g., caused by nearby ephemeral streams) during their point counts using a simple 1-10 scale. We incorporate these simple 1-10 estimates of background noise into our models to describe variation in detection probability between sampling locations. Here, you can observe that background noise begins to obscure Blackpoll Warbler songs at a mean background noise value of ~5. Observers formerly used an app to record background decibel levels, but the app measurements were also affected by the amount of bird song! So quiet places with lots of singing birds ended up having the same background noise decibel levels as locations next to extremely loud rushing water where observers couldn’t hear many birds.

Analyzing the Data

Community scientists enter their own survey data into a sophisticated online database (managed by the Forest Ecosystem Monitoring Cooperative) and mail in their original datasheets. Each summer, our interns proof these datasheets (and the online data entries) for transcription errors and unusual observations–contacting the observers when needed. The Mountain Birdwatch Chief Scientist, Jason Hill, is a quantitative ecologist who currently analyzes Mountain Birdwatch data using hierarchical binomial mixture models in a Bayesian framework–all of the results presented in the State of the Mountain Birds Report are generated from these models. By taking advantage of sophisticated Bayesian statistical models, we can transform the simple counts into estimates of density – the number of individuals per unit area – while accounting for the reality that not all individual birds present during a survey will be detected by the observer. These flexible models allow us to easily handle missing data (e.g., because not all routes get surveyed each year) and to account for variation in observer acuity, variation in weather conditions that affect the behavior of birds, and seasonal and daily changes in the frequency of singing.

Each species is analyzed separately, using the full suite of the Mountain Birdwatch data with two exceptions–Fox Sparrow and Boreal Chickadee models only include data from sampling stations located north of 43.6 degrees latitude (the approximate southern breeding boundary of both species in our region). We conduct a model selection procedure in a frequentist framework using the pcount function of the unmarked package in program R, and fit the final model for each species in a Bayesian framework using JAGS. Final inferences are drawn from the posteriors of a model with 3000 saved iterations (1000 iterations saved from three Markov chain Monte Carlo processes) after a thinning rate of 1 in 100. Each model takes several days to run on a 24-core desktop computer with 32 GB of memory. For more details see Hill and Lloyd (2017).

Posterior predictive check. All models are predictive…but are they good at it? From each species’ model, we generate and compare 100,000 simulated data sets to our observed data set to insure that our model makes accurate predictions. The ratio of the simulated data that over- and under-predict the observed data should be P = 0.5 (in a perfect world). Here, P = 0.39 for our Fox Sparrow model indicates a well-fitting model that makes reasonable predictions.

Interested in reading more about how these data are used to help protect species and their habitat? Read about how we translate science into action.

State of Mountain Birds