Background information

Analyses of Shorebird Migration Monitoring Data

This analysis includes results from the International Shorebird Survey (ISS), the Atlantic Canada Shorebird Survey (ACSS), and the Ontario Shorebird Survey (OSS).

In 1974, the Manomet Center for Conservation Sciences organized the ISS to gather information on shorebirds and the wetlands they use. The OSS and ACSS, overseen by Environment and Climate Change Canada, were also initiated at this time to address similar objectives. These surveys are now also used to monitor trends in relative abundance of shorebirds during migration at the continental and regional level. The integrated shorebird survey analyses produce results for trends and annual indices for 28 species of shorebirds with varying levels of precision. Volunteers have carried out well over 100,000 surveys, with on average 1300 added each year by up to 100 surveyors. Surveyors are asked to census a location using ISS Guidelines three times per month during key migration periods in both spring and fall. Because these birds are migrants, counts at a site vary dramatically through time, and timing of peak passage varies among species; in contrast to previous analyses based on mean peak counts, this analysis was based on raw counts during the peak (i.e., central 95%) of the species-specific migration period.

Statistical Methods

In this analysis, trends were estimated with a hierarchical Bayesian model using Markov Chain Monte Carlo sampling. The geographic strata for the analysis grouped individual survey sites within biologically relevant regions. The model assumed that counts follow an over-dispersed, Poisson distribution, and it included terms for a long-term log-linear trend, year-effects, day-of-season effect (a generalized additive smoothing term), and site-level abundance. The trend terms within each stratum were estimated as hierarchical random effects distributed around a mean, continental trend. In effect, trend estimates were allowed to vary among the strata, but were partially shrunk towards the continental trend for strata with relatively sparse data. This means that the continental trend estimates are more strongly influenced by rates of change in the strata with more data. Although this relative weighting of stratum-level trends is somewhat arbitrary, it is not currently possible to estimate the proportion of a species' continental population migrating through each stratum. The year effects were estimated as random effects -- annual departures from the stratum-level long-term trend lines -- and were constrained to be identical among strata (i.e., the hierarchical structure allowed them to vary among years but not among strata). Trend estimates were calculated as the transformed slope of a log-linear regression of the annual indices over time. The 95% credible intervals on the trend estimates were calculated from the full posterior distribution of regression slopes. Therefore, the trends partly incorporate the estimated year-effect fluctuations into the short- and long-term estimates of population change.

For more information on ECCC's shorebird programs, visit the Shorebird surveys web page.