Surveillance des populations aviaires boréales : comment estimer les tendances et les trajectoires àpartir de données bruyantes?

International audienceubstantial effort has been dedicated to developing reliable monitoring schemes for North American bird populations,but our ability to monitor bird populations in the boreal forest remains limited because of the sparsity of long-term data sets, particularlyin northerly regions. Given the importance of the boreal forest for many migratory birds, we set out to (1) summarize the main challengesassociated with monitoring avian populations, (2) describe the available statistical tools for population monitoring and their applications,and (3) identify future directions to overcome current challenges in monitoring bird populations in the boreal forest. Defining anddelineating populations of interest and identifying the drivers that affect those populations present the greatest current challenges. Thisis because migratory birds may be affected by many population-limiting processes at different stages of their annual life cycles. Thesefactors are often hierarchically structured and can influence populations at the local, regional, or continental scales. Some of thechallenges associated with delineating populations and identifying population drivers can be addressed via the plethora of samplingand analytic methods available to examine population change over time. Choosing the proper analytic methods depends on the goalsof the study and the nature of the data such as single or multiple populations, repeated occurrence or count-based surveys, ordemographic rates. Recent advances in hierarchical and integrated population models make these analytic approaches some of themost promising avenues for the development of future methods. However, these tools require large data sets, and acquiring sufficientdata on bird populations and potential explanatory variables is difficult in the boreal forest. If the current challenges to monitoringbirds in the boreal forest are to be overcome, serious effort should be dedicated to integrating existing data and making them accessible.Enhancing survey effort through multispecies surveys will also play an important role. Implementing spatially balanced sampling planswith a rotating panel design could balance the trade-offs between spatial versus temporal replication at an affordable cost. Improvingthe accessibility of environmental covariates that are spatially and temporally explicit would also enable development of mechanisticpopulation models that improve our understanding of migratory bird population dynamics. Finally, given that long-term monitoringprograms can take many decades before delivering reliable population trends and that organizational priorities often change over time,we suggest that collaborative efforts will help ensure the long-term survival of new monitoring programs