Estimation of Coast-Wide Population Trends of Marbled Murrelets in Canada Using a Bayesian Hierarchical Model

• Published in: PloS one 2015-08, Vol.10 (8), p.e0134891-e0134891
• Main Authors: Bertram, Douglas F, Drever, Mark C, McAllister, Murdoch K, Schroeder, Bernard K, Lindsay, David J, Faust, Deborah A
• Format: Article
• Language: English
• Subjects: Air traffic control; Analysis; Animal behavior; Animals; Aquatic birds; Auks; Bayes Theorem; Bayesian analysis; Birds; Brachyramphus; Brachyramphus marmoratus; Breeding; British Columbia; Canada; Charadriiformes; Coastal zone; Conservation; Conservation of Natural Resources; Ecosystem; Endangered & extinct species; Endangered species; Environmental monitoring; Female; Forest watersheds; Forests; Geography; Interannual variability; Male; Monitoring; Multivariate Analysis; Nesting; Nesting Behavior; Old growth; Old-growth forests; Population Dynamics; Population statistics; Predatory Behavior; Prey; Probability; Radar; Regression Analysis; Risk taking; Threatened species; Time Factors; Trends; Watersheds; Wildlife conservation
• ISSN: 1932-6203; 1932-6203
• DOI: 10.1371/journal.pone.0134891

Species at risk with secretive breeding behaviours, low densities, and wide geographic range pose a significant challenge to conservation actions because population trends are difficult to detect. Such is the case with the Marbled Murrelet (Brachyramphus marmoratus), a seabird listed as 'Threatened' by the Species at Risk Act in Canada largely due to the loss of its old growth forest nesting habitat. We report the first estimates of population trend of Marbled Murrelets in Canada derived from a monitoring program that uses marine radar to detect birds as they enter forest watersheds during 923 dawn surveys at 58 radar monitoring stations within the six Marbled Murrelet Conservation Regions on coastal British Columbia, Canada, 1996-2013. Temporal trends in radar counts were analyzed with a hierarchical Bayesian multivariate modeling approach that controlled for variation in tilt of the radar unit and day of year, included year-specific deviations from the overall trend ('year effects'), and allowed for trends to be estimated at three spatial scales. A negative overall trend of -1.6%/yr (95% credibility interval: -3.2%, 0.01%) indicated moderate evidence for a coast-wide decline, although trends varied strongly among the six conservation regions. Negative annual trends were detected in East Vancouver Island (-9%/yr) and South Mainland Coast (-3%/yr) Conservation Regions. Over a quarter of the year effects were significantly different from zero, and the estimated standard deviation in common-shared year effects between sites within each region was about 50% per year. This large common-shared interannual variation in counts may have been caused by regional movements of birds related to changes in marine conditions that affect the availability of prey.