Retrospective analysis of Trumpeter Swan Cygnus buccinator decline in Yellowstone National Park, USA

Abstract

The Trumpeter Swans Cygnus buccinator that nest and winter in Yellowstone National Park (YNP) played an important role in the survival of the species when unregulated shooting and habitat loss threatened them with extirpation. While conservation measures saved the Trumpeter Swan, and their numbers have increased greatly across North America, the abundance and productivity of YNP’s resident Trumpeter Swans declined from about 1960 until 2010, when captive-bred releases began to supplement their numbers. Many hypotheses for the initial decline in YNP Trumpeter Swans exist, including human disturbance at nesting areas, changes in habitat quality, predation and lower Trumpeter Swan abundance in the broader geographic region reducing the immigration of swans into YNP. This study used long-term historical Trumpeter Swan monitoring data and existing covariate data to explore competing hypotheses about possible factors associated with temporal and spatial variation in swan abundance and reproductive success in YNP during 1931–2011. Bayesian reversible jump Markov chain Monte Carlo (RJMCMC) methods were used to evaluate whether covariates representing swan decline hypotheses explained variation in annual, wetland-level patterns of where swans were absent (Absent), present but without fledged young (Present), and bred successfully with fledged young (Successful) each year. Model covariates that explained variation in wetland status supported several of the hypotheses for Trumpeter Swan decline. Wetlands within YNP were more likely to have Trumpeter Swans Present as opposed to Absent during 1931–1959 in years when the total abundance of Trumpeter Swans in the broader geographic area around YNP was greater. During 1960–2011, wetlands within YNP were more likely to have Trumpeter Swans Present as opposed to Absent when estimated Grizzly Bear Ursus arctos abundance was lower and when YNP recorded fewer annual visitors, although these covariates correlated strongly with time, making it difficult to distinguish between the underlying causes of temporal trends and preventing stronger inferences from being made. The lakes, rivers and wetlands identified quantitatively in this study as being the most likely to have swans Present and/or Successful can be a useful tool to help YNP staff manage important swan habitat or justify future management actions.