Forest inventory attributes predict the presence of cavities suitable for nesting by wood ducks

Secondary cavity-nesting birds such as waterfowl and raptors rely on tree cavities developed principally through decay and damage processes or excavation by woodpeckers. Forest and wildlife managers are tasked with maintaining and producing these essential habitat components through forest management practices. Generating predictions about where cavities have developed based on commonly collected forest-inventory data would aid in the conservation of important bird species. Wood ducks (Aix sponsa) are a common and wellstudied example, though until recently, population-management efforts have primarily focused on artificial nesting structures as opposed to influencing forest-management decisions. We measured and inspected 7,869 trees and 1,186 potential cavities to determine their suitability for use by nesting wood ducks in forests of north-central Minnesota during 2016-2018. Fifteen logistic regression models using treeand stand-level forest attributes were compared and tested for their utility in predicting whether trees had developed suitable cavities. Our top model was additive and included 3 tree-level predictors: diameter at breast height (DBH), health status, and species. We also found some support for including an interaction between DBH and health status, but it was not in our top model. The top model predicted whether trees had suitable cavities well, with an average area under the receiver-operating-characteristic curve of 0.85. For every 1cm increase in DBH, the odds that a given tree would have a suitable cavity increased by 7.3% (95% CI; 6.0-8.7%). Dead and declining trees were more likely to have suitable cavities than live-healthy trees, with 834% (483-1420%) and 477% (276-807%) higher odds, respectively. When comparing 7 common deciduous species with cavities, sugar maple (Acer saccharum) and American basswood (Tilia americana) were most likely to have developed cavities. These results can be applied to existing forest-inventory datasets to predict the availability of cavities in the landscape and to maximize conservation benefits for wood ducks and other large-bodied secondary cavity-nesting species.