Errors associated with two methods of assessing tree hollow occurrence and abundance in Eucalyptus obliqua forest, Tasmania

Tree hollows provide important habitat for fauna, but difficulties associated with detecting tree hollows can impede the formulation of appropriate management action. This paper examines the accuracy and errors associated with two methods commonly used to assess the presence/absence or abundance of tree hollows; 'ground-based' surveys and 'tree-felling' surveys. Three hundred and forty-six trees in Tasmania's State Forest were surveyed for hollows both before and after being felled. In order to assess the type and frequency of errors associated with each method, the fate of every potential hollow identified during the ground-based survey was determined after the tree had been felled. Three main types of error were identified: hollows misidentified during the ground-based survey; hollows not detected during ground-based surveys; and hollows not found during tree-felling surveys. Bayesian models were used to examine the association between site, tree and hollow variables and the occurrence of these errors. The likelihood that a tree would have a misidentified hollow increased with the number of potential hollows that were seen during the ground-based survey. Smaller hollows were more likely to be misidentified during ground-based surveys than larger hollows, particularly in trees that have little dead wood. For both of the survey methods the rate at which hollows were not detected was found to increase with tree size and hollow abundance. Hollows located in the branches were more likely to remain undetected during both survey methods, while small hollows, hollows high off the ground and those facing upwards were more likely to be missed during ground-based surveys. Although ground-based surveys provide inaccurate hollow counts, they are useful for assessing relative rather than true hollow abundance, which is valuable in the selection of habitat trees for retention in production forests. Tree-felling surveys provide a more accurate measure of actual hollow counts, particularly when a correction is made for the proportion of tree that cannot be surveyed. Tree-felling surveys are therefore useful for doing research such as developing models predicting hollow abundance.