Search

Abstract: This paper details predictive models of tree-level hollow incidence for State forests in central and eastern Victoria. Models are based on the hollows component of the Statewide Forest Resource Inventory (SFRI), the most comprehensive database of hollow incidence in Australia. A two-stage methodology was used at the individual tree-level that in its first-stage generated a statement of the probability of hollows presence, and in its second-stage estimated the size of hollows. The hierarchical nature of SFRI data prompted the search for statistical methodology capable of explicitly modelling a complicated error structure. Subsequently, generalised linear mixed models (GLMMs) were used to estimate first-stage, tree-level models in the presence of spatial and nested dependence. The developed models were statistically and biologically plausible and performed well when validated using independent data. By overcoming the limitations facing previous research associated with insufficient data and inappropriate statistical methodology viable predictive models were developed. Tree-level models will help us understand patterns of hollow incidence, can be used for predictive purposes when individual-tree information is available, and can improve habitat tree retention guidelines. [Copyright &y& Elsevier]

Hollow-bearing trees are an important ecological resource for many forest-dwelling species throughout the world. Assessments of the abundance and characteristics of tree hollows usually involve ground-based surveys. These are likely to be biased because of the distance over which the assessments are made. We demonstrate the use of double-sampling theory to correct this bias in order to provide the most efficient estimate of hollow abundance. A sample of 40 Eucalyptus leucoxylon (yellow gum) trees located at Yarra Bend Park, Melbourne, were climbed and assessed for hollows through close-up inspection. The results were compared to ground-based surveys conducted by individuals with varying levels of experience in assessing tree hollows. Strong correlations (r=0.57−0.83) were found between climbing and ground surveys and amongst ground surveys. Those ground surveyors who took more time with their surveys generally identified a higher proportion of hollows (Spearman’s r2=0.72), although all ground surveys underestimated hollow frequency. The proportion of hollows identified by each observer (range 0.09–0.44) remained approximately constant across all tree diameter size classes, indicating systematic underestimation of hollow frequency. The relative efficiency of ground surveys in comparison to climbing surveys is dependent on the time taken for each survey method and the strength of the relationship between them. Although climbing surveys provided the standard for hollow detection, in all cases, the quicker ground surveys were found to increase the efficiency of hollow assessments. Therefore, it was determined that efficient surveys of hollow occurrence could be undertaken by inexperienced surveyors with periodic climbing surveys to measure and correct for bias. [Copyright &y& Elsevier]