Introduction 20

Natural eucalypt regeneration has been shown to occur in relatively unmodified remnant woodland vegetation but to be highly restricted in modified remnants and grazed pastures in many dry agricultural districts of Australia. This includes the south eastern temperate grazing region (Nadolny 1995; Dorrough and Moxham 2005; Spooner and Briggs 2008; Fischer et al. 2009; Gibbons et al. 2010; Weinberg et al. 2011); the West Australian wheat belt (Yates et al. 1994a; Norton et al. 1995; Saunders et al. 2003); the Mt Lofty Ranges bioregion of South Australia (Ottewell et al. 2010); the ACT (Landsberg et al. 1990) and the Northern Tablelands of NSW (Curtis 1990; Landsberg et al. 1990). While a number of studies have assessed the condition of woodland and dry forest remnants in the dry agricultural district of Tasmania, the focus of these studies was not on eucalypt regeneration but rather on tree health (Davidson et al. 2007; Close et al. 2008); age, shape, disturbance and proximity of remnants (Gilfedder and Kirkpatrick 1998); rare and threatened species (Kirkpatrick and Gilfedder 1995); bryophyte diversity (Pharo et al. 2005); and bird species composition and richness (MacDonald and Kirkpatrick 2003). This current study focuses on the relationship between remnant vegetation condition and eucalypt regeneration.

There are many measures of ‘health’ or ‘condition’ of remnant vegetation that attempt to integrate the range of attributes (including regeneration) that contribute to biodiversity. These are generally qualitative and have their critics because of this. For example Gibbons and Freudenberger (2006) consider ‘condition’ to be a “value-

laden concept that requires data to be interpreted through a ‘value prism’ along a continuum from ‘good to ‘bad’.” Keith and Gorrod (2006) suggested that vegetation condition has three main facets with values based around aesthetics, production and biodiversity and that the context, meaning and scope of ‘condition’ needs to be explicitly articulated in each application. For the purposes of this thesis vegetation condition is being defined within the context of biodiversity values, that is, the “capacity of native vegetation to sustain local populations of native plants and animals” (Keith and Gorrod 2006), with a particular focus on eucalypts.

McElhinny et al. (2006) suggest that ‘condition’ (used in the context of biodiversity) should be measured quantitatively at a stand, or “site” scale. This is also the scale at which restoration efforts are most likely to be made (Yates and Hobbs 1997a). Assessments of condition made at individual sites need to be compared to each other and to the range of conditions across a study area, as condition is a relative rather than absolute concept (Gibbons and Freudenberger 2006). Therefore, an index which combines data from multiple attributes into a single score is a useful tool. Indices allow sites to be ranked according to their assessed condition and for their potential contribution to biodiversity (Parkes et al. 2003; McElhinny et al. 2005). They may also be useful as a monitoring tool over time, especially at sites that undergo

restoration treatments or a change in management regime (Hobbs and Norton 1996). The method used by government agencies to assess vegetation condition in Tasmania (Michaels 2006) has been adapted from the ‘Habitat Hectares’ method developed by Parkes et al. (2003) and involves assessing site-based and landscape components against a benchmark defined for a particular TASVEG vegetation community (as described and mapped in the TASVEG statewide mapping, Harris and Kitchener

2005). The vegetation condition and landscape context scores are added to produce a single condition score which gives an indication of the degree to which the site differs from the ‘natural’ or benchmark state (Parkes et al. 2003; Michaels 2006). The ‘Habitat Hectares’ assessment method is subjective, does not provide continuous quantitative data on the amount of regeneration present at a site and therefore makes it difficult to relate regeneration to other vegetation attributes and management regimes. A benchmark approach is also problematic as it does not take into account that a particular vegetation type may have a range of ‘natural’ stable states

(McCarthy et al. 2004).

During his PhD studies, Chris McElhinny from Australian National University developed a structural complexity index that was based on the quantitative

measurement of a range of attributes in dry eucalypt woodland and forests. Structural complexity is a measure that integrates the range of micro-environments (or

microhabitats) available to organisms. The greater the range of structural

components in an ecosystem, the greater the variety of resources (and microhabitats) and thus a greater number of plants and animals can utilize these resources

(Lindenmayer et al. 2000; McElhinny et al. 2005; Fischer et al. 2006).

The structural complexity index was not based on expert opinion or any idea of what a “natural” state is. It is mathematically and conceptually simple and is based on the range of condition in a region of interest (McElhinny 2005). A comprehensive review of the links between structural attributes, habitat provision and biodiversity (McElhinny et al. 2005) and the methodology of constructing the index (McElhinny et al. 2006) have been reported.

McElhinny (2005) initially measured a comprehensive suite of over 70 structural attributes at 48 sites across the South-eastern Highlands Bioregion in the ACT and NSW. The set was reduced to 13 core attributes through redundancy analysis of correlations between measured attributes. The aim was that each core attribute had low kurtosis (peakedness) in its distribution among sites, distinguished effectively between site types (woodland and dry sclerophyll forest), functioned as a surrogate for other co-related attributes and was efficient to measure in the field. Each core attribute contributed a potential 10 points, the scores for each attribute were added, and then the total expressed as a percentage of the mathematical maximum (130). A sensitivity analysis revealed that there was no need for weighting of attributes and thus the final index is a simple, additive, non weighted score that rates forest or woodland at each site relative to the range of observed attribute levels across all sites. Being able to assess each individual attribute relative to the range of reference values also provides a useful method of determining whether remedial action needs to be taken to improve that attribute at a particular site. For use in regions outside that in which it was developed, the index would need to be calibrated to a new set of reference sites in the region of interest. This would maintain the ability of the index to distinguish between sites (pers.comm C.McElhinny).

The McElhinny index of structural complexity was calibrated to Tasmanian

conditions and used in this study to assess the condition of dry forest and woodland remnants, with particular reference to eucalypt regeneration and the stand attributes associated with it. The research questions examined were:

1. What stand scale factors are associated with eucalypt regeneration in remnant vegetation?

2. What levels of condition and disturbance of remnants support or inhibit eucalypt regeneration?

3. How can the McElhinny et al. (2006) structural complexity index be calibrated to the range of Tasmanian dry forest and woodland conditions and how effective were decisions made in calibrating the index for such purpose?

4. Is the observed quantity of eucalypt regeneration sufficient or is remedial management needed to improve eucalypt regeneration in Tasmanian dry forests and woodlands?