Monitoring rangeland biodiversity – plants as...

24
Monitoring rangeland biodiversity – plants as indicators JILL LANDSBERG 1,2 & GABRIEL CROWLEY 2 1 Tropical Savannas CRC, School of Tropical Biology, James Cook University, PO Box 6811, Cairns, QLD 4870, Australia (Email: [email protected]), and 2 Queensland Parks and Wildlife Service and School of Tropical Biology, James Cook University, PO Box 6811, Cairns, QLD 4870, Australia. Abstract Because plants are relatively sensitive to many of the pressures acting on rangeland biota and also relatively amenable to measurement, metrics based on plants have considerable potential to be highly efficient indicators of the status of rangeland biodiversity. To achieve this potential, plant indicators need to be embedded in an appropriate monitoring and interpretive framework. We recommend a hierarchical approach that incorporates low intensity, broad-scale monitoring to provide context, and highly intensive, local-scale monitoring to provide understanding of process. A recent report commissioned by the National Land & Water Resources Audit recommends a core set of 11 indicators, seven of which rely on measurements of plants. These are: (i) progress to a comprehensive, adequate and representative reserve system; and trends in (ii) the extent of clearing, (iii) the cover of native perennial ground layer vegetation, (iv) the distribution and abundance of exotic plant species, (v) the distribution and abundance of fire-sensitive, and (vi) grazing-sensitive species, and (vii) the distribution and abundance of listed threatened entitities. Assessments of the current state of knowledge about these indicators for two case-study regions (the Gascoyne- Murchison Strategy area and Cape York Peninsula) showed that it would be possible to directly monitor most of them at regional scales, but current monitoring programs fall short of achieving this. We recommend two valuable additions to the core set of indicators: (a) explicit indicators of grazing pressure and fire regimes and (b) explicit inclusion of minimally disturbed reference areas to provide a standard against which to assess the biotic integrity of areas under pressure. Introduction Context: building on the Audit report A great deal has already been said and published about potential indicators for monitoring rangeland biodiversity. Fortunately, much of it has been encapsulated in a recent major report for the National Land and Water Resources Audit by the Tropical Savannas Cooperative Research Centre (Whitehead et al. 2001). The purpose of the report was to develop an analytical framework for monitoring aimed at State, Territory, and Federal Governments. The framework (Whitehead 2001a) was accompanied by a draft manual for monitoring (Woinarski 2001a) and substantive literature reviews of: changes in status of rangeland biodiversity and threatening processes (Woinarski 2001b); pastoral monitoring programs and their current and potential contribution to biodiversity monitoring (Fisher 2001); existing biodiversity monitoring programs (Whitehead 2001b), and relevant international experience (Beggs 2001). Central to the framework is a set of eleven core indicators of the status of rangeland biodiversity and the processes that threaten its persistence (Whitehead 2001; Table 1). 1 Our aim in the current paper is not to revisit ground already covered in this comprehensive report. Instead we plan to focus on the plant-related indicators it proposes, with the intention of identifying what we see as some of the key issues that have hindered their adoption for monitoring of rangeland biodiversity, and suggesting potential avenues for resolving these issues. In doing this, we are

Transcript of Monitoring rangeland biodiversity – plants as...

Monitoring rangeland biodiversity – plants as indicators JILL LANDSBERG1,2 & GABRIEL CROWLEY2

1Tropical Savannas CRC, School of Tropical Biology, James Cook University, PO Box 6811, Cairns, QLD 4870, Australia (Email: [email protected]), and 2Queensland Parks and Wildlife Service and School of Tropical Biology, James Cook University, PO Box 6811, Cairns, QLD 4870, Australia.

Abstract Because plants are relatively sensitive to many of the pressures acting on rangeland biota and also relatively amenable to measurement, metrics based on plants have considerable potential to be highly efficient indicators of the status of rangeland biodiversity. To achieve this potential, plant indicators need to be embedded in an appropriate monitoring and interpretive framework. We recommend a hierarchical approach that incorporates low intensity, broad-scale monitoring to provide context, and highly intensive, local-scale monitoring to provide understanding of process. A recent report commissioned by the National Land & Water Resources Audit recommends a core set of 11 indicators, seven of which rely on measurements of plants. These are: (i) progress to a comprehensive, adequate and representative reserve system; and trends in (ii) the extent of clearing, (iii) the cover of native perennial ground layer vegetation, (iv) the distribution and abundance of exotic plant species, (v) the distribution and abundance of fire-sensitive, and (vi) grazing-sensitive species, and (vii) the distribution and abundance of listed threatened entitities. Assessments of the current state of knowledge about these indicators for two case-study regions (the Gascoyne-Murchison Strategy area and Cape York Peninsula) showed that it would be possible to directly monitor most of them at regional scales, but current monitoring programs fall short of achieving this. We recommend two valuable additions to the core set of indicators: (a) explicit indicators of grazing pressure and fire regimes and (b) explicit inclusion of minimally disturbed reference areas to provide a standard against which to assess the biotic integrity of areas under pressure.

Introduction

Context: building on the Audit report A great deal has already been said and published about potential indicators for monitoring rangeland biodiversity. Fortunately, much of it has been encapsulated in a recent major report for the National Land and Water Resources Audit by the Tropical Savannas Cooperative Research Centre (Whitehead et al. 2001). The purpose of the report was to develop an analytical framework for monitoring aimed at State, Territory, and Federal Governments. The framework (Whitehead 2001a) was accompanied by a draft manual for monitoring (Woinarski 2001a) and substantive literature reviews of: changes in status of rangeland biodiversity and threatening processes (Woinarski 2001b); pastoral monitoring programs and their current and potential contribution to biodiversity monitoring (Fisher 2001); existing biodiversity monitoring programs (Whitehead 2001b), and relevant international experience (Beggs 2001). Central to the framework is a set of eleven core indicators of the status of rangeland biodiversity and the processes that threaten its persistence (Whitehead 2001; Table 1).

1

Our aim in the current paper is not to revisit ground already covered in this comprehensive report. Instead we plan to focus on the plant-related indicators it proposes, with the intention of identifying what we see as some of the key issues that have hindered their adoption for monitoring of rangeland biodiversity, and suggesting potential avenues for resolving these issues. In doing this, we are

mindful of the State, Territory and national focus of the framework for which the indicators were developed, and the problems already identified by practitioners of pastoral monitoring programs (Fisher 2001). We are also mindful of the benefits of using specific examples from the two case-study regions used in the Audit report, Cape York Peninsula (Crowley & Fisher 2001) and the Gascoyne-Murchison strategy area (Hopkins & Watson 2001). Both regions share many of the features that typify Australian rangelands (relative remoteness, low population density, and extensive land use) while also providing instructive contrasts in climate, vegetation, land use, pastoral management and processes threatening biodiversity (Table 2).

Before looking at the individual indicators, however, we draw attention to what we see as important concepts about plants as biodiversity indicators and approaches to their measurement.

Important concepts

Plants as biodiversity indicators Biodiversity indicators can be regarded as sets of environmental, pressure and biotic attributes whose measurement signals the effects on biodiversity of pressures caused by human activities. They should be embedded in a well-develop interpretive framework and have meaning beyond the individual attributes being measured (Saunders et al. 1998). Plant-related attributes are important descriptors of each of these dimensions.

Collectively, plants reflect physical environmental variables such as climate, topography, and soils. Their response to the physical environment flows on to influence biophysical variables such as pastoral productivity, biotic richness and flammability. Indicators that focus on plants collectively, i.e., on vegetation, are most amenable to expression as environmental metrics. Productivity, as monitored by greenness indices derived from remotely sensed data, is an example of one such metric. So too is the use of vegetation mapping as the primary tool for differentiating regional ecosystems (Wilson et al. 2002).

Plants are also good subjects for pressure metrics, because they are the primary target for most of the processes threatening rangeland biodiversity: grazing, clearing, weed invasions, inappropriate fire regimes and global change all act most directly on the composition and structure of plant communities, and only secondarily on other elements of biodiversity. Pressure metrics can be derived for plants collectively (e.g. extent of clearance, trends in woody or ground layer vegetation, wildfire scars) and individually (e.g. trends in populations of grazing- or fire-sensitive plant species).

Biotic metrics rely on counts of species or individuals. Because plants are sensitive to many of the processes threatening rangeland biota, and are also relatively amenable to monitoring, metrics based on the identity and relative abundance of different plant species – i.e. floristic composition - have considerable potential as surrogates for less amenable elements of biodiversity and/or for providing early warning of potentially catastrophic change. For example, Pringle (2001) showed that, in the northeastern goldfields region of Western Australia, grazing impacts on floristic composition were considerably more widespread than impacts on landscape pattern and process, which were concentrated near watering points where livestock congregated. (It is noteworthy that changes in ant species were also concentrated near watering points, possibly because they were responding primarily to changes in landscape function.) Yet many current monitoring programs differentiate only between species perceived to be functionally important, such as pasture plants and woody vegetation, and neglect that substantial proportion of the rangeland flora that is not perennial (Fisher 2001; Whitehead 2001b). Why?

2

ST Garnett & GM Crowley
I find that these terms/concepts do not stick in my mind, and I was not clear when you were introducing a new measure. I think italicising them helps.
ST Garnett & GM Crowley
Not sure I understand this. Can you say what it is used for/how?

The most comprehensive monitoring programs in Australian rangelands have a strong pastoral focus (Fisher 2001; Whitehead 2001b). Pastoral monitoring programs concentrate on vegetation cover and perennial plant abundance. Short-lived plants are seldom included, except perhaps as a single functional group labelled “forbs” or “ephemerals”. This is because (a) they are less important than perennials for tracking long-term change in pasture condition, (b) their inclusion would be time-consuming because there are so many species involved, (c) their identification requires specialist expertise, and (d) their populations are spatially and temporally variable. These reasons are undeniably valid in the context of pastoral productivity, but they are far less valid when the specific context is biodiversity.

Some of the more “difficult” plants - those that are short-lived and not well known - may be particularly sensitive indicators of grazing impacts on rangeland biota. For example, when Landsberg and colleagues (Landsberg et al. 2002) compared plant diversity patterns on pastoral lands with those on comparable undeveloped lands in northwest South Australia, they found two contrasting patterns among species that declined under grazing. The first group consisted of species known to be pastoral “decreasers”. These were species that declined where grazing impacts were concentrated near watering points. They tended to be palatable, drought-hardy perennial species. Though their abundance was clearly affected by grazing, healthy populations of these pastoral decreasers persisted in areas of paddocks where grazing was less intense. The second group of plant responders was unexpected: it consisted of a suite of species that were always rare in paddocks, and were only ever abundant outside the pastoral lands, where livestock had never grazed. Many of the species in this more sensitive group were short-lived and little known. Some were found only outside the pastoral lands. Hence these short-lived, often-neglected, plant species may be among the most sensitive indicators of grazing impact on biodiversity more generally.

Plants are not sensitive to all threats, however. Several significant threats to rangeland biota, most notably introduced predators, act directly on fauna and only indirectly, if at all, on rangeland plants. Plant-related indicators have limited utility for monitoring impacts of such threats, except as they relate to environmental metrics that influence animal habitats. Thus, although a set of plant-related indicators should be an important component of any biodiversity monitoring framework for the rangelands, it will never be sufficient for all purposes.

In summary, environmental metrics such as those describing tree and ground cover provide valuable information about the environmental envelope inside which biotic interactions are played out. Pressure metrics such as those describing the impacts of grazing, clearing, weed invasions and fire regimes on vegetation structure and composition, provide valuable information about the direct impacts of these pressures on plant diversity, and their indirect impacts on the fauna for which vegetation provides habitat and food. Biotic metrics describing the abundance and distribution of individual, threat-sensitive plant species may be particularly useful for predicting potentially damaging impacts before they have become widespread among less-sensitive biotic elements.

Need for a hierarchical approach These different kinds of metric illustrate the tension between a requirement for less-intensive but broader-scale monitoring to provide context, and highly-intensive but local-scale monitoring to provide details of processes. We follow Noss (1990) in suggesting that the various ecological scales at which biodiversity could be described constitute a nested hierarchy, beginning with coarse inventory of broad-scale indicators, then overlaying data on distribution of pressure indicators to identify biologically significant areas; then formulating specific questions about what might be occurring in these areas; and lastly, designing and implementing specific studies to address these questions.

3

ST Garnett & GM Crowley
Or some other comment that indicates you would have expected them there.
ST Garnett & GM Crowley
can you be more specific here. Are we talking about direct measures of stress (eg physiological) or pressure indicators.
ST Garnett & GM Crowley
For me it was too confusing to grapple with yet another set of terminology. I have tried to rewrite this section keeping its sense but using “following Noss” to allow us to continue with the original terminology. I may have got this wrong however. So read the lot first then see what you think of it.

The plant-related indicators identified in the Audit report (Table 1) span this hierarchy. Broad-scale indicators most amenable to coarse-scale inventory are indicators 2 and 4 (clearing and ground cover). The other indicators are all fine-scale, high-resolution indicators that provide insight into what is occurring in relatively small areas. The core set does not include indicators of all pressures, however, encompassing only measures of clearance and exotic plants, but not of changed fire regimes or grazing. We suggest that the core set would benefit from expansion to include explicit means of monitoring these additional pressures. Like clearing, fire regimes can be monitored directly (see detailed discussion of the specific indicators) so there is no need for surrogate indicators of this pressure. But grazing cannot be monitored directly, though it is an even more ubiquitous pressure on rangeland biodiversity than fire. Thus we suggest that indicator(s) of the spatial distribution of grazing, at the very least at broad spatial scales sufficient for identifying areas of high and low stress, should be a high priority in an expanded list of core indicators. Again following Noss (1990), assessments of concentrations of irreplaceable biodiversity (e.g. areas with high number of endemic species) could be coupled with high risk of impoverishment due to anthropogenic pressures to identify areas at greatest risk.

The value of biodiversity reference areas The pluralistic nature of biodiversity adds another dimension of difficulty to interpreting data about indicators. There is increasing recognition that biodiversity cannot be adequately described by simple algebraic indices (e.g. richness, evenness and diversity) that treat all species as equivalent (Cousins 1991; Landsberg et al. 1999b). There is less agreement, however, about alternatives. Angermeier and Karr (1994) argued convincingly that biological integrity - the degree of intactness of a system relative to a benchmark state - would provide a more effective primary management goal than diversity. By definition, naturally evolved assemblages possess integrity, but random assemblages do not. Adding exotic species from distant populations may increase local diversity, but it reduces integrity. Furthermore, many changes in diversity can be evaluated objectively only on the basis of changes in integrity. For example, continuous grazing of a system evolved under intermittent or light grazing may increase local diversity, but eliminate unique entities. While such a change may be interpreted as either a gain or loss in diversity, there is no ambiguity about integrity, which is clearly reduced.

Biological integrity can only be quantified relative to a reference site, which is a special form of control site representing naturally evolved, dynamic reference conditions against which modified states can be compared. Hence reference sites need to be minimally influenced by all putative threatening processes, not just those under study; they also need to represent the characteristic landscapes and habitats of all the areas to be compared with them, not just those that are amenable to statistical replication; and they need to be large enough to support the larger-scale processes that sustain biodiversity, despite the additional environmental variation this inevitably brings. A network of biodiversity reference sites across a region could have immediate illustrative value for land managers, by showing “what it can be like” for representative country types (Pringle & Landsberg ms). They could also provide regional standards for computing and comparing national change statistics, such as proportional decline in threat-sensitive species. However, apart from the aquatic examples cited by Angermeier and Karr (1994), we are aware of few attempts to develop operational metrics based on the concept of biological integrity, though a study by Majer and Beeston (1996) of ant communities in Western Australia is a noteworthy exception. Identifying a reference site not only presupposes that the site is in prime condition, but that management of the site is assured of maintaining that condition. Ideally sites should be located on conservation reserves or in areas covered by conservation agreements or other such devices.

4

Results

Potential for using the indicators in the case-study regions There are three aspects to using the indicators: identification, implementation and interpretation. Firstly, we must identify which elements to measure, and how they respond to different pressures. For example, eucalypt woodlands of monsoonal Australia have been considered highly stable and largely insensitive to changing fire regimes, whereas grasslands are rapidly contracting worldwide (Bowman 1988; Crowley and Garnett 1998). Monitoring of widespread eucalypt woodlands may be less informative about the impact of changed fire regimes than monitoring of the less extensive grasslands. Identification of informative taxa and vegetation characteristics has been patchy through the rangelands. In the discussion that follows we describe current knowledge of potential indicators and how this should be supplemented.

Once appropriate indicators are identified, a system of collecting and collating data about them must be implemented. Schemes of information collection are in place for some indicators, but are again patchy. Most rangeland areas suffer from logistical barriers to regular monitoring. However, some creative solutions are available. We explore the adequacy of the current programs, and make suggestions for new initiatives.

Interpretation of the collected data is fraught with difficulties. There are standard considerations, such as separating trends from noise, but additional problems through much of the rangelands, with its temporally variable environments impeding the assessment of absences and changes in abundance. Inclusion of minimally-disturbed reference areas would go some way toward mitigating such assessment difficulties, by providing a means by which current state could be standardised relative to a reference condition subject to the same temporal variation. However, scavenging data collected for other purposes means that reference sites may not be available, and there may be inconsistencies in methodologies, such as the different thresholds for tree cover used in monitoring vegetation clearance in different states. Hence apart from stressing the value of reference areas we make few recommendations regarding interpretation, merely flag it as an essential aspect of ensuring indicators are properly applied.

The discussion below focuses on using the indicators in the Gascoyne-Murchison Strategy (GMS) area and the Cape York Peninsula (CYP) bioregion.

Potential for locating reference areas We suspect that one of the main hurdles to wider adoption of the concept of measuring biological integrity is the perceived difficulty of selecting reference sites. What then are the prospects for locating biodiversity reference areas in the two case-study regions?

They appear to be reasonable in the GMS area, at least for the Murchison-catchment where reference areas were identified for all but one major vegetation type (Curry et al. 1994). The Murchison report also indicates a moderate number of areas greater than 5 km from water within particular vegetation types, where grazing impacts are expected to be relatively light (but see Landsberg et al. 2002 for limitations of this approach). However, reference areas are not currently included in the Western Australian Rangeland Monitoring System (WARMS). Instead, most sites are deliberately located in the zone of greatest anticipated grazing-induced change, usually between 1.5 and 3 km from livestock watering points. In the GMS area only 4% of the 686 WARMS sites are more than 5 km from water (Watson & Thomas 2002). This reflects the primary focus of WARMS on pastoral condition. Watson and Thomas (2002) have identified the following gaps in need of consideration were WARMS to be expanded for more general environmental monitoring:

5

ST Garnett & GM Crowley
From here there is a lot of shorthand. I will not edit language too much for this section as it looks as if you are still chewing over it.

over representation of pastorally productive areas and under representation of areas considered pastorally robust or of low pastoral value; restriction to land held under pastoral tenure; and exclusion of vegetation types of potential importance for regional biodiversity (e.g. ephemeral wetlands, riparian areas); low spatial coverage (Watson & Thomas 2002). We suggest lack of reference sites against which to standardise biodiversity metrics is an additional gap reducing the utility of WARMS sites for monitoring biodiversity.

There are even more difficulties associated with identifying reference areas in CYP, because minimally disturbed reference areas are likely to be hard to find. The most pervasive threats - inappropriate fire regimes, weed incursions, feral cattle and pigs - are largely independent of tenure boundaries. Furthermore, most National Parks have a history of use as grazing leases prior to their acquisition. Although complete eradication of feral animals and weeds from all conservation areas may not be a realistic management aim, eradication in some areas of especially high conservation value is being attempted (e.g. Box 1). If studies such as these show that uncontrolled grazing and other pressures have major impacts on threatened biota, they may provide impetus for implementing other targeted eradication programs. Should such programs be undertaken, there could be major benefits for regional biodiversity monitoring if they were concentrated in potential reference areas.

Indicator 1. Progress to a Comprehensive, Adequate and Representative reserve system

A reserve system is most appropriate for conserving biodiversity where alternative land uses would otherwise threaten its persistence. CAR (Comprehensive, Adequate and Representative) assessment examines whether the lands set aside in reserve systems are properly selected and managed to maximise biodiversity conservation. Acquisition programs are currently underway to supplement the reserve systems in both the GMS area and CYP bioregion (Crowley & Fisher 2001; Hopkins & Watson 2001). For example, the Department of Conservation and Land Management has purchased eight grazing leases and parts of nine other properties since December 1998 and land acquisitions are continuing (Hopkins & Watson 2001). However, we focus on CAR assessment on CYP as an example of the application of this indicator.

Comprehensiveness is based on mappable units, and assesses whether all units are included in reserves. Vegetation mapping units are usually selected, as these are more widely available, accurate and consistent than maps of animal distributions, and vegetation units are to some extent considered surrogates for animal habitats. Vegetation maps for Cape York Peninsula have been produced at 1:250 000 (Neldner 1999), and form the basis of comprehensiveness analysis currently being undertaken for the CYP bioregion (S.T. Garnett pers. comm. 2002). A concurrent analysis using micro-catchments, largely based on geology and climate (Nix et al. 2001) is also being undertaken, with the intention of comparing the results of both analyses, to assess the comprehensiveness of the current reserve system on Cape York Peninsula, and to assist in decisions about potential acquisitions to it.

Adequacy of the reserve system to maintain ecological viability and integrity of populations, species and communities is most complicated where appropriate management regimes are poorly understood. Optimum fire regime has been a matter of conjecture on Cape York Peninsula, and there is a paucity of information about damage caused by feral pigs and cattle, though both are known to be widespread and locally abundant on some reserves. Plant-related indicators that are useful to assess adequacy are the extent to which endangered and of concern regional ecosystems are represented, and the ongoing maintenance of grasslands within the reserve systems.

Representativeness measures whether areas in reserves should reasonably reflect the biotic diversity of the communities. This measure recognises the shortcomings of simple mappable units to represent habitats of many species. On CYP, representativeness of the more widespread plant 6

ST Garnett & GM Crowley
Not sure why you have chose not to include CAR as this is almost entirely a plant community based measure, and is underway on CYP.

species can be gleaned from vegetation mapping units. Herbarium records (HERBRECS) are probably also adequate for assessing widespread species. However for species that are less common, restricted in distribution or important as habitat indicators, information gained from these resources should be supplemented by targeted searches (Table 3; see also discussion on indicator 10).

Indicator 2. Trends in the extent of clearing of native vegetation Clearing of native woody vegetation is an issue of national concern and the focus of a collaborative, nationally coordinated, State-run monitoring program (Barson et al. 2000). However, the main focus is on clearing of forests and woodlands, and the canopy-cover protocols used to define these effectively exclude the vast areas of rangeland classified as open woodland, shrublands or grasslands.

Much of CYP bioregion is likely to be an exception. This is firstly because many parts of the bioregion have moderately high woody cover, and partly because the Queensland Statewide Landcover and Tree Study (SLATS) can detect changes in woody vegetation with 8-9% tree cover. Prior to SLATS the perception of clearing on CYP was that, although little had occurred, it none-the-less posed a threat to three of the region’s smallest but most fertile regional ecosystems (Sattler & Williams (1999). Summaries of more recent data collected by the Environmental Protection Agency (Wilson et al. 2002) confirm both assessments: little clearing has occurred (99% of the bioregion has not been cleared) but the limited clearing that has occurred has been concentrated in fertile regional ecosystems of limited extent.

Existing Statewide landcover data cannot be used to monitor clearing rates in the GMS area, because the State’s current landcover protocols do not provide data at an appropriate canopy-cover resolution. If it were applied, it would likely show very little clearing has occurred, and hence that there has been little landscape-scale loss of biodiversity relative to the pre-clearing benchmark condition. However, the possibility remains that any clearing that has occurred (e.g. associated with irrigated horticulture) may have targeted ecosystem types of limited extent; hence there could be loss of distinctive biodiversity as appears to have occurred in CPA.

Remote sensing using Land Cover Change Analysis (LCCA) has been undertaken in a number of Western Australian rangeland regions, including two Landsat scenes around Leonora and Carnarvon in the GMS area (Watson et al. 2001). While outputs from the technique could be used to report on trends in clearing and cover, no systematic reporting has yet been undertaken in the GMS area.

Indicator 4. Cover of native perennial ground layer vegetation Although remote sensing is routinely used to assess cover of ground layer vegetation in some rangeland jurisdictions (particularly South Australia and the Northern Territory), this is not the case for either of the case-study regions. The two main operational approaches are the grazing gradient approach (Pickup et al. 1998), which has most applicability to arid regions with widely spaced watering points, and Land Cover Change Analysis (LCCA) (Karfs 1999), which has proved more applicable for savanna regions.

An early version of the grazing-gradient approach was tested in the Murchison and Carnarvon regions of Western Australia (Cridland & Stafford Smith 1993) but no consistent grazing-induced change could be discerned. The authors attributed this to the diversity of land systems and the number of waterpoints per paddock. Some experimental development of the LCCA technique has

7

ST Garnett & GM Crowley
Are you sure SLATS is available for CYP. Last I heard \(a few years ago I admit\) I was told it wasn’t\)

also been undertaken in the GMS area (for the two Landsat scenes near Leonora and Carnarvon); the results were promising but there has been no follow-up (Watson et al. 2001).

Neither approach has been tested on CYP. The LCCA technique could be applicable, given it has been adopted for routine monitoring in parts of the tropical savannas of the Northern Territory (Karfs 1999). To date, however, it has been most successful for detecting trends in relatively uniform grassland environments, and less successful for trend detection in more variable land system types (Karfs pers. comm.), such as those that dominate across much of Cape York Peninsula. The extent of burning of Cape York Peninsula (> 80% of the total area burnt during 2002; P Thomson pers. comm.) could further complicate the potential for using remote-sensed assessment to detect relatively subtle, grazing-related shifts in ground cover in this region.

We are not aware of any routinely collected ground-based data that could be used to map regional trends in ground cover in either case-study region. However, relative cover change has been estimated for the GMS area, using the WARMS monitoring sites. Although ground cover was not assessed directly, an index of shrub cover was estimated from field measurements of canopy width (Watson & Thomas 2002). Shrubland sites predominate in the GMS area (650 of 686 sites); hence shrub cover was seen as reasonable surrogate for the perennial ground layer vegetation. The estimated shrub cover was greater than actual because it was based on maximum canopy width and assumed 100% foliage cover. Hence the cover estimates were used to assess relative change between two monitoring dates rather than to make inferences about absolute cover on any particular date. Watson and Thomas (2002) used these estimates to assess relative cover change for 223 WARMS sites that had been assessed twice: once between 1993-7 and a second time between 1999-2001. Estimated canopy area increased during the interval on 96% of sites, with an average increase in canopy area per site of 81%. The increase was consistent across different vegetation groups and was attributed to unusually high rainfall during the assessment period. The lack of reference sites made it impossible to determine whether grazing management was also influential.

Just as the paucity of reference areas included in WARMS means there is little ecological justification for using the data to make inferences about the impact of grazing on biodiversity, the relatively low density of sample plots means there is no statistical justification for using the data for regional mapping. Though the WARMS sampling effort is moderately intensive, with around 460 plots in the combined Gascoyne & Murchison bioregions, it equates to plot densities of only 1 plot per 1000 km2 (both estimates from Table 5 in Fisher 2001), a density much too low to provide sufficient statistical rigour for making inferences about spatial coverage (Webster & Oliver 1990). The data are useful for compiling regional narratives, however (sensu Pringle & Landsberg ms) and could also be appropriate for inclusion in a standardised national database developed to allow cross-regional comparisons of selected subsets of data. Given that regional reference sites may exist, their inclusion in future monitoring could facilitate the standardisation of trends in cover change at individual sites against regional reference sites.

Despite the shortcomings of WARMS plots for biodiversity monitoring, they far exceed any routine monitoring currently being undertaken on Cape York Peninsula. The existing State pasture monitoring programs, QGraze or Traps, do not include any plots on CYP (Fisher 2001). Queensland Parks and Wildlife Service has recently established a small number of plots where all dominant ground cover plants are assessed (c. 25 in 2001: Crowley & Fisher 2001), but their establishment is very recent and repeat measurements have not yet been undertaken. There are also too few of them to provide a basis for regional mapping. Crowley and Fisher (2001) suggest that reassessment of about 700 of the inventory sites used for mapping of the region’s vegetation could provide a statistically adequate sample for mapping, but no repeat measurements have been undertaken to date.

8

One specific study of east-central CYP indicated that management-induced cover change may be extensive. This study compared 64 sites surveyed in 1964 and 1993, and identified species (including Themeda triandra, Sorghum plumosum, Heteropogon spp.) and communities (grasslands) most susceptible to changing land management (Crowley & Garnett 1998). Any future monitoring of perennial ground layer vegetation should therefore include these species. It should also include reference sites to allow management effects to be distinguished from natural seasonal variation.

Indicator 5. Exotic plant species We are not aware of any currently collected data that could be used to quantify regional status or trend of exotic plant species in either region, though environmental weeds appear to pose serious risks to biodiversity in Cape York Peninsula. A list of 10 significant weeds with potential to threaten environments in the region was developed as part of CYPLUS, and includes a number of weeds of national significance (Table 3). Crowley and Fisher (2001) found there was adequate monitoring of only one of these 10 species. Given that most of them are spread by grazing animals, in their feed and/or vehicles, or by water, and that much of the region is grazed by livestock or feral animals, and/or seasonally inundated, there appears a high potential for these weeds to spread. One of the 10 strategies funded under Cape York Peninsula NHT Plan addresses feral animals and weeds. As yet, however, there has been no systematic regional-wide assessment of the distribution and abundance of exotic plants, though a report due in several months will provide a compilation of information gathered during extensive control programs (Peter James, DNRM, Cooktown, pers. comm.).

In the Gascoyne-Murchison region exotic plant species are assessed at the plot scale in the rangeland monitoring program, but not with sufficient spatial coverage for mapping purposes (Hopkins & Watson 2001). Few exotic species were recorded in the condition survey of the Murchison River catchment (Currey et al. 1994), so they may pose less of a threat to biodiversity than environmental weeds currently pose for Cape York Peninsula.

Indicator 6. Fire-sensitive plant species and communities Though the distribution of fire-sensitive species and communities has not been explicitly investigated in either case study region, monitoring data on fire regimes are available, in the form regular satellite-based assessment of fire history, extent and timing (Hopkins & Watson 2001; Cape York Peninsula Development Association nd). Fires are uncommon throughout much of the Gascoyne-Murchison, except in the limited spinifex grasslands, and we have no information about the distribution or environmental correlates of fire-sensitive vegetation types.

Fires are widespread and frequent throughout Cape York Peninsula, where fire is recognised as an important influence on biodiversity (Cape York Peninsula Development Association nd; Crowley & Fisher 2001). Interpretation of what constitutes fire sensitivity is complicated, however, because much of the region is naturally fire-prone; hence plant species and communities respond to fire regimes that are not adequately described in terms of simple presence or absence of wildfire. Fire-sensitive vegetation classes are being developed by Queensland Parks and Wildlife Service as part of their ongoing fire management planning. Based on regional ecosystem units, such classification is made on a property by property basis. Detailed ground-truthing of mapped fire-sensitive units and observations of their responses to contemporary fire regimes are required to improve this system. Lists of fire-sensitive taxa are yet to be compiled.

It should be possible, however, to make inferences about likely impacts of current fire regimes on particular savanna communities, using a combination of existing mapping (e.g. Fox et al. 2001), 9

documented responses of major savanna vegetation types to fire regimes (Russell-Smith et al. 2001), and satellite-based monitoring of fire regimes (Allan et al. 2001). Two contrasting types of fire-sensitive plant communities are currently of greatest concern (Russell-Smith et al. 2001), and should therefore be a priority for such analysis.

(a) The first is those restricted vegetation types, including rainforest and riverine gallery forest, and heathlands of rocky rangelands, which contain many obligate seeder species that can be eliminated by too frequent fires.

(b) The second is those more widespread vegetation types, such as the mesic savannas and seasonally inundated wetlands, where woody thickening and invasion occur if fires are too infrequent, or not sufficiently intense.

Indicator 7. Grazing-sensitive plants This indicator remains one of the most problematic for reporting at regional scales, because there are inherent constraints to making cross-regional comparisons from data on individual indicator species. There are two reasons. One is that the distribution of many rangeland plant species is localised (restricted extent) or sparse (widespread, but with low local abundance). For example, Landsberg and colleagues found that 72% of the ground layer species identified along eight different grazing gradients were found at only one gradient, and none at all were found on all eight. Even within individual gradients, 45% of species occurred on one or two of the six sites that constituted a gradient, and 46% were locally uncommon at every site where they were found (Landsberg et al. ms.). Hence, very few rangeland plant species are likely to be sufficiently widespread or common to have potential as indicators in spatially or environmentally disparate locations. The second inherent problem is that many plant species show differential responses to grazing depending on the nature of the grazing pressure, and the plant assemblage in which they occur. This occurs because apparency and palatability to grazing animals are both context-sensitive, and so too is the ability of plant species to tolerate grazing (Landsberg et al. 1999b). For example, in the grazing gradient study referred to above, two relatively widespread and common grasses, Aristida contorta and Eragrostis dielsii showed mixed responses to grazing: on mulga gradients where more palatable grass species were common, both tended to increase in response to grazing, while on chenopod gradients where no grass species were common, both species declined with grazing (Landsberg et al. ms.).

Attempts to identify indicator response types, rather than species, for monitoring grazing-induced change have met with limited success. For a mulga grassland community, species with ‘large, erect tussocks branching above ground’ showed potential as response types indicative of light grazing and ‘small, sprawling basal tussocks’ showed potential as indicators of heavy grazing; however, no comparable trait combinations were identified for a gidgee shrubland community (Landsberg et al. 1999a). One of the most widespread grass species showing this decreaser syndrome is also one of the most consistent in its response: Themeda triandra was once widely distributed in semi-arid, temperate and tropical rangelands, and generally decreases in response to moderate to heavy grazing pressure in most environments where it occurs. Localised responses are known for many other pasture species in particular environments, and these are widely promoted as indicators of pasture condition in those environments (e.g. Mitchell & Wilcox 1994; Partridge 1999). The problems with using these species as indicators of biodiversity condition are two-fold. Firstly, because they are useful pasture species they are almost invariable relatively insensitive to grazing: by the time known pasture indicators are in decline many other less well-known species may be lost from the local species pool (Landsberg et al. 2002). Secondly, their local abundance depends as much on site potential as the cumulative effects of grazing: hence without benchmark sites the

10

information provided by monitoring their abundance is likely to be difficult to interpret and insensitive to all but gross trends.

What this means is that detailed monitoring against local reference areas is required to (a) identify the local pool of grazing-sensitive species; and (b) to separate grazing-related trends in their abundance from natural variation. Woinarski (2001a) suggests this should be done for one or several nominated species or species-groups for each environment of interest. However, given (a), we suggest that it would be more efficient and effective to monitor as wide a range of candidate species as possible, to maximise the prospects of identifying the most grazing-sensitive responders.

It may be possible, with sufficiently intensive ground-based monitoring of both treatment plots and adequate controls or reference sites, to derive regionally-comparable metrics, e.g. proportion of the regional species pool in decline. However, the intensity of sampling that would be needed to do this for each of the main vegetation types in a region is likely to limit the number of regions and/or vegetation types in which it could be conducted. Development of regional narratives linked to locality-specific information may be a more realistic reporting aim for monitoring this indicator. Current monitoring plots would need to be supplemented with adequate reference sites to achieve even this rather narrow objective. Linking locality-specific monitoring like this with aerial-based monitoring of grazing pressure (see Pringle & Landsberg ms) may prove more informative and cost-effective than attempting to achieve spurious cross-regional comparability in trends shown by individual species.

Watson and Thomas (2002) used data from WARMS plots in the GMS area to explore the utility of pastorally-defined plant response groups. They used published sources and local knowledge to assign the 71 perennial species monitored on WARMS plots to four categories based on postulated responses to grazing: Decreasers (33% of species), Intermediates (44% of species), Increasers (13% of species) and unassigned species (10%). They found little difference between categories in population growth rate, canopy growth, frequency of occurrence, and attribute this to the overall improvement in the condition of most sites over the assessment period (which was one of unusually high rainfall). The only substantial difference within response groups was for canopy size, where the overall increase was less for Decreaser species than for the others. The authors suggest that, although this may reflect a differential response due to grazing sensitivity, it was not sufficient to influence other population parameters. However they also note that the allocation of species to grazing response groups was difficult and is currently under review.

Indicator 10. Listed threatened plant species and communities Commonwealth regulation of environmental issues has been greatly expanded with the introduction of the Environment Protection and Biodiversity Conservation Act 1999 (EPBC Act), which places special emphasis on listed threatened entities (species and communities) as matters of national environmental significance (Environment Australia, nd-a). Hence their inclusion as biodiversity indicators is important within both regulatory and planning contexts. For species, however, the value of the lists as indicators is limited by bias in the types and locations of the species included in them, and the bureaucratic nature of the listing process, which ensures that the lists are very slow to change (Possingham et al. in press). Furthermore, changes to the list are usually due to changes in knowledge (resulting from inventory surveys) rather than change in actual status, particularly for plant species. And lastly, bulked indices of numbers of threatened entities per site or region are ambiguous as indicators of biodiversity status. For example, increases in numbers of listed species in an area could indicate an improvement in their status (identification of populations not previously reported from the area) or deterioration (additions of local species to the threatened lists).

11

At the national level, nominations to add or subtract species or communities to threatened lists are assessed by the Threatened Species Scientific Committee, which uses current IUCN criteria for guidance in determining conservation status and provides detailed justifications for all determinations (IUCN 2001; Environment Australia, nd-b). However, the vast majority of currently listed entities were assessed before the EPBC Act commenced and did not require documentary evidence of the rationale for their listing. For example, around 90% of the plant species currently listed as endangered were transferred from pre-EPBC Act lists (Environment Australia, nd-c). Recent work on the listed plants of Cape York Peninsula (Landsberg & Clarkson unpublished) shows many sources of potential list bias and error. These include:

• discrepancies between State and Commonwealth lists, apparently due to poor information flow

• over-representation of some taxa, particularly orchids, palms and cycads (Table 5)

• selective listing of some species on the basis of restricted distributions, without evidence of threatening processes, while many comparably restricted species are not listed

• listing of species thought to be uncommon on the basis of herbarium records, but shown to be common and/or widespread when searched for

While the first three problems may eventually be resolved if old lists are reviewed and new assessment procedures come to dominate the lists, the last problem is symptomatic of a broader issue relating to the types of entities most likely to be listed. Most plant species tend to be assessed against distributional, rather than population, criteria. For example the two IUCN criteria most commonly used for assessing plant species are criterion B1(ii) (small distribution coupled with inferred continuing decline) and criterion D1 (extremely restricted distribution) (Gardenfors 2001). Hence the plant species most likely to be listed are those that are localised in their distribution. Sparsely distributed species (those that are widespread but everywhere have low local abundance) are unlikely to be listed because detection of their rarity is only apparent from population data, and these are seldom available. Yet many rangeland plants are sparsely distributed (Landsberg et al. ms, and discussion under indicator 7). Although such plants may be highly susceptible to widespread threats such as grazing and inappropriate fire regimes, this is unlikely to be reflected in changes in the status of any listed species.

Listing of threatened ecological communities is a much newer, hence less tested, endeavour. As far as we are aware, Australia is the only country to include threatened ecological communities in its formal listing process. As the category achieved conservation prominence only after the inception of the EPBC Act (Environment Australia nd-a) there are still relatively few ecological communities formally listed as nationally threatened. Criteria for listing communities may be based on declining distribution or declining condition (Environment Australia nd-d). For those communities listed on the basis of declining distribution (usually in the form of clearing) list status essentially duplicates information provided under indicator 2. For many rangeland communities, listing on the basis of declining condition has potential to be far more informative. Several of the communities listed since the inception of the EPBC Act fall into this category; hence monitoring their status and trend may provide a useful indication of the status of the biota they include. Examples of such communities include (Environment Australia nd-e):

• Semi-evergreen vine thickets of the Brigalow Belt;

• The community of native species dependent on natural discharge of groundwater from the Great Artesian Basin; and

12

• Bluegrass (Dichanthium spp.) dominant grasslands of the Brigalow Belt Bioregions

There are no nationally listed ecological communities in either case-study region, however, nor in most other rangeland bioregions.

Conclusions Assessments of the current state of knowledge of the plant-related indicators proposed in the Audit report (Table 1) show that it would be possible to directly monitor many of them at regional scales, and use the results to assess national status and trend in plant diversity. However the only monitoring program that currently provides that level of information is the statewide landcover monitoring in Queensland, from which relevant data on tree clearing could be extracted for regional ecosystems in Queensland savannas. The Western Australian Rangeland Monitoring System provides an extensive network of ground-based plots that can be used for making inferences about some of the indicators. However significant gaps in the system limit the utility of the plots for monitoring biodiversity. Lists of threatened plant species and communities are maintained at national and State scales, and could also be incorporated into biodiversity monitoring programs without collecting additional data. Unfortunately their utility for making inferences about biodiversity status and trend is limited by the nature of the listing process.

Current monitoring programs do not provide sufficient information about most of the plot-based indicators to allow statistically meaningful extrapolation for regional reporting of their condition. Impediments are partly to do with resourcing (implementation of new programs would be required in most cases, and/or considerable supplementation of existing plot-based programs); partly to do with technological limitations (remote-sensed assessment of ground cover may require further technical development before it could be applied across diverse savannas, or intensively developed arid regions); and partly to do with knowledge limitations, particularly about how to link plot-based, species-level information to regional trends in biodiversity status.

Two additions to the core-set of indicators would go some way toward assisting implementation of many of the plot-based indicators.

(a) Indicators of the main pressures acting on rangeland biota should be explicitly included in the core set, to provide a spatial and environmental context for selecting where to locate ground-based monitoring sites. This means including indicators of grazing pressure and fire regimes that can be monitored at sufficiently broad-scales to provide a context for identifying meaningful locations for case-studies, and other forms of ground-based monitoring.

(b) For ground-based indicators, there should be an explicit requirement that trends are assessed relative to reference sites, particularly for indicators about which the natural biological integrity is uncertain. Reference conditions are already implied for indicators of tree clearing and weed invasion, because the natural condition is assumed to be the null state (no clearing or weeds). Reference sites are critical for interpreting trends due to grazing and fire regimes, because so little is known of the state in which biodiversity existed before changes in these pressures were imposed.

References Allan G., Dyer R., Johnson A., Russell-Smith & Yates C. (2001) Monitoring fire regimes. In:

Savanna Burning - understanding and using fire in northern Australia. (Dyer et al. eds) pp. 102-121. Tropical Savannas CRC, Darwin.

13

Angermeier P.L. & Karr J. R. (1994) Biological integrity versus biological diversity as policy directives. BioScience 44, 690-696.

Barson M., Randall L. & Bordas V. (2000) Land Cover Change in Australia. Results of the collaborative Bureau of Rural Sciences - State agencies’ Project on Remote Sensing of Land Cover Change. Bureau of Rural Sciences, Canberra.

Beggs K. (2001) Approaches to broad scale monitoring of biological diversity: a brief review of international experience. In: Developing an analytical framework for monitoring biodiversity in Australia’s rangelands. A report by the Tropical Savannas CRC for the National Land and Water Resources Audit (P. Whitehead et al. eds) pp. 407-428. Tropical Savannas CRC, Darwin.

Bowman D.M.J.S. 1988. Stability amid turmoil? Towards an ecology of north Australian eucalypt forests. Proc. Ecol. Soc. Aust. 15, 149-158.

Cape York Peninsula Development Association (nd) Sustainable Fire Management Project Cape York. Retrieved 10th December 2002 from [http://www.cypda.com.au/projects/fire//fire.htm]

Cousins S. H. (1991) Species diversity measurement: choosing the right index. Trends in Ecology & Evolution 6,

Cridland S. & Stafford Smith M. (1993) Development and dissemination of design methods for rangeland paddocks with maximise animal production and minimise land degradation. Misc. Publ. 42/93, Department of Agriculture, Western Australia.

Crowley G. M. & Garnett S. T. (1998) Vegetation change in the grasslands and grassy woodlands of east-central Cape York Peninsula, Australia. Pac. Cons. Biol. 4, 132-48.

Crowley G. M. & Garnett S. T. (2001). Growth, seed production and effect of defoliation in an early flowering perennial grass, Alloteropsis semialata (Poaceae), on Cape York Peninsula, Australia. Aust. J. Bot. 49, 735-43.

Crowley G.M. & Fisher A. (2001) Biodiversity monitoring in Cape York Peninsula. In: Developing an analytical framework for monitoring biodiversity in Australia’s rangelands. A report by the Tropical Savannas CRC for the National Land and Water Resources Audit (P. Whitehead et al. eds) pp. 429-450. Tropical Savannas CRC, Darwin.

Currey P.J., Payne A.L., Leighton K.A., Hennig P. & Blood D.A. (1994) An inventory and condition survey of the Murchison River catchment and surrounds, Western Australia. Tech. Bull. 84. Department of Agriculture, Perth, Western Australia.

Environment Australia (nd-a) EPBC Act- Biodiversity Conservation Information. Retrieved 10th December 2002 from [http://www.ea.gov.au/epbc/biodiversityconservation/].

Environment Australia (nd-b) Threatened Species & Ecological Communities Nominations. Retrieved 10th December 2002 from [http://www.ea.gov.au/biodiversity/threatened/nominations/index.html].

14

Environment Australia (nd-c) Flora that are ENDANGERED. Retrieved 10th December 2002 from [http://www.ea.gov.au/cgi-bin/sprat/public/publicthreatenedlist.pl?wanted=flora#ENDANGERED FLORA]

Environment Australia (nd-d) Ecological communities nomination form. Retrieved 10th December 2002 from [http://www.ea.gov.au/biodiversity/threatened/nominations/index.html]

Environment Australia (nd-e) Listings since commencement of EPBC Act to December 2002. Retrieved 10th December 2002 from http://www.ea.gov.au/biodiversity/threatened/species/listings-todate.html]

Fisher A. (2001) A review of pastoral monitoring programs and their real and potential contribution to biodiversity monitoring. In: Developing an analytical framework for monitoring biodiversity in Australia’s rangelands. A report by the Tropical Savannas CRC for the National Land and Water Resources Audit (P. Whitehead et al. eds) pp. 311-366. Tropical Savannas CRC, Darwin.

Fox I.D., Neldner V.J., Wilson G.W. & Bannink P.J. (2001) The Vegetation of the Australian Tropical Savannas. Environmental Protection Agency, Brisbane.

Gardenfors, U. (2001) Classifying threatened species at national versus global levels. Trends in Ecology & Evolution 16, 511-515

Hopkins A. & Watson I. (2001) Biodiversity monitoring in the Gascoigne-Murchison strategy area. In: Developing an analytical framework for monitoring biodiversity in Australia’s rangelands. A report by the Tropical Savannas CRC for the National Land and Water Resources Audit (P. Whitehead et al. eds) pp. 451-462. Tropical Savannas CRC, Darwin.

IUCN (2001) IUCN Red List Categories: Version 3.1 Prepared by the IUCN Species survival Commission. IUCN, Gland, Switzerland & Cambridge, UK

Landsberg J., Lavorel S. & Stol J. (1999a) Grazing response groups among understorey plants in arid rangelands. J. Veg. Sci. 10, 693-96.

Landsberg J., O’Connor T. & Freudenberger D. (1999b) The impacts of livestock grazing on biodiversity in natural ecosystems. In: Nutritional Ecology of Herbivores. (H-J Jung & G.C. Fahey eds) pp. 752-77. American Society of Animal Science, Savoy, IL.

Landsberg J., James C.D., Maconochie J., Nicholls A.O., Stol J. & Tynan R. (2002) Scale-related effects of grazing on native plant communities in an arid rangeland region of South Australia. J. Appl. Ecol. 39, 427-44.

Landsberg J., James C.D., Morton S.R., Muller W.J. & Stol J. (ms) Trends in abundance and composition of plant species along grazing gradients in Australian rangelands. In review.

Majer J. D. & Beeston G. (1996) The biodiversity integrity index: an illustration using ants in Western Australia. Cons. Biol. 10, 65-73

Mitchell A. A. & Wilcox D. G. (1994) Arid Shrubland Plants of Western Australia 2nd Edition. University of Western Australia Press, Perth.

Neldner V. J. (1999) Cape York Peninsula In: The Conservation Status of Queensland's Bioregional Ecosystems. (P. Sattler and R. D. Williams eds) pp. 3/1-3/85. Environmental Protection Agency, Brisbane.

Noss R.F. (1990) Indicators for monitoring biodiversity: a hierarchical approach. Cons. Biol. 4, 355-64.

15

Pickup, G., Bastin, G.N. & Chewings, V.H. (1998) Identifying trends in land degradation in non-equilibrium rangelands. Journal of Applied Ecology 35, 365-377.

Possingham H. P., Andelman S. J., Burgman M. A., Medellin R., Master L. L. & Keith D. A. (in press) Limits to the use of threatened species lists. Trends in Ecology & Evolution

Pringle H.J.R & Landsberg J. (ms) Watering points and domestic livestock grazing: patterns and processes affecting rangeland biodiversity. Submitted to Aust. Ecol.

Pringle H.J.R. (2001) Grazing impacts in rangelands: assessment of two contrasting landscape types in arid Western Australia from different land management perspectives. PhD thesis, Australian Nation University.

Russell-Smith J., Start A. & Woinarski J. (2001) Effects of fire in the landscape. In: Savanna Burning - understanding and using fire in northern Australia. (Dyer et al. eds) pp. 29-49. Tropical Savannas CRC, Darwin.

Sattler P. & Williams R. (eds) (1999) The Conservation Status of Queensland’s Bioregional Ecosystems. Environmental Protection Agency, Brisbane.

Saunders D., Margules C. & Hill B. (1998) Environmental Indicators for National State of the Environment Reporting - Biodiversity, Australia: State of the Environment (Environmental Indicator Reports), Department of the Environment, Canberra.

Smith N.M. (2002) Weeds of the Wet/Dry Tropics of Australia. A field guide. Environment Centre NT Inc., Darwin.

Trollope W. S. W. & Potgeiter A. L. F. (1986). Estimating grass fuel loads with a disc pasture meter in the Kruger National Park. J. Grassl. Soc. S. Africa 4, 148-152.

Watson I. & Thomas P. (2002) The Western Australian Rangeland Monitoring System (WARMS) in the Gascoyne-Murchison Strategy (GMS) area, including an assessment of recent changes to shrub populations. Unpublished report prepared for the Regional Environmental Management Program of the Gascoyne-Murchison Strategy.

Watson I., Blood D., Novelly P., Thomas P. & van Vreeswyk S. (2001) Rangeland monitoring, resource inventory, condition assessment and lease inspection activities in Western Australia conducted by the Department of Agriculture. Unpublished report prepared for the Rangeland Theme of the National Land and Water Resources Audit.

Webster R. & Oliver M.A. (1990) Statistical Methods in Soil and Land Resource Survey. Oxford Uni. Press, Oxford.

Whitehead P. (2001a) The framework. In: Developing an analytical framework for monitoring biodiversity in Australia’s rangelands. A report by the Tropical Savannas CRC for the National Land and Water Resources Audit (P. Whitehead et al. eds) pp. 1-70. Tropical Savannas CRC, Darwin.

16

Whitehead P. (2001b) A review of information gathered from existing biodiversity monitoring programs. In: Developing an analytical framework for monitoring biodiversity in Australia’s rangelands. A report by the Tropical Savannas CRC for the National Land and Water Resources Audit (P. Whitehead et al. eds) pp. 367-406. Tropical Savannas CRC, Darwin.

Whitehead P., Woinarski J., Fisher A., Fensham R. & Briggs K. (eds) (2001) Developing an analytical framework for monitoring biodiversity in Australia’s rangelands. A report by the Tropical Savannas CRC for the National Land and Water Resources Audit. Tropical Savannas CRC, Darwin.

Wilson B.A., Neldner V.J. & Accad A. (2002) The extent and status of remnant vegetation in Queensland and its implication for statewide vegetation management and legislation. Rangel. J. 24, 6-35.

Woinarski J. (2001a) A manual for biodiversity monitoring. In: Developing an analytical framework for monitoring biodiversity in Australia’s rangelands. A report by the Tropical Savannas CRC for the National Land and Water Resources Audit (P. Whitehead et al. eds) pp. 463-485. Tropical Savannas CRC, Darwin.

Woinarski J. (2001b) A review of changes in status and threatening processes. In: Developing an analytical framework for monitoring biodiversity in Australia’s rangelands. A report by the Tropical Savannas CRC for the National Land and Water Resources Audit (P. Whitehead et al. eds) pp. 71-310. Tropical Savannas CRC, Darwin.

17

Table 1. Core indicators proposed by Whitehead et al. (2001) for monitoring rangeland biodiversity. The indicators are all interlinked, but those to which plant measures most directly contribute are shown in bold font. Indicator 1. Progress to a Comprehensive, Adequate and Representative reserve system. Indicator 2. Trends in the extent of clearing of native vegetation. Indicator 3. Landscape functionality. Indicator 4. Trends in the cover of native perennial grass/ native perennial ground layer vegetation. Indicator 5. Trends in the distribution and abundance of exotic plant species. Indicator 6. Trends in the distribution, abundance, and condition of fire-sensitive plant species and

communities. Indicator 7. Trends in the distribution and abundance of grazing-sensitive plants. Indicator 8. Trends in the distribution and abundance of susceptible mammals. Indicator 9. Trends in the distribution and abundance of susceptible birds. Indicator 10. Trends in the distribution and abundance of listed threatened species and the distribution and

condition of listed threatened communities. Indicator 11. Trends in the intensity of land use.

18

Table 2. Characteristics of the case-study regions referred to for examples of implementation issues (Curry et al. 1994; CYPLUS reports; Sattler & Williams 1999; Crowley & Fisher 2001; Woinarski 2001b; Hopkins & Watson 2001). Regional characteristics Cape York Peninsula bioregion Gascoyne-Murchison Strategy area Location far north Queensland west-central Western Australia Area 137,200 km2 c. 340,000 km2 Information base Cape York Peninsula Land Use Study

(CYPLUS) natural resource description; Vegetation mapping at 1:250 000

Regional range surveys, land system mapping; and biogeographic survey; c. 460 rangeland monitoring plots; Vegetation mapping at 1:250 000

Planning context Commonwealth & Qld government funded Cape York NHT Plan

Commonwealth & WA government funded Gascoyne-Murchison Strategy

Climate Monsoonal: 60% <1000mm pa Reliable intense wet season, predictable

but extended dry season

Arid & semi-arid: average c. 200 mm pa Winter dominant but unreliable, summer rains

highly variable Vegetation Highly diverse flora (>3000 spp)

Eucalypt woodlands 64% Melaleuca woodlands 14% Rainforest 6% (rich in endemics) Heathlands 3% Grasslands 6%

Highly diverse flora (>2000 spp) Mostly shrublands or low woodlands naturally

depauperate in grasses. Mulga (Acacia aneura) and cotton bush

(Ptilotus obovatus) the most ubiquitous perennials

Population >18,000 people, 60% indigenous c. 70% live in towns c. 4% (700) people live on pastoral

properties

Land tenure Pastoral lease 57% Aboriginal lands 15% National Park 10% Mining & forestry 4% Freehold c. 4%

Pastoral lease Freehold National Park Mining lease

Economy Mining 52% GRP, 12% of workforce Public sector 20% GRP, 46% workforce Pastoralism, cropping & fishing 5% GRP,

6% workforce

Mining Pastoralism Tourism

Pastoral industry Based on beef production, increasingly for live export through Weipa;

Predominant management system is year-round grazing on mainly native pastures of low carrying capacity (4-55 ha/beast);

Many properties unfenced, with a strong reliance on natural, often seasonal, surface water;

Use of fire & supplements important management tools

Based on wool production, but subject to declining terms of trade;

Predominant management system is year-round grazing on native pastures of low-variable carrying capacity (7-30 ha/dse);

Properties fenced and provided with bore water, but deteriorating infrastructure is leading to ineffective enclosure;

Fires uncommon

Constraints to development

Isolation, uncertainty over leases, infertile soils, poor infrastructure, restricted wet season access, divergent aspirations

Declining terms of trade, seasonal failure, low natural carrying capacity, irreversible degradation of some eroded land types

Main threats to biodiversity

Inappropriate fire regimes leading to woody thickening,

Pressure to intensify agricultural systems, Grazing by domestic cattle, and feral

cattle, pigs & horses (distribution influenced by seasonal variation in water supply & habitat)

Rooting & trampling by pigs Serious environmental weeds (sicklepod,

rubber vine, exotic grasses)

Sheep grazing (distribution influenced by fences and water location)

Grazing by feral goats and kangaroos (distribution not influenced by fences and less influence by water location)

Grazing by rabbits (independent of water location)

Predation by feral foxes & cats

19

Table 3. Assessment of representativeness of the reserve system on Cape York Peninsula for exemplary plant species. Group Examples Priority Appropriate measures V eg et at

ion

m H er ba riu D et ai

led

Widespread, dominant species Eucalyptus tetrodonta Low Yes Yes - Widespread, non-dominant species

Cochlospermum gillivrayi Low Yes Yes -

More restricted, common species Eucalyptus setosa Medium Yes Yes Yes Habitat indicators Planichloa nervilemma Medium Yes Yes Yes Rare and threatened species Jedda multicaulis High - Yes Yes

20

Table 4. Significant environmental weeds present on rangelands in Cape York Peninsula (Crowley & Fisher 2001), and the threats they pose (Smith 2002). (WON = declared Weed of National Significance.) Common name Scientific name Distribution and threat Rubber vine Cryptostegia grandiflora WON; common along many river systems in central and southern

CYP; smothers vegetation replacing native species, particularly in heavily grazed areas fringing river systems

Sickle pod Senna obtusifolia Common in central and southern CYP in disturbed areas, often in dense swards along flood plains

Grader grass Themeda quadrivalvis Prominent along roadsides, heavily grazed and disturbed areas in much of CYP; can invade native grassland and seriously reduce diversity

Lion’s tail Leonotis nepetifolia Restricted locations in central CYP; has ability to develop large colonies that displace native species, particularly along riverbanks and floodplains

Hymenachne Hymenachne amplexicaulis WON; present in central and southern CYP; has ability spread rapidly across flood plains and form dense stands, which smother native vegetation.

Hyptis, Horehound Hyptis suaveolens Widespread and common throughout CYP, particularly in grazed and disturbed areas; forms dense thickets particularly on flood plain margins

Gamba grass Andropogon gayanus Present in Bamaga, Weipa, Cooktown and Laura districts; can invade native grasslands and alter fire regimes causing serious damage to native woody species and other biota

Vetiver grass Vetiveria Planted to stabilise edges of road cuttings. Declared in Cook Shire. Limited potential for spread as reproductively sterile.

Salvinia Salvinia Present in restricted wetland areas in western CYP; potential to spread in seasonal floodwaters

Water Hyacinth Eichhornia Present in restricted wetland areas in southern and western CYP; potential to spread in seasonal floodwaters

21

Table 5. Representation of families and life forms of listed plant species on Cape York Peninsula, compared with the total flora (Landsberg unpublished) Listed species All species recorded in bioregion Total number 69 3, 338 Orchids (Orchidaceae) 30% 5% Palms (Arecaceae) 6% <1% Grasses (Poaceae) 7% 9% Cycads (Cycadaceae) 6% <1% Other trees 10% - Other shrubs 17% - Other vines, creepers & twiners 12% - Other epiphytes 1% - Other herbs 4% - Ferns 6% -

22

Box 1. Example of strategic control of livestock and feral animals to determine conservation benefit in Cape York Peninsula Monitoring the impact of pigs and cattle on cockatoo grass at Artemis Station, Cape York Peninsula. Authors: Gabriel Crowley, Stephen Garnett and Susan Shephard, , Queensland Parks and Wildlife Service, Cairns. Background Ecological studies of the endangered golden-shouldered parrot Psephotus chrysopterygius on Artemis station commenced in 1992. In these studies, cockatoo grass Alloteropsis semialata was identified as an important food plant for the parrots, which feed on its seeds in the early wet season when other food sources are scarce. It is also palatable to cattle and appears to be selectively eaten by pigs, which dig up the plants to chew on their bases. Mild levels of defoliation comparable to light cattle grazing were found to reduce seeding for up to two years (Crowley & Garnett 2001). The effects of pigs on persistence of Cockatoo Grass and seed availability are unknown, as is whether there is likely to be a detrimental effect on feeding and nesting of the golden-shouldered parrot. The current project aims to determine the extent of impacts and, hence, whether additional controls on pig numbers or cattle movements are required. Information on ground layer vegetation, fuel load, fire history and grazing regime will also be collected. Methodology Plot design and layout Eight groups of three sites were established on Artemis station in December 2001. Each group of sites was within a single vegetation community and had similar ground layer composition and canopy composition and structure. Each site in a group was allocated to one of three treatments (unfenced, fenced to exclude cows and fenced to exclude cows and pigs). Fenced sites were approximately 30 x 20 m. At each site, a 25 x 4 m plot was permanently marked with star pickets, and divided into 20 1.25 x 4 m segments to allow stratified sampling of ground layer vegetation.

Vegetation community Regional ecosystem no.

Groups of sites

Lemon-scented ti-tree woodland (on hill slopes) (3.3.48) 1 Broad-leaved ti-tree low open woodlands in seepage areas (on hill sides) (3.3.49) 1 Broad-leaved ti-tree low open woodlands in on flats in sand ridge country 3.3.50 4 Messmate woodland on low lying sand ridges 3.5.7 2 Total 8

Timetable All measurements are to be taken annually. In the early dry season (usually May), ground layer composition will be recorded. In the late dry season (August to October) fuel load will be monitored. Cockatoo grass will be measured within one month of the first wet season rain (December/January). Site history Records will be kept of the incidence of fire and the approximate stocking densities maintained between each monitoring period. Presence of disturbance signs, including animal droppings, defoliation and pig diggings will be recorded in each quadrat when ground cover is monitored. Daily rainfall record will be collected at Artemis homestead. Ground layer vegetation Each time ground layer vegetation is monitored, one 0.25 x 0.25 m quadrat is to be randomly located in each of the 20 plot segments at each site. In each quadrat, presence of all ground layer species (or genera where individual species are not determined) will be recorded, also noting which three taxa have the greatest above ground biomass in each quadrat. Changes in ground layer vegetation, and effects of treatments will be tracked using multidimensional scaling. Fuel loads Fuel loads will be measured using the disc and stick method (Trollope & Potgeiter 1986). Within each site, but outside the 25m x 4 m plot, 50 fuel depth measures will be made of 100% cured fuel. Calibration with fuel loads will be undertaken in the same vegetation types close to each group of sites. Changes in fuel loads and effects of treatments will be assessed using repeated measures analysis. Cockatoo grass Ten plants of cockatoo grass have been randomly selected for monitoring in each plot, and permanently encircled using a loop of wire forced into the ground. As an index of biomass (Crowley & Garnett 2001), vegetative plant height and number of tillers will be recorded for each plant. Changes in biomass index will be assessed using repeated measures analysis. Outcomes At the end of three years (2004), preliminary analyses will be undertaken to assess the relative impacts of pigs and cattle. If deleterious impacts are evident, recommendations will be made about management of pigs and cattle on the property, and the relevance of these recommendation to tropical savannas in general will be assessed. Monitoring may 23

then continue as an index of ground layer health. If no impacts are detected, monitoring will cease. If results are inconclusive, monitoring at this experimental level may continue until results are conclusive.

24