Biodiversity publications archive

Landscape planning for biodiversity conservation in agricultural regions: A case study from the Wheatbelt of Western Australia

Biodiversity Technical Paper, No. 2
Robert J. Lambeck, CSIRO Division of Wildlife and Ecology
Commonwealth of Australia, 1999
ISBN 0 6422 1423 9

Chapter 2 - Retaining biodiversity in agricultural landscapes (continued)

2.5 Strategic enhancement: using focal species to define landscape requirements for nature conservation

Strategic enhancement approaches generally aim to conserve locally indigenous species in their natural habitat. In order to achieve this, it is necessary to determine the spatial, compositional and functional characteristics required to meet the needs of the biota that we wish to conserve.

If these characteristics are currently inadequate then a strategic approach will require the reconstruction of habitat and changes to management practices in order to provide the minimum requirements of the plants and animals that are at risk. This section presents a procedure for addressing the needs of the species which occur in any given landscape. This procedure is then applied to the Wallatin Creek Catchment. While the reintroduction of species was not considered for the study area, the issue of reintroductions is considered briefly in Section 2.6.

When assessing the capacity of a landscape to retain its biota we need to consider two aspects of that landscape which I term 'landscape adequacy' and 'landscape viability'. An adequate landscape is one which contains all of the necessary resources to meet the immediate needs of the species present in that landscape. These resources, however, may not be available in sufficient quantities or appropriate configurations to support enough individuals to ensure persistence in the face of natural or anthropogenic catastrophes, or under conditions of demographic, environmental, stochastic, and genetic variability (Shaffer 1987). A viable landscape, on the other hand, is one that not only contains the resources required to meet the immediate needs of the individuals present, but can also support sufficient numbers such that the populations of the species present are able to persist for some specified period of time.

The importance of this distinction became apparent when attempting to design a landscape to protect conservation values at a scale as small as the Wallatin sub-catchment (26000 ha). Management at the sub-catchment scale could potentially provide all of the resources required by the species that occurred there, but could not provide them in sufficient quantities to support viable populations. Even if the whole catchment was revegetated, it would be unlikely to support viable populations of naturally uncommon species. Under these circumstances, failure of the biota to persist in the catchment would not be due to the quality of the catchment, but due to the poor quality of adjoining catchments. The catchment would be 'adequate' but not 'viable'.

Clearly, if we are to maintain the biological diversity of our production landscapes, it will be necessary to firstly define the composition, quantities and configuration of landscape elements that must be present to meet the immediate needs of the flora and fauna (landscape adequacy) and secondly, to define the area over which this solution needs to be implemented to support viable populations of the species present (landscape viability). Achievement of the latter will require a regional approach to conservation planning. In the following sections, I present a procedure for defining landscape adequacy and discuss the implications of this approach for defining landscape viability. Appendix 1 considers the implications of applying the approach presented here at a regional scale.

2.5.1 Species or pattern and process as the basis for conservation planning?

There has been considerable debate in the ecological literature about whether the requirements of single species should serve as the basis for determining landscape adequacy, or whether the analysis of landscape pattern and process should underpin conservation planning (Franklin 1993; Hansen et al. 1993; Orians 1993; Franklin 1994; Hobbs 1994; Tracy & Brussard 1994). Species-based approaches have taken the form of either single-species studies, often targeted at rare or vulnerable species, or the study of groups of species which are considered to represent components of biodiversity (Soulé & Wilcox 1980; Simberloff 1988; Wilson & Peter 1988; Pimm & Gilpin 1989; Brussard 1991; Kohm 1991). Species-based approaches have been criticised on the grounds that they do not provide whole landscape solutions to conservation problems; that they cannot be conducted at a rate sufficient to deal with the urgency of the threats; and that they consume a disproportionate amount of conservation funding (Franklin 1993; Hobbs 1994; Walker 1995). Consequently, critics of single-species studies are calling for approaches that consider higher levels of organisation such as ecosystems and landscapes (Noss 1983; Noss & Harris 1986; Noss 1987; Gosselink et al. 1990; Dyer & Holland 1991; Salwasser 1991; Franklin 1993; Hobbs 1994). These alternative approaches are based on the recognition that species requirements are not independent of the landscape in which they occur and that landscape mosaics strongly influence the long-term viability of species within those landscapes (Janzen 1983; Newmark 1985; Saunders et al. 1991; Anglestam 1992; Hobbs 1993, 1994).

While approaches that consider pattern and processes at a landscape scale help to identify the elements that need to be present in a landscape, they are unable to define the appropriate quantity and distribution of those elements. Such approaches have tended, by and large, to be descriptive. While they can identify relationships between landscape patterns and community measures such as species diversity or species richness, they are unable to define the composition, configuration and quantity of landscape features required for a landscape to retain its biota. Nor are they able to specify which species will be retained and which will be lost. In other words, they may be useful for addressing a general enhancement objective, as discussed in the previous section, but they are inadequate for defining the characteristics necessary to ensure the persistence of all species in a nominated location.

Ultimately, questions such as what type of pattern is required in a landscape, or at what rate a given process should proceed, cannot be answered without reference to the needs of the species in that landscape. Given this, it is clear that we cannot ignore the requirements of species if we wish to define the characteristics of a landscape which will ensure their retention. The challenge then is to find an efficient means of meeting the needs of all species without studying each one individually. In order to overcome this dilemma, proponents of single-species studies have developed the concept of 'umbrella' species (Murphy & Wilcox 1986; Noss 1990; Cutler 1991; Ryti 1992; Hanley 1993; Launer & Murphy 1994; Williams & Gaston 1994). These are species whose requirements for persistence are believed to encapsulate those of an array of additional species.

The attractiveness of umbrella species to land managers is obvious. If it is indeed possible to manage a whole community or ecosystem by focusing on the needs of one or a few species, then the seemingly intractable problem of considering the needs of all species is resolved. Species as diverse as owls (Franklin 1994; Kavanagh 1991), desert tortoises (Tracy & Brussard 1994), black-tailed deer (Hanley 1993), gliding possums (Kavanagh 1991) and butterflies (Launer & Murphy 1994) have been proposed to serve an umbrella function for the ecosystems in which they occur. However, given that the majority of species within an ecosystem have widely differing habitat requirements, it seems unlikely that any single species could serve as an umbrella for all other species. As Franklin (1994) points out, landscapes designed and managed around the needs of single species may fail to capture other critical elements of the ecosystems in which they occur. It would therefore appear that if the concept of umbrella species is to be useful, it will be necessary to search for multi-species approaches which identify a set of species whose spatial, compositional and functional requirements encompass those of all other species in the region.

In the following sections, I present a new method for selecting, from the total pool of species in a landscape, a subset of 'focal species' (Lambeck 1997) whose requirements for persistence define the attributes that must be present in a landscape if it is to meet the needs of the remaining biota. The approach extends the umbrella species concept by clearly linking the species that are declining with the threats that are causing that decline. A suite of species that are considered most sensitive to each of the different threats are identified and the requirements of each of these species is used to define the landscape attributes and management regimes that are needed to ameliorate the perceived threats. Area-limited species are used to define the minimum acceptable size of the patch types they occupy, dispersal-limited species define configuration and connectivity, resource-limited species define habitat composition, and 'process-limited' species define the appropriate management regimes for threatening processes. The needs of these focal species can be used to develop explicit guidelines regarding the composition, quantity and configuration of patch types required in the landscape and the management regimes that must be applied to the resulting design. The procedure for defining these focal species firstly requires identification of focal groups, the members of which are vulnerable to the same threats. The most demanding or most sensitive species from within each group are then selected. These become the focal species whose requirements define the limits within which the perceived threats must be managed.

2.5.2 Identifying the focal groups

In order to select the focal species, it is necessary to firstly identify the processes responsible for declining population sizes. Species which are considered susceptible to similar threatening processes are grouped and then, for each threat, the species that requires the most comprehensive response is identified. In the fragmented agricultural landscapes of Western Australia the major threats have been identified as the loss and fragmentation of habitat; the loss of critical resources; habitat degradation due to stock grazing and weed invasion; and inappropriate rates and intensities of ecosystem processes such as fire, nutrient cycling and predation (Hobbs et al., 1993). Grazing pressure from rabbits or the presence of invasive diseases such as Phytophthora may also be threats in some areas. Figure 6 outlines the sequence of decisions made, firstly to identify groups of species whose vulnerability is attributable to common threats and subsequently to identify those species whose requirements for mitigating the threat encompass those of the other species. The outcome of this selection process is a suite of 'focal' species whose requirements for management or habitat reconstruction encapsulate the needs of all other species.

At risk or secure?

The first dichotomy in Figure 6 differentiates between those species considered secure in the current landscape and those expected to be lost in the absence of action. Species considered secure are removed from the selection process; if the status of a species is in doubt, it should remain in the analysis. Secure species may re-enter the analysis subsequently if their presence is identified as being the cause of vulnerability of some other species.

Figure 6: Figure 6. Decision tree for developing landscape designs and management guidelines based on the focal species.

Figure 6: Figure 6. Decision tree for developing landscape designs and management guidelines based on the focal species.

Note: The limiting processes identified in the figure are simply illustrative. These may vary from one landscape to another. The procedure for applying this approach is described in the text. (Sections 2.5.2 - 2.5.3)

Reconstruction or within-remnant management?

The second decision differentiates between species whose persistence in the landscape depends on some form of habitat reconstruction and species which would be able to persist in the current landscape provided biophysical processes were managed in a different way. This dichotomy reflects a distinction between the relative importance of pattern and process. Generally, species which could persist under the current landscape configuration if the landscape was managed differently are those sensitive to the rates of particular processes, or to changes in the intensity and frequency of those processes. In Australian agricultural landscapes these processes include altered fire regimes, predation by introduced foxes and cats, grazing of native vegetation by stock and other herbivores, and competition between native plants and exotic weeds. Species threatened by these changes can be considered 'process-limited'. The remaining species are those primarily constrained by landscape patterns which limit the availability of suitable habitat or critical resources. Such species will rely on landscape reconstruction to meet their needs.

Species requiring reconstruction

Species identified as requiring landscape reconstruction are further assessed to determine whether they are limited by (i) a shortage of critical resources, (ii) an inability to move between suitable habitat patches, or (iii) insufficient habitat to meet their resource needs.

i) Resource-limited species

For resource-limited species, the number of individuals that a region can support is determined by the carrying capacity at the time of lowest resource availability. Species limited by a resource bottleneck may exhibit a significant population response to the enhancement of resources at the time of greatest shortage. For example, many nectarivorous birds utilise a sequence of nectar sources throughout the year. Depletion of these resources at any stage in this sequence constrains their population size (Lambeck 1995). A rehabilitation response targeted at alleviating the bottleneck should increase the local carrying capacity for nectarivorous species. In such circumstances, a relatively local strategic restoration action may produce a greater population response than would a major landscape reconstruction if the latter failed to explicitly address the resource shortage. Species that may commonly be identified as being resource limited include birds or mammals that require tree hollows for nesting or roosting or species which depend on the sequential availability of seasonally variable food resources such as nectar or fruit.

ii) Dispersal-limited species

Dispersal-limited species are those for which there are suitable habitat patches which could potentially support small populations, but the patches are beyond the distance over which individuals can move or are separated by a matrix that is too hostile to permit movement. If individual populations are too small to be viable in their own right, the combination of stochastic and anthropogenic impacts can result in rates of local extinction that exceed rates of recolonisation. Such species will require increased connectivity between habitat patches either by the provision of corridors or by reducing the 'resistance' of the intervening matrix (Knaapen et al. 1992). This would require the adoption of land uses or management practices in the matrix which are less hostile to nature conservation. Species which are most likely to be dispersal limited will be sedentary habitat specialists which occur in low densities and have low mobility relative to the distances between suitable habitat patches.

iii) Area-limited species

Area-limited species are those for which the patches of appropriate habitat are simply too small to support a breeding pair or, in the case of colonial species, a functional social group. Area-limited species are, in reality, resource limited but are considered in this category if the limiting resource is not obvious or quantifiable. Habitat patches are therefore used as a surrogate for resources (Hansen et al. 1993) and it is assumed that there is a minimum patch size of a given quality that will provide sufficient resources to support a pair or group. In general, species which occur higher in the food chain are more likely to be area limited than will be species in lower trophic orders. Similarly, species which have a greater reliance on the maintenance of a cohesive social structure may have greater area requirements than do solitary species as any given habitat patch will have to support a greater number of social individuals.

Species requiring management of ecosystem processes

While the spatial and compositional characteristics of habitat may not be the primary limiting factor for many species, they may still be vulnerable because of inappropriate rates or intensities of important ecosystem processes. Throughout Australia, a number of threatening processes other than land clearance contribute to a reduction in the diversity of native ecosystems. These include predation by foxes and cats, grazing by stock and rabbits, invasion by weeds and disease, chemical and nutrient drifts from adjoining farmland, and inappropriate fire regimes. Species which are limited by processes such as these are grouped into categories which reflect the range of threats that occur in the region being managed.

2.5.3 Selecting the focal species

After completing the above decision-making process, all species considered at risk will be allocated to at least one of four major categories: area-limited, resource-limited, dispersal-limited, or process-limited. The process-limited group will be further subdivided according to the number of processes that require management. Some species may occur in more than one category. The species in each category are then ranked in order of their requirements for area, connectivity, resources or management, respectively.

Area requirements

Area-limited species are categorised on the basis of the dominant patch types that they utilise. For each patch type, species are ranked according to the size of the smallest patch in which they are observed to occur. The species with the largest minimum occupied area for a particular patch type is identified as the focal species for that patch type. Its spatial requirements are then used to define the minimum suitable size for that patch type. Any patch large enough to support a breeding pair, or social group, of the focal species is assumed to be large enough to support individuals of all other species that utilise that patch type. For any given region there will be as many focal species which define minimum patch area as there are patch types.

A difficulty which arises when attempting to determine adequate patch area is the definition of what constitutes a patch. It is widely recognised that patchiness in a landscape is hierarchical, with patches defined at one scale, being subdividable into smaller patches when examined at a finer scale. In agricultural landscapes, the dominant vegetation associations will generally provide a useful basis for defining patchiness. These associations tend to have dominant species and characteristic structural attributes which can be easily recognised by land managers. These associations also tend to correlate with soil types and other landform features with which many land-holders are familiar. Consequently they provide useful 'building blocks' for landscape design and reconstruction. In some circumstances vulnerable species may be responding to a finer landscape grain than that of the vegetation association. Attempts to reconstruct vegetation formations should therefore ensure that this finer scale variability is incorporated where appropriate.

Connectivity requirements

For the majority of species, dispersal is one of the least understood aspects of their ecology. The approach taken to define the characteristics necessary for dispersal would ideally follow that used in the previous section to define area requirements. Species would be ranked according to the minimum width, length and structural requirements of the connecting vegetation through which they are known to move. The species with the greatest need for wide corridors or with the least inclination to move along corridors would become the focal species for defining corridor width and length, respectively. Similarly, species with the most demanding structural requirements would be used to define the structural attributes of the connecting vegetation.

Because dispersal data are rarely available, presence/absence data can be used to determine the inter-patch distance beyond which seemingly suitable habitat is unoccupied. For example, Cale (unpublished data) found for a range of bird species that seemingly suitable patches remained unoccupied if they were too isolated. For each patch type in a landscape, the minimum acceptable distance between patches can be defined by the species with the shortest distance beyond which an otherwise suitable patch is not occupied.

Resource requirements

Resource-limited species are those for which critical resources can be identified and be shown to limit the carrying capacity of consumer species in the region. Where a number of species utilise the same resource base, the resource must be increased to a level sufficient to meet the needs of the least abundant consumer (Lambeck 1995). This species becomes the focal species for defining the appropriate level of that resource. Resources that are commonly limiting include nest hollows for birds and mammals, nectar supplies for birds, insects and some small mammals such as the Honey Possum (Tarsipes rostratus), or appropriate microhabitats for invertebrates such as Mygalomorph spiders (Main 1987).

Management requirements

Having categorised species according to their needs for management of threatening processes they are then ranked in terms of their vulnerability to those threats. Those species most vulnerable to, or most dependent upon a given process become the focal species for defining the intensity, rate or frequency at which that process should be managed. For example, the species most deleteriously affected by weed invasion will define the level of weed control required and the species most vulnerable to feral predators will define the appropriate level of predator control.

2.5.4 Applying the focal species approach to Wallatin Creek Catchment

The application of the focal species approach would ideally be based on a comprehensive survey of the area to be studied and a knowledge of the status and requirements of all species that occurred there. Obviously this level of knowledge will never be available for any location. Consequently, the analysis presented here is based on a combination of survey data and 'expert opinion'. This opinion was obtained through a series of small workshops or through consultation with individuals who had a knowledge of the area and expertise in particular taxonomic groups which occurred in the region. The participants were asked to identify species they considered at risk and to identify the likely threats. This process necessarily introduces an element of subjectivity to the analysis given that the expertise of the participants is usually derived from a combination of rigorous study, anecdotal observations and interpretation of observations made elsewhere.

Identifying the focal groups

At risk or secure?

Members of the expert panels were asked to nominate species they considered potentially at risk in the central wheatbelt if no management action was taken. This analysis identified 19 species of birds, 2 species of mammals, 5 reptiles, 7 frogs, 12 invertebrates and 11 plant species that were potentially vulnerable (see Appendix 2 for a list of vulnerable species and their perceived threats). These species were then categorised according to the factors that limit their distribution and abundance. Factors considered relevant for the study area were (i) insufficient habitat (ii) habitat isolation (iii) insufficient resources (iv) feral predators (v) weeds (vi) grazing by stock and (vii) inappropriate fire regimes.

Reconstruction or management?

The categories identified above can be separated into two broad groups which reflect the need for habitat reconstruction, on the one hand, or the management of threatening process on the other. Species of birds, reptiles and invertebrates were identified as requiring both habitat reconstruction and management of ecosystem processes, whereas mammals and plants primarily required more appropriate management of the habitat that is available. For example, predation by foxes and cats was perceived to be the main factor limiting mammal distribution and abundance, while stock grazing and weeds were the greatest potential threats to the vulnerable plants.

Species requiring habitat reconstruction

Species requiring habitat reconstruction were partitioned according to whether they are (i) area limited, (ii) dispersal limited or (iii) resource limited.

The species falling into each of these categories are listed in Appendix 3.

i) Area-limited species

Ten species of birds were considered potentially area limited (Table 1). These species were grouped into three broad habitat categories (woodland, shrubland/mallee, and heathland) and then ranked in order of the minimum patch size in which they were known to occur.

The species with the greatest area requirement for each patch type was considered the 'focal' species for that patch type. For woodlands, both the sittella and jacky winter had similar area requirements and hence were equally appropriate as focal species. Western yellow robins and crested bellbirds had similar minimum area requirements for the shrubland/mallee habitat type but these were marginally less than those of the shy hylacola which was only recorded from five patches, none of which were less than 25 ha. Very little is known about the field wren (Calamanthus fuliginosus). It has only been recorded on the largest remnant in the catchment (1100ha) in a heathland patch that is approximately 25 ha in size. It is not possible to ascertain whether its presence on this remnant is attributable to the size of the remnant or to the size of the habitat patch as there are very few other remnants with equivalent amounts of heath. In the absence of better information this figure of 25 ha will be used to specify the minimum suitable area of heathland.

Several invertebrate species were also identified initially as being area limited. These included Mygalomorph spiders and several scorpion species (Main 1987; Smith 1995). However, the primary spatial limitation for these species was considered to be amount of suitable microhabitat. This suggests that the partitioning of the landscape at the resolution of dominant vegetation types is too coarse for these types of species. Under these circumstances it was considered that these species be viewed as being 'resource-limited' with microhabitat being the limiting resource. These species will therefore be considered further in the section on resource-limited species.

While vulnerable plants were generally considered to be threatened by the impacts of stock or inappropriate fire regimes they could also be considered to be area limited if they require populations to have sufficient numbers of individuals to maintain genetic diversity. While this type of information is rarely available, it has been suggested that more than 500 individual Eucalyptus regnans are required to ensure adequate genetic diversity for that species (Ashton 1975). Two tree species which are potentially at risk in the catchment are salmon gums (E. salmonophloia) and gimlets (E. salubris). These species occur at average densities of 96 and 102 trees per hectare respectively (Lambeck 1995; Sarre et al. 1995). This suggests that the minimum area required to support 500 individuals would be 5.2 ha for salmon gums and 4.9 ha for gimlets. These area requirements are considerably less than those of the most demanding woodland birds.

On the basis of the requirements of species that are area-limited, the patch sizes considered necessary to support the most demanding species are (i)greater than 23 ha for woodland, (ii) greater than 25 ha for shrubland/mallee and (iii) greater than 25 ha for heathland. It was assumed that any patch big enough to be occupied by these focal species would also be large enough to support all other species that use those patch types.

Table 1: Area-limited bird species ranked in order of the minimum patch size in which they regularly occur
Woodland Shrubland/Mallee Shrub/Heathland
Species Area (ha) Species Area (ha) Species Area (ha)
Jacky Winter 23 Shy hylacola 25 Field wren 25
Sittella 23 Crested bellbird 16 Blue-breasted fair-wren 3
White-eared honeyeater 4 Western yellow robin 22    
    White-eared honeyeater 13    
    Southern scrub robin 22    
    Inland thornbill 5    
ii) Dispersal-limited species

Information on dispersal is available for a limited number of species in the Kellerberrin district. Most of this information is for movements of common, highly mobile vertebrate species (eg, Arnold et al. 1991; Saunders & de Rebeira 1991). Very little is known about the less common species which are the most vulnerable. The bird species considered most likely to be dispersal limited include shy hylacolas and field wrens which appear to prefer the interior of habitat patches. Both of these species occur in only a few remnants in the study area. Consequently, virtually nothing is known about their requirements for movement.

In the absence of direct measures of movement, estimates of potential dispersal ability can be obtained by measuring the distances between suitable habitat patches occupied by dispersal-limited species. For a given species the inter-patch distance beyond which seemingly suitable habitat remains unoccupied can provide a rough estimate of distances beyond which that species will not move. The best information available for the study area comes from Cale (1994) who identified the minimum patch size occupied by particular bird species and the inter-patch distances beyond which they are unlikely to occupy an otherwise suitable patch (eg. Figure 7).

Figure 7: Pattern of shrubland patch occupancy by western yellow robins (Eopsaltria griseogularis).

Figure 7: Pattern of shrubland patch occupancy by western yellow robins (Eopsaltria griseogularis).

Note: This species was never found in patches of scrubland less than 20 ha or in patches more than 2km from the nearest occupied patch.

Source: Data from Cale (1994)

Estimates of inter-patch distances between occupied and unoccupied sites in the study area were available for only a handful of bird species. Of these, the species which displayed the greatest dispersal limitation was the western yellow robin. As can be seen from Figure 7, this species was not found in remnants more than two kilometres from the nearest occupied patch. While this suggests that it would not be a good idea to locate a patch more than two kilometres from an existing one, it does not guarantee that patches less than two kilometres apart will be occupied. In fact, some patches that are large enough to be occupied and are less than two kilometres from the next nearest patch, do not have western yellow robins on them.

Of the species examined, the yellow robin would appear to be the focal species for defining the minimum inter-patch distance. However, because no information is available for field wrens and hylacolas, any recommendation for a 2 km inter-patch distance should be treated with caution. It should not be assumed that remnants that occur within 2 km are redundant and can therefore be cleared.

Similarly, little is known about the dispersal characteristics of small vertebrates or of invertebrates. Individuals of many of these groups are unlikely to disperse between remnants in an agricultural landscape. The maintenance of population continuity for these species will only be possible if the linear vegetation along road verges, waterways and fence lines provides habitat which can support resident populations. Under these circumstances, this linear vegetation should be viewed as linear habitat and should be designed to meet the habitat needs of dispersal-limited species.

Scorpions (Smith 1995) and some spiders (Main 1987) were identified as potentially dispersal-limited species. Of the scorpions, Cercophonius michaelseni appears to be a very slow disperser. This species has not been recorded for a period of 5 years following a fire, in spite of the fact that adjoining populations in unburnt habitat occur less than 20m away. This failure to disperse could also be attributable to the fact that the vegetation has only partly regenerated over this period and may not yet constitute suitable habitat. Some trapdoor spiders are known to have limited dispersion powers. Anidiops, Idiosoma and Teyl sp. B tend to aggregate around the matriarchal nest if space is available (Main 1987) although their capacity to survive and disperse in the absence of such space is not known.

Because of our limited knowledge of the dispersal requirements of most of the species in the landscape, it is not possible to make unequivocal statements about the characteristics of connecting vegetation. Recommendations will therefore be based on a combination of data, where it is available, and general principles, where it is not.

For birds, it appears that remnants should be no more than 2km apart and that connecting vegetation should be sufficiently wide for interior habitat specialists to utilise them. In the case of Hylacolas and Field Wrens this may require corridor widths greater than 50m. It is unlikely that the implementation of such a recommendation for all linkages between all remnants would be feasible in the short term. Hence it is suggested that linear vegetation of this quality should be preferentially established between sites occupied by these species and the nearest suitable habitat patches.

If linear vegetation is to act as habitat for small mammals, reptiles and invertebrates it must be ungrazed, relatively free of weeds and have a range of habitat types. Fencing is therefore a primary consideration. In order to reduce edge effects linear vegetation should have a minimum width of 30 m if it consists of heath species, with the width increasing as the plant density decreases. Where woodland species are used with little understory, linear vegetation should ideally exceed 60 m. The linking vegetation should contain the structural and compositional characteristics of the local vegetation types. Where woodland species are used, substantial clumps of middle canopy and understory species should be included.

These recommendations should not be taken to imply that strips of vegetation less than 30 m have no value. Narrow strips can be used by a large number of species but they will simply have a lower probability of meeting the needs of the more dispersal-limited species or of species that use the linear vegetation as habitat.

iii) Resource-limited species

The only species in the region for which resources have been demonstrated to be limiting are honeyeaters (Lambeck 1995) and predatory invertebrates (Main 1987; Smith 1995). Competition for tree hollows between various parrot species may have contributed to the demise of regent parrots and western rosellas but, because these species no longer occur in the study area, the availability of tree hollows appears to no longer be a threat to the remaining species.

Populations of honeyeaters are restricted by the depletion of nectar resources in late summer and autumn. At this time, many of the nectar-dependent species leave the area, while the more generalist species become largely insectivorous. The capacity of the remnants to support honeyeaters at this time is lower than their carrying capacity when nectar is abundant.

There are two possible responses to this problem. One is to consider these honeyeater species nectar limited and therefore increase the availability of nectar-producing plants at the limiting times of the year. The alternative is to consider them to be insectivores at this time and increase the amount of habitat available on the assumption that more or bigger patches will provide more insects and therefore support more individuals. The former response will be the most cost-effective if it is possible to provide a rich food source in a relatively small area at the appropriate time.

If the alternative response of increasing the habitat area is being considered, these species should be included in the area-limited analysis described above.

In the Wallatin case study, it was decided that the most efficient response would be to enhance access to nectar over summer by increasing the availability of summer and autumn flowering species such as mallee eucalypts, the acorn banksia (Banksia prionotes) and the Epacrid, Astroloma serratifolium. The acorn banksia produces copious amounts of nectar in late summer and autumn. In the Wallatin Catchment it is confined to a single small patch and a few scattered trees (Lambeck & Saunders 1993; Lambeck 1995).

As discussed previously, some spiders and scorpions were considered to be area-limited, but the limitation was at the level of micro-habitat rather than the dominant vegetation associations. We can therefore consider these microhabitats to be a limiting resource and must attempt to ensure that they are appropriately represented in any reconstructed vegetation. It is therefore assumed that an area of a given vegetation type that is large enough for vulnerable bird species will also be of sufficient size for any invertebrates that occupy that patch type, provided the appropriate microhabitat requirements are present.

Of the vulnerable trapdoor spiders Idiosoma nigrum requires stable litter mats for burrow sites in York gum/Jam wattle (Eucalyptus loxophleba / Acacia acuminata) woodland; Aganippe sp. D prefers flood-prone depressions and flats which support myrtaceous shrub heaths. Teyl sp. B and two species of an undescribed diplurine genus prefer open patches within the litter matrix (Main 1987).

An important issue to address in future is whether these microhabitat attributes need to be created as part of the habitat reconstruction, or whether they will develop, over time, of their own accord if degrading processes such as stock impacts and weed invasion are excluded from the reconstructed patches.

Species requiring management

Species requiring management were categorised according to the processes that were considered responsible for their vulnerability (Appendix 3). Undoubtably a primary threat to nature conservation in the wheatbelt of Western Australia is that posed by an inexorably rising saline water table which threatens all remnant vegetation in the low-lying parts of the landscape. However, it is recognised that salinity is also the major threat to agricultural production and hence its control will be an integral part of an agricultural management strategy. Salinity is not considered to be a threat to biodiversity in this analysis, purely because it is assumed that it will be managed to protect agricultural production. Guidelines for managing hydrology are therefore presented in Chapter 3. The main threatening processes in the Wallatin Creek Catchment were therefore considered to be (i) grazing by stock, (ii) predation by foxes and cats, (iii) inappropriate fire regimes, and (iv) invasion of remnant vegetation by weeds. For each of these processes, species were ranked in terms of their vulnerability to or dependence upon the process.

i) Grazing

While no specific data are available to assess the relative vulnerability of the plant species threatened by grazing, it is apparent that a number of species are sufficiently uncommon that they should be protected from any level of grazing. The poor recruitment of woodland species, including Salmon gums, Gimlets and Banksias, in the presence of stock suggests that there is no level of regular grazing by stock that is acceptable in woodlands. If occasional grazing is to be allowed in woodlands, it must be at intervals that are sufficiently long for germination to occur and for saplings to grow to a stage where they are resilient to stock.

Heath and shrubland communities have also been severely degraded by stock, resulting in marked changes in both plant and animal species diversity (Hobbs & Atkins 1988; Abensperg-Traun 1992; Scougall et al. 1993). These changes are a consequence of direct grazing of plants, compaction of soil, and destruction of microhabitats. Given these impacts, the precautionary approach is to exclude stock entirely from remnant vegetation. Because remnant vegetation has been traditionally used to provide shelter for stock this strategy will require the planting of new woody perennial species in other parts of the landscape to provide alternative shelter.

Rabbits have also been identified as limiting the recruitment of many plant species in other parts of Australia (Lange 1983; Foran. 1985). However, their numbers in the study area are currently not high enough to present a major threat. Given their potential to become a threat if populations increase, it would be advisable to maintain a program of baiting and warren ripping.

ii) Feral predators

Four species of birds, two mammal species and numerous reptile species were considered potentially vulnerable to predation by foxes and cats (Appendix 3). These species were selected, not because of empirical evidence of predation, but because their behaviour increases their vulnerability. The four bird species (field wren, shy hylacola, bush stone curlew, and southern scrub robin) are either terrestrial and nest on the ground, or occupy low shrub habitat. The western brush wallaby (Macropus irma) is currently known from only one remnant in the study area, where it has been sighted in patches of dense Allocasuarina adjoining open woodland. This species is endemic to the south-west of Western Australia and anecdotal observations suggest that it may be declining in both distribution and abundance in the wheatbelt. The ash-grey mouse (Pseudomys albocinereus) is also uncommon, having been captured in only a few locations where deep yellow sands occur. This species has been found to be a common prey item of cats elsewhere in its distribution (Risbey 1997). While it is clear that both cats and foxes prey upon reptiles, there is currently no information to suggest that any particular reptile species are at risk from predation. However, there is the potential that species which are uncommon for reasons other than predation may be vulnerable if predators are able to drive small isolated populations to local extinction. The mountain devil (Moloch horridus) for example, is uncommon in the study area and local populations may be susceptible to predation.

The species in this study that appears most obviously to be at risk from fox predation is the western brush wallaby. Studies of feral predators elsewhere have indicated that even very low predator numbers can have significant impacts on mammal survival (Spencer 1991; Horsup & Evans 1993). This suggests that it will be necessary to remove virtually all foxes from Durokoppin Nature Reserve in which the remaining wallabies occur. If this reserve is unable to support viable populations of this species even in the absence of predation, it will be necessary to extend the distribution of this species beyond the reserve. This will require the control of feral predators over a much larger area. The brush wallaby can therefore be considered the focal species for fox control as its requirements for predator management will encompass the needs of other species that are also vulnerable to fox predation. Initial baiting of the reserve should be intensive. In a fox control program at a number of granite outcrops to the south of Kellerberrin, the state conservation agency (CALM) laid fowl eggs injected with 1080 (sodium monofluoroacetate) at 50m intervals around the perimeter of the reserves (Kinnear et al. 1988). This study found that foxes killed by early baiting were quickly replaced by others moving into the area and hence an intensive baiting regime was required.

It may be possible that baiting intensity could be reduced if baits are distributed more widely throughout the landscape. However, such a strategy has not been adopted in agricultural regions and its feasibility has not been assessed. Whatever strategy is adopted, baiting must be implemented in a coordinated manner to ensure that all farmers are aware of the presence of baits and can take the necessary actions to protect farm dogs.

Little is known about the impact of predation by cats in this area, however there is anecdotal evidence to suggest that cat numbers may increase as fox numbers decline (J. Short unpublished data). Such an increase may potentially impact on small mammals such as the ash-grey mouse. In the absence of information about the risk that cats pose, a precautionary approach would be to adopt a strategy which aimed for simultaneous control of cats and foxes. There is still some debate about which methods are most effective for cat control and in some instances it appears that the use of a range of bait types and trapping methods may be most effective (J. Short pers. comm.). If baiting or trapping for either cats or foxes is to be undertaken it is essential to consult the appropriate management authorities to ensure that relevant regulations are complied with.

iii) Fire

Limited information is available about the importance of fire in the maintenance of plant or animal populations in the catchment. Animals identified as being potentially vulnerable to fire were reptiles and some species of scorpions (Smith 1995) and spiders (Main 1987). While the majority of species in the region would be vulnerable to extreme fire frequencies, the assessment of risk was based on the current regime of very infrequent fires.

Reptile species which seem to be vulnerable to single fire events include an arboreal gecko (Diplodactylus spinigerus), an agamid lizard (Ctenophorus reticulatus) which occurs primarily in mallee communities, and a skink (Ctenotis pantherinus) which occupies heath/shrubland. Both Ctenophorus and Ctenotis disappeared from study sites following experimental fires and population recovery was dependent upon recruitment from adjoining unburnt areas (G. T. Smith unpublished data). A period of approximately 4–5 years was required before mature adults of these species reappeared after fire in remnants where there was adjoining unburnt habitat. While data are not available for D. Spinigerous, its arboreal habit suggests that it would also be sensitive to fire.

The scorpion, Cercophonius michaelseni also disappeared from its highly flammable heath habitat following an experimental fire. Individuals of this species were able to survive the fire but were unable to persist in subsequent seasons (Smith 1995). Given the extremely slow rate of dispersal by C. michaelseni back into the burnt habitat, it would appear that long inter-fire intervals would be required for this species to recolonise areas from which it had become locally extinct. The trapdoor spider Anidiops villosus showed a similar response with adults surviving the fire itself, but suffering progressive mortality and no recruitment following the fire due to inadequate shade and litter and possibly also due to reduced prey and increased vulnerability to predation (Main 1995). This species is also likely to be a slow disperser and hence many years may be required for recolonisation to occur, provided there are adjacent unaffected populations.

The persistence of many plant species may also depend on the frequency and intensity of fire. An important factor which determines the impact of fire on plant population viability is the life history attributes of species affected. Species which recover by re-sprouting may be more resilient to frequent fires or high intensity burns. Of the potentially vulnerable plant species in the study area, Grevillea dryandroides and Halgania tomentosa re-sprout after fire and hence may only be at risk if the fire frequency is sufficient to deplete below ground nutrient stores. By contrast, fire-sensitive species which rely on the germination of seeds stored in the canopy or in the soil will be particularly vulnerable to fire frequencies which do not allow sufficient time for post-fire recruits to reach maturity and replenish the seed bank. Verticordia hauganii is an obligate re-seeder and may be vulnerable to a sequence of burns at intervals less than the two to three years following germination that this species may require to set seed. Fire ephemerals which set seed within a short time of a fire are unlikely to be at risk as they will replenish seed bank before the area has sufficient vegetation to support a subsequent burn. Flannel flowers (Actinotus superbus) are post fire annuals which only appear for a few years following a burn (Hester & Hobbs 1992). Their seeds presumably remain viable in the soil seed bank for many decades as they will readily appear following fire in areas that have not been burnt for long periods of time.

Other fire-sensitive species may take longer to reach maturity and therefore may be more vulnerable to frequent fires. Grevillea eriostachya, for example, may be killed by an intense burn and may not flower for up to 4 years post-fire. For this species, several more years may be required before an adequate soil seed bank can be established. This would suggest that a minimum inter-fire interval of at least 5–6 years may be required to ensure that local populations of this species were not threatened. Numerous Banksia species are also killed by fire and may take several years before they produce flowers and accumulate sufficient seeds to restore the population following fire (Gill & McMahon 1986). Where such species occur their requirements should be used to specify the minimum appropriate inter-fire period.

Short-lived, fire-dependent species that do not have a long lasting seed bank should be used to identify the maximum acceptable inter-fire period. For example, Banksia coccinea, a species which occurs in the southern coastal areas of Western Australia requires fire to stimulate seed release and germination but senesces after approximately 15 years. This species therefore requires a fire frequency of less than that time interval. The only Banksia species in the study area is Banksia prionotes which senesces after approximately 40 years. However, this is one of the few Banksias which readily recruits in the absence of fire.

For the Kellerberrin area, there is no clear evidence to suggest that a frequent fire regime is necessary or appropriate. Some plant species appear to be potentially vulnerable if fire intervals are less than 5–6 years. Some animal species, on the other hand appear to require much longer intervals between fires to allow for the regeneration of appropriate habitat and for their slow rates of dispersal into those regenerating habitats. In the absence of evidence to the contrary it is recommended that fire be used as a management tool very infrequently and only where there is evidence that vegetation communities are senescing. For most of the vegetation types in the region, this would suggest that there is no need to burn at interval of less than 50 years. It is important to recognise however that the basis for such recommendations is somewhat limited and that classification of species in terms of their responses to fire would be an important aid to managers. Gill and Nicholls (1989) provide a methodology for monitoring fire sensitive plants and point out the importance of using monitoring as a tool for learning more about the fire responses of the flora.

In areas where there is little or no information about the specific needs of the biota, land managers have traditionally reverted to general management principles. These principles suggest that the optimal approach is to ensure a fire regime that is as varied as possible in terms of season, intensity, frequency, patch size and availability of adjacent unburnt areas for recolonisation. While this type of strategy may be possible in large areas of continuous habitat where a range of options are available, it will not be possible to adopt such an approach in small remnants of native vegetation. Many such remnants are too small to apply a spatially and temporally variable fire regime and a hit-and-miss approach based on general principles will have a high probability of contributing to the local extinction of some species. If these remnants are to be considered to be a part of the conservation network, and if fire plays a critical part in the dynamics of the plants and animals that occupy them, it will be necessary to gain a better understanding of the fire responses of the species present and to develop fire management strategies which meet their needs.

iv) Weeds

The presence of weeds in vegetation remnants can have a significant influence on plant communities by inhibiting the regeneration of native plant species (Hobbs & Atkins 1991), by changing micro-climatic conditions and by altering fuel loads and hence fire dynamics. The resulting changes in plant community composition and structure can subsequently influence animal diversity due to modified habitat characteristics. In the study area, none of the vulnerable species were considered to be directly threatened by weed invasion. Consequently no species have been nominated to define appropriate levels of weed control. In spite of this, it is apparent that weeds have the potential to threaten plant species and communities and hence should be closely monitored and controlled if they are known to be highly invasive and where they appear to be threatening native plant communities. Control of weeds is likely to be critically important when trying to re-establish native vegetation (Panetta & Groves 1990).

Key

   Links to another web site
   Opens a pop-up window