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)

2.10 Guidelines for implementation

The outcome of the focal species approach is a list of species whose requirements for key habitat and resource variables define different attributes that must be present if a landscape is to meet the needs of its constituent flora and fauna. Ideally, the list would include focal species to define the minimum area of each patch type; species to define the minium width, length and structure of connecting vegetation; species to define appropriate levels of critical limiting resources; and species to define the appropriate rates or intensities of each potentially threatening process. Insufficient information is currently available to identify focal species for all of these variables. While it is possible to use the methodology to identify the minimum size of each patch type, general principles and anecdotal observations will have to be invoked to define the appropriate configurations of patches and the attributes of connecting vegetation.

In the Wallatin Creek case study, maps of abiotic attributes, including soils and landforms, which correlate with the relevant patch types, were used to identify positions in the landscape which were suitable for reconstruction of the main habitat types. Existing vegetation types were mapped using satellite imagery. Air-photo interpretation and ground surveys were used to validate the satellite-derived maps and to identify significant vegetation associations within each formation. The resultant maps were incorporated into a geographic information system (GIS) and interrogated to identify all vegetation patches which failed to meet the spatial requirements identified by the focal species analysis. GIS routines were developed which specified areas which needed to be added to these patches in order to meet the minimum area requirement (Figure 11).

Figure 11: Map of the Wallatin Catchment indicating areas requiring habitat reconstruction to meet the needs of area-limited focal species.

Figure 11: Map of the Wallatin Catchment indicating areas requiring habitat reconstruction to meet the needs of area-limited focal species.

This analysis revealed that 61 of the 113 mapped habitat patches in the catchment did not meet the minimum size required by the focal species. Expansion of these patches to meet these minimum sizes would require the revegetation of 1,121 ha. or 4.3% of the catchment. When this area is combined with the total area already in remnant vegetation (1,925 ha. or 7.4% of the catchment), the total area required to produce an adequate landscape for nature conservation is 3,046 ha or 11.7% of the catchment. The proportion of each landform type that requires revegetation is shown in Table 2.

Table 2: The amount of each landform type that is required to expand small vegetation patches to a size where they meet the minimum area requirements defined by the focal species
Landform Area (hectares) % of each landform required
Ulva 194.3 4.6
Booraan 270.9 4.2
Collgar 168.7 5.2
Merredin 85.1 1.2
Belka 85.8 10.6
Danberrin 316.7 7.3

This analysis does not take into account the area required for providing linear vegetation to link the remnant vegetation. In the Wallatin Catchment, the majority of remnants are connected to some degree, although the width of the connecting vegetation varies from a few meters to up to 15m. For this exercise, it was considered that a general recommendation be adopted which ensures that all remnants are connected by linear vegetation at least 30m wide with greater than 60m being the preferred width for open vegetation types such as woodlands and a minimum of 50m for linear vegetation linking sites occupied by dispersal-limited species to nearby suitable habitat patches. Any reconstruction of connecting vegetation should ideally use local species planted according to soil type.

From the procedure described above, the following guideline was developed to allocate land to nature conservation:

This commitment was considered necessary because the existing remnant vegetation appears to be inadequate for meeting the objective of ensuring no further loss of species. The evidence for this concern comes from the continuing decline of species that are not threatened by processes such as predation and fire (Saunders 1989). Consequently all patches should be retained and should serve as the framework to which new habitat should be added.

The following guidelines are for habitat expansion based on the results of the focal species analysis:

Three additional guidelines were provided which recommend that suitable areas be allocated to reconstruction with salmon gum/gimlet woodland and Banksia woodland. This reflects their under-representation in the catchment (Section 2.8):

These guidelines formed the nature conservation input to the integrated planning exercise described in Chapter 3. Further details about the role of these guidelines in the planning process are presented in that chapter.

2.11 Moving the goal posts: the consequences of implementation

While focal species can be used to determine the attributes required in a landscape, it must be remembered that the initial assessment of risk considered only immediate threats. It is possible that a species currently limited by landscape configuration may, when the configuration is altered, change in numbers only to a level whereby a new limit is imposed by another factor. In addition, changes in landscape pattern may result in altered species responses to the new landscape configuration. For example, as the number of patches in a landscape increases, it may be possible for individuals of a species to occupy smaller patches than they could when fewer patches were available. Similarly, changes in the quality of corridors may alter the minimum inter-patch distance over which individuals of some species can move. Not only will interactions between species and their habitat change as a result of changing configurations, but interactions between species may also change as a result of different species responding in different ways to altered configurations.

These unpredictable interactions, together with the obvious risks associated with the initial assumption that the needs of the focal species encapsulate the needs of all other species, make the establishment of a monitoring process critically important. The monitoring program must be designed to test the underlying assumptions and must have a capacity to detect deviations from predicted responses at the earliest possible time. The monitoring program must focus primarily on the focal species, but must also consider the responses of a suite of additional non-focal taxa. These additional species should be selected to represent a range of life-history characteristics in a variety of taxonomic groups. Using this approach, a strategic monitoring program based on a limited set of species would provide an indication of the changes occurring in the landscape in response to the management actions.

Any monitoring strategy will obviously require the allocation of additional resources. However, the failure to implement a monitoring strategy will ensure that we will fail to learn from our actions. The focal-species approach provides a theoretical basis for a monitoring strategy which tests clearly stated hypotheses and assumptions and enables an assessment of performance against nominated objectives. Because general enhancement approaches are unable to specify the outcome expected, monitoring will simply indicate that changes are occurring but will not be able to specify what the causes of those changes are or what remedial actions should be taken in response to unexpected changes. The level of resources required for monitoring will not be trivial which provides yet another reason for adopting a bioregional scale approach to conservation planning with a carefully designed monitoring program distributed strategically throughout the region.

2.12 Transportability of solutions

The hope of many conservation managers is that it will be possible to identify a limited set of landscape variables which, if present, will ensure the retention of all of the biota in the landscape that they are managing. In other words, they are seeking relatively simple templates that specify a proportion of the landscape that should be allocated to native vegetation in order to ensure the persistence of the biota. Expectations that general templates can be developed stem from an overly simplified interpretation of a body of theory that developed in the 1960s. This theory is based on the species-area relationship developed by MacArthur and Wilson (1967) which showed that the number of species on oceanic islands increased with the size of the island and decreased with the degree of isolation from source populations. This theory had intuitive appeal to conservation biologists who were viewing terrestrial vegetation remnants as islands in a sea of crops.

The limitation of this theory in a modified terrestrial environment is that it does not take into account the combination of habitat heterogeneity (patchiness) in the pre-clearing landscape and the selective clearing of that landscape. Particular patterns of clearing in a patchy landscape can have impacts on biodiversity that are disproportionate to the area modified. For example, the selective removal of small resource-rich areas, or of refuges used by animals during times of environmental stress, may cause a disproportionately large decline in species richness. Conversely, the removal of large resource-poor areas may have a smaller impact on species numbers than would be expected on the basis of a simple relationship between species numbers and area. These factors are particularly pertinent in the highly fragmented wheatbelt of Western Australia which is renowned for its species richness combined with high levels of local endemism.

In order to test whether the nature conservation recommendations produced for the Wallatin Catchment were relevant elsewhere, a preliminary survey was conducted in the Dumbleyung Shire, approximately 200 km to the south of the current study. The objective of the survey was to determine patterns of habitat use of species that were considered vulnerable in Wallatin, and of species that have disappeared from Wallatin but still persist in Dumbleyung. The area surveyed in Dumbleyung had a similar configuration of remnants to the Wallatin Catchment, with approximately 10% cover of remnant vegetation. Both areas were also similar with respect to the structural attributes of the dominant vegetation associations although there were differences in plant species composition.

It was immediately apparent from even a superficial survey that there were significant differences in patterns of patch use by bird species in the two areas. For example, western yellow robins and southern scrub robins occurred in much smaller patches of shrubland in Dumbleyung than was the case in Wallatin. Rufous tree creepers, which no longer occur in Wallatin were found in 4 ha degraded patches of woodland. Patches of habitat which are suitable for these species in Dumbleyung are obviously not adequate in Wallatin. There are many equivalent patches in Wallatin Catchment and the areas surrounding it and yet they are not occupied by these species.

The reason for this discrepancy is likely to be attributable, at least in part, to differences in landscape pattern at a regional scale. While the sites are similar at a local scale, the two locations are very different when viewed from a regional perspective. The distribution of vegetation remnants in the Kellerberrin shire, where Wallatin Creek is located, is roughly the same throughout the shire. In other words, the Wallatin site is broadly typical of the rest of the region. The Dumbleyung site, on the other hand, was located within 15 km of extensive bushland which includes Dongolocking Nature Reserve. The close proximity of this high quality habitat is likely to be an important factor in determining the patterns of habitat use observed at Dumbleyung. Clearing history may also influence the observed habitat patterns and any detailed comparisons between areas should therefore take into account the time since clearing.

The fact that a solution generated in one part of the wheatbelt clearly cannot be transported to another is likely to concern conservation managers. However, it is critical that this fact is acknowledged as any attempts to take solutions from one location and apply them to another will produce results which will be inadequate in some instances and excessive in others. The focal species approach does not provide a template that can be simply transported to a new location. Rather, it provides a procedure for deriving landscape designs and management guidelines in any landscape.

While it is apparent that the results derived from Wallatin Creek could not be applied 200 km away, it may be reasonable to expect that the solution could be usefully applied in the catchment immediately adjacent to Wallatin Creek, given the similarities in the flora and fauna and in landscape configuration. If it can be applied to the next sub-catchment, could it be applied to the one beyond that? This raises a critical question for conservation planning: To what extent can a solution be legitimately extended beyond the location where it was generated?

A potential solution to this problem could be found if it is possible to partition a large region, such as the wheatbelt of Western Australia, into smaller sub-regions which are internally homogeneous, not only with respect to environmental variables, as is the case with bioregions, but also with respect to their landscape configuration. These sub-regions, which could be termed Conservation Management Zones (CMZs) would form a rational framework for conservation planning.

If such Conservation Management Zones can be identified, then a solution derived from one location within a CMZ could more reliably be applied to the remainder of that zone.

Such an analysis of landscape pattern would help to define the appropriate scale for conservation planning. In areas that are relatively uniform with respect to both biophysical and anthropogenic parameters, it may be possible to develop solutions that can be applied over a large region. On the other hand, where patterns of clearing are more spatially variable it may be necessary to develop plans at a smaller spatial scale. Appendix 1 presents a framework for planning for nature conservation at a regional scale.

2.13 The role of science and data adequacy

Three major problems underpin the acquisition of knowledge for managing biological diversity in production landscapes. The first of these is the absence of comprehensive survey data and a detailed knowledge of the ecology of the biota we wish to protect. Consequently we have limited knowledge of what we are trying to conserve, or of how the species that we wish to protect are using the landscape that we are managing. Unfortunately the acquisition of comprehensive survey data and detailed ecological understanding is extremely labor intensive and expensive (Burbidge 1991). Consequently, the information that is available to managers is neither comprehensive, nor detailed. Where detail does exist, it invariably deals with a very small component of the total problem.

The second difficulty stems from a perception that the acquisition of information for management should be based on sound scientific methodology. Such scientific rigour demands replicated experiments with appropriate controls and large sample sizes to provide sufficient statistical power. As a consequence, practitioners of 'good' science often focus on those aspects of a landscape that can provide the required rigour. We therefore count those things which are easily counted and describe events commonly, and hence reliably, observed. The result is a wealth of knowledge about the common and conspicuous features in the landscape. However, when that landscape is deteriorating, this is not necessarily the knowledge that managers require. We do not need to manage those features in a landscape that are secure. Unfortunately, the components that are threatened, and therefore do need to be managed, are often rare and hence less amenable to rigorous scientific investigation.

The third fundamental problem again relates to the practice of science. The scientific method has traditionally focussed on deriving general principles from a series of observations that appear to exhibit common patterns. We therefore gather data from a number of specific events in particular locations and derive a general principle which we believe should hold over a range of circumstances. Unfortunately, while it is relatively easy to progress from a series of site-specific observations to a general principle, it is more difficult to return from the general principle to a prescription for a new location. Ultimately, general principles enable us to make statements about the relative merits of alternative actions, but they cannot provide the absolute values that managers require when they are competing for limited physical and monetary resources. For example, a general principle which states that 'more is better than less' for nature conservation is of little use to a conservation manager confronted with a land owner who will only sacrifice the minimum possible amount of productive land. General principles can tell us a direction in which to proceed, but provide little guidance as to how far to go.

The resolution of these problems requires firstly, a recognition that they are problems and secondly, a conscious effort to broaden the horizons of ecological science from the tradition of description to the challenge of prescription. Given the complexities of ecological systems and the spatial and temporal variability both within and between landscapes it is unlikely that efficient 'template' solutions can be developed that will hold far beyond the source of their development. Consequently, the aim of science in conservation management should not be to provide 'answers' that apply to all circumstances, but rather, to provide procedures for arriving at an answer that is appropriate for the situation at hand in the most efficient manner possible. The focal species approach presented in this section represents an attempt to provide such a procedure. The 'answer' generated by the procedure will depend upon the objectives, the location, and the degree of biological impoverishment in the area being considered.

This alternative view of the role of science in conservation management brings into question the type of data required for making management decisions. If our objective is to prevent the loss of elements at risk in our landscapes then we need to focus on those elements. This requires increasing attention to the needs of the less common features in our landscapes. This is not to advocate a return to single species studies of rare and endangered species for their own sake. Rather, it is an argument for focusing on those vulnerable elements of a landscape which, if managed effectively, can deliver the greatest benefit to the widest range of additional species.

If a focal species approach to nature conservation is to be adopted, there are a number of data sets that must be acquired. The first of these is a map which partitions the landscape into biologically meaningful units which can be used as the building blocks for landscape design. These units will invariably be a compromise between the true complexity of the ecological system being managed and the capacity of managers to incorporate that complexity into their management regime. For the current study the dominant vegetation associations were considered appropriate units for landscape assessment and planning. Subsequent design was based on creating different configurations of these units which meet the needs of the focal species.

A range of methodologies are available to create such maps including satellite remote sensing, air photo interpretation and ground survey. Often the combination of these methods can provide more reliable data. In an attempt to improve the efficiency of the mapping process, predictive models of landscape pattern are often generated to enable knowledge generated from one location to inform us about conditions elsewhere. The reliability of such predictive modelling relies on the strength of the relationship between the predictive variables and the response variable. For example, vegetation classification from satellite imagery relies on unique and consistent spectral responses from the different vegetation types that we wish to map. The strength of this relationship will depend on the degree of variation within patch types relative to that between patch types. If patches are internally uniform and significantly different from other patches, such technology can produce relatively reliable results. However, when the patches are themselves variable and the differences between patches are subtle, the reliability of these predictive approaches will diminish. By combining satellite imagery with models of landform pattern, it may be possible to improve the quality of the result. For example, areas which are not separable on the basis of their spectral properties may be distinguished on the basis of their position in the landscape. It has yet to be demonstrated that this will improve the quality of our broad-scale mapping capacity. However, given its potential to significantly improve that capacity this represents an area of research that warrants further investigation. Whatever approach is adopted, ground truthing of the results to assess their accuracy is essential.

The second data set that is required is a list of species that are potentially vulnerable in the region being managed. Species or communities should be classified as vulnerable if there is evidence of decline over time. This should apply equally to species which are currently common but declining and those which are rare. Rarity per se is not necessarily an indicator of vulnerability. Some species may be naturally rare and have stable populations. However, in the absence of information on population changes over time it would be sensible to adopt a cautious approach and consider rarity to indicate potential vulnerability.

Having identified the vulnerable species in a landscape it is necessary to have a capacity to derive basic population parameters for key species. As described in Section 2.5, the focal species can currently only deliver 'adequate' landscapes unless applied over an area that is large enough to have a high probability of also being viable. Specification of what is a sufficiently large area will require an assessment of the population viability of the focal species and hence estimates of appropriate demographic variables for these species.

If further investigation of the focal species approach reveals that there are characteristic types of species that are regularly identified as playing an umbrella role, then it may be possible to design efficient survey procedures that specifically assess the status of these types of species. These species are likely to be relatively sedentary habitat specialists. Surveys should therefore be strategically designed to target these less common species and should have a capacity to be repeated at appropriate intervals to assess population trajectories. The routine collection of species-based information at a catchment scale will clearly not be possible in all cases further indicating the need to conduct such analyses at larger, regional scales. The strategic acquisition of data at these larger scales may be cost effective in the long run if such information can contribute to the development of strategies that require less effort per unit area to achieve the desired conservation outcome.

Key

   Links to another web site
   Opens a pop-up window