A Simple Method for Predicting the Consequences of ... - Springer Link

DOI: 10.1007/s002670010230

A Simple Method for Predicting the Consequences of Land Management in Urban Habitats CHRISTOPHER H. YOUNG* PETER J. JARVIS School of Applied Sciences University of Wolverhampton Wulfruna Street Wolverhampton WV1 1SB, UK ABSTRACT / Land management in urban areas is characterized by the diversity of its goals and its physical expression in the landscape, as well as by the frequency and often rapidity of change. Deliberate or accidental landscape alterations lead to changes in habitat, some of which may be viewed as environmentally beneficial, others as detrimental. Evaluating what is there and how changes may fit into the landscape context is therefore essential if informed land-management decisions are to be made. The method presented here uses a simple ecological evaluation technique, employing a restricted number of evaluation criteria, to

Changes to the urban environment result from both human activities and natural processes. In terms of human influence, landscape change occurs deliberately or accidentally, either as a by-product of general urban development or else as a result of specific directed management. These kinds of changes encompass not only large strategic developments but also impact on communities and individuals alike through the development of small land parcels that add diversity to the urban landscape. As pressure on limited resources increases, many more people need (and want) to be informed about the potential environmental consequences, including ecological consequences, of this change, both from an official landscape-planning perspective and from a more sociocultural perspective. Predicting the effects of change is a notoriously difficult exercise, yet it is “perhaps the most important role for the ecologist in the field of planning” (Daniels 1988). Traditional predictive modeling methods rely KEY WORDS: Urban habitats; GIS; Ecological evaluation; Decisionmaking; Prediction *Author to whom correspondence should be addressed; email: [email protected]

Environmental Management Vol. 28, No. 3, pp. 375–387

gather a spatially complete data set. A geographical information system (GIS) is then used to combine the resulting scores into a habitat value index (HVI). Using examples from Wolverhampton in the United Kingdom, existing real-world data are then applied to land-management scenarios to predict probable landscape ecological consequences of habitat alteration. The method provides an ecologically relevant, spatially complete evaluation of a large, diverse area in a short period of time. This means that contextual effects of land-management decisions can be quickly visualized and remedial or mitigating measures incorporated at an early stage without the requirement for complex modeling and prior to the detailed ecological survey. The strengths of the method lie in providing a detailed information baseline that evaluates all habitats, not just the traditional “quality” habitats, in a manner that is accessible to all potential users—from interested individuals to professional planners.

heavily on statistical probability and mathematical complexity to model habitats under a range of input parameters resulting in a diverse range of highly intricate modelling approaches. For example, Bolger and others (1997) use stepwise logistic regression to aid in the prediction of urbanization on breeding bird abundance. While the results of such analyses are undeniably extremely useful, to a nonexpert their methodologies are difficult to understand and their numerical outputs are not particularly accessible (see Baker 1989, for a full review of traditional approaches to modelling landscape change). In order to ensure the full implications of land-use decisions are conveyed to all interested parties, visualizing the spatial extent and context of change is vital. Increasingly, approaches using geographical information systems (GIS) are being employed to accommodate this spatial aspect. In a habitat management context GIS methods are frequently used in conservation and natural areas planning (Merrill and others 1995), risk assessment (Dale and others 1998), or in the prediction of environmental change (Van Der Meulin and others 1991). Approaches such as these allow both temporal and spatial change to be represented in an accessible manner using the display capabilities and the built-in analysis tools in tandem. ©

2001 Springer-Verlag New York Inc.

376

C. H. Young and P. J. Jarvis

Few studies, however, have used these capabilities to extend the predictive approach into urban areas at a scale that reflects the habitat mosaic of the built environment. Indeed, in practice, many even exclude it altogether. For example, Canters and others (1991) predict the effects of a rapid rail route on both biotic and abiotic components of the landscape working in 1-km ⫻ 1-km grid cells and leaving predominantly urban grid cells as unmapped. One recent exception to this lack of detail is found in the method of Villa and others (1996), who use a GIS-based approach to combine a set of mapped landscape attributes and quantitatively expressed management priorities. Although promising, this approach is spatially restricted at the opposite end of the scale, looking at options for a single park area, and appears neither to offer provision for use outside of the particular location nor to have the visual impact of many of the larger studies. The simple method we present here provides data for quantifying habitat alterations at a scale appropriate for urban environments. The method uses a suite of criteria to calculate a habitat value index (HVI) for all individual habitat patches within a defined area, with a 1-km ⫻ 1-km grid square being used here. Using the IDRISI GIS, actual or anticipated habitat alterations are superimposed upon the landscape, and habitat values are applied to these new patches allowing each patch to be considered both in its own right and relative to its landscape context.

Methods Study Area The study area lies approximately 2 km to the northwest of Wolverhampton town center in the West Midlands area of Great Britain. Specifically, the mapping and evaluation focuses on the Ordnance Survey grid squares SJ 8900, 8901, 9000, and 9001, incorporating the area of the racecourse and its environs (Figure 1). This area has a diverse mix of habitats ranging from seminatural woodland and grassland to industrialized, commercial, and residential areas, allowing data to be collected for a range of commonly encountered urban habitats.

plementary urban-specific detail was added through the use of categories from a variety of existing sources (Sukopp and others 1980, Wittig and Schreiber 1983, Gilbert 1989, Sisinni and O’Hea-Anderson 1993). Selection of Criteria Four criteria were used to derive the score for each patch: individual structural elements, indicator species, general habitat structure, and aesthetics. The first three criteria are objective and are found in many evaluation approaches (Wittig and Schreiber 1983, Dickman 1987, Flather and others 1992). Individual structural elements are small features that contribute to the overall habitat diversity yet are too small to be classed as a habitat in their own right. For example, a single tree adds diversity to an area of grassland yet cannot be justified as being a separate habitat patch. Indicator species are applied in this instance to mean an indicator of site species diversity and not an indicator of a particular habitat or site condition per se. They were chosen to represent a simple cross section of the frequently occurring flora species found in the West Midlands. General habitat structure is a simple summary measure of the apparent similarity of the general site appearance expressed in terms of vegetation structure. The fourth criterion, aesthetics, subjectively summarizes many landscape and social functions that may otherwise be difficult to quantify (Ahern 1991) and allows the ecological measures to be placed in a landscape context that may be more accessible to a nonecologist. The structural elements and the indicator species are termed the assessment criteria (Table 1) and are used as the basis of the scoring for each individual patch. The general structure and aesthetics are used as weighting criteria (Table 2) and are used as the basis for determining the relative weights. Habitat type was also used as a weight. Although this is not strictly a criterion, it is used in order to incorporate the inherent importance of habitat type in determining the value of a habitat patch. The specific habitat classification contained 104 possible habitat types; for the purposes of simplicity of display and explanation, we have reclassified these into 10 general habitat types.

Patch Identification and Classification Habitat patches were provisionally identified using air photo interpretation and delineated on a 1:5000 base map. Each patch was then assigned a habitat category from a classification based on the UK phase-1 categories (Nature Conservancy Council 1990). As this classification was designed for seminatural areas, sup-

Field Method In the field the provisionally identified patch boundaries and classification were confirmed; then, in each of the separate habitat patches, the data were collected by noting the presence of each criterion or criterion subcategories on a prepared checklist. A standard ap-

Predicting Consequences of Urban Land Management

377

Figure 1. (A) Location of study area within the West Midlands conurbation, (B) Original habitat map of the Valley Park study area grid squares. Grid square SJ 9000 is highlighted in black.

proach was ensured by walking the longest axis of a site; where this was not possible, the data were collected by walking the patch perimeter or viewing the patch from

as many vantage points as possible. In the office the field maps were digitized onto the GIS and the evaluation data were input to the GIS database.

378

Table 1.


Evaluation criteria categories

Assessment criteria Structural elements

Indicator species

Table 2.

Weighting criteria categories

Weighting criteria Criteria subcategories 1. 2. 3. 4. 5. 6. 7. 8. 9. 10. 11. 12. 13. 14. 15. 16. 17. 18. 19. 20. 21. 22. 23. 1. 2. 3. 4. 5. 6. 7. 8. 9. 10. 11. 12. 13. 14. 15. 16. 17. 18. 19. 20. 21. 22. 23.

Wood/woodlike parks Row(s) of trees Group(s) of trees Single trees Hedges and shrubbery Areas of tall herbs/grasses Hay meadow Pasture Park grass/ornamental lawn Non-formal short grass area Trodden area communities Annual areas Bare areas Reed communities Communities of floating leaf plants Wall communities Hollow trees of soft or dead wood Very varied habitat mosaic Walls, broken stone or rubble Pronounced contours Artificial structure Paved area/pathway Water feature Oak (Quercus spp.) Hawthorn (Crataegus monogyna) Birch (Betula spp.) Willow (Salix spp.) Bramble (Rubus fruticosus agg.) Ivy (Hedera helix) Buddleia (Buddleia spp.) Daisy (Bellis perennis) Dandelion (Taraxacum officinale) Buttercup (Ranunculus spp.) Red clover (Trifolium pratense) White clover (Trifolium repens) Thistles (Cirsium spp.) Nettle (Urtica dioica) Foxglove (Digitalis purpurea) Poppy (Papaver spp.) Japanese knotweed (Fallopia japonica) Yellow iris (Iris pseudacorus) Meadowsweet (Filipendula ulmaria) Reeds (Phragmites spp.) Horsetails (Equisetum spp.) Rushes (Juncus spp.) Fern spp.

Conversion of Field Totals to Score Classes In order to combine totals of different criteria, scores were derived through assigning the actual totals of the different assessment criteria to score classes using a straightforward conversion (Table 3). The same score classes were used for both structural elements and indicator species.

General habitat structure

Aesthetics

General habitat categories (derived from 104 specific habitat subdivisions)

Criteria subcategories 1. 2. 3. 4. 5. 6. 7. 8. 9. 10. 1. 2. 3. 1. 2. 3. 4. 5. 6. 7. 8. 9. 10.

Construction Bare ground Moss/lichens/seedlings Open annual (ruderal) herbs Closed herbs and grasses Open scrub communities Closed scrub communities Open, developing woodland Mature woodland Open water Negative Neutral Positive Woodland and scrub Grassland and marsh Tall herb and fern Heathland Swamp, marginal and inundation Open water Rock exposure and waste Miscellaneous Industrial and commercial Residential

“Quality” and “Non-Quality” Weights Each patch could have only one characteristic in each weighting criterion: a single habitat type, a single general structural category, and a single aesthetics category. Based upon information in the general ecological literature (e.g., Gilbert 1989, Rebele 1994) and also experience of habitats in an urban context, these were then determined to be either quality (giving a weight of 2) or non-quality (giving a weight of 1). The ten general habitat structure categories were split into high and low quality with categories 1–5 (artificial construction to closed herbs and grasses) considered as low quality (weight ⫽ 1) and categories 6 –10 (open scrub to water feature) considered as high quality (weight ⫽ 2). For aesthetics, the categories 0 –1 were considered as non-quality (weight ⫽ 1) and category 2 as quality (weight ⫽ 2). The 104 specific habitat classification categories were examined individually and determined to be either quality, e.g., seminatural broadleaved woodland and scrub, or non-quality, e.g., industrial buildings. For the general classification used here, categories 1– 6 were determined to be quality and 7–10 non-quality. Combination of Scores and Weights to Derive the HVI Within each patch the indicator species scores and the structural elements scores were multiplied by the


Table 3.

379

Score classes and points scores for indicator species and habitat structural elements

Structural elements score classes

Indicator species score classes

0 elements ⫽ 0 points 1–4 elements ⫽ 1 point 5–8 elements ⫽ 2 points 9–12 elements ⫽ 3 points 13–16 elements ⫽ 4 points 17–20 elements ⫽ 5 points ⬎20 elements ⫽ 6 points

0 indicator species ⫽ 0 points 1–4 indicator species ⫽ 1 point 5–8 indicator species ⫽ 2 points 9–12 indicator species ⫽ 3 points 13–16 indicator species ⫽ 4 points 17–20 indicator species ⫽ 5 points ⬎20 indicator species ⫽ 6 points

Figure 2. Criteria weighting and score combination methodology.

total of simple quality weights to derive the HVI (Figure 2). From these values an image of the spatial distribution of the HVI could then be displayed. The values of the HVI are in themselves notional, but they allow quantitative comparisons to be made between patches, sites, and whole urban landscapes and can indicate, again quantitatively, the predicted extent of improvement or deterioration of habitat quality as land-use changes are implemented. Prediction The simple approach to prediction we adopted takes information about the type of habitat change antici-

pated from the implementation of management plans and overlays any changes in patch boundary and patch types. Real-world criteria totals were then applied to these changed areas and compared with the original situation. The totals were taken from the existing database of habitats located within the study area (877 patches for the four study grid squares), ensuring values typical of real habitats found in the local area. Where different habitat types are being envisaged for which data are not available, values can be quickly generated from visits to sites elsewhere that are similar to the end point of the proposed habitat alteration. The values were then converted to the assessment criteria

380


Figure 3. SJ 9000 negative (area A) and positive (area B) habitat change. Specific habitat areas mentioned in the text include: canal (C), racecourse (Rc), factory (F), and railway marshalling yards (R). For the original habitat distribution see Figure 1.

scores and weighted accordingly to generate the HVI for the new habitat patches. Prediction of the effects of habitat alteration is illustrated with two hypothetical examples based on different areas within one of the study area grid squares, SJ 9000. Area A (Figure 3) shows the location of the negative development that covers a structurally diverse and species-rich tall herb community that has arisen on abandoned allotments in the northeast section of the grid square. This new industrial area links into the existing small-scale industrial units and railway marshalling yards and consists of buildings, grounds, and roadways. The positive example (Figure 3, area B) shows the planting of a wedge-shaped area of amenity broadleaf woodland immediately to the south of the racecourse. This planting on what is poor semi-improved grassland has two main benefits: first, it expands and enhances the small concentration of existing semi-natural scrub, grassland, and tall herb patches in the immediate vicinity; and second, it provides both a visual and noise screen for the areas next to the racecourse.

Results Negative Change For area A, the two assessment criteria, structural elements (Figure 4) and indicator species (Figure 5), showed a sharp reduction in value from the original data. The structural elements totals fell from 11 to

values of 6 and 4 for the road and industrial area, respectively. Indicator species showed an even larger decline from two patches of 12 and 7 to two patches with values of 3 for the road and 1 for the industrial areas. General habitat structure (Figure 6) showed the largest change in any weighting criterion from a predominantly open scrub community to bare ground and construction. In contrast, aesthetics (Figure 7) showed only a small change from largely neutral with a small negative area, to a wholly negative area. In combination with the change in habitat type, the net effect on the HVI (Figure 8) was to produce a decrease from an area with a predominant HVI of 30 to a series of small patches with HVIs of mostly 6 and 9. In addition to the large decline, there was also fragmentation of the tall herb area into several smaller patches. The display allows the context of the landscape change to be seen. It is here that the influence of the new patches is most obvious, as the area had originally provided a patch of quality habitat on the margins of the built environment, as well as acting as a wide link between the canal corridor and the thin sliver of woodland. This has now been removed, hindering dispersal and splitting the single, relatively large corridor into two significantly narrower corridors. Therefore, the rough tall herb area has been lost not only as a valuable area in its own right but also as a component of the wider landscape.


381

Figure 4. Structural elements: (A) original, and (B) after changes.

Positive Change The two assessment criteria showed different responses to the positive change. The structural elements total (Figure 4) actually decreased from 7 to 5, but this gave no change in score as both values fell within the score class 2 (5– 8 structural elements). In contrast, the indicator species total (Figure 5) increased from 6 to 10, with a consequent increase in scores from 2 to 3. Although in terms of the general habitat classification

weightings there was no change, the overall effect was the introduction of a quality habitat into an area previously occupied by a non-quality habitat. The other two weighting criteria both showed this change with general habitat structure (Figure 6) changing from closed herbs and grasses (weight ⫽ 1) to an open scrub community (weight ⫽ 2), and aesthetics (Figure 7) changing from neutral (weight ⫽ 1) to positive (weight ⫽ 2). The overall effect on the HVI (Figure 8) is a change

382


Figure 5. Indicator species: (A) original, and (B) after changes.

from a low score of 12 to a relatively high score of 30. This compares favorably with the highest HVI score of 42 found along the canal and in the small patch of scrub in the southeast corner. The new patch is a high value habitat in its own right at the center of the grid square where these habitats are otherwise lacking. It also links into the lower-scoring, but still quality, area behind the factory and fringing the southern margins of the racecourse as well as the residential area to the

southwest. As with the negative example, the context is all-important since there is now a high-value area where previously there was none. This links with the adjacent lower-scoring HVI patches providing an important source and sink for flora and fauna in the immediate neighborhood. The location of the new high HVI area also provides a focal point from which possible further sympathetic management could be encouraged to link the low-scoring racecourse to the canal corridor.


383

Figure 6. General habitat structure: (A) original, and (B) after changes.

Discussion In this study predictions involve a perimeter-constrained or at-a-site approach, i.e., changes within a set area. This contrasts with the range expansions and contractions of specific species or habitats which has been the primary application of GIS to aid ecological prediction (e.g., Pereira and Duckstein 1993, Schippers and others 1996). Prediction in these cases examined

the effects of habitat change on animal populations. The HVI approach identifies and predicts the spatial adequacy of habitat provision which underlies movements of populations. The important feature of this method is that the magnitude of the loss or gain from the within-patch changes can be identified along with the effect this has on the wider landscape. The examples we have used here concentrate on the 1-km2 scale but changes could

384


Figure 7. Aesthetics: (A) original, and (B) after changes.

be viewed at different spatial scales depending on the requirements of the individual study. For example sitebased studies could view the individual patch and its immediate surrounds while a larger development could be viewed in the context of an area-wide assessment. Whatever scale is required, the results can be clearly visualized and the data itself can be easily extracted from the GIS database. The impact of changes on both the individual crite-

ria and the overall HVI can be easily seen. Large relative spatial and temporal changes are obvious and can be identified quickly from the GIS-generated imagery without the need for knowledge of precise totals, although these can be quickly accessed from the database if required. A key feature is flexibility since the limitations of what assessment or weighting criteria could be used are restricted only by the aims of the individual study. For


385

Figure 8. Habitat value index: (A) original, and (B) after changes.

example, in terms of indicator species, the list used here is targeted at the West Midlands but in other regions different species dominate (Gilbert 1989, Hodge and Harmer 1996), and where appropriate

other weighting criteria, e.g., rarity of habitat or presence of a rare species, could be applied instead of, or in addition to, those outlined previously. Care should be taken, however, that not too many criteria are used,

386


otherwise the resultant index becomes more complex and the impact of changes in individual criteria becomes diluted. From an administrative perspective many offices already possess the appropriate levels of resources necessary to undertake this simple predictive modeling, therefore implementation of this approach could be undertaken at a minimal cost. Most interested bodies already use, or have access to, appropriate software and base maps, and while the use of aerial photographic interpretation reduces the time spent in the field, the fieldwork aspects, including the mapping, can be undertaken solely in the field. For example, the HVI has been applied to a 400-m ⫻ 400-m site-based case study outside the 2-km ⫻ 2-km study area allowing the production of relevant baseline GIS images in less than a day starting from a blank map with just the bare site outlines (Young 1999), therefore providing a quick, spatially complete, directly comparable survey where traditional methods would have required substantial field studies. Importantly, this level of technology and the scale of application are such that the process is accessible to small groups or even individuals for whom even single patches of habitat have a range of roles (Harrison and others 1987). There are many potential applications of this type of spatially complete, comparative survey and modeling. An application currently being investigated is using the results of the HVI to direct survey effort and place detailed botanical quadrat studies in their local context. Other obvious uses are in urban land-use planning, natural areas planning, site-specific management, and directing application of resources, e.g., tree planting schemes. Where proposed habitat alterations are known, alternative scenarios could be developed to take account of more sympathetic development with the particular aim of increasing structural element diversity, species diversity, and quality habitat connectivity. Likewise different stages of habitat development could be used to project changes at different points in time, thereby guiding the temporal as well as spatial aspects of habitat management. Using the rapid updating ability of the GIS, different scenarios could be incorporated from different user groups, allowing these scenarios to be modified or viewed in parallel throughout the development discussion process but without recourse to more complex or costly modeling processes.

ical boundaries due to their location within the landscape. This combination of intrinsic and contextual value gives added importance to even the smallest patch, a feature that is compounded further in urban areas where the day-to-day interaction of people and the environment is at its most intense. Patch values will inevitably change as land-management occurs; therefore a spatially complete but simple modeling approach, such as the HVI demonstrated here, can ensure that land-use or habitat change can take place in an informed manner with appropriate compensatory or remedial action planned and visualised from an early stage.

Acknowledgments This research was funded by the School of Applied Sciences, University of Wolverhampton. We are particularly grateful to Prof. Ian Trueman and Dr. Ian Hooper for assistance in clarifying many of the details involved in this study.

Literature Cited Ahern, J. 1991. Planning for an extensive open space system: Linking landscape structure and function. Landscape and Urban Planning 21:131–145. Baker, W. L. 1989. A review of models of landscape change. Landscape Ecology 2(2):111–133. Bolger, D.T., T.A. Scott, and J.T. Rottenberry. 1997. Breeding bird abundance in an urbanizing landscape in coastal southern California. Conservation Biology 11(2):406 – 421. Canters, K.J., C.P. den Herder, A.A. de Veer, P.W.M. Veelenturf, and R.W. de Waal. 1991. Landscape ecological mapping of the Netherlands. Landscape Ecology 5(3): 145–162. Dale, V. H., A.H. King, L.K. Mann, R.A. Washington-Allen, and R.A. McCord. 1998. Assessing land-use impacts on natural resources. Environmental Management 22(2):203– 211. Daniels, R.E. 1988. The role of ecology in planning: some misconceptions. Landscape and Urban Planning 15:291– 300. Dickman, C.R. 1987. Habitat fragmentation and vertebrate species richness in an urban environment. Journal of Applied Ecology 24:337–351. Flather, C.H., S.J. Brady, and D.B. Inkley. 1992. Regional habitat appraisals of wildlife communities: A landscape-level evaluation of a resource planning model using avian distribution data. Landscape Ecology 7(2):137–147.

Conclusions

Gilbert, O.L. 1989. The ecology of urban habitats. Chapman & Hall, London, 369 pp.

The results of this study show that individual patches have a value that extends beyond their phys-

Harrison, C., M. Limb, and J. Burgess. 1987. Nature in the city—popular values for a living world. Journal of Environmental Management 25:347–362.


387

Hodge, S.J., and R. Harmer. 1996. Woody colonization on unmanaged urban and ex-industrial sites. Forestry 69(3): 245–261.

Sisinni, S.M., and M. O’Hea-Anderson. 1993. Methods and results of natural resource assessments in New York City, New York. Landscape and Urban Planning 25:95–114.

Merrill, T., R.G. Wright, and J.M. Scott. 1995. Using ecological criteria to evaluate wilderness planning options in Idaho. Environmental Management 19(6):815– 825.

Sukopp, H., W. Kunick, and C. Schneider. 1980. Biotope mapping in the built-up areas of West Berlin II. Garten und Landschaft 80(7):565–569.

Nature Conservancy Council. 1990. Handbook for phase-1 habitat survey—a technique for environmental audit. N.C.C., Peterborough, UK.

Van Der Meulen, F., J.V. Witter, and S.M. Arens. 1991. The use of a GIS in assessing the impacts of sea-level rise on nature conservation along the Dutch coast: 1990 –2090. Landscape Ecology 6(1/2):105–113.

Pereira, J. M. C., and L. Duckstein. 1993. A multiple criteria decision-making approach to GIS-based land-suitability evaluation. International Journal of GIS 7(5):407– 424. Rebele, F. 1994. Urban ecology and special features of urban ecosystems. Global Ecology and Biogeography Letters 4:173–187. Schippers, P., J. Verboom, J.P. Knaapen, and R.C. van Apeldoorn. 1996. Dispersal and habitat connectivity in complex heterogeneous landscapes: An analysis with a GIS-based random walk model. Ecography 19:97–106.

Villa, F., M. Ceroni, and A. Mazza. 1996. A GIS-based method for multi-objective evaluation of park vegetation. Landscape and Urban Planning 35:203–212. Wittig, R., and K-F. Schreiber. 1983. A quick method for assessing the importance of open spaces in towns for urban nature conservation. Biological Conservation 26:57– 64. Young, C.H. 1999. Greening the city: habitat evaluation in the Black Country. Unpublished PhD thesis. Wolverhampton University, Wolverhampton, UK, 190 pp.