5.4 Scale in ecological management
Scale is a rather complex term with many different meanings. We can talk of ‘map scale’ (e.g. 1:50,000), which is the ratio of a distance unit on the ground to a distance unit on a map. Scale can also mean the extent of a study area (e.g. a 20 by 20km area or the area within a national park boundary). More generally, scale is often used to refer to the level of spatial detail used to conduct an analysis. This is the definition we will use in the following discussion. Scale may be expressed in terms of raster grid size: grids may be composed of 20 by 20 metre pixels at a fine scale or 50 by 50 metre pixels at a coarser scale. We can also talk of vector data scales: a 1:25,000 vector map layer of a river network will contain more detail than a 1:100,000 map layer.
Why scale matters
Many of the GIS operations that are used to support environmental decision-making are affected by scale – hence its importance in ecological management. An analysis conducted at one scale will not necessarily produce the same results as one conducted at a different scale. This affects many types of GIS operation commonly used in ecological management:
- Measuring perimeters, areas, and lengths: Imagine we wish to measure the total length of a river network. We might find the total length is 57.3km based on a 1:100,000 scale vector map layer of rivers, but 58.2km based on a 1:25,000 scale vector map layer. In other words, the length of the river network appears greater at more detailed scales.
- Measuring landscape patterns: Imagine we wish to calculate relative richness (a measure of the number of different habitat types present in an area) from a raster grid depicting land cover. If we calculated relative richness using a 9 by 9 pixel window, we would probably find slightly higher relative richness values than if we used a 3 by 3 pixel window for the calculation.
- Comparing two or more map layers: Imagine we were to examine the relationship between night-time temperature and the distribution of a lizard species for a habitat suitability model. We may find a significant relationship between temperature and the presence of the lizard using 100 by 100 metre raster grids as the basis for our investigation, but not using 1km by 1km raster grids.
In short, the results of a great many GIS operations in ecological management can be influenced by the scale of analysis used.
Measuring the effects of scale
What can be done about this? One answer is to carry out a piece of analysis at many different scales and then measure how the results are influenced by changes in scale. For example, we could investigate the perimeter of a vegetation patch using different resolutions of raster grid, to see if the pixel size of the raster grid influences our estimate of its perimeter. The advantage of doing analysis at several different scales is that we can find out whether our results only apply at one particular scale, or whether they hold true no matter what the scale of our analysis.
The concept of fractals is often used as a theoretical basis for this type of investigation. A fractal object is one which contains the same patterns at different levels of magnification, a property sometimes known as self-similarity. The high-tide line around an island may show the same pattern of gradual variation, whether viewed for the whole island or magnified and viewed for just a small portion of the island. It is possible to calculate the fractal dimension of a map feature’s properties, such as the perimeter of a polygon or the length of a line (see illustration below).
Apart from fractals, there are other techniques that have been used to look at how scale affects the result of GIS analysis. For example, a measure known as lacunarity has been used to assess patterns of habitat suitability and gaps in tree canopy at different scales.
Which scale is best?
There is no simple answer as to which scale is the most appropriate for a given analysis. When considering a single species, many GIS analysts working in ecology would suggest that the scale should be selected based on how that species perceives its environment. If an average fox territory is about 4 square km, then an appropriate scale for analysing fox habitats may be to use 2 by 2km grid squares. In contrast, for a beetle, an appropriate scale of analysis may be to assess 1 by 1 metre grid squares.
Q. How useful do you think this advice is in practice?
In practice, it is not always possible to choose an analysis scale in this way. The only data available may be at quite coarse scales and the management problem being addressed may affect many different species, not just one. Furthermore, it may not be obvious exactly which scale is most useful for a given species if its behaviour changes over its life cycle. In short, scale remains a thorny problem in most environmental uses of GIS.
References (Essential reading for this learning object indicated by *)
For an example of a GIS-based analysis of some species distributions that looks at the effects of scale, see this article: Collingham, Y. C., Wadsworth, R. A., Huntley, B., and Hulme, P. E. (2000) Predicting the spatial distribution of non-indigenous riparian weeds: issues of spatial scale and extent. Journal of Applied Ecology 37 s1, 13-27