3 Methods of landscape structure analysis

Use of Geographic Information Systems (GIS) is required to analyse landscape structure using landscape metrics. GIS is necessary due to the need to evaluate a large amount of spatial information (such as land use information, habitat types, soil types) and in order to overlay and intersect this information with other information, enabling the parameters of landscape structure to be calculated. In addition, spatial reference units (e.g., natural or administrative units or regular fishnets) are required. Only by overlaying georeferenced spatial data and computing partial complex mathematical formulas can landscape structures in large areas be analysed.

A number of specialised software programmes are now available for calculating landscape metrics. One of the first programmes to appear on the market was FRAGSTATS (McGarigal and Marks, 1995), followed by PatchAnalyst (Rempel, 2008) and V-LATE (Tiede).

As the data basis, land use data from official land use surveys or remote sensing data, especially for large areas (e.g., Groom et al., 2006), are often used. One of the main problems in analysing landscape structure is that the landscape elements need to be delineated and defined, which may be extremely difficult and arbitrary in some types of landscape. In reality, it is often not easy to delineate a landscape element because sometimes no clear line distinguishes a landscape element from a neighbouring element. Some authors have proposed considering the landscape as gradients (McGarigal and Cushman, 2005*; Bolliger et al., 2007), e.g., the transition zones between patches. Indeed, in most studies, the delineation of landscape elements is a simplification of reality, which depends on data source, scale and, ultimately, the interpreter.

The thematic and geometric resolution influences the results of analyses using landscape metrics (Baldwin et al., 2004; Castilla et al., 2009; Mas et al., 2010). It is crucial that the scale of investigation (Wiens, 1989, 386) and the spatial resolution of the data correspond to one another (Corry, 2005*, 606). Fine-scale or multi-scale methods may be more informative than those based on only one, or very coarse, scales. For example, in a study by Lawler and Edwards (2002, 242), a coarse resolution of 30 m and higher seemed to be insufficient for statements about bird habitats. The need for multi-scale studies is also illustrated by the fact that studies of habitats often provide different results at different scales for the same species (Corry, 2005, 606). The decision which information on land use classes should be included in the landscape structure analysis (thematic resolution) is dependent on the aim of the investigation. The information depth of the data, such as land use/land cover, should meet the necessary habitat use or habitat types for the investigated species. Sometimes the similarity of types of landscape element is also important for biodiversity at the landscape scale.

Another simplification of reality is that, in most cases of landscape structure analysis, the underlying relief is not considered. In GIS analyses and calculations of landscape metrics, only planimetric areas and distances are calculated. With extreme reliefs in particular, this can lead to differing results from calculations with “real” areas and distances (Hoechstetter et al., 2008; Jenness, 2004; Blaschke et al., 2004; Dorner et al., 2002).

These limitations should be borne in mind and viewed with caution when comparing results from different areas and studies. Care must be taken that the data sources, methods and scales are indeed comparable. The selection of landscape metrics as indicators must also ensure they reflect the real demands of the species under investigation (Dormann et al., 2004*, 70-71).