The question of scale

The problem of scale is central in ecology (Levin, 1992). Ecological processes occur and, hence, can be analysed on different scales of time and space. Spatial and temporal scales can be considered as hierarchically structured whereby both scales are positive correlated (Delcourt et al., 1982). For instance, roe deer (Capreolus capreolus) distri-bution can be analysed over the whole area of Europe. Analysis of changes over such broad regions make sense rather on a larger temporal scale, such as years, decades or centuries. Dynamics of large-scale roe deer distribution may have been caused by re-lated longer-term changes in vegetational, soil or climatic conditions. Such long-term dynamics in landscape conditions are appropriately observed on a large spatial scale. In contrast, dynamics that influence roe deer distribution in the Bavarian Forest National Park, a small region in Central Europe, may be rather observed on a seasonal, weekly, diurnal or even hourly scale. In order to relate environmental conditions to such small spatial patterns of roe deer distribution, landscape must be characterised on a much smaller spatial scale e.g. in terms of vegetation structure.

Besides the hierarchy and the positive correlation of temporal and spatial scales, scale has two related aspects which are extent and grain. The extent corresponds to the domain of interest, usually it is the size of study area. The grain is the minimum resolution of the data.

The recognition of the importance of scale greatly influenced the analysis of habitat

Figure 1.1: The analysis of habitat selection of animals is a matter of scales. Johnson (1980) structured the selection process in four hierarchical levels.

Figure 1.2: Spatial and temporal scales in habitat selection are highly correlated. Data can be provided on high or low resolution on spatial and temporal scale. The resolution of data for an analysis must correspond to the research question. At very small scales the energy intake of an individual can be captured, which would not be possible at broad scales. On this scale, spatial dynamics of a population or species can be studied. (Source:

Gaillard et al., 2010)

Johnson (1980) the selection process can be ordered in four hierarchical levels (Fig.

1.1). The first-order selection occurs on the largest range, this is the geographical space a species occupy (In Fig. 1.1 it is limited to the study area.). Selection of second-order is the choice of home range of an individual or a group within the above mentioned geographical space. Third-order selection is the choice of a site within the home range and fourth-order selection the actual utilization of an item (resource) at that site. Wiens (1989) adapted the positive correlation of spatial and temporal scales and emphasized that Johnson’s spatial levels are directly linked to temporal levels (Fig.

1.2). Hence, the four levels can be transferred to the temporal scale: species persistence

> population/individual persistence > feeding/resting period > handling time. This spatio-temporal hierarchy helps to clarify on which level a study intends to provide new insights. It further gives implication of the degree of resolution (grain) the data needs to have to be useful for answering the research question. The resolution of data can vary in time (e.g. time-intervals of GPS-location) and space (e.g. grain size of a map). The optimal scale on which habitat selection should be studied depends on the question the researcher has (Boyce, 2006). If individual behaviour is the scope of the study the resolution of the data needs to be small (Fig. 1.2). If the interest lies in general patterns of the population, resolution can be broader. The question of whether there is an “optimal” scale has not fully answered yet and is related to finding the characteristic scales at which resources most restrictively affect an species’ fitness (Mayor et al., 2009).

then, studies on habitat selection are usually conducted on multiple scales (Mayor et al., 2009). Most studies concluded that results can not directly up” or “scaled-down” between different scales because of scale-dependent behavioural responses to environmental factors (Bissonette, 1997; Boyce, 2006; Moreau et al., 2012). For example, there has been found non-linear differences in the effect of factors between different scales (Johnson et al., 2002). Consequently, habitat selection on a single scale can not be transferred to another scale. A comprehensive insight can be obtained when analysing habitat selection on multiple scales to define the relevant factors for each scale.

Rettie and Messier (2000) suggested that the factors of different scales also operate hierarchically in habitat selection, whereby factors of larger scales constrain selection on smaller scales. The hierarchy theory is intuitive as habitat selection varies with landscape features that are usually spatially nested in a hierarchical form. For exam-ple, climate and geographical conditions constrain landscape features on a very large scale, whereas availability of vegetation types influence habitat selection on a rather small scale. Due to the nested structure of the effects of factors on scales, factors that operate on a large scale are assumed to operate on each successive smaller scale, also.

Consequently, limiting factors that affect species’ fitness should be avoided at coarse spatial scale, as these operate impairingly on each successive smaller scale. However, Mayor et al. (2009) argued that studies on habitat selection not necessarily support this

“top-down” view of selection. Also, a “bottom-up” view is possible, when selection on a small scale entail changes in patterns on large scale.

Concluding, results of habitat selection studies must be interpreted in the light of the focal scale only. A study on multiple scales provide a more comprehensive insight to movement behaviour of the animals. The determination of the scale involves the determination of the extent and grain of the data, e.g. the landscape representation.

In the next section the necessity of a sensible choice of the right scale is motivated, as it has direct implications for the definition of availability. The definition of availability, in turn, is related to the evaluation of use and selection.