As a result of this type of phenomena, the dynamics of
natural populations depend on the degree of environmental
heterogeneity, i.e., the rate at which key environmental properties
change in space. This is called habitat fragmentation
[68–70] and can be quantified through several ad hoc parameters.
A usual procedure is subdividing a landscape image in
square kernels of 9, 25 . . . 49, etc., pixels and computing a
fragmentation index (FI) of the type
FI 5 (n 2 1)/(c 2 1)
where n is the number of different classes present in kernels
of c pixels considered (9, 25 . . . 49, etc). In the frame of a
study on the spatial distribution of pesticides at the lower
Colorado River basin in Argentina [71], we inspected a thematic
mapper satellite image of the Ascasubi area, where farming
activities started early last century (Fig. 2). Pioneering
farmers selected the best land to start activities, and the successive
occupation of surrounding areas also followed this
trend. High-quality land was subdivided to a greater extent
than poor-quality land in successive farmer generations. Also,
farming is more intensive in highly valued and highly fragmented
land (Fig. 2a, lower left) than in lower quality areas (Fig. 2b, upper right), where activities are more extensive.
Pesticide applications in older farming areas occur on smaller
patches and at higher and more frequent doses than in larger
farming units at the newer areas.