Chapter 1

What is a landscape?

A forest bird, a beetle and a wolf can live in the same valley and yet inhabit very different worlds. To the beetle, a road is an impassable desert; to the wolf it is a nuisance crossed in a few seconds. What they share is that where things are matters as much as how much of them there is. Landscape ecology is the science of that "where": of how the spatial arrangement of the environment shapes ecological processes, and of how those processes, in turn, reshape the arrangement.

In this chapter

• the origins and the central paradigm of landscape ecology;
• the patch–corridor–matrix model and its alternatives;
• two theories from biogeography that explain why spatial structure matters: island biogeography and metapopulations;
• why habitat depends on the organism;
• how a landscape becomes a map, and the kinds of maps Landschaft reads.

A science of spatial heterogeneity

The term Landschaftsökologie, landscape ecology, was coined by the German geographer Carl Troll in 1939, while interpreting aerial photographs: seen from above, vegetation, soils, water and land use formed mosaics whose parts clearly influenced one another (Troll, 1939). The name of this package comes from the same word, Landschaft.

For decades the field grew mostly in Europe, close to geography, land-use planning and the study of cultural landscapes. In the 1980s a second tradition emerged in North America, centred on spatial heterogeneity: how to quantify it, what causes it and what its consequences are (Risser et al., 1984; Forman & Godron, 1986; Turner, 1989). The two traditions have since converged (Wu & Hobbs, 2002), and the quantitative toolbox of this course comes mostly from the second one.

What, then, is a landscape? Not a fixed size. A widely used definition is an area that is spatially heterogeneous in at least one factor of interest (Turner & Gardner, 2015). A landscape can be a few square metres for a ground beetle and thousands of square kilometres for a migratory ungulate. The organism or process under study, not the observer, sets the scale (Wiens, 1989), an idea we develop in chapter 2.

Pattern, process and scale

The central paradigm of landscape ecology links three words (Turner, 1989):

• Pattern is the spatial arrangement of the landscape: the amount of each land cover and how it is distributed, broken up, clumped or interspersed.
• Process is what happens in it: the dispersal of seeds and animals, the spread of fire, pests and invasive species, the flow of water and nutrients, the colonisation and extinction of populations.
• Scale is the spatial and temporal dimension at which pattern and process are observed.

Pattern and process form a feedback loop. Pattern constrains process: a continuous forest lets a fire run while a mosaic of fields and woods stops it; a line of hedges channels the movement of small mammals. Process creates pattern: fire, logging, agriculture and urban growth are what break a forest into pieces in the first place. Both depend on scale: a landscape that looks fragmented at 10 m resolution can look continuous at 1 km.

Most of this course is about describing pattern with numbers and maps. Keep in mind why we do it: pattern matters because of the processes it controls. A metric that cannot be linked to a process is a number without an ecological meaning (Li & Wu, 2004).

Patches, corridors and the matrix

The most influential mental model of a landscape is the patch–corridor–matrix model (Forman & Godron, 1986; Forman, 1995):

Element	Definition	Example
Patch	a relatively homogeneous area that differs from its surroundings	a woodlot in farmland, a pond, a town
Corridor	a strip that differs from the land on both sides	a hedgerow, a riparian forest, a road
Matrix	the most extensive and connected element, which controls the dynamics of the landscape	the farmland around the woodlots

The model is simple, visual and useful: it names the elements whose size, shape, number and arrangement landscape metrics quantify, and it frames conservation questions in terms everyone understands (keep large patches, maintain corridors, soften the matrix).

It also has limits. It treats the landscape as sharp-edged pieces of homogeneous habitat in a hostile background, as islands in an ocean. Real landscapes often change gradually: tree cover thins out, grazing pressure varies continuously, the matrix itself offers food and shelter to many species. The gradient model (McGarigal & Cushman, 2005) and the variegation model (McIntyre & Hobbs, 1999) describe landscapes as continuous surfaces rather than patches. Landschaft covers both views: patch-based tools on categorical maps, and tools that work on continuous maps (percentage of tree cover, resistance to movement) where the gradient view fits better.

Two theories that explain why structure matters

Why should the arrangement of habitat affect how many species live in a landscape, or whether a population persists? Two theories from biogeography and population ecology give the classic answers.

Island biogeography

MacArthur and Wilson (1967) proposed that the number of species on an island is a dynamic balance between immigration of new species and extinction of those present. Immigration falls with the island's isolation (distance from the mainland); extinction falls with its area (larger islands hold larger populations, which are less likely to die out by chance). Large, near islands therefore hold more species than small, remote ones.

Conservation biologists soon read habitat fragments as islands in a sea of farmland, and derived rules for reserve design (large rather than small, close rather than far, connected rather than isolated). The analogy is imperfect: the matrix is not an ocean, it can be crossed, it supplies species of its own and it drives edge effects deep into the fragments (Laurance, 2008). But the two variables of the theory, area and isolation, remain the backbone of fragmentation research, and many landscape metrics measure one or the other.

Metapopulations

A metapopulation is a set of local populations in separate habitat patches, linked by dispersal (Levins, 1969; Hanski, 1998). Local populations go extinct from time to time; empty patches are recolonised by dispersers from occupied ones. The metapopulation persists as long as colonisation keeps pace with extinction, which depends on the size of the patches (extinction) and their connectivity (colonisation). A patch that is too isolated stays empty even when it is perfectly good habitat.

This is the theoretical root of the connectivity chapters of this course (6, 7, 8): the indices used to rank patches and links for conservation are, in essence, estimates of how much habitat an organism can reach.

Habitat is species-specific

"Habitat" is not a land-cover class. It is the set of resources and conditions that allow a given species to live and reproduce (Hall et al., 1997). A broad-leaved forest is habitat for a woodpecker, a barrier for a grassland butterfly and irrelevant for a generalist fox. Two consequences follow for every analysis in this course:

1. The habitat map is a hypothesis. Deciding which classes count as habitat, and for which species, is the first and most consequential decision of a study. Justify it with the natural history of the species.
2. One landscape, many patterns. The same land-cover map gives different patches, fragmentation and connectivity for different species, as the example below shows.

From a landscape to a map

Landscape analyses work on maps, and a map is itself a model: somebody decided the classes, the resolution and the extent, and a classifier or an interpreter assigned every cell to a class with some error. Errors in the map propagate into every metric computed from it, sometimes strongly (Langford et al., 2006). Know your map: its source, date, resolution, legend and accuracy.

Landschaft works on raster maps: grids of cells, each holding a number. Four kinds appear in this course (core concepts has the technical details):

Map	Cells hold	Used for	Chapters
categorical	a class code (land cover)	landscape metrics	5, 10
binary habitat	0 missing, 1 background, 2 habitat	patterns, fragmentation, change	3, 4, 9, 10
continuous	a quantity (% tree cover, resistance)	grayscale fragmentation, connectivity	3, 6, 8
patch	one id per habitat patch	habitat networks	7

Try it: one map, two species

The example land-cover map is a 50 × 50 km extract of CORINE Land Cover, a European map with 44 classes, at 25 m resolution:

import numpy as np
import landschaft as ls

landcover = ls.load_raster(ls.example_data("clc"))
landcover.shape, landcover.rio.resolution(), landcover.rio.crs

Which classes are present, and how much of the map does each cover?

codes, cells = np.unique(landcover.values, return_counts=True)
share = dict(zip(codes.tolist(), (100 * cells / cells.sum()).round(1).tolist()))
share

Half of the map is class 12 (non-irrigated arable land): an agricultural landscape. Forests (23 broad-leaved, 24 coniferous, 25 mixed) cover about 20%, pastures (18) about 7%.

Now build the habitat of two imaginary species. A forest bird lives in the three forest classes; a grassland insect in pastures (18) and natural grassland (26):

forest = ls.reclassify(landcover, foreground=[23, 24, 25])
grassland = ls.reclassify(landcover, foreground=[18, 26])

for name, habitat in [("forest bird", forest), ("grassland insect", grassland)]:
    table = ls.Landscape(habitat).class_metrics(["pland", "np", "area_mn"])
    row = table[table["class"] == 2].iloc[0]          # class 2 = habitat
    print(f"{name:17s} {row.pland:5.1f}% of the map in {row.np:.0f} patches "
          f"of {row.area_mn:.0f} ha on average")

The same landscape offers the forest bird three times as much habitat as the grassland insect, in patches almost three times larger. Every later analysis (fragmentation, connectivity, restoration) would reach different conclusions for the two species. pland, np and area_mn are landscape metrics: the percentage of the landscape, the number of patches and their mean area.

Three ways of seeing a landscape

The tools of Landschaft look at a landscape in three complementary ways, which organise the rest of the course:

View	The landscape is...	Questions	Chapters
cells and neighbourhoods	a grid; each cell is described by what surrounds it	Is this cell interior or edge? How dense is habitat around it?	3, 4
patches	a set of discrete objects with area, shape and position	How many patches? How large, complex, aggregated, diverse?	5
networks and flows	nodes linked by movement, or a surface that conducts it	How much habitat is reachable? Which patches and routes matter?	6–9

No view is the right one in general: each answers different questions, and a good study often combines them.

Key points

• Landscape ecology studies how spatial pattern affects ecological process, and vice versa, at scales set by the organism or process.
• The patch–corridor–matrix model names the elements that landscape metrics quantify; gradient models describe landscapes that change continuously.
• Island biogeography and metapopulation theory explain why the area and isolation of habitat, and hence its arrangement, matter.
• Habitat is species-specific: the habitat map is the first hypothesis of any study.
• Maps are models with errors; know their source, resolution and legend.

Check your understanding

1. A colleague says a 5 km² area "is too small to be a landscape". How would you answer?
2. Why is a patch of farmland not equivalent to an ocean for a forest species? Give two reasons.
3. In the example above, which species' habitat is more fragmented? Which single metric would you look at first, and why is it not enough?

Answers

1. A landscape is defined by heterogeneity relevant to the organism or process, not by size. For a small mammal or an insect, 5 km² contains many habitat patches and a complex matrix; it is a landscape.
2. The matrix can be crossed (with some cost or risk), and it is not neutral: it provides resources to some species, it is a source of predators, competitors and invasive species, and it causes edge effects inside the fragments.
3. The grassland insect has less habitat, in more patches relative to its area and in smaller ones. Mean patch area alone mixes two things: the amount of habitat and how it is subdivided. Chapter 3 explains why those two must be separated.

References

• Forman, R. T. T. (1995). Land Mosaics: The Ecology of Landscapes and Regions. Cambridge University Press.
• Forman, R. T. T., & Godron, M. (1986). Landscape Ecology. Wiley.
• Hall, L. S., Krausman, P. R., & Morrison, M. L. (1997). The habitat concept and a plea for standard terminology. Wildlife Society Bulletin, 25(1), 173–182.
• Hanski, I. (1998). Metapopulation dynamics. Nature, 396, 41–49.
• Langford, W. T., Gergel, S. E., Dietterich, T. G., & Cohen, W. (2006). Map misclassification can cause large errors in landscape pattern indices: examples from habitat fragmentation. Ecosystems, 9, 474–488.
• Laurance, W. F. (2008). Theory meets reality: how habitat fragmentation research has transcended island biogeographic theory. Biological Conservation, 141(7), 1731–1744.
• Levins, R. (1969). Some demographic and genetic consequences of environmental heterogeneity for biological control. Bulletin of the Entomological Society of America, 15(3), 237–240.
• Li, H., & Wu, J. (2004). Use and misuse of landscape indices. Landscape Ecology, 19, 389–399.
• MacArthur, R. H., & Wilson, E. O. (1967). The Theory of Island Biogeography. Princeton University Press.
• McGarigal, K., & Cushman, S. A. (2005). The gradient concept of landscape structure. In J. A. Wiens & M. R. Moss (Eds.), Issues and Perspectives in Landscape Ecology (pp. 112–119). Cambridge University Press.
• McIntyre, S., & Hobbs, R. (1999). A framework for conceptualizing human effects on landscapes and its relevance to management and research models. Conservation Biology, 13(6), 1282–1292.
• Risser, P. G., Karr, J. R., & Forman, R. T. T. (1984). Landscape Ecology: Directions and Approaches. Special Publication 2, Illinois Natural History Survey.
• Troll, C. (1939). Luftbildplan und ökologische Bodenforschung. Zeitschrift der Gesellschaft für Erdkunde zu Berlin, 241–298.
• Turner, M. G. (1989). Landscape ecology: the effect of pattern on process. Annual Review of Ecology and Systematics, 20, 171–197.
• Turner, M. G., & Gardner, R. H. (2015). Landscape Ecology in Theory and Practice: Pattern and Process (2nd ed.). Springer.
• Wiens, J. A. (1989). Spatial scaling in ecology. Functional Ecology, 3(4), 385–397.
• Wu, J., & Hobbs, R. (2002). Key issues and research priorities in landscape ecology: an idiosyncratic synthesis. Landscape Ecology, 17, 355–365.