ArticlePDF Available

Ecological classification and mapping for landscape management and science: Foundations for the description of patterns and processes


Abstract and Figures

There is growing demand for biogeographical landscape classifications and ecological maps that describe patterns of spatially co-varying biotic and abiotic ecosystem components. This demand is fuelled by increasing data availability and processing capacity, by institutional practices of land and water resource management and planning and by the growth of transdisciplinary science that requires the development of a shared conceptual framework through which to view landscape character and behaviour. Despite the widespread use of ecological maps, and the extent to which they have become embedded in institutional practice, policy and law, no standard approach to ecosystem mapping has emerged, such that there are many valid ways of mapping the same landscape. Consensus is possible only when there is agreement on the spatial entities to be mapped. We propose a way of defining such entities and identifying them in any given landscape. Landscapes are conceived in terms of a conceptual biophysical template that constrains a wide range of ecological processes at various hierarchical levels. The template is conceived as comprising co-evolved associations of soils, vegetation, topography and hydrology that form a dynamic mosaic characteristic of a particular topographic, climatic and geological context that is continually being shaped by many perturbations. We synthesise themes from vegetation, soil and river sciences, using hierarchy theory to frame a perspective that facilitates the definition of mappable landscape entities at three hierarchical levels of organisation. These entities are conceived as archetypal structural-functional units, with form and process linked in conceptual models that underpin each archetype. We describe how our approach has been used to map ecological entities in Kruger National Park, South Africa, showing how the proposed framework integrates key system components, providing transparent foundations for transdisciplinary approaches to landscape management and science.
Content may be subject to copyright.
Ecological classification
and mapping for landscape
management and science:
Foundations for the description
of patterns and processes
Carola Cullum
University of the Witwatersrand, Johannesburg, South Africa; University of Auckland, New Zealand
Kevin H. Rogers
University of the Witwatersrand, Johannesburg, South Africa
Gary Brierley
University of Auckland, New Zealand; University of the Witwatersrand, Johannesburg, South Africa
Ed T.F. Witkowski
University of the Witwatersrand, Johannesburg, South Africa
There is growing demand for biogeographical landscape classifications and ecological maps that describe
patterns of spatially co-varying biotic and abiotic ecosystem components. This demand is fuelled by increasing
data availability and processing capacity, by institutional practices of land and water resource management
and planning and by the growth of transdisciplinary science that requires the development of a shared
conceptual framework through which to view landscape character and behaviour. Despite the widespread
use of ecological maps, and the extent to which they have become embedded in institutional practice, policy
and law, no standard approach to ecosystem mapping has emerged, such that there are many valid ways of
mapping the same landscape. Consensus is possible only when there is agreement on the spatial entities to be
mapped. We propose a way of defining such entities and identifying them in any given landscape. Landscapes
are conceived in terms of a conceptual biophysical template that constrains a wide range of ecological
processes at various hierarchical levels. The template is conceived as comprising co-evolved associations of
soils, vegetation, topography and hydrology that form a dynamic mosaic characteristic of a particular
topographic, climatic and geological context that is continually being shaped by many perturbations. We
synthesise themes from vegetation, soil and river sciences, using hierarchy theory to frame a perspective that
facilitates the definition of mappable landscape entities at three hierarchical levels of organisation. These
entities are conceived as archetypal structural-functional units, with form and process linked in conceptual
models that underpin each archetype. We describe how our approach has been used to map ecological
entities in Kruger National Park, South Africa, showing how the proposed framework integrates key system
Corresponding author:
Carola Cullum, School of Environment, University of Auckland, Private Bag 92019, Auckland, New Zealand.
Progress in Physical Geography
2016, Vol. 40(1) 38–65
ª The Author(s) 2015
Reprints and permission:
DOI: 10.1177/0309133315611573
at The University of Auckland Library on January 30, 2016ppg.sagepub.comDownloaded from
components, providing transparent foundations for transdisciplinary approaches to landscape management
and science.
archetypes, ecological mapping, ecosystem mapping, landscape classification, landscape hierarchy
I Introduction
Most landscapes are highly organised, with reg-
ularities in time and space that occur at many
scales. Indeed, the spatial co-variance of envi-
ronmental attributes such as climate, soil, biota
and topography is routinely exploited for a wide
range of purposes. Soil surveyors use terrain and
vegetation as indicators of soil patterns (e.g.
Moore et al., 1993), whilst hydrological model-
lers use topography and soils to infer water
flows (e.g. Beven and Kirkby, 1979). Ecologists
and vegetation scientists also recognise the cou-
pling of climate, vegetation, soil and topogra-
phical patterns in their efforts to map species
distributions, plant communities and resources
available to organisms or to model fluxes
of materials such as carbon or nutrients (e.g.
Austin, 2002; Franklin, 1995; Pedrotti, 2013;
Porporato et al., 2003).
Increasing data availability and processing
capacities have fuelled demands for biogeogra-
phical landscape classifications and ecological
maps that describe these patterns of spatially
co-varying biotic and abiotic ecosystem compo-
nents. These spatial clusters of ecosystem com-
ponents are also often conceived as interacting,
mutually dependent and co-evolved complexes
that form discrete ecosystems (Rowe, 1961;
Wiken, 1986). Hence this type of ecological
mapping is sometimes called ‘ecosystem map-
ping’ (e.g. Bailey, 1996; Pedrotti, 2013; Sayre
et al., 2009). Land units placed in the same class
are assumed to be ‘ecologically equivalent’,
supporting similar suites of species, sustained
by the same dominant processes and likely to
respond in similar ways to management initia-
tives and climate or other environmental changes
(e.g. Bailey, 1983; Loveland and Merchant,
2004; McMahon et al., 2004; Olson et al.,
2001; Omernik, 1987; Sayre et al., 2009). This
ecological equivalence justifies the selection
of individual sites for assessment, monitoring,
experimentation and/or measurement and the
subsequent extrapolation of results to all areas
within the same ecological class (MacMillan
et al., 2003).
Such maps and classifications are now
widely used both to inform land and water
resource management and planning as well as
to provide contextual information for scientific
research projects. For example, at global scales,
biogeographical ‘ecoregions’ that contain dis-
tinct communities of flora and fauna have been
identified to guide the setting of conservation
priorities (e.g. Abell et al., 2008; Olson et al.,
2001; Udvardy, 1975). At regional and local
scales, ‘habitat’ maps are used in conservation
management to stratify biodiversity inventories,
condition assessments and monitoring, whilst at
national, continental and global scales they are
used to inform conservation and land-use plan-
ning (e.g. South Africa: Driver et al., 2005;
Kleynhans et al., 2005; New Zealand: Singers
and Rogers, 2014). In many parts of the world,
these ecological maps are entrenched in envi-
ronmental management practices and policies
and are used to demonstrate compliance with
policies, standards and laws (MacMillan et al.,
2003). For example, in Europe, habitat maps
form the basis of the Natura 2000 network of
sites protected under the 1992 Habitats Direc-
tive (see
legislation/habitatsdirective/index_en.htm EEA,
Cullum et al. 39
at The University of Auckland Library on January 30, 2016ppg.sagepub.comDownloaded from
Despite the widespread use of ecological
maps, and the extent to which they have become
embedded in institutional practice, policy and
law, no standard approach to ecosystem map-
ping has emerged (e.g. McMahon et al., 2004;
Omernik, 2004). First, the environmental attri-
butes and scales of observation that are used
to define ecological units vary between different
regions (e.g. Omernik, 2004) such that it does
not make sense to formulate a standardised or
comprehensive taxonomy of ecoregions. Sec-
ond, although there is general consensus on the
types of environmental factors deemed relevant
(topography, climate, vegetation, lithology),
there are numerous metrics available to describe
each of these dimensions and many ways of
combining them (e.g. Shary et al., 2002). Each
combination of these metrics and methods can
generate different findings (e.g. Behrens et al.,
2010; Deng and Wilson, 2006; Hazeu et al.,
2011; Rossum and Lavin, 2000). Indeed, land-
scape ecologists have demonstrated how the
patterns we perceive depend on the scale of
observation and the attributes selected, such that
there are many valid ways of mapping the same
landscape (e.g. Behrens et al., 2010; Buyan-
tuyev and Wu, 2007; Levin, 1992; Schneider
and Kay, 1994). Thus the definition and delinea-
tion of ecological classes is always contestable,
with boundaries that are far from self-evident
and which cannot be verified independently of
the process of map production.
The multiplicity of equally valid ecological
maps is a classic example of the Modifiable
Areal Unit Problem (MAUP) (Openshaw, 1984):
different patterns are perceived at different
scales of observation and when the same data
are aggregated in different ways. The MAUP
and associated issues are most easily resolved
when there is agreement on the entities that
are being mapped (Fotheringham, 1989; Open-
shaw, 1984). For example, if the mapped enti-
ties are individual trees, then the size and
characteristics of the trees informs appropriate
choices of image resolution and descriptive
variables. Since individual trees usually have
incontestable boundaries, different approaches
and methods can be evaluated using conven-
tional groundtruthing techniques (e.g. Congal-
ton and Green, 2008).
What, then, are the entities being described in
ecological maps? Many ecological mappers
claim to map ‘ecosystems’ following Rowe’s
geographical concept of an ecosystem, which
was subsequently developed by Wiken and
A perceptible ecosystem is a topographic unit, a vol-
ume of land and air plus organic contents extended
areally over a particular part of the Earth’s surface for
a particular time (Rowe, 1961, cited in Bailey, 1996).
Ecological land classification is a process of deli-
neating and classifying ecologically distinctive areas
of the Earth’s surface. Each area can be viewed as a dis-
crete system which has resulted from the mesh and
interplay of the geologic, landform, soil, vegetative,
climatic, wildlife, water, and human factors which may
be present (Wiken, 1986: 4).
Ecosystem: An area of any size with an association
of physical and biological components so organized
that a change in any one component will bring about
a change in the other components and the operation
of the whole system (Bailey, 1996: 167).
However, such geographically based concep-
tions of ecosystems have been seriously con-
tested, since they suggest that ecosystems are:
Discrete and easily identifiable. How-
ever, plant and animal species do not
necessarily form communities that occupy
areas with clear geographical boundaries
and interact only within these boundaries.
Similarly, fluxes of materials and energy
are not confined within discrete loca-
tions, but interact across space, time and
scales (e.g. Currie, 2011; O’Neill, 2001;
Perry, 2002). Unlike closed systems,
which have distinct boundaries, ecosys-
tems are always open, relying on inputs
of energy to hold them in positions that
are often far from equilibrium and so
40 Progress in Physical Geography 40(1)
at The University of Auckland Library on January 30, 2016ppg.sagepub.comDownloaded from
always interacting with the environment
in which they occur (e.g. Holling, 1973;
O’Neill et al., 1986). Furthermore, most
ecosystems are highly modified by humans,
such that many of the boundaries evi-
dent in many landscapes are not equivalent
to ecosystem boundaries (e.g. domesti-
cated landscapes and ecosystems (Kareiva
et al., 2007; Tockner et al., 2011)).
Either static or at equilibrium. Conver-
sely, ecosystems are in constant flux, in
response to stochastic disturbance events
that can potentially shift systems into a
new state (Scheffer et al., 2001; Turner
et al., 1993). In other words, ecosystems
are dynamic and are continually chang-
ing in space and time, so that the geogra-
phical boundaries of ecological units are
not fixed.
Contain interactions between system com-
ponents that are regular and predictable.
Ecosystems are now widely acknowl-
edged to be compl ex systems, charac-
terised by emergent properties and
behaviour that result from inherently
uncertain inter actions between system
components within and across multiple
scales (Funtowicz and Ravetz, 1993;
Holling, 2001; Kay et a l., 1999; Levin,
1998; Nowotny et al., 2001; Pickett and
Cadenasso, 2002; Ulanowicz, 2009).
There is now widespread consensus around
conceptualisations of ecosystems as open sys-
tems, with overlapping, indeterminable bound-
aries. Interactions within and between these
open systems operate over various spatial and
temporal scales. There are many instances in
which no key attributes can be isolated, no opti-
mal scales of observation can be determined and
neither functional nor geographical boundaries
can be drawn around discrete systems. Such per-
spectives suggest that there are no definitive
ways of delineating ecosystems. Thus it makes
little sense to conceive of ecosystems as self-
evident ‘entities’ that can be invoked to solve
the MAUP.
Rather than assuming that landscapes can
be decomposed into self-evident entities, we
acknowledge that all ecological classifications
and maps are based upon conceptualisations
of reality that are inherently contestable. Thus
agreement on ecological maps can only be
reached if users agree on the utility and credibil-
ity of the underlying conceptualisation (e.g.
Cash et al., 2003). It is therefore imperative that
the conceptualisations underpinning ecological
maps are transparent and clearly articulated,
allowing a consensus to be built on solid
In this paper we propose a conceptualisation
of landscapes that can be used to identify spatial
entities for ecological mapping that aims to
inform landscape management and science. The
conceptualisation is designed to:
Inform ecological landscape classifica-
tion and wall to wall mapping of land-
scape units, using remotely-sensed data
and geospatial analysis. Recognising that
ecological classifications are inherently
contestable, assumptions are articulated,
defended by a narrative that aims to offer
credibility and relevance to a wide range
of end users and purposes.
Integrate many system components and
processes, so that the conceptualisation
and resultant maps are relevant to a wide
range of applications, integrating the
perspectives of different agencies and
disciplines and thus avoiding fragmented
approaches to ecological science, man-
agement and policy making.
Provide a basis for extrapolating site
behaviour across time and space, so that
observations, experience and foresighted
responses from sample and monitoring
sites can be transferred t o other l oca-
tions. This requirement implies t hat the
conceptualisation must be process-based,
Cullum et al. 41
at The University of Auckland Library on January 30, 2016ppg.sagepub.comDownloaded from
relating observed features used for classi-
fication to the contemporary processes
that sustain them and determine their
future trajectory.
Be adaptable, so that spatial units can be
tailored to accommodate the various
scales and attributes characteristic of
individual landscapes.
We start by reviewing existing ways of con-
ceptualising landscapes, identifying ways in
which hierarchy theory, vegetation mapping,
soil mapping and river classification contribute
to our approach. We then describe our concep-
tualisations of a biophysical template, the hier-
archy used to define mappable entities and the
way archetypes can be developed and used to
describe particular landscapes.
Finally, we describe how our approach has
been used to map ecological entities in Kruger
National Park (KNP), South Africa. We demon-
strate how our proposed approach to ecological
mapping contributes towards the development
of a defensible basis for ecological mapping that
integrates key system components, providing a
solid foundation for integrated approaches to
landscape management and science.
II Conceptual frameworks for the
description of ecological patterns
1 Introduction
The conceptualisations of landscapes considered
below are all widely used to underpin wall to wall
landscape classifications in which the entire
landscape is partitioned into classes that can be
mapped at least semi-automatic ally wit hin a
GIS s ystem. All can take advantage of the
increasing availability of high r esolution satel-
lite data to capture environmental gradients,
reducing the nee d for exte ns ive fiel d surve ys
that are both expensive and impra ctical over
large areas. Lastly, they are all applied in land
and conservation management and planning.
We start with hierarchy theory, a general con-
ceptualisation that implicitly or explicitly underlies
many other approaches. This conceptualisation is
somewhat different from the other perspectives
reviewed, presenting an empty framework through
which landscapes can be viewed rather than sug-
gesting the entities, scales and attributes that can
or should be used to describe them.
We then consider framings that are used
within specific disciplines that aim to describe
and predict the spatial distribution of the key
system components: vegetation, soil, topogra-
phy and water.
2 Hierarchy theory
Hierarchy theory offers a heuristic framework
that can be applied to systems that are near-
decomposable, both vertically and horizontally,
into subsystems that each contain more or less
distinct clusters of components (Allen and Starr,
1982; Salthe, 1985). The framework suggests
that, at each vertical level of organisation, sys-
tem components can be decomposed into their
parts, which form separate subsystems at a
lower organisational level.
Different sets of processes dominate at each
organisational level within a given system,
resulting in different types of entities. This
means that different conceptual models, meth-
ods of observation and analytical tools are
appropriate at each level (e.g. Church, 1996).
Each level of organisation is associated with a
characteristic spatial and temporal scale domain
that is determined by the rates of the processes
that produce the observed patterns. Higher
levels are characterised by patterns evident at
coarse spatial and temporal scales, produced
by processes that operate at relatively slow
rates, whilst faster processes characterise lower
levels of organisation and are responsible for
the patterns observed at finer scales (Allen and
Starr, 1982; Salthe, 1985).
Interactions between system components are
constrained by both their vertical and their
42 Progress in Physical Geography 40(1)
at The University of Auckland Library on January 30, 2016ppg.sagepub.comDownloaded from
horizontal position in the hierarchy (Allen and
Starr, 1982; Figure 1). Vertical constraints oper-
ate both top-down, where higher-level contexts
constrain the range of variability occurring at
lower levels and bottom-up, where low-level
mechanisms are responsible for the patterns that
emerge at higher levels. Horizontal constraints
arise from relationships with other entities on
the same organisational level and include con-
straints arising from configuration and connec-
tivity (or lack of connectivity) (Fryirs, 2013).
Since these constraints favour some interactions
and outcomes over others, potential exists for
the development of a self-organising system and
recurring patterns that can be observed at higher
levels of organisation (see Favis-Mortlock,
2013, for an overview of self-organising sys-
tems). The recurrence of these patterns means
that they are somewhat predictable and can be
described in terms of general principles.
Viewing landscapes in terms of a conceptual
hierarchy that defines entities at multiple scales
of observation, each associated with different
hierarchical levels, helps to resolve the issues
of scale and attribute selection that otherwise
haunt the description of ecological patterns
(e.g. Fotheringham, 1989; Jelinski and Wu,
1996; Openshaw, 1984; Parsons and Thoms,
Figure 1. Illustration of the key concepts in hierarchy theory. Hierarchy theory proposes a heuristic
framework in which a system is considered to be near-decomposable in both vertical and horizontal
dimensions. Hierarchical relationships describe the relationships between entities within and between each
hierarchical level. For example, spatial entities at lower levels of organisation are contained within higher level
entities and are adjacent to other entities at the same level. The character and behaviour of entities at the
focal level of interest is contextually constrained, both horizontally by neighbourhood relationships and
vertically by top-down and bottom-up relationships. After Wu (1999).
Cullum et al. 43
at The University of Auckland Library on January 30, 2016ppg.sagepub.comDownloaded from
2007). Such conceptualisations also inform dis-
cussions about scaling up and down between
different levels, even though explicit rules for
such scaling may be elusive in many circum-
stances (e.g. Wu and David, 2002).
3 The classification of river systems
River systems are highly organised, containing
structures and patterns that are repeatedly
observed at multiple scales in many locations
(e.g. Cullum et al., 2008; Fonstad and Marcus,
2010; Grabowski et al., 2014). Relationships
between the area drained, the number of streams
(and hence stream density), stream length and
hillslope gradient are expressed in Hack and
Horton’s laws, with constants that are character-
istic of particular geoclimatic zones (Hack,
1957; Horton, 1945). Repeating broad-scale
patterns of hills and valleys result from long-
term fluvial erosion, forming drainage network
patterns that reflect the geology and climate of
a particular setting (Howard, 1967; Huggett,
At catchment scales, Schumm (1977)
described a general downstream transition from
a zone of sediment production in the headwater
region, through a zone dominated by sediment
transport to an accumulation zone along the
lower river reaches. The biological implications
of this physical gradient have also been articu-
lated, notably by Vannote et al. (1980) in their
‘River Continuum Concept’. These clinal pat-
terns are produced by downhill fluxes of water,
transporting sediment, nutrients and organic
material, all of which can be involved in multi-
ple, overlapping series of cascading interac-
tions. These patterns, and the processes that
generate them, are conceptualised by Petts and
Amoros (1996) as the ‘fluvial hydrosystem’,
which operates in all four dimensions and at
many scales (Thorp et al., 2006; see also Ward,
1989). Although deviations are often observed
(e.g. Junk et al., 1989; Montgomery, 1999;
Poole, 2002; Thorp et al., 2006), this idealised
pattern provides a useful starting point for the
description and classification of river systems.
In due course, the generalised pattern can be
modified as local circumstances are taken into
The systematic arrangement of river systems
outlined above underpins many conceptual fra-
meworks for process-based river classifications
(e.g. Bohn and Kershner, 2002; Brierley and
Fryirs, 2005; Frissell et al., 1986; Rogers and
O’Keefe, 2003; Rosgen, 1994; Thorp et al.,
2010; see also Tadaki et al., 2014). Such classi-
fications are now routinely used for a wide
range of management applications, including
the assessment and monitoring of water quality,
biodiversity and ecological integrity (Buffing-
ton and Montgomery, 2013; Melles et al.,
2012; Olden et al., 2012).
In river classification, structural-functional
units are typically delineated at scales associ-
ated with various organisational levels within
a nested hierarchy. These units are then grouped
into classes based upon observable characteris-
tics that are indicators of particular processes.
All these classifications show remarkable agree-
ment, both in how best to decompose river
systems vertically into different levels of organi-
sation and in the processes and controls that are
hypothesised to operate at each level (Cullum
et al., 2008). For example, most define a regional
level in which climate, geology and history have
shaped characteristic structures of the drainage
network (e.g. relief, valley spacing, network geo-
metry, etc.). Within individual catchments, river
segments are identified, contrasting headwater,
transitional and downstream zones in terms of
sediment production, transport and deposition and
resultant morphology. At lower levels of organisa-
tion, individual reaches are characterised in terms
of channel and floodplain geometry and the pres-
ence of particular geomorph ic features.
This is a classic example of how the adoption
of a hierarchy can help to tame issues of scale
and attribute selection by specifying the differ-
ent types of entity that are visible within
44 Progress in Physical Geography 40(1)
at The University of Auckland Library on January 30, 2016ppg.sagepub.comDownloaded from
different scale domains. Once these entities are
agreed, the selection of appropriate attributes
and mapping scales is clarified, depending as
much on the intrinsic nature of the entities as
on subjective decisions by the cartographer
(e.g. Fotheringham, 1989; Jelinski and Wu,
1996; Openshaw, 1984).
In fluvially incised landscapes, the drainage
network determines the spatial configuration
and morphology of hillslopes and hillslope
units. Thus conceptualisations and tools devel-
oped to describe and analyse the spatial organi-
sation of river systems are also pertinent to the
description of terrestrial systems, and, indeed,
entire landscapes.
4 Digital soil and landform mapping
Soil-landscape models are underpinned by Jen-
ny’s (1941) theory that identifies soil forming
factors as climate, relief, organic matter, parent
material and time (Huggett, 1975). Assuming
that climate, parent material and time are con-
stant, the model posits relationships between soil
distributions and hillslope morphology (Gerrard,
1992; Paton et al., 1995). Soil scientists describe
these patterns in terms of ‘catenas’ or ‘topose-
quences of co-varying soils, vegetation and
slope morphology (Huggett, 1975; Milne, 1935).
Soil surveyors are expert in mentally con-
structing soil-landscape models that describe and
explain the catenas observed in any given setting,
so that obser ved pat terns of ge omorphology
and vegetation can be used to infer less easily
observed soil distributions. In traditional soil
surveys, soil-landscape models are rarely made
explicit. However, in digital soil mapping (DSM),
soil distributions are mapped by translating the
mental soil-landscape models tacitly used by soil
surveyors into explicit rule sets that can be used to
classify remotely-sensed imagery and terrain
models (e.g. Bui, 2004; Deng, 2007; Lagacherie
and McBratney, 2006; Scull et al., 2003).
The vast literature on DSM includes consid-
erable self-reflection, tackling issues such as the
integration of different forms of knowledge
(e.g. Shi et al., 2009), fuzzy classification
(Burrough, 1989; Qi et al., 2006), the effects
of various scale and attribute choices (e.g. Deng
and Wilson, 2006; Fisher et al., 2004; Schmidt
and Andrew, 2005; Wood, 1996; Zhu et al.,
2008) and the advantages and disadvantages
of different mapping methods (e.g. Bryan,
2006; Deng, 2007; Park and Burt, 2002). It is
evident that DSM and landform mappers are
intensely aware of the slippery issues of scale,
attribute selection and other methodological
choices that can dramatically affect mapping
results. The following insights and approaches
are of particular relevance to our current concerns.
It is now widely recognised that there can
never be a universally applicable set of DSM
procedures since the relative importance of
different land- and soil-forming factors and
processes differs between landscapes (e.g. Mac-
Millan and Shary, 2009; Moller et al., 2008;
Shary et al., 2002). Furthermore, hillslopes vary
in size, according to the horizontal spacing of
major ridges and valleys, or ‘topographic grain’
(Pike, 1988; Wood and Snell, 1960). This varia-
bility means that different combinations of
topographical variables, thresholds and scales
of observation are needed to describe similar
landforms in different landscapes, so that meth-
ods are not directly transferable between loca-
tions (Behrens et al., 2010).
Fuzzy classification methods are widely used
to deal with the imprecision in both the class and
positional boundaries of landforms and associ-
ated soils (Burrough, 1989). In fuzzy classifica-
tions, the degree of membership to each class is
calculated separately for each image object (i.e.
a pixel or group of pixels). Classes are neither
mutually exclusive nor exhaustive, so it is pos-
sible for a single spatial object to fit equally well
(or badly) into two or more overlapping classes.
Degrees of class membership cannot only be
used to describe blurred boundaries, but also
to assess dissimilarity between class members,
thereby addressing some of the uncertainties
Cullum et al. 45
at The University of Auckland Library on January 30, 2016ppg.sagepub.comDownloaded from
associated with spatial extrapolation (Benz
et al., 2004).
Although the soil-landscape model is usually
applied at scales associated with individual
hillslopes, the model has also been applied at
regional scales to inform inventories of agricul-
tural potential and development opportunities.
Christian not only identified ‘land units’ and
catenal sequences of these units at hillslope
scales (‘land systems’), but also introduced the
concept of ‘land types’ (Christian, 1958; Chris-
tian and Stewart, 1953). Land types are areas in
which the same (or very similar) land systems
occur on most hillslopes. These concepts are
widely used in South Africa (see MacVicar
et al., 1974:
html). They form the basis for ecological clas-
sification in KNP (see Venter, 1990). This
approach provides an elegant way of generalis-
ing landscape patterns without recourse to
averages that disguise important local differ-
ences. Instead, typical catenas are described for
each land type, detailing the soils, vegetation
and morphology associated with each catenal
unit in the hillslope sequence (Figure 2).
5 Vegetation mapping
There are two contrasting approaches to classi-
fying and mapping vegetation, which we call
the ‘phytosociological’ and the ‘gradient analy-
sis’ approaches.
The (mainly European) phytosociological
approach aims to describe ‘associations’ of
plant species that commonly occur together,
based on field surveys of plots selected as
homogenous examples of a particular type of
vegetation (Braun-Blanquet, 1928). Although
there are continuing debates around the most
useful ways of defining and characterising plant
Figure 2. Land types and hillslope units in southern Kruger National Park. Venter (1990) classifies KNP into
land types, each of which he describes in terms of typical catenas of hillslope units. The principle is illustrated
with reference to two land types found in the granites and basalts of southern KNP. This method of mapping
allows details of the local heterogeneity found in repeating catenal patterns to be preserved in a more
generalised area-class map. KNP: Kruger National Park. Diagrams after Venter (1990).
46 Progress in Physical Geography 40(1)
at The University of Auckland Library on January 30, 2016ppg.sagepub.comDownloaded from
associations, all seek to describe recurrent com-
binations of plant species that form visually dis-
cernible patches that are relatively internally
homogenous and differ from their neighbours
(see Ewald, 2003; Willner, 2006). Each plant
community is conceived as occupying a partic-
ular environmental niche, such that mapping
plant communities effectively also maps the
various combinations of abiotic variables that
constrain plant distributions. Over time, differ-
ent plant communities occupy the same niche,
forming a successional series that ends in a rel-
atively stable plant community. In space, plant
associations often occur in repeating series that
are strongly related to hillslope position (cf.
soil-vegetation catenas discussed above). Phy-
tosociologists therefore describe vegetation
patterns at various levels in a hierarchy that
ranges from individual plants, through associa-
tions (communities of particular species assem-
blages) and series of these associations in time
(plant successions) and space (catenal hillslope
sequences) to regional patterns that reflect dif-
ferences in climate and geology (for overviews
of modern phytosociology see Blasi and Fron-
doni, 2011; Pedrotti, 2013).
In order to address the challenges posed by
anthropological disruptions to natural patterns,
the abiotic conditions associated with remnant
vegetation patches deemed ‘natural’ are some-
times extrapolated to map ‘potential natural
vegetation’ (PNV) (Ku
chler, 1964; Tu
xen, 1956;
Westhoff and van der Maarel, 1978). In this way,
the phytosociological approach has been extended
from inductive mapping based on field plots
(releve´s) to include deductive mapping based
on the overlay of climatic, lithological and topo-
graphical data, using methods similar to those
adopted in DSM (see for example Blasi and
Frondoni, 2011; Mucina, 2010). As in DSM,
recurring catenal patterns of vegetation associa-
tions are used to characterise regional patterns.
The phytosociological approach has been
strongly criticised on both conceptual and metho-
dological grounds (e.g. Carrio´n, 2010; Chiarucci
et al., 2010). Conceptual concerns centre around
issues of scale, heteroge neit y and inherent
uncertainty, contesting the existence of discrete,
self-evident vegetation patches, the possibility
of constructing a taxonomy of plant communities
and the occurrence of deterministic successions
to relatively stable end-points (Ewald, 2003;
Loidi and Ferna´ndez-Gonza´lez, 2012; Waterton,
By contrast, the (mainly Anglo-American)
gradient analysis approach seeks to describe
vegetation patterns by relating the abundance
of particular species to various environmental
gradients (Whittaker, 1967). Rather than pro-
viding a framework for the ecological description
of a territory for planning and management pur-
poses, these plant ecologists ultimately seek
explanations and rules for plant community
assemblage (Cody and Diamond, 1975; Grime
and Pierce, 1999; Weiher and Keddy, 2001).
Ecological niches are conceived as overlapping
rather than as contiguous patches and as being
controlled by a myriad of factors acting at many
scales. The approach has now been broadened
to include population processes, intraspecific
variation, microhabitat mosaics, animal and
microbial interactions and stochastic events
alongside abiotic environmental gradients as
potential determinants of species distributions
(see for example Keddy, 2007).
Debates between proponents of the two
approaches echo old contests between Clement-
sian and Gleasonian views of plant ecology (in
which Clements described plant communities
as distinct entities composed of interdependent
species, whilst Gleason saw plant communities
as the serendipitous result of the individual
responses of each species to spatially variable
environmental conditions (Clements, 1916;
Gleason, 1926). The debates are also reminiscent
of tensions between the ‘European’ and ‘Ameri-
can’ schools of landscape ecology (the former
emphasises relatively homogenous landscape
entities and the descriptions of patterns per se,
whilst the latter focuses on heterogeneity, issues
Cullum et al. 47
at The University of Auckland Library on January 30, 2016ppg.sagepub.comDownloaded from
of scale and the effects of pattern on process (see
Bastian, 2001; Wu, 2012; Wu and Hobbs, 2002).
Both perspectives are useful. Whereas the
‘American’, plant ecology and landscape ecol-
ogy approaches are best suited to improving our
understanding of the mechanistic processes
underlying the formation of plant communities
and ecological patterns, the ‘European’ approaches
offer pragmatic methods for the description of
these communities in ways that are useful for
landscape planning and management. Indeed,
whilst phytosociology had been suffering from
scientific isolation, the approach has now been
revitalised as the foundation for the identifica-
tion and classification of European habitats, as
enshrined in European law in Annex 1 of the
Habitats Directive. In the context of the Eur-
opean Directive, ‘habitat’ is defined somewhat
differently from its usage in ecology (EEA,
2014). Rather than referring to the environmen-
tal resources that support a particular organism
or species, ‘habitat’ takes on a geospatial mean-
ing similar to that of the geographically defined
ecosystem discussed in the introduction to this
paper. For example, Bunce et al. define habitat
as ‘an element of the land surface that can be
consistently defined spatially in the f ield in
order to define the principal environments in
which organisms live (Bunce et al., 2005: 12).
The European habitat definitions lean heavily
on data from the 4.36 million releve´s in Europe
(Schamine´e et al., 2009). Indeed, some 60% of
the Annex 1 habitats are based on plant commu-
nities (Bunce et al., 2013).
European habitats are used to inform and
enforce conservation and land-use planning pol-
icy throughout the EU, providing an excellent
example of how phytosociology can inform prag-
matic approaches to identifying pieces of land
that are then treated as ecologically equivalent.
6 Synthesis
Each of the conceptualisations reviewed faces
the same issues of scale, local variability and the
inherent uncertainties associated with complex
ecological systems, but deals with these chal-
lenges in slightly different ways. Thus synthesis-
ing the approaches allows us both to integrate
across different disciplines and also to integrate
the lessons learned within each discipline.
Hierarchy theory offers a way of conceptua-
lising the world in terms of entities at different
levels, each of which occupy particular domains
in space and time. If agreement is reached on
such a conceptualisation, issues of scale and
attribute selection are clarified, since the pat-
terns being described can be interpreted in terms
of recognisable entities with intrinsic scales and
attributes. By introducing the concept of top-
down constraints and bottom-up mechanisms,
hierarchy theory also neatly reconciles per-
spectives that stress sy stem complexity and
potentially unique combinations of spatial and
temporal contingencies with perspectives that
emphasise confor mity to general laws and
principles (Mitchell, 2009).
However, although hierarchy theory offers a
useful heuristic framework for describing and
explaining horizontal and vertical relationships
between landscape entities, it does not attempt
to specify what those entities might be or how
landscapes might be decomposed. In other
words, the hierarchical framework needs to be
populated before it can provide a conceptualisa-
tion capable of informing ecological mapping.
River system classifications and soil-landscape
models suggest how the hierarchy might be
populated. The two ways of conceptualising
landscape structure are intimately linked, since
the distribution and morphology of hillslopes
is just the opposite side of the coin of the dis-
tribution and morphology of rivers. As valley
and channel morphology varies systematically
from headwaters to estuaries, so does hillslope
morphology and landscape relief, the factors
that largely determine the nature of hillslope
vegetation-soil catenas. Indeed, Huggett described
the soil-landscape model as ‘a systematic expan-
sion of the proposition that the valley basin ...
48 Progress in Physical Geography 40(1)
at The University of Auckland Library on January 30, 2016ppg.sagepub.comDownloaded from
is the basic organizational unit of soil systems’
(Huggett, 1975: 7).
The experience of DSM and landform map-
pers confirms that no standard set of variables
and scales are able to describe hillslope soil
units, toposequences of these units or the varia-
bility in topography associated with different
physiographic zones. The fuzzy mapping tech-
niques that have been developed and widely
applied in these fields offer excellent tools to
deal with imprecise class and positional bound-
aries. DSM also demonstrates how the tacit
mental models that were once locked inside the
minds of field soil scientists and geomorpholo-
gists can be translated into transparent and con-
testable rule sets that inform automated mapping.
Both soil and vegetation mappers describe
catenal patterns on hillslopes, and use these
repeating patterns to characterise broader zones,
so that regional patterns can be described in
ways that avoid averaging out important differ-
ences over large areas.
Lastly, the revitalisation of phytosociological
methods reveals a pragmatic approach to describ-
ing vegetation patterns for the purposes of land
and conservation management and planning.
III Towards a new conceptualisation
of landscapes to underpin ecological
1 Introduction
In this section, we synthesise learnings from the
reviewed approaches to landscape classification
and mapping, developing a new conceptualisation
to underpin such tasks. The conceptualisation
has three key components:
Rather than attempting to encapsulate an
entire ecosystem (however defined), it
describes landscapes as a biophysical
template that constrains many different
types of socio-ecological processes.
A conceptual landscape hierarchy situ-
ates hillslopes within river systems,
offering a perspective that integrates
aquatic and terrestrial perspectives and
in which the selection of scales and attri-
butes used to define landscape units is
largely informed by the characteristics
of the landscape itself.
Archetypes are used to mediate between
particular landscapes and mental models,
principles and laws that link observed form
to generative processes. The archetypes
that have been refined and adapted to suit
particular circumstances are then used to
develop rule sets for classification and
mapping purposes.
2 Ecological clusters conceived as
a biophysical template
Rather than attempting to capture entire ecosys-
tems, or focusing on species’ populations and
population dynamics, we conceive the Earth’s
surface in terms of a biophysical ‘template’ of
spatially coupled complexes of soils, vegetation,
hydrology and topography (e.g. Caylor et al.,
2005). Such a notional template does not deter-
mine the species or population of plants and ani-
mals that use the resources it provides, but it does
limit the range of available possibilities, acting as
an environmental filter that constrains a wide
range of socio-ecological processes at many
scales (see Phillips, 2004; Poff, 1997). Focusing
on this template moves initial discussion away
from the distributions of species (which are often
mobile) and subjective perceptions of ecosys-
tems (which have indeterminable boundaries).
Instead, attention is focused on the construc-
tion of the place-based understandings that are
essential for integrated science, management
and land-use planning (Brierley et al., 2013).
However, this template is challenging to
describe and map. As geomorphologists, soil,
river and vegetation scientists have all discov-
ered, each location demands mapping in terms
of its own scale and set of variables (e.g.
Cullum et al. 49
at The University of Auckland Library on January 30, 2016ppg.sagepub.comDownloaded from
Behrens et al., 2010; Hengl, 2006; MacMillan
and Shary, 2009; Simon et al., 2007). Many
authors have explained how the use of a concep-
tual hierarchy helps address the vexed issues of
scale and attribute selection for the description
and mapping of landscape patterns (e.g. Fother-
ingham, 1989; Jelinski and Wu, 1996; Open-
shaw, 1984). The MAUP is solved by user
consensus on the recognisability and utility of the
spatial entities defined by the hierarchy. Once
such a consensus is reached, suitable mapping
scales and attributes can be determined based
on the size and characteristics of these entities.
3 A conceptual landscape hierarchy
In our conceptual hierarchy, the toposequences
of soil-vegetation associations that are described
both by soil scientists and phytosociologists
are placed within the context of the river sys-
tems described in geomorphological river clas-
sifications. Catenal elements (the lowest level
of the hierarchy) are equivalent to the soil
units described by soil mappers or the land
units described by Christian (Christian, 1958;
Christian and Stewart, 1953), but in our con-
ceptualisation also include channel reaches.
Catenal elements are systematically arranged
in toposequences within a catchment, as described
by the soil-landscape models used in DSM and
the catenal series described by phytosociolo-
gists. The character and behaviour of a catch-
ment emerges from the nature and connectivity
of the catenal elements it contains, but is con-
strained by its position within the river network
and by the geoclimatic history that shapes the
networks within a particular physiographic
zone, as described by river science (Figure 3).
The conceptual hierarchy defines the entities
to be mapped. For example, once it is agreed
that we are trying to map repeating catenal
sequences of vegetation patches associated with
particular hillslope positions, then these patches
can usually be observed in aerial photographs or
other imagery or detected through analysis of
relationships between topographic and vegeta-
tion metrics. The particular metrics and scales
relevant to the description of these patterns can
then be determined with reference to the partic-
ular landscape being described. This approach
allows the choice of variables and scales used
to define landscape units to be informed by the
character of individual landscapes, rather than
being based on data availability, on tacit deci-
sions made by cartographers or expert panels
or imposed by a standardised taxonomy and rule
set that is unlikely to suit every landscape
equally well.
4 Archetypes: Fuzzy approaches
for a fuzzy world
The conceptualisation of landscapes as a mosaic
of relatively homogeneous patches is both per-
vasive and very useful (Turner, 2005). Although
clearly bounded, discrete, internally homoge-
nous units are easy to represent, analyse and
manage, the internal heterogeneity within patches
and between patches is truncated as information
about how variables change through space and
time is sacrificed to provide easily manipulated
summaries (Cushman et al., 2010). However,
these summaries are often poor reflections of
reality: not only are patch transitions often
blurred, but small within-patch variations can
have large effects both on the distributions of
individual system components and on overall
patch character and behaviour (McGarigal and
Cushman, 2005; Phillips, 2004). Furthermore,
landscapes constantly adjust and evolve over
multiple temporal scales, influenced by histori-
cal and contextual contingencies. Indeed, the
prevalence and importance of contingencies in
time and space suggests that every location is
potentially unique, undermining the assumptions
of homogeneity and ecological equivalence that
underpin extrapolations between landscape units
assigned to the same class (e.g. Beven, 1999;
Kennedy, 1979; Phillips, 2007). Thus tension
exists between the convenience of aggregated
50 Progress in Physical Geography 40(1)
at The University of Auckland Library on January 30, 2016ppg.sagepub.comDownloaded from
summaries and the potential importance of local
However, despite th e fac t that loca l cir cum-
stances can disrupt broad patterns and/or intro-
duce a wide range of variations, underlying
regularities can be discerned in most land-
scapes. In other words, landscapes are highly
organised in space, but they are not precisely
organised. As Kennedy put it, we live in a
‘naughty’ world that often disobeys the ‘rules’
that we would like to impose upon it (Kennedy,
1979). The reality of a ‘naughty’ world does not
mean that approximate descriptions are worth-
less, but instead means that conceptualisations
must be flexible, accommodating not only
differences between locations, but also the varia-
bility of the same location over time. Accuracy
and precision are traded off against the greater
need for generality and the t ransferability of
knowledge and experience (Levins, 1966).
Figure 3. A hierarchy of structural-functional landscape units for the ecological mapping of savanna land-
scapes. At the lowest level of organisation the template consists of geographically distinct patches of vege-
tation and soils found in particular hillslope positions, each with a characteristic hydrological regime. These
patches form repeating series along hillslopes (catenal level of organisation). In turn, hillslope characteristics
change according to their position within a river network (catchment level of organisation). Lastly, the
character of river networks (and the hillslopes they contain) changes between physiographic zones that are
characterised by different geologies and climates (physiographic zone level of organisation).
Cullum et al. 51
at The University of Auckland Library on January 30, 2016ppg.sagepub.comDownloaded from
It is therefore unlikely that scientific under-
standing and management actions can ever be
applied precisely to any given area. Furthermore,
the inherent uncertainty of the outcomes of eco-
logical processes means that even useful
hypotheses and theories are likely to be falsi-
fied in some circums tances (e.g. Harris and
Heathwaite, 2012). This suggests that we need
to learn to live with approximations and to
become more relaxed with vagueness. For
example, reflecting on 15 years of experience
implementing adaptive management in KNP,
Biggs and Roux concluded that the institu-
tional culture of SANParks had learned to be
more ‘relaxed with complex issues’ and that
‘muddling through’ had become ‘more accep-
table, even desirable’ in the face of scepticism
about the possibility of ‘perfect’ management
control (Biggs and Roux, 2013: 2).
Field scientists are adept at applying general
conceptual models to specific circumstances
(e.g. Burt, 2005). For example, soil surveyors,
vegetation mappers and geomorphologists all
move easily between the general and the partic-
ular. Landscapes are interpreted though the
application of a priori general principles and
tacit mental models (such as the soil-landscape
model discussed above), noting where local
deviations demand that the idealised model
needs to be adapted to accommodate local cir-
cumstances to build particular, place-based
understandings (Brierley et al., 2013; Phillips,
2012; Wohl, 2013).
In scientific discourse, vagueness is usually
avoided and despised since it is the antithesis
of the precision and conceptual clarity that is
demanded by traditional approaches to empiri-
cal enquiry (Strunz, 2012). Whereas the scien-
tific method aims to separate truth from
falsehood or nonsense through rigorous testing
of precisely articulated hypotheses (Popper,
1959), field-based scientists are comfortable in
a world of approximations and uncertainty.
This world of vagueness and approximation
is familiar territory, since it is the world we all
inhabit and in which we are comfortable talking
and thinking about using mental models that can
shift to accommodate reality and ‘fuzzy logic’
that allows for partial truths that cannot be rig-
orously tested (Zadeh, 1996).
Although this type of reasoning appears at
first sight to lack the rigour of conventional sci-
entific argument, this is far from true:
Fuzzy logic is not fuzzy. Basically, fuzzy logic is a pre-
cise logic of imprecision and approximate reasoning.
More specifically, fuzzy logic may be viewed as an
attempt at formalization/mechanization of two remark-
able human capabilities. First, the capability to converse,
reason and make rational decisions in an environment of
imprecision, uncertainty, incompleteness of informa-
tion, conflicting information, partiality of truth and par-
tiality of possibility in short, in an environment of
imperfect information. And second, the capability to
perform a wide variety of physical and mental tasks
without any measurements and any computations
(Zadeh, 2008: 2751).
Recognising the value of vagueness does not
mean that traditional science is rejected, merely
that our toolkit for viewing, describing and
explaining the world is extended, reaching into
the now well-established field of ‘soft comput-
ing’, which is tolerant of imprecision, uncer-
tainty, partial truth and approximation (see
According to Zadeh, the cornerstones of
fuzzy logic are:
Graduation everything is allowed to be
a matter of degree (or ‘fuzzy’).
Granulation attributes are clumped in
terms of their similarity, proximity, func-
tionality, suitability for purpose or in
terms of some other criterion.
Generalised constraint of possibility,
probability or truth.
Precisiation a process in which an
object becomes progressively more pre-
cisely defined, either in terms of meaning
or value (in the mathematical, not the
normative sense of the word).
52 Progress in Physical Geography 40(1)
at The University of Auckland Library on January 30, 2016ppg.sagepub.comDownloaded from
This type of reasoning is similar to that used
by field scientists ‘reading’ the landscape, inter-
preting new landscapes against a backdrop of
accumulated experience and knowledge (Brier-
ley et al., 2013; Fryirs and Brierley, 2012; Wohl,
2013). The links made between form and
process are not precise or easily described in
process models. Instead they are adaptable
approximations, capable of incorporating incon-
sistencies, incomplete information and partial
truths. Such framings are just as Zadeh describes
fuzzy logic: the process of adapting a general
model to particular circumstances is a process
of precisiation, whilst recognising contextual
constraints on attributes and processes is a form
of articulating generalised constraints.
In our approach to ecological mapping an
archetype, or a set of archetypes, is a conceptual
example of the whole of a class of landscape
units. Archetypes can be conceived and repre-
sented in many ways: they can be real examples,
abstract mental constructs or theoretical con-
cepts. Archetypes are deliberately vague, embra-
cing fuzzy logic as the best way of describing a
world that only approximately corresponds to the
idealised models and representations that we use
to describe it (Zadeh, 2008).
A single archetype can be expressed and used
in many ways, so that it can link or generate:
Observations and analysis of particular
Conceptual models that express relation-
ships between observed forms and the pro-
cesses that create and sustain them. Such
models can also be used to guide fore-
sighting (loose predictions) of future beha-
viour and responses to various scenarios.
Rule sets for the classification and map-
ping of ecological units from remotely-
sensed data.
Archetypes for the units defined by the land-
scape hierarchy are initially developed a priori,
based on existing conceptual models (ranging
from textbook examples to local knowledge and
expert appraisals in the field), together with
field observations and data analyses. They are
then subject to continual revision through an
iterative process of map-making, in which the
character of the mapped landscape contributes
to an ongoing ‘precisiation’ of the archetype
as it is adapted to suit local circumstances.
The use of fuzzy mapping techniques allows
membership values to indicate the extent to
which any particular parcel of land deviates
from the class archetype. Understanding the
spatial distribution of such deviation opens pos-
sibilities for describing local heterogeneity and
boundary transitions as well as opening possibi-
lities for assessing the extent to which concep-
tual form-process models are likely to apply to
particular tracts of land.
We now illustrate the application of our
approach to ecological classification and map-
ping in the KNP, South Africa.
IV Application of our proposed
approach to ecological mapping
in the Kruger National Park,
South Africa
1 Introduction
The flat savannas of the KNP in South Africa lie
between the Great Escarpment to the east and the
Lebombo mountains that border Mozambique to
the east (Figure 4). The high biodiversity that
makes this park one of the world’s foremost con-
servation areas results from the diverse geology
and differences in rainfall that occur in this 2 mil-
lion hectare park. Distinct ecoregions occur in
which particular combinations of geology and cli-
mate support spatial clusters of interdependent
soils and vegetation, linked to specific hydrological
regimes typical of the zone. Each region also has a
distinctive pattern of landscape dissection, with
characteristic hillslope and channel morphology.
Given the semi-arid, water-controlled eco-
systems and the consequent tight spatial
Cullum et al. 53
at The University of Auckland Library on January 30, 2016ppg.sagepub.comDownloaded from
coupling of hydrological, geomorphic and bio-
tic processes, together with the high abiotic and
biotic diversity present in the park, the history of
landscape classification (e.g. Gertenbach, 1983;
Venter, 1990), the strong links between science
and conservation management and the well-
established and large interdisciplinary research
community that exists in KNP, the park offers
excellent scope for trialling our new approach
to ecological classification and mapping.
2 The biophysical template
We conceived the park in terms of a biophysical
template that could be described in terms of
catenal sequences of patches of co-varying
vegetation, soils and topography, as described
by Venter (1990). Each patch in the hillslope
sequence (catena) was considered as a distinct
functional unit, or ‘process domain’ (sensu
Montgomery, 1999), with a water budget that
both arises from and controls interactions
between water inputs, soils, vegetation, animals
and humans. Since water limits many ecosystem
processes in semi-arid systems, by definition
these landscape units integrate a wide range of
highly interdependent system components.
3 The landscape hierarchy
Applying our conceptual hierarchy to KNP, we
conceived of each soil-vegetation patch as a ‘cate-
nal element’ that formed part of a toposequence
on the hillslopes of each subcatchment. At the
highest level of the hierarchy that we considered,
hillslope and catchment morphology were used to
identify physiographic zones (Figure 5).
This hierarchy informed the detection and
delineation of landscape elements by defining
the entities that we sought to describe and map
at each hierarchical level. For example, the first
step in defining catenal elements was to identify
combinations of vegetation and topographic
attributes that were strongly associated with var-
ious hillslope positions, in a manner similar to the
phytosociological approach described by Gillet
and Gallandat (1996). This involved an iterative
interplay between visual inspection of imagery
and statistical analysis of variables describing
woody cover and topographic variables such as
gradient and curvature (Cullum 2015; Cullum
and Rogers, 2011). This approach allowed the
process of adapting (or precisifying) each arche-
type to suit local circumstances to be informed by
the landscape itself. Thus the choice of attributes
and scales used to describe catenal elements
varied between set tings (Figure 6).
The scales at which attributes were measured
and patches were mapped was informed by
landscape ‘grain’, which is the typical size of
Figure 4. Location and geology of Kruger National
Park (KNP). Within KNP, the geology generally fol-
lows a north-south strike, from the youngest rhyo-
lites in the east, across the basalts and the narrow
ridge of sandstone and shales to the older granites in
the west. Throughout the length of the park, the
granites are intruded by igneous rocks, notably gab-
bro (1:250,000 Council for Geoscience 1986)..
54 Progress in Physical Geography 40(1)
at The University of Auckland Library on January 30, 2016ppg.sagepub.comDownloaded from
landscape units at each hierarchical level of
organisation (Figure 7). At the level of catenal
elements, landscape grain is reflected in the size
of the hillslope patches that form repeating
toposequences. At the scale of entire hillslopes
and subcatchments, landscape grain is equiva-
lent to the horizontal spacing of major ridges
and valleys (described as ‘topographic grain’
by Pike, 1988; Wood and Snell, 1960). At
higher levels, landscape grain describes the bio-
physical heterogeneity of entire regions in terms
of the size and diversity of physiographic zones.
Differences in landscape grain have substan-
tial implications for the constructionof ecological
maps, since the optimum scales of observation
for the same level of the landscape hierarchy var-
ies between different landscapes. For example, in
finely dissected landscapes with high stream den-
sity and relatively short hillslopes (e.g. the south-
ern granites of KNP), a much smaller support
distance is needed to calculate the changes in gra-
dients and curvature that characterise different
catenal elements than in landscapes with low
stream density, large catchments and long hill-
slopes (e.g. the southern basalts of KNP).
Although the same sequence of catenal ele-
ments was repeated on many hillslopes within
each physiographic zone, in some cases the pat-
tern was much clearer than in others and there
were also many exceptions to the general rule.
Fuzzy classification of archetypes not only
helped us to address this variability, but also
suggested ways in which the heterogeneity of
each landscape could be described.
Figure 5. A hierarchy of structural-functional units for ecological mapping in savanna landscapes. a) Distinct
sequences of catenal elements recur throughout savanna landscapes. b) Similar toposequences of catenal
elements frequently recur in the sub-catchments associated with the stream segments shown on the 1:50,000
topographical map. c) Within a physiographic zone, catchments in a particular network position tend to have
similar relief and geometry, containing recurring assemblages of catenal elements that are configured in
sequences that typify the zone. d) Different suites of catenal elements are seen in the granites and basalts,
reflecting the different patterns of landscape dissection in these physiographic zones. Photos: Cullum, Nov
2006; Brierley, Mar 2010..
Cullum et al. 55
at The University of Auckland Library on January 30, 2016ppg.sagepub.comDownloaded from
In most landscapes, both class and positional
boundaries are blurred and/or contested and
even the identification of distinguishing charac-
teristics is likely to be imprecise (Burrough,
1989). Fuzzy classification methods are there-
fore far better suited to landscape classification
and mapping than are conventional crisp classi-
fication techniques. In our approach, all spatial
units (e.g. pixels or image objects) are assessed
in terms of their similarity to all archetypes,
and classified according to the archetype that
they most closely resemble. Classes are neither
mutually exclusive nor exhaustive, so it is possi-
ble for a single spatial object to fit equally well (or
badly) into two or more overlapping classes (Bur-
rough, 1989). Thus archetypes not only mediate
between general and particular conceptions of
reality, but also between these conceptualisations
and their representation in a map.
If ecological maps are to be used for the spatial
extrapolation of the anticipated behaviour of
landscape units, then classifications must be
‘process-based’: the processes responsible for the
anticipated behaviour must be described and dis-
tinctions between classes must reflect differences
in the processes that are associated with different
outcomes. This is achieved if class archetypes are
based on conceptual models that link observed
patterns and generative processes, providing a
sound basis for spatial extrapolation.
Furthermore, by acknowledging that not all
class members are equally representative of the
whole class, archetypal classifications allow the
assessment of the range of variability that
occurs within a class. Understanding this range
of variability has important implications for
both land management and policy, since it is a
major source of uncertainty in predictions and
foresighting. Furthermore, the degree of varia-
bility within a class can be taken into account
when choosing sample or reference sites and
when extrapolating observations or experimen-
tal results to other sites in the same class. For
example, archetypes for the southern granites
of KNP suggest that subcatchments contain
crests with sandy soils and woody vegetation
that lie above midslopes with relatively clayey
soils and relatively open, grassy vegetation.
However, although this pattern is often found,
many subcatchments do not conform to the
Figure 6. Different suites of vegetation attributes are used to describe catenal elements in different settings.
Granitic catenas near Pretoriuskop (a) tend to have less pronounced grassy midslopes than those further west,
towards Skukuza (b). In the Pretoriuskop area, the dominant woody species on both crests and midslopes is
Terminalia sericea, whilst in the Skukuza area,Combretum species dominate the crests andAcacia species dominate
midslopes. These differences demand different variables and rule sets to describe the same catenal elements in
each area. Photos: a) Rogers, Apr 2005; b) Cullum, Apr 2009..
56 Progress in Physical Geography 40(1)
at The University of Auckland Library on January 30, 2016ppg.sagepub.comDownloaded from
archetype, having relatively grassy crests and/or
woody midslopes or even containing large rocky
outcrops that are almost devoid of vegetation. In
these situations, conceptual models describing
the hydrological fluxes that generate and sustain
the observed catenal elements are unlikely to
apply. Mapping the catenal units present in a
landscape enables such anomalies to be
identified, indicating locations where modelling
assumptions are less likely to apply and where
spatial extrapolations are likely to be unreliable.
V Discussion
We have presented an approach to landscape
classification and ecological mapping that aims
Figure 7. Landscape grainat various hierarchical levels of organisation.Thegrain of a landscape is the typical size
of units at a particular level of organisation, reflecting scales that are intrinsic to particular landscapes. Although
the detection of landscape grain is undoubtedly influenced by cartographic decisions such as the minimum size of
mapping units, the structure of the landscape itself plays the major role in determining landscape grain..
Cullum et al. 57
at The University of Auckland Library on January 30, 2016ppg.sagepub.comDownloaded from
to be transparent, adaptable and process-based,
integrating a wide range of ecological patterns
and processes in a way that is likely to be seen
as credible and relevant to a wide range of end
users, for a wide range of purposes.
The conceptualisation of landscapes in terms
of a biophysical template that can be observed
through the lens of a hierarchy that integrates
river networks and hillslopes exposes the core
assumptions of this approach. The conceptuali-
sation sacrifices the inclusion of fauna and bio-
logical processes such as competition, predation
and dispersal in order to gain the convenience of
a geographical, generally applicable and spa-
tially explicit approach that is based on the anal-
ysis of more readily available imagery. Based
on the experience and practice of soil, vegeta-
tion and hydrological mapping, we assume that
the co-variation of topography and vegetation
cover indicates patterns of similarly co-varying
water and soils. We also assume that the distribu-
tion of these geocomplexes of vegetation, soils,
water and topography constrains a wide range
of ecological processes.
Rather than assuming that the conditions with
a particular landscape unit determine a particular
suite of ecological processes, we invoke hierar-
chy theory to suggest that higher-level contextual
constraints limit the possibilities generated by
lower-level mechanisms within a range of varia-
bility that is characteristic of the landscape class.
Once maps are produced, further research is
required to explore the limits of class variability
and the various ‘faces’ of the same landscape unit
though time, documenting seasonal differences
as well as typical responses to perturbations such
as fire, flood, grazing or activities such as harvest-
ing natural resources, agriculture or settlement.
The mechanisms and constraints that limit
within-class variability can be articulated in con-
ceptual models that link form and process, pro-
viding justification for the assumption of
‘ecological equivalence’ and the spatial extrapo-
lation of foresighted future behaviour predicted
for a particular archetype across all members of
the class based on that archetype. We recognise
that local features, processes and evolutionary
narratives can be more or less similar to archety-
pal mental models. We suggest that it is only by
considering the effect of these variations on an
underlying process model that it becomes possi-
ble to hypothesise which differences are likely to
‘matter’. Thus locations where ecological sur-
prises can be expected can potentially be identi-
fied and mapped, even if system complexity
dictates that they can never be eliminated (Har-
ris and Heathwaite, 2012). Therefore the use of
archetypes not only accommodates the range
of variability f ound within each class, it could
potentially also sugges t ways in which t his
variability might be quantified.
We believe that the inherent vagueness of
archetypes is a strength rather than a weakness,
allowing very general archetypes to be adapted
as they are ‘precisified’ to reflect the attributes
and scales appropriate to a particular setting.
In this way, general principles, text book exam-
ples, tacit mental models and hypothesised con-
ceptual models can all inform the development
of specific place-based understandings.
Whilst not wanting to suggest that ours is the
only valid approach to ecological mapping, we
propose that laying bare the assumptions underly-
ing such mapping allows these assumptions to be
debated and contested, rather than being imposed
by experts or created through ‘black box statisti-
cal analysis. This transparency falls within the
emerging discourses of critical cartography (e.g.
Crampton, 2010) and critical physical geography
(Blue et al., 2012; Lave et al., 2014; Tadaki et al.,
2015). This approach moves debate away from
contestations about patch boundaries, reframing
discussions around the credibility and usefulness
of the conceptualisation that underpins landscape
classification and mapping.
It is increasingly recognised that consensus
around shared conceptualisations is a condition
for sharing knowledge and experience across
disciplines and management agencies (Pickett
et al., 2007; Stirzaker et al., 2011). It is therefore
58 Progress in Physical Geography 40(1)
at The University of Auckland Library on January 30, 2016ppg.sagepub.comDownloaded from
essential that proposed conceptualisations are
justified in terms of narratives that demonstrate
their relevance and credibility (Cash et al.,
2003). The relevance of our approach to diverse
end users for a wide range of applications stems
from the description of a biophysical template
that constrains a very wide range of ecological
processes. The credibility of our approach to
ecological mapping gains legitimacy and cred-
ibility in two ways. First, it draws on the widely
accepted theory that underpins soil surveys and
river classifications. Second, it uses the actual
attributes and scales that characterise the land-
scapes being described or mapped, rather than
relying on subjective expert assessment or
‘black box’ statistical techniques.
VI Conclusion
The need for transdisciplinary frameworks for
ecological mapping and conservation manage-
ment has never been greater. As diverse groups
of stakeholders, managers and scientists come
together to look after entire catchments, pro-
tected reserves or small areas adopted by local
communities, shared conceptualisations and
visualisations are needed to facilitate commu-
nication and to f rame management practices.
Integrated perspectives are also demanded
within earth and environmental sciences (e.g.
Paola et al., 2006), where approaches are conver-
ging to create new fields that dissolve traditional
boundaries (e.g. ecohydrology (Rodriguez-
Iturbe, 2000), biogeo morpholo gy (Stallins ,
2006) and hydropedology (Bouma, 2006)).
We have synthesised perspectives from soil,
vegetation, ecology and river science, produc-
ing a conceptualisation of the Earth’s surface
in terms of a biophysical template that con-
strains a myriad of ecological processes. We
have shown how this template can be described
in terms of a conceptual hierarchy of landscape
units. Applying these concepts to KNP, we have
demonstrated how the broad archetypes defined
in the hierarchy can be precisiated to create
definitions of landscape units in a given setting
that can then inform rule sets for mapping and
the construction of conceptual models which
link observed patterns to generative processes.
We have illustrated how the process of precisia-
tion is guided by the character of the landscape
itself, rather than relying solely on expert inter-
pretations or ‘black box’ statistical analysis.
The process of constructing place-based
understandings is necessari ly iterative a nd
ongoing, as new knowledge can continually
be added to the local archetypes. Furthermore,
the archetypes can be viewed from different
perspectives as the implications of the local
constraints on d ifferent socio-ecological pro-
cesses are explored, providing a platform for
integrating different types of knowle dge. How-
ever, recognising that many such platforms
could be constructe d, we offer a way of ensur-
ing that the foundations of the platform are
transparent, allowing e nd users to ensure the
relevance and utility of this approach for their
particular applications.
Thanks to SANParks Scientific Services and espe-
cially to Izak Smit, my research coordinator in
KNP. Thanks are also due to t he Andrew Mellon
Foundation and the endowment of the Carnegie
Institution for Science who facilitated the collection
of LiDAR data by the Carnegie Airborne Observa-
tory, from which some of analyses in this paper are
The author(s) disclosed receipt of the following finan-
cial support for the research, authorship, and/or publi-
cation of this article: This research was funded by the
South African Water Research Commission (K5/
1790) and by a Short Term Doctoral Student Intern-
ship Grant from the Faculty of Science, University
of the Witwatersrand, Johannesburg. Thanks are also
due to the School of Environment at the University of
Auckland, who generously hosted the lead author as a
visiting scholar.
Cullum et al. 59
at The University of Auckland Library on January 30, 2016ppg.sagepub.comDownloaded from
Abell R, Thieme ML, Revenga C, e t al. (2008) Fresh-
water ecoregions of the world: A new map of bio-
geographic units for freshwater biodiversity conservation.
Bioscience 58(5): 403–414.
Allen T and Starr T (1982) Hierarchy. Chicago: University
of Chicago Press.
Austin MP (2002) Spatial prediction of species distribution:
An interface between ecological theory and statistical
modelling. Ecological Modelling 157(2–3): 101–118.
Bailey RG (1983) Delineation of ecosystem regions.
Environmental Management 7(4): 365–373.
Bailey RG (1996) Ecosystem Geography: From ecor-
egions to sites. New York: Springer.
Bastian O (2001) Landscape ecology: Towards a unified
discipline? Landscape Ecology 16(8): 757–766.
Behrens T, Zhu AX, Schmidt K, et al. (2010) Multi-scale
digital terrain analysis and feature selection for digital
soil mapping. Geoderma 155(3–4): 175–185.
Benz UC, Hofmann P, Willhauck G, et al. (2004) Multi-
resolution, object-oriented fuzzy analysis of remote sen-
sing data for GIS-ready information. ISPRS Journal of
Photogrammetry and Remote Sensing 58(3–4): 239–258.
Beven KJ (1999) Uniqueness of place and process repre-
sentations in hydrological modelling. Hydrology and
Earth System Sciences 4: 203–213.
Beven KJ and Kirkby MJ (1979) A physically based
variable contributing area model of basin hydrology.
Hydrological Sciences Bulletin 24(1): 43–69.
Biggs HC and Roux DJ (2013) Adaptive Management in
and beyond SANParks: A short interpretation of
experiences and possible trajectories. In: Proceedings
of the 11th Savannah Science Networking Conference,
Skukuza, South Africa. Available from: www.sanparks.
4March13.pdf (accessed 4 July 20 14).
Blasi C and Frondoni R (2011) Modern perspectives for
plant sociology: The case of ecological land classifi-
cation and the ecoregions of Italy. Plant Biosystems
145(Supp.1): 30–37.
Blue B, Gregory C, McFarlane K, et al. (2012) Freshwater
geographies: Experimenting with knowing and doing
geography differently. New Zealand Geographer 68(1):
Bohn B and Kershner JL (2002) Establishing aquatic
restoration priorities using a watershed approach. Jour-
nal of Environmental Management 64(4): 355–363.
Bouma J (2006) Hydropedology as a powerful tool for
environmental policy research. Geoderma 131(3–4):
Braun-Blanquet J (1928) Pflanzensoziologie, Grundzu
der Vegetationskunde. Biologische Studienbu
cher, vol.
7. Berlin: Springer.
Brierley G, Fryirs K, Cullum C, et al. (2013) Reading the
landscape: Integrating the theory and practice of geo-
morphology to develop place-based understandings of
river systems. Progress in Physical Geography 37(5):
Brierley GJ and Fryirs KA (2005) Geomorphology and
River Management: Applications of the river styles
framework. Malden, MA: Blackwell.
Bryan BA (2006) Synergistic techniques for better under-
standing and classifying the environmental structure of
landscapes. Environmental Management 37(1): 126–140.
Buffington JM and Montgomery DR (2013) Geomorphic
classification of rivers. In: Shroder JF (ed.), Treatise on
Geomorphology. San Diego, CA: Academic Press,
Bui EN (2004) Soil survey as a knowledge system. Geo-
derma 120(1–2): 17–26.
Bunce RGH, Groom GB, Jongman RHG, et al. (2005)
Handbook for Surveillance and Monitoring of European
Habitats. Wageningen, NL: Alterra Report 1219.
Bunce RGH, Bogers MMB, Evans D, et al. (2013) The
significance of habitats as indicators of biodiversity and
their links to species. Ecological Indicators Biodi-
versity Monitoring 33: 19–25.
Burrough PA (1989) Fuzzy mathematical methods for soil
survey and land evaluation. Journal of Soil Science
40(3): 477–492.
Burt T (2005) General/Particular. In: Castree N, Rogers
AP and Sherman D (eds), Questioning Geography.
Chichester, West Sussex; Hoboken, NJ: Blackwell-
Wiley, 117–130.
Buyantuyev A and Wu J (2007) Effects of thematic reso-
lution on landscape pattern analysis. Landscape Ecol-
ogy 22(1): 7–13.
Carrio´n JS (2010) The concepts of potential natural
vegetation (PNV) and other abstractions (trying to pick
up fish with wet hands). Journal of Biogeography
37(11): 2213–2215.
Cash DW, Clark WC, Alcock F, et al. (2003) Knowledge
systems for sustainable development. Proceedings of
the National Academy of Sciences of the United States
of America 100: 8086–8091.
60 Progress in Physical Geography 40(1)
at The University of Auckland Library on January 30, 2016ppg.sagepub.comDownloaded from
Caylor KK, Manfreda S and Rodriguez-Iturbe I (2005) On
the coupled geomorphological and ecohydrological
organization of river basins. Advances in Water
Resources 28(1): 69–86.
Chiarucci A, Arau
jo MB, Decocq G, et al. (2010) The
concept of potential natural vegetation: An epitaph?
Journal of Vegetation Science 21: 1172–1178.
Christian CS (1958) The concept of land units and land
system. Proceedings of the 9th Pacific Science Con-
gress 20: 74–81.
Christian CS and Stewart GA (1953) GeneralReporton
the Survey of the Katherine-Darwin Region, 1946.
CSIRO Land Research Series, 1. Melbourne, Aus-
tralia: CSIRO.
Church M (1996) Space, time and the mountain: How do
we order what we see? In: Rhoads BL and Thorn TE
(eds), The Scientific Nature of Geomorphology.Chi-
chester: John Wiley & Sons Ltd, 147–170.
Clements FE (1916) Plant Succession: An analysis of the
development of vegetation. Washington, DC: Carnegie
Institute, Publication 242.
Cody M and Diamond J (1975) Ecology and Evolution of
Communities. Cambridge, MA: Belknap Press of Har-
vard University Press.
Congalton RG and Green K (2008) Assessing the Accuracy
of Remotely Sensed Data: Principles and practices, 2nd
edn. Boca Raton, CA: CRC Press.
Crampton JW (20 10) Mapping: A critical introduction to
cartography and GIS, 1st edn. Malden, MA: Wiley-
Cullum CJ (2015) Integrating stream networks and land-
scape mosaics in a new conceptualisation of savanna
landscapes. PhD thesis. University of the Witwa-
tersrand, South Africa.
Cullum C and Rogers K (2011) A Framework for the
Classification of Drainage Networks in Savanna
Landscapes. Report No. TT 498/11. Pretoria, SA:
Water Research Commission.
Cullum C, Brierley G and Thoms M (2008) The spatial
organization of river systems. In: Brierley GJ and Fryirs
K(eds),River Futures. An integrative scientific
approach to river repair. Washington, DC: Island Press.
Currie WS (2011) Units of nature or processes across
scales? The ecosystem concept at age 75. New Phy-
tologist 190(1): 21–34.
Cushman SA, Huettmann F, Gutzweiler K, et al. (2010) The
gradient paradigm: A conceptual and analytical frame-
work for landscape ecology. In: Cushman SA and
Huettmann F (eds), Spatial Complexity, Informatics, and
Wildlife Conservation. Tokyo: Springer Japan, 83–108.
Deng YX (2007) New trends in digital terrain analysis:
Landform definition, representation and classification.
Progress in Physical Geography 31(4): 405–419.
Deng YX and Wilson JP (2006) The role of attribute
selection in GIS representations of the biophysical
environment. Annals of the Association of American
Geographers 96: 47–63.
Driver A, Maze K, Rouget M, et al. (2005) National
Spatial Biodiversity Assessment 2004: Priorities for
biodiversity conservation in South Africa. Strelizia, 14,
Pretoria, South Africa: South African National Biodi-
versity Institute.
EEA (2014) Terrestrial Habitat Mapping in Europe: An
overview. EEA Technical report No.1/2014. Lux-
embourg: European Environment Agency.
Ewald J (2003) A critique for phytosociology. Journal of
Vegetation Science 14(2): 291–296.
Favis-Mortlock D (2013) Systems and complexity in geo-
morphology. In: Shroder JF (ed.), Treatise on Geomor-
phology. San Diego, CA: Academic Press, 257–270.
Fisher P, Wood J and Cheng T (2004) Where is Helvellyn?
Fuzziness of multi-scale landscape morphometry.
Transactions of the Institute of British Geographers
29(1): 106–128.
Fonstad MA and Marcus WA (2010) High resolution,
basin extent observations and implications for under-
standing river form and process. Earth Surface Pro-
cesses and Landforms 35(6): 680–698.
Fotheringham AS (1989) Scale-independent spatial anal-
ysis. In: Goodchild M and Gopal S (eds), The Accuracy
of Spatial Data Bases. London: Taylor & Francis,
Franklin J (1995) Predictive vegetation mapping: Geo-
graphic modelling of biospatial patterns in relation to
environmental gradients. Progress in Physical Geo-
graphy 19(4): 474–499.
Frissell CA, Liss WJ, Warren CE, et al. (1986) A hier-
archical framework for stream habitat classification:
Viewing streams in a watershed context. Environmen-
tal Management 10(2): 199–214.
Fryirs K (2013) (Dis)connectivity in catchment sediment
cascades: A fresh look at the sediment delivery problem.
Earth Surface Processes and Landforms 38(1): 30–46.
Fryirs KA and Brierley GJ (2012) Geomorphic Analysis of
River Systems: An approach to reading the landscape.
Chichester, UK; Hoboken, NJ: Wiley-Blackwell.
Cullum et al. 61
at The University of Auckland Library on January 30, 2016ppg.sagepub.comDownloaded from
Funtowicz SO and Ravetz JR (1993) Science for the post-
normal age. Futures 25(7): 739–755.
Gerrard J (1992) Soil Geomorphology: An integration of
pedology and geomorphology. London: Chapman and
Gertenbach WPD (1983) Landscapes of the Kruger
National Park. Koedoe 26(1): 9–121.
Gillet F and Gallandat J-D (1996) Integrated synusial
phytosociology: Some notes on a new, multiscalar
approach to vegetation analysis. Journal of Vegetation
Science 7(1): 13–18.
Gleason HA (1926) The individualistic concept of the
plant association. Bulletin of the Torrey Botanical Club
53(1): 7–26.
Grabowski RC, Surian N and Gurnell AM (2014)
Characterizing geomorphological change to support
sustainable river restoration and management: Char-
acterizing geomorphological change in rivers. Wiley
Interdisciplinary Reviews: Water 1(5): 483–512.
Grime JP and Pierce S (1999) The Evolutionary Strategies
that Shape Ecosystems. Cambridge, UK: Cambridge
University Press.
Hack JT (1957) Studies of longitudinal stream profiles in
Virginia and Maryland. US Geological Survey Pro-
fessional Paper 294B: 45–97.
Harris GP and Heathwaite AL (2012) Why is achieving
good ecological outcomes in rivers so difficult?
Freshwater Biology 57(Supp. S1): 91–107.
Hazeu GW, Metzger MJ, Mu
cher CA, et al. (2011) Eur-
opean environmental stratifications and typologies: An
overview. Agriculture, Ecosystems & Environment
142(1–2): 29–39.
Hengl T (2006) Finding the right pixel size. Computers
and Geosciences 32(9): 1283–1298.
Holling CS (1973) Resilience and stability of ecological
systems. Annual Review of Ecology & Systematics 4:
Holling CS (2001) Understanding the complexity of eco-
nomic, ecological and social systems. Ecosystems 4:
Horton R (1945) Erosional development of streams and
their drainage basins: Hydrophysical approach to
quantitative morphology. Bulletin of the Geological
Society of America 56(3): 275–370.
Howard AD (1967) Drainage analysis in geologic inter-
pretation: A summation. In: Coates DR and Vitek JD
(eds), Thresholds in Geomorphology. London, UK:
George Allen and Unwin, 227–258.
Huggett RJ (1975) Soil landscape systems: Model of soil
genesis. Geoderma 13(1): 1–22.
Jelinski DE and Wu J (1996) The modifiable areal unit
problem and implications for landscape ecology.
Landscape Ecology 11(3): 129–140.
Jenny H (1941) Factors of Soil Formation. New York:
McGraw Hill.
Junk WJ, Bayley PB and Sparks RE (1989) The flood pulse
concept in river-floodplain systems. Canadian Special
Publication of Fisheries and Aquatic Sciences 106:
Kareiva P, Watts S, McDonald R, et al. (2007) Domes-
ticated nature: Shaping landscapes and ecosystems for
human welfare. Science 316(5833): 1866–1869.
Kay JJ, Regier HA, Boyle M, et al. (1999) An ecosystem
approach for sustainability: Addressing the challenge
of complexity. Futures 31(7): 721–742.
Keddy P (2007) Plants and Vegetation: Origins, pro-
cesses, consequences. Cambridge, UK: Cambridge
University Press.
Kennedy BA (1979) A naughty world. Transactions of the
Institute of British Geographers 4(4): 550–558.
Kleynhans CJ, Thirion C and Moolman J (2005) A Level I
River Ecoregion Classification System for South
Africa, Lesotho and Swaziland. N/000/00/REQ0104.
Pretoria, SA: Resource Quality Services, Department
of Water Affairs and Forestry.
chler AW (1964) Potential Natural Vegetation of the
Conterminous United States. Special Publication No.
36. New York: American Geographical Society.
Lagacherie P and McBratney AB (2006) Spatial soil
information systems and spatial soil inference systems:
Perspectives for digital soil mapping. In: Lagacherie P,
McBratney AB and Voltz M (eds), Digital Soil Map-
ping: An introductory perspective. Developments in
Soil Science, vol. 31. Amsterdam: Elsevier, 3–22.
Lave R, Wilson MW, Barron ES, et al. (2014) Interven-
tion: Critical physical geography. The Canadian Geo-
grapher/Le Ge
ographe Canadien 58(1): 1–10.
Levin SA (1992) The problem of pattern and scale in
ecology. Ecology 73(6): 1943–1967.
Levin SA (1998) Ecosystems and the biosphere as com-
plex adaptive systems. Ecosystems 1(5): 431–436.
Levins R (1966) The strategy of model building in popu-
lation ecology. American Science 54(4): 421–431.
Loidi J and Ferna´ndez-Gonza´lez F (2012) Potential natural
vegetation: Reburying or reboring? Journal of Vege-
tation Science 23(3): 596–604.
62 Progress in Physical Geography 40(1)
at The University of Auckland Library on January 30, 2016ppg.sagepub.comDownloaded from
Loveland TR and Merchant JM (2004) Ecoregions
and ecoregionalization: Geographical and ecological
perspectives. Environmental Management 34(Supp.1):
McGarigal K and Cushman SA (2005) The gradient concept
of landscape structure. In: Weins JA and Moss MR (eds),
Issues and Perspectives in Landscape Ecology.Cam-
bridge, UK: Cambridge University Press, pp. 112–119.
McMahon G, Wiken EB and Gauthier DA (2004) Toward
a scientifically rigorous basis for developing mapped
ecological regions. Environmental Management
34(Supp. 1): S111–S124.
MacMillan RA and Shary PA (2009) Landforms and
landform elements in geomorphometry. In: Hengl T
and Reuter HI (eds), Geomorphometry Concepts,
Software, Applications. Developments in Soil Science
No. 33. Elsevier, Amsterdam, pp. 227–254.
MacMillan RA, Martin TC, Earle TJ, et al. (2003) Automated
analysis and classification of landforms using high-
resolution digital elevation data: Applications and issues.
Canadian Journal of Remote Sensing 29(5): 592–606.
MacVicar GN, Scotney DM, Skinner TE, et al. (1974) A
classification of land (climate, terrain form, soil) pri-
marily for rainfed agriculture. South Africa Journal of
Agriculture Extension 3: 8–24.
Melles SJ, Jones NE and Schmidt B (2012) Review of
theoretical developments in stream ecology and their
influence on stream classification and conservation
planning. Freshwater Biology 57(3): 415–434.
Milne G (1935) Some suggested units for classification
and mapping, particularly for East African soils. Soil
Research 4(2): 183–198.
Mitchell S (2009) Unsimple Truths: Science, complexity
and policy. Chicago, IL: University of Chicago Press.
Moller M, Volk M, Friedrich K, et al. (2008) Placing soil-
genesis and transport processes into a landscape con-
text: A multiscale terrain-analysis approach. Journal of
Plant Nutrition and Soil Science/Zeitschrift Fur
Pflanzenernahrung Und Bodenkunde 171(3): 419–430.
Montgomery DR (1999) Process domains and the river
continuum. Journal of the American Water Resources
Association 35(2): 397–410.
Moore ID, Gessler PE, Nielsen GA, et al. (1993) Soil
attribute prediction using terrain analysis. Soil Science
Society of America Journal 57(2): 443–452.
Mucina L (2010) Floristic-phytosociological approach,
potential natural vegetation and survival of prejudice.
Lazaroa 31: 173–182.
Nowotny H, Scott P and Gibbons M (2001) Re-Thinking
Science: Knowledge and the public in an age of
uncertainty. Cambridge, UK: Polity Press.
Olden JD, Kennard MJ and Pusey BJ (2012) A framework
for hydrologic classification with a review of meth-
odologies and applications in ecohydrology. Ecohy-
drology 5(4): 503–518.
Olson DM, Dinerstein E, Wikramanayake ED, et al. (2001)
Terrestrial ecoregions of the world: A new map of life
on earth. Bioscience 51(11): 933–938.
Omernik JM (1987) Ecoregions of the conterminous
United States. Annals of the Association of American
Geographers 77(1): 118–125.
Omernik JM (2004) Perspectives on the nature and defi-
nition of ecological regions. Environmental Manage-
ment 34(1): S27–S38.
O’Neill RV (2001) Is it time to bury the ecosystem con-
cept? (With full military honors of course! ). Ecology
82(12): 3275–3284.
O’Neill RV, DeAngelis DL, Waide JB, et al. (1986) A
Hierarchical Concept of Ecosystems. Princeton, New
Jersey: Princeton University Press.
Openshaw S (1984) The Modifiable Areal Unit Problem
Concepts and Techniques in Modern Geography.
Norwich, UK: Geo Books.
Paola C, Foufoula-Georgiou E, Dietrich WE, et al.
(2006) Toward a unified s cience of the Earth’s
surface: Opportunities for synthesis among hydrology,
geomorphology, geochemistry and ecology. Water
Resources Research 42: W03S10. DOI: 10.1029/
Park SJ and Burt TP (2002) Identification and character-
ization of pedogeomorphological processes on a hill-
slope. Soil Science Society of America Journal 66(6):
Parsons ME and Thoms MC (2007) Hierarchical pat-
terns of large woody debris distribution and
macroinvertebrate-environment associations in river
ecosystems. Geomorphology 89: 127–146.
Paton TR, Humpreys GS and Mitchell PB (1995) Soils: A
new global view. London: UCL Press.
Pedrotti F (2013) Plant and Vegetation Mapping. Berlin
and Heidelberg, Germany: Springer.
Perry GLW (2002) Landscapes, space and equilibrium:
Shifting viewpoints. Progress in Physical Geography
26(3): 339–359.
Petts GE and Amoros C (1996) Fluvial hydrosystems.
London: Chapman & Hall.
Cullum et al. 63
at The University of Auckland Library on January 30, 2016ppg.sagepub.comDownloaded from
Phillips JD (2004) Laws, contingencies, irreversible
divergence and physical geography. The Professional
Geographer 56(1): 37–43.
Phillips JD (2007) The perfect landscape. Geomorphology
84: 159–169.
Phillips JD (2012) Storytelling in earth sciences: The eight
basic plots. Earth-Science Reviews 115(3): 153–162.
Pickett STA and Cadenasso ML (2002) The ecosystem as
a multidimensional concept: Meaning, model, and
metaphor. Ecosystems 5(1): 1–10.
Pickett STA, Kolasa J and Jones CG (2007) Ecological
Understanding: The nature of theory and the theory of
nature, 2nd edn. San Diego, CA: Elsevier.
Pike R (1988) The geometric signature: Quantifying
landslide-terrain types from digital elevation models.
Mathematical Geology 20(5): 491–511.
Poff NL (1997) Landscape filters and species traits:
Towards mechanistic understanding and prediction in
stream ecology. Journal of the North American Ben-
thological Society 16(2): 391–409.
Poole GC (2002) Fluvial landscape ecology: Addressing
uniqueness within the river discontinuum. Freshwater
Biology 47(4): 641–660.
Popper KR (1959) The Logic of Scientific Discovery.
Primary Material: Book Subject(s): Logic. Science-
Methodology. London: Hutchinson.
Porporato A, D’Odorico P, Laio F, et al. (2003) Hydro-
logic controls on soil carbon and nitrogen cycles: I .
Modeling scheme. Advances in Water Resources
26(1): 45–58.
Qi F, Zhu AX, Harrower M, et al. (2006) Fuzzy soil
mapping based on prototype category theory. Geo-
derma 136(3-4): 774–787.
Rodriguez-Iturbe I (2000) Ecohydrology: A hydrologic
perspective of climate-soil-vegetation dynamics. Water
Resources Research 36(1): 3–9.
Rogers KH and O’Keefe J (2003) River heterogeneity:
Ecosystem structure, function and management. In: Du
Toit JT, Rogers K and Biggs HC (eds), The Kruger
Experience: Ecology and management of savanna
heterogeneity. Washington, DC: Island Press, 189–218.
Rosgen DL (1994) A classification of natural rivers.
Catena 22(3): 169–199.
Rossum S and Lavin S (2000) Where are the great plains?
A cartographic analysis. The Professional Geographer
52(3): 543–552.
Rowe JS (1961) The level-of-integration concept and
ecology. Ecology 42(2): 420–427.
Salthe SN (1985) Evolving Hierarchical Systems. New
York: Columbia University Press.
Sayre R, Comer P, Warner H, et al. (2009) ANewMapof
Standardized Terrestrial Ecosystems of the Conterminous
United States. US Geological Survey Professional Paper
No. 1768. Reston, VA: US Geological Survey.
Schamine´e JHJ, Hennekens SM, Chytry
M, et al. (2009)
Vegetation-plot data and databases in Europe: An
overview. Preslia 81(3): 173–185.
Scheffer M, Carpenter S, Foley JA, et al. (2001) Catastrophic
shifts in ecosystems. Nature 413(6856): 591–596.
Schmidt J and Andrew R (2005) Multi-scale landform
characterization. Area 37(3): 341–350.
Schneider ED and Kay JJ (1994) Complexity and ther-
modynamics: Towards a new ecology. Futures 26(6):
Schumm SA (1977)
The Fluvial System. New York: John
Wiley and Sons.
Scull P, Franklin J, Chadwick OA, et al. (2003) Predictive
soil mapping: A review. Progress in Physical Geo-
graphy 27(2): 171–197.
Shary PA, Sharaya LS and Mitusov AV (2002) Funda-
mental quantitative methods of land surface analysis.
Geoderma 107(1–2): 1–32.
Shi X, Long R, Dekett R, et al. (2009) Integrating d if-
ferent types of knowledge for digital soil mapping.
Soil Science Society of America J ournal 73(5):
Simon A, Doyle M, Kondolf M, et al. (2007) Critical
evaluation of how the Rosgen classification and asso-
ciated ‘natural channel design’ methods fail to integrate
and quantify fluvial processes and channel response.
Journal of the American Water Resources Association
43(5): 1117–1131.
Singers NJD and Rogers GM (2014) A Classification
of New Zealand’s Terrestrial Ecosystems. Science
for Conservation, 325. Wellington, New Zealand:
Department of Conservation.
Stallins JA (2006) Geomorphology and ecology: Unifying
themes for complex systems in biogeomorphology.
Geomorphology 77(3-4): 207–216.
Stirzaker RJ, Roux DJ and Biggs HC (2011) Learning to
bridge the gap between adaptive management and
organisational culture. Koedoe 53(2): 28–33.
Strunz S (2012) Is conceptual vagueness an asset?
Arguments from philosophy of science applied to the
concept of resilience. Ecological Economics 76:
64 Progress in Physical Geography 40(1)
at The University of Auckland Library on January 30, 2016ppg.sagepub.comDownloaded from
Tadaki M, Brierley G, Dickson M, et al. (2015) Cultivating
critical practices in physical geography. The Geo-
graphical Journal 181(2): 160–171.
Tadaki M, Brierley G and Cullum C (2014) River classifi-
cation: Theory, practice, politics. Wiley Interdisciplinary
Reviews: Water 1(4): 349–367.
Thorp JH, Thoms MC and Delong MD (2006) The riverine
ecosystem synthesis: Biocomplexity in river networks
across space and time. River Research and Applica-
tions 22(2): 123–147.
Thorp JH, Thoms MC and Delong MD (2010) The
Riverine Ecosystem Synthesis: Toward conceptual
cohesiveness in rive r science.Amsterdam,The
Netherlands: Elsevier.
Tockner K, Pusch M, Gessner J, et al. (2011) Domesticated
ecosystems and novel communities: Challenges for the
management of large rivers. Ecohydrology & Hydro-
biology 11(3–4): 167–174.
Turner M, Romme WH, Gardner RH, et al. (1993) A
revised concept of landscape equilibrium: Disturbance
and stability on scaled landscapes. Landscape Ecology
8: 213–227.
Turner MG (2005) Landscape ecology in North America:
Past, present, and future. Ecology 86(8): 1967–1974.
xen R (1956) Die heutige potentielle natu
rliche Vege-
tation als Gegenstand der Vegetationskartierung. Ange-
wandte Pflanzensoziologie, 13. Stolzenau, Germany:
Udvardy MD (1975) A Classification of the Biogeo-
graphical Provinces of the World.IUCNOccasional
paper No.18. Moreges, Switzerland: International Union
for Conservation of Nature and Natural Resources.
Ulanowicz RE (2009) A third Window: Natural life beyond
Newton and Darwin. West Conshohocken, PA: Tem-
pleton Foundation Press.
Vannote RL, Minshall GW, Cummins KW, et al. (1980)
River continuum concept. Canadian Journal of Fish-
eries and Aquatic Sciences 37(1): 130–137.
Venter FJ (1990) A classification for land management
planning in the Kruger National Park. PhD thesis.
University of Pretoria, South Africa.
Ward JV (1989) The four-dimensional nature of lotic
ecosystems. Journal of the North American Bentholo-
gical Society 8(1): 2–8.
Waterton C (2002) From field to fantasy: Classifying
nature, constructing Europe. Social Studies of Science
32(2): 177–204.
Weiher E and Keddy P (2001) Ecological Assembly Rules:
Perspectives, advances, retreats. Cambridge, UK:
Cambridge University Press.
Westhoff V and van der Maarel E (1978) The Braun-
Blanquet approach. In: Whittaker RH (ed.), Classifi-
cation of Plant Communities. The Hague, The Neth-
erlands: Junk, 287–297.
Whittaker RH (1967) Gradient analysis of vegetation.
Biological Reviews 4(2): 207–264.
Wiken EB (1986) Terrestrial ecozones of Canada . Eco-
logical Land Classification Series, No.23. Lands
Directorate, Ottawa: Environment Canada.
Willner W (2006) The association concept revisited.
Phytocoenologia 36(1): 67–76.
Wohl E (2013) The complexity of the real world in the
context of the field tradition in geomorphology. Geo-
morphology 200: 50–58.
Wood JD (1996) The geomorphological characterisation
of digital elevation models. PhD Thesis. University of
Leicester, UK.
Wood WF and Snell JB (1960) A Quantitative System for
Classifying Landforms. Natick, MA: Quartermaster
Research and Engineering Command, Defense Tech-
nical Information Center.
Wu J (1999) Hierarchy and scaling: Extrapolating infor-
mation along a scaling ladder. Canadian Journal of
Remote Sensing 25(4): 367–380.
Wu J and Hobbs R (2002) Key issues and research prio-
rities in landscape ecology: An idiosyncratic synthesis.
Landscape Ecology 17(4): 355–365.
Wu JG (2012) Key concepts and research topics in land-
scape ecology revisited: 30 years after the Allerton
Park workshop. Landscape Ecology 28(1): 1–11.
Wu JG and David JL (2002) A spatially explicit hier-
archical approach to modelling complex ecological
systems: Theory and applications. Ecological Model-
ling 153(1–2): 7–26.
Zadeh LA (1996) Fuzzy logic ¼ computing with
words. IEEE Transactions on Fuzzy Systems 4(2):
Zadeh LA (2008) Is there a need for fuzzy logic? Infor-
mation Sciences 178(13): 2751–2779.
Zhu AX, Burt JE, Smith M, et al. (2008) The impact of
neighbourhood size on terrain derivatives and digital
soil mapping. In: Zhou C, Lees BG and Tang G (eds),
Advances in Digital Terrain Analysis. Heidelberg,
Springer, pp. 333–348.
Cullum et al. 65
at The University of Auckland Library on January 30, 2016ppg.sagepub.comDownloaded from
... The landscape maps are developed by determination of spatial configuration of these components in the form of environmental variables in accordance with available geospatial data and knowledge. These landscape concepts are used for the hierarchical classification by applying the holistic approach [14]. In general, each region of the world has adopted different landscape typology, depending on regional or local characteristics of ecosystems, landscape concepts, and taxonomic organization approaches, or adapted to the needs of land management. ...
... This example also presents the possibilities of data synthesis of optical satellite images and digital elevation models (DEM) for landscape objects recognition. DEM and derived datasets (slope, aspect, ruggedness, topographic indices, river basin, and shady relief) have been used by researchers for geomorphological and geomorphometric landscape studies [14]. In fact, landforms such as uplands, slopes, terraces, and valleys have strong links with lithology and geomorphology, which make them good variables for landscape identification in different scales. ...
Full-text available
An analysis of the landscape spatial structure and diversity in the mountain ranges of Northeast Siberia is essential to assess how tundra and boreal landscapes may respond to climate change and anthropogenic impacts in the vast mountainous permafrost of the Arctic regions. In addition, a precise landscape map is required for knowledge-based territorial planning and management. In this article, we aimed to explore and enhanced methods to analyse and map the permafrost landscape in Orulgan Ridge. The Google Earth Engine cloud platform was used to generate vegetation cover maps based on multi-fusion classification of Sentinel 2 MSI and Landsat 8 OLI time series data. Phenological features based on the monthly median values of time series Normalized Difference Vegetation Index (NDVI), Green Normalized Difference Vegetation Index (GNDVI), and Normalized Difference Moisture Index (NDMI) were used to recognize geobotanical units according to the hierarchical concept of permafrost landscapes by the Support Vector Machine (SVM) classifier. In addition, geomorphological variables of megarelief (mountains and river valleys) were identified using the GIS-based terrain analysis and landform classification of the ASTER GDEM scenes mosaic. The resulting environmental variables made it possible to categorize nine classes of mountain permafrost landscapes. The result obtained was compared with previous permafrost landscape maps, which revealed a significant difference in distribution and spatial structure of intrazonal valleys and mountain tundra landscapes. Analysis of the landscape structure revealed a significant distribution of classes of mountain Larix-sparse forests and tundra. Landscape diversity was described by six longitudinal and latitudinal landscape hypsometric profiles. River valleys allow boreal–taiga landscapes to move up to high-mountainous regions. The features of the landscape structure and diversity of the ridge are noted, which, along with the specific spatial organization of vegetation and relief, can be of key importance for environmental monitoring and the study of regional variability of climatic changes.
... Ground-cover classification maps have become a key resource in institutional, policy and law-making practice, but so far, there is a lack of conceptual frameworks and agreeable standards for map based monitoring [97]. Structuring ground-cover classes (i.e., in a hierarchical approach) can help with choosing appropriate class level detail and allow transferring across scales [97]. ...
... Ground-cover classification maps have become a key resource in institutional, policy and law-making practice, but so far, there is a lack of conceptual frameworks and agreeable standards for map based monitoring [97]. Structuring ground-cover classes (i.e., in a hierarchical approach) can help with choosing appropriate class level detail and allow transferring across scales [97]. In our study, we found a structured selection approach, in line with adaptive monitoring goals [98], appropriate to find the optimal class level detail. ...
Full-text available
The Arctic is under great pressure due to climate change. Drones are increasingly used as a tool in ecology and may be especially valuable in rapidly changing and remote landscapes, as can be found in the Arctic. For effective applications of drones, decisions of both ecological and technical character are needed. Here, we provide our method planning workflow for generating ground-cover maps with drones for ecological monitoring purposes. The workflow includes the selection of variables, layer resolutions, ground-cover classes and the development and validation of models. We implemented this workflow in a case study of the Arctic tundra to develop vegetation maps, including disturbed vegetation, at three study sites in Svalbard. For each site, we generated a high-resolution map of tundra vegetation using supervised random forest (RF) classifiers based on four spectral bands, the normalized difference vegetation index (NDVI) and three types of terrain variables—all derived from drone imagery. Our classifiers distinguished up to 15 different ground-cover classes, including two classes that identify vegetation state changes due to disturbance caused by herbivory (i.e., goose grubbing) and winter damage (i.e., ‘rain-on-snow’ and thaw-freeze). Areas classified as goose grubbing or winter damage had lower NDVI values than their undisturbed counterparts. The predictive ability of site-specific RF models was good (macro-F1 scores between 83% and 85%), but the area of the grubbing class was overestimated in parts of the moss tundra. A direct transfer of the models between study sites was not possible (macro-F1 scores under 50%). We show that drone image analysis can be an asset for studying future vegetation state changes on local scales in Arctic tundra ecosystems and encourage ecologists to use our tailored workflow to integrate drone mapping into long-term monitoring programs.
... One of the applications of Remote Sensing is the use of imagery to classify and delineate different objects and land cover types on the Earth's surface, a process that involves collecting field data from a series of samples as an input for training a classification model (Zou and Greenberg, 2019;Prentice et al., 2021). Classification in Remote Sensing involves the categorization of response functions recorded in imagery as representations of real-world objects, according to their spectral similarity to the initial values overlapping the samples, which can provide detailed information about land-cover, specifically in a mixed forest-grassland (Corbane et al., 2015;Cullum et al., 2016;Hamylton et al., 2020;Zhang et al., 2021). ...
Full-text available
Modern UAS (Unmanned Aerial Vehicles) or just drones have emerged with the primary goal of producing maps and imagery with extremely high spatial resolution. The refined information provides a good opportunity to quantify the distribution of vegetation across heterogeneous landscapes, revealing an important strategy for biodiversity conservation. We investigate whether computer vision and machine learning techniques (Object-Based Image Analysis—OBIA method, associated with Random Forest classifier) are effective to classify heterogeneous vegetation arising from ultrahigh-resolution data generated by UAS images. We focus our fieldwork in a highly diverse, seasonally dry, complex mountaintop vegetation system, the campo rupestre or rupestrian grassland, located at Serra do Cipó, Espinhaço Range, Southeastern Brazil. According to our results, all classifications received general accuracy above 0.95, indicating that the methodological approach enabled the identification of subtle variations in species composition, the capture of detailed vegetation and landscape features, and the recognition of vegetation types’ phenophases. Therefore, our study demonstrated that the machine learning approach and combination between OBIA method and Random Forest classifier, generated extremely high accuracy classification, reducing the misclassified pixels, and providing valuable data for the classification of complex vegetation systems such as the campo rupestre mountaintop grassland.
... We hope that the LEC system constructed by this research can provide theoretical support for decision-makers, government, engineering construction personnel, and other researchers [56]. Each landscape patch unit is a discrete system generated by grid and geology, geomorphology, soil, vegetation, climate, wildlife, water, man, and many other factors [57]. Not all landscapes can be (easily) decomposed into a set of structural-functional units that can be clearly and unambiguously delineated and linked to explanatory conceptual models [58]. ...
Full-text available
Landscape Ecological Classification (LEC) is the premise and foundation of landscape ecology research. The current research on LEC of Mining Cities in the Semi-arid Steppe (MCSS) is relatively low. Moreover, the question of how to classify the mining landscape into ecologically significant landscape units at a scale suitable for ecological management has not been clear. The research results are as follows: (1) Google Earth, Gaode map, Baidu map, various high-resolution images, unmanned aerial vehicle, and field investigation were used to investigate the landscape types. Based on the land classification of the occurrence model, integrating theories of landscape ecology, mining, ecology, geography, and land resources, this study constructed the LEC system for MCSS using the top-down decomposition classification method, including 4 types of landscape kingdom, 16 types of landscape class, 62 types of landscape family, and more than 200 types of landscape species. (2) Based on LEC, we found the landscape type evolution characteristics of MCSS. Both the open-pit landscape and the dumping landscape were constantly expanding, and ecological restoration of the mining area was carried out simultaneously with coal mining. The trend of Change Intensity (CI) of mining industrial square landscape and industrial storage landscape was very similar. The development of coal has driven the development of the regional industry. The expansion intensity of the town commercial and residential service landscape was gradually decreasing, and the motivation for town expansion was insufficient. The research area was a typical landscape evolution mode of “human advance and grassland retreat”. However, the intensity of humans occupying grassland was decreasing. This study provides a reference for the research of LEC in the semi-arid steppe and provides a theoretical basis for the landscape ecological assessment, planning, and management of mining cities.
... The floristic composition of a plant community in a geographical region occurs through the processes of adaptation, competition, and natural selection [46,47]. The pattern of a few abundant species, often referred to as a dominant species, and many rarer species are defining characteristics of communities worldwide [48]. ...