REVIEWS
Ecological Monographs, 84(4), 2014, pp. 533–561
Ó 2014 by the Ecological Society of America
EcoVeg: a new approach to vegetation description and classification
DON FABER-LANGENDOEN,
1,11
TODD KEELER-WOLF,
2
DEL MEIDINGER,
3,12
DAVE TART,
4
BRUCE HOAGLAND,
5
CARMEN
JOSSE,
1
GONZALO NAVARRO,
6
SERGUEI PONOMARENKO,
7
JEAN-PIERRE SAUCIER,
8
ALAN WEAKLEY,
9
AND PATRICK COMER
10
1
NatureServe, Conservation Science Division, 4600 North Fairfax Drive, Arlington, Virginia 22203 USA
2
Biogeographic Data Branch, California Department of Fish and Game, Sacramento, California 95814 USA
3
British Columbia Ministry of Forests and Range, Research Branch, Victoria, British Columbia V8W 9C2 Canada
4
USDA Forest Service, Intermountain Region, Natural Resources, Ogden, Utah 84401 USA
5
Oklahoma Biological Survey and Department of Geography, University of Oklahoma, Norman, Oklahoma 73019 USA
6
Universidad Cato
´
lica Boliviana San Pablo, Unidad Acade
´
mica Regional Cochabamba Departamento de Ciencias
Exactas e Ingenierı
´
as, Carrera de Ingenierı
´
a Ambiental, Cochabamba, Bolivia
7
Ecological Integrity Branch, Parks Canada, Rue Eddy, Gatineau, Quebec K1A 0M5 Canada
8
Ministe
`
re des Ressources Naturelles 2700, Rue Einstein, Bureau B-1-185, Quebec City, Quebec G1P 3W8 Canada
9
North Carolina Botanic Garden, University of North Carolina, Chapel Hill, North Carolina 27599 USA
10
NatureServe, 2108 55th Street, Boulder, Colorado 80301 USA
Abstract. A vegetation classification approach is needed that can describe the diversity of
terrestrial ecosystems and their transformations over large time frames, span the full range of
spatial and geographic scales across the globe, and provide knowledge of reference conditions
and current states of ecosystems required to make decisions about conservation and resource
management. We summarize the scientific basis for EcoVeg, a physiognomic-floristic-ecological
classification approach that applies to existing vegetation, both cultural (planted and dominated
by human processes) and natural (spontaneously formed and dominated by nonhuman
ecological processes). The classification is based on a set of vegetation criteria, including
physiognomy (growth forms, structure) and floristics (compositional similarity and character-
istic species combinations), in conjunction with ecological characteristics, including site factors,
disturbance, bioclimate, and biogeography. For natural vegetation, the rationale for the upper
levels (formation types) is based on the relation between global-scale vegetation patterns and
macroclimate, hydrology, and substrate. The rationale for the middle levels is based on scaling
from regional formations (divisions) to regional floristic-physiognomic types (macrogroup and
group) that respond to meso-scale biogeographic, climatic, disturbance, and site factors. Finally,
the lower levels (alliance and association) are defined by detailed floristic composition that
responds to local to regional topo-edaphic and disturbance gradients. For cultural vegetation,
the rationale is similar, but types are based on distinctive vegetation physiognomy and floristics
that reflect human activities. The hierarchy provides a structure that organizes regional/
continental vegetation patterns in the context of global patterns. A formal nomenclature is
provided, along with a descriptive template that provides the differentiating criteria for each type
at all levels of the hierarchy. Formation types have been described for the globe; divisions and
macrogroups for North America, Latin America and Africa; groups, alliances and associations
for the United States, parts of Canada, Latin America and, in partnership with other
classifications that share these levels, many other parts of the globe.
Key words: biogeography; Canadian National Vegetation Classification; cultural vegetation; ecosys-
tem; floristics; growth form; International Vegetation Classification; natural vegetation; novel ecosystem;
ruderal vegetation; U.S. National Vegetation Classification; vegetation type.
INTRODUCTION
There never has been greater need than now to
systematically inventory, classify, and map the incred-
ible diversity of vegetation and ecosystems on Earth as
land managers, conservationists, and policy makers are
facing ever intensifying land uses and degraded land-
Manuscript received 19 December 2013; revised 5 March
2014; accepted 15 April 2014; final version received 12 May
2014. Corresponding Editor (ad hoc): N. Christensen.
11
E-mail: don_faber-langendoen@natureserve.org
12
Present address: Meidinger Ecological Consultants, 639
Vanalman Avenue, Victoria, British Columbia V8Z 3A8 Canada.
533
See last page for erratum on nomenclature
scapes. The implications of global change for biodiver-
sity, ecological processes, and ecosystem services are
profound, even as historic natural systems are replaced
by new or novel ecosystems. A paramount need for
assessing these alterations is a system of vegetation
classification that is operable at multiple spatial and
temporal scales of resolution. Although vegetati on-
based classifications are often eschewed because of the
heterogeneous and dynamic nature of vegetation,
modelers of both climate and land-cover changes also
recognize the merits of describing the dynamics of
vegetation types (Leemans 1997, Williams et al. 2000,
Mitchell 2005, Willis and Birks 2006, Beckage et al.
2008, Chiarucci et al. 2010, Williams and Baker 2011).
In addition, ecologists and conservation scientists need
real-time knowledge of ecosystem structure and compo-
sition in order to characterize reference conditions and
natural disturbance dynamics across the landscape
(Swetnam et al. 1999, Scott et al. 2002, Stoddard et al.
2006, Leu et al. 2008, Keene et al. 2009, Tierney et al.
2009, Thompson et al. 2013).
Vegetation ecologists acknowledge the need for more
comprehensive systematic approaches to both vegeta-
tion survey and classification (e.g., Chytry´ et al. 2011).
Although vegetation classifications are a priority in
many parts of the world, the systems devised cater to
national or subcontinental interests and scale (e.g.,
Curtis 1959, Rodwell 1991–2000, Sawyer et al. 2009,
Navarro 2011, Chytry´ 2012), thereby limiting the need
for the classifiers to account for the worldwide diversity
of vegetation patterns. A globally applicable classifica-
tion system is lacking.
We present a hierarchical classification that integrates
biogeography, bioclimatology, and land-cover data into
a scientifically based global vegetation classification for
the interpretation of vegetation pattern at all scales. Our
methodology, which we term the EcoVeg approach,
provides a repeatable scientific system for the develop-
ment and description of vegetation types. The goal is to
systematically classify e xisting vegetation, reflecti ng
both ecological and human processes and applicable
from the global to local scale. While we do not argue
that this is the only vegetation classification approach to
use, we will show that it does address an important set of
current needs and solves other classification shortcom-
ings.
The EcoVeg a pproach builds on the traditional
physiognomic-floristic-ecological classifications that
have been developed over many years (e.g., Ru
¨
bel
1930, as cited in Shimwell 1971, Whittaker 1962,
Westhoff 1967, Webb et al. 1970, Beard 1973, Werger
and Sprangers 1982, Borhidi 1991, Adam 1992). These
classifications suggested ways in which multiple criteria
for vegetation classification could be used to organize
vegetation patterns along ecological lines. Common to
these authors’ perspectives is that both floristic and
physiognomic units should be constructed in the context
of ecological relationships. As Warming stated early in
the last century (1909:142), ‘‘Why not use each growth
form [lichen, moss, herb, dwarf-shrub, shrub, tree] as a
foundation upon which to build a special class? The
following classes could then be distinguished: that of
forest formations, of bush-formations, of shrub-forma-
tions, of dwarf-shrub formations, of perennial-herb
formations, of moss-formations, and of alga-forma-
tions ...from a morphological standpoint this would
possess a certain interest, but from a phytogeographical
one it must be dismissed, because it would involve the
separation of formations that are oecologically closely
allied.’’ (Emphasis added.)
Similarly, Daubenmire (1968:252) observed that ‘‘...a
‘needle-leaved coniferous forest’ category would em-
brace the Pinus elliottii forests of Cuba, the Pinus
ponderosa forests of Colorado, the Sequoia sempervirens
forests of California, the Picea glauca forests of Yukon
Territory, etc. Collectively these share nothing in
common from the synecologic standpoint ...thus it is
clear that physiognomy by itself lumps vegetation types
that are vastly different in their ecological relations, and
so results in an artificial classification. Then the opposite
difficulty is illustrated by Warming’s placement of salt
marshes dominated by shrubby Salicornia in a different
category from salt marshes dominated by herbaceous
species of the same genus ...all this is not to deny that
physiognomy can serve a useful purpose in defining
major plant groupings, but it is useful only when
ecologic and other considerations are allowed to govern
its application.’’
Floristic approaches, such as those of Braun-Blanquet
or Daubenmire, often give strong consideration to
ecological relationships when assessing vegetation types
(Westhoff and van der Maarel 1973:619). In fact, the
historic association concept typically includes habitat
conditions (Mueller-Dombois and Ellenberg 1974, Will-
ner 2006, Jennings et al. 2009). Thus, both physiognomic
and floristic chara cteristics can provide the biotic
information needed for defining vegetation classification
units, and the organization of their relationships can be
assessed by their ecological, dynamic, and geographic
(chorology) relevance (see also Pignatti et al. [1994]).
There are other vegetation characteristics that are still
being explored for their role in vegetation classification.
For example, inductive approaches to characterizing
plant functional traits are now being gathered at the
same scale as floristic data, such that classifications may
benefit from considering these traits for classification
(Box 1981, Cramer 1997, Gillison 2013). Although these
fine-scale traits can be used to characterize environmen-
tally adaptive aspects of plants, more research is needed
to understand how they might be incorporated with
physiognomic and floristic criteria for classification
purposes. We note how information on functional traits
could be used to extend our approach.
One of the challenges of a global classification is to
provide guidance for type recognition across all scales.
An a ccurate , concise definition for a gl obal-sc ale
DON FABER-LANGENDOEN ET AL.534
Ecological Monographs
Vol. 84, No. 4
REVIEWS
tropical dry forest formation may be difficult, given the
enormous diversity of tropical forests across the globe.
Because the principles of the EcoVeg hierarchy ulti-
mately bring global and local scales together, there is the
opportunity for ongoing refinement and improvement of
type definitions using both top-down and bottom-up
methods (e.g., Miles et al. 2006).
No other global vegetation classification approach
that we know of is based on an integration of
physiognomy and floristics across all vegetation types
at multiple scales. Perhaps the closest in scope to our
approach is that of Di Gregorio and Jansen (1996), but
theirs is a comprehensive descriptive method with
multiple attri butes organized around a few core
categories. There are also a number of important
continental or subcontinental physiognomic-floristic
classifications, though they differ from our approach
in a variety of ways, including in North America (Brown
et al. 1979, Brown 1982), the former USSR (Komarov
Botanical Institute; Aleksandrova 1973), in Australia
(Specht et al. 1974, Specht and Specht 2001), in Europe,
the physiognomic-floristic classification that guides the
Natural Vegetation Map of Europe (Bohn et al. 2000–
2003) and EUNIS (European Na ture Informati on
System; see Davies et al. [20 04]), with its various
predecessors (e.g., Devillers et al. 1991 [CORINE
biotopes manual] and Devillers and Devillers-Terschu-
ren 1996). Apart from Brown et al. (1979), the units and
hierarchy neither attempt to represent global patterns of
vegetation nor contain multiple scales of global to local
types. We therefore developed the EcoVeg approach to
address this need.
M
ETHODS
Working group
The classification approach presented here is the
product of numerous efforts that began in the 1970s
and 1980s to establish global ecological classification
frameworks (see Grossman et al. [1998] for a brief
overview of various physiognomic, floristic, and phys-
iognomic-floristic approaches). At that time, any effort
to develop a systematic vegetation classification in the
United States had to first respond to challenges to
community-unit concepts raised by continuum-based
theories (Austin 1985), disenchantment with strongly
floristic approaches based on the character species
concepts of the Braun-Blanquet metho d (Mueller-
Dombois and Ellenberg 1974:208–209), which led to a
greater emphasis on the characteristic species combina-
tion (Chytry´ and Tichy´ 2003), and a desire to explore
more bio-geo-ecosystem approaches (Driscoll et al.
1984, Pojar et al. 1987, Bailey 1989a, Rowe and Barnes
1994). Concurrently, however, improvements in remote
sensing technology and spatial modeling tools created a
demand for a classification system that was consistent,
repeatable, and operable at multiple spatial scales for
characterization of vegetation-ecosystem patterns from
both the ground and the air (Lowry et al. 2007).
In 2003, the Vegetation Subcommittee of the United
States Federal Geographic Data Committee (FGDC
2008) sponsored the Hierarchy Revisions Working
Group (HRWG) to address shortcomings in the
UNESCO (1973) physiognomic-ecologic classification
that formed the basis for the first iteration of the United
States National Vegetation Classification (USNVC; see
Grossman et al. [1998]). UNESCO (1973) was intended
to facilitate global vegetation classification and mapping
using remote sensing imagery. Because of an FGDC
requisite that international standards for vegetation
classification be considered during development of the
USNVC, the HRWG was composed of vegetation
ecologists from across the western hemisphere (see
Appendix A for a list of members), and it sought peer
review from international experts. The HRWG focused
on the conceptual development of the upper and middle
level units of the hierarchy to complement the lower
level (alliance and association) units already put in place
by Grossman et al. (1998). The FGDC (2008) and
Jennings et al. (2009) provided guidance for the
collection of vegetation field data from plots and for
the description and analysis of lower level units of the
classification, guidance that is relevant to the descrip-
tions of all levels of the hierarchy presented here. They
also provided a glossary of terms. A brief introduction
to the USNVC was given in Faber-Langendoen et al.
(2009), Franklin et al. (2012), and Kent (2012). An
introduction and systematic description of the global
formation types is provided in Faber-Langendoen et al.
(2012).
The terrestrial vegetation focus of EcoVeg is based on
assumptions that vegetation represents the majority of
primary production of terrestrial ecosystems, is readily
observable, and to a large degree integrates the biotic
response to a variety of abiotic and disturbance factors
at local, regional, and global scales. Thus, the approach
adopts a bio-ecosystems (Walter 1985), as opposed to a
geo-ecosystems (Rowe and Barnes 1994) approach, and
is largely synonymous with the natural or ecological
community concept used by a variety of state agencies
and organizations within the United States (Grossman
et al. 1998).
The conceptual development and description of
EcoVeg units draws on important ecological products
from bioclimatic (e.g., Holdridge 1947, Pojar et al. 1987,
Rivas-Martı
´
nez 1996–2011, Metzger et al. 2012), bio-
geographic (e.g., Takhtajan 1986, Rivas-Martı
´
nez et al.
2011), and soils classifications (USDA 1999, Eswaran et
al. 2003) to facilitate the understanding of vegetation
patterns. In addition, the multiscale structure of the
EcoVeg approach is compatible with existing land cover
classifications that are utilized at national, continental,
and global scales (see USGS National Land Cover
Database [data available online];
13
Loveland et al. 1991,
13
http://landcover.usgs.gov/natllandcover.php
November 2014 535ECOLOGICAL VEGETATION CLASSIFICATION
REVIEWS
2000, Di Gregorio an d Jansen 1996, USGS 2001,
Bontemps et al. 2009, Fry et al. 2011).
Methodological principles
We contend that an operable vegetation classification
scheme must have the capacity to describe existing
vegetation patterns, including both cultural (planted and
dominated by human processes) and natural (spontane-
ously formed and dominated by ecological processes, cf.
van der Maarel 2005); describe vegetation types at
multiple thematic scales, from thematically coarse
formations (biomes) to fine-scale associations (bio-
topes); provide a readily interpretable inventory of
vegetation and ecosystem patterns within and across
landscape/ecoregional/watershed units; document status
and trends of vegetation and ecosystems (e.g., trends in
extent, such as range shifts, or trends in condition);
facilitate interpretation of long-term (even paleoecolog-
ical) change in vegetation with short-term change of
existing vegetation, based on multiple vegetation criteria
(growth forms, structure, floristics, etc.); and document
the real-time shifts in ecosystem states caused by
invasive species, land use, and climate change.
Based on these contentions, the EcoVeg approach
contains nine core principles.
1) The classification is based on existing vegetation
types, defined as the plant cover, including both
floristic composition and vegetation structure, docu-
mented at a specific location and time, under
specified ecological conditions, and preferably de-
scribed at an optimal time during the growing season
(Tart et al. 2005, FGDC 2008, Jennings et al. 2009).
This is in contrast to potential vegetation concepts
(Ku
¨
chler 1964, Daubenmire 1968, Loidi and Ferna
´
n-
dez-Gonza
´
lez 2012), which rely on assumptions
regarding vegetation successional stages, the presence
of selected late successional plant species, and
ecological species groups related to soils, topography,
and climatic factors in the description of vegetation
types. The two concepts are related in that mature
examples of existing vege tat ion may repres ent
expected states of potential vegetation types (Loidi
and Ferna
´
ndez-Gonza
´
lez 2012).
2) Vegetation types are characterized by full floristic
and growth form (physiognomic) composition, which
together express ecological and biogeographical
relations. Floristic data can provide joint species
responses to environment and disturbance, both in
the short and long term. These responses can be
indicators of environmental change, disturbance
regime shifts, and anthropogenic alterations. Growth
forms reflect ecological and evolutionary pressures
and processes; thus the composition of growth forms
expresses both the long-term and immediate set of
abiotic variables influenc ing v egetation s tructure
(Whittaker 1975, Werger and Sprangers 1982, Adam
1992). For example, Box (1981) defined 90 plant
growth-forms based on structural types (e.g., tree,
shrub, etc.), leaf form (e.g., broad-leaved macro-
phyll), relative plant and leaf size, and seasonal
activity pattern (e.g., summer green) for predictively
mapping world biomes or vegetation formations.
Thus, regional patterns of growth forms and species
constitute distinctive patterns that can be used to
define regional biomes, reflecting a long-term adjust-
ment of vegetation to sites (Williams et al. 2000). At
the same time , natural disturbances and human
activities can rapidly alter the growth forms and
species composition.
3) Vegetation characteristics are the product of natural
and cultural processes. Cultural processes are human
activities with purposeful, direct vegetation manage-
ment objectives that produce distinct suites of species
and growth forms (e.g., orchards, vineyards, row
crops, gardens, forest plantations). Natural processes
are ecologically driven and lead to more or less
spontaneous vegetation patterns (Ku
¨
chler 1969,
Westhoff and van der Maarel 1973, Di Gregorio
and Jansen 1996, van der Maarel 2005). See also
Basic categories of the EcoVeg approach: Natural and
cultural vegetation.
4) Characterizing and describing vegetation types is best
accomplished using plot data, including both vege-
tation and other ecological data, which is collected
and compiled using systematic protocols and survey
techniques. Data management tools, including bo-
tanical databases, vegetation plot databases, and
vegetation classification databases (Westhoff and van
der Maarel 1973, Dengler et al. 2011, Peet et al.
2012), are essential for these activities.
5) Vegetation types can be defined using a number of
differentiating criteria, including diagnostic, constant
and dominant species, dominant and diagnostic
growth forms, and compositional similarity. The
most useful criteria are those that express ecological
and biogeographical relationships and that clearly
distinguish types (Warming 1909, Curtis 1959,
Westhoff and van der Maarel 1973, Pignatti et al.
1994, Dierschke 1997, Willner 2006). These criteria
should be defined for application in the field or lab,
so that recognizable field characteristics are provided
to ensure consistent identification using keys and
other tools (De Ca
´
ceres and Wiser 2012). To that
end, types are preferably defined as extensive
concepts (the class concepts of Whittaker
[1962:114–118]). Extensive concepts describe the full
membership or range of variation of a type in
relation to other types (e.g., as shown in Austin
[2013: Fig. 3.4]), as compared to intensive or nodal
concepts, where the membership or range of varia-
tion in the type is based on selected typical plots, but
such variation may exclude intermediate plots.
Intensive concepts may provide a first approximation
of a type, which can be later expanded with increased
knowledge of the type. There will always be
DON FABER-LANGENDOEN ET AL.536
Ecological Monographs
Vol. 84, No. 4
REVIEWS
difficulties in assigning stands to a type as one type
transitions to another along a gradient, even when
differentiating criteria are well understood.
6) Classification and field recogni tion of vegetati on
types is a distinct process that differs from vegetation
mapping in that all vegetation maps are constrained
by issues related to scale and technical limitations of
mapping, which may restrict the ability of the map
legend to show all vegetation types in a mapped area.
Conversely, mapping units can describe spatial
relat ionships among types not described by the
classification (e.g., dry-dune–wet-swale type relation-
ships). Because vegetation maps are often developed
to study the geographic distribution, extent, and
landscape patterns of vegetation, the linkages be-
tween recognizable field characteristics of vegetation
classification an d vegetation mapping s hould be
established (Tart et al. 2005).
7) Differentiating criteria for vegetation types can be
arranged hierarchically from upper levels primarily
based on general growth forms to middle levels based
on specific growth form and floristics that includes
suites of general and regional combinations of
characteristic species, and lower levels based primar-
ily on regional to local floristics. At all levels,
vegetation provides the primary criteria for descrip-
tions within the hierarchy, but the organization may
be based on the ecological and biogeographical
relations expressed by the vegetation (Ru
¨
bel 1930
as cited in Shimwell 1971, Whittaker 1962, Westhoff
1967, Pignatti et al. 1994, Brown et al. 1998).
8) An integrated hierarchy of vegetation types is best
established by considering ea ch l evel as both
independent and interconnected in a nested relation-
ship; that is, criteria selected to differentiate levels in
the hierarchy are sufficient to define and distinguish
types of a particular level, thereby preventing it from
being arbitrarily defined by the level immediately
above or below in the hierarchy. Thus, the EcoVeg
method is both top-down and bottom-up. It is largely
an inductive method, in so far as it rarely poses
formal hypotheses, and the method of induction can
be appropriately applied at any level (Mentis 1988).
Opportunities for hypothesis testing (e.g., how
relationships among vegetation types are established
based on their physiognomy, floristics, and ecology)
may emerge as the hierarchy is further developed.
9) A coordinating body should be established to oversee
the recognition and integration of new classification
units. A coordinating body is needed, given the more
or less continuous nature of vegetation patterns and
the potential for both multiple overlapping or uneven
concepts. Even when such concepts are published,
they may be difficult to reconcile with other
independently published types. Proposed vegetation
types, whether previously published or not, should
undergo a peer review process specific to the
classification standard conducted by a coordinating
body. Th e coordinating body can review types
published in independent publications to determine
their relation to the standard set of vegetation types
to ensure that all scientific contributions to the
classification are considered, but the coordinating
body provides the critical role of ensuring that
published types are clearly distinguished from each
other as much as possible. This assures that an
authoritative version of the classification is main-
tained at all times, and it prevents potential
confusion over duplication, overlap, and uneven
scaling among types. A coordinated approach makes
a standardized classification system readily available
to practitioners, policy makers, and others (e.g.,
Rodwell 1991–2000, Grossman et al. 1998, Davies et
al. 2004, Mucina and Rutherford 2006, Jennings et
al. 2009).
These principles are essential when a consistent and
comprehensive set of vegetation types is needed that are
organized by vegetation and ecological relationships
within a global framework. We encourage local or
regional inventories and classifications to retain the core
methodological principles 1–5 on defining types, even if
their classification hierarchy is structured differently, in
order to permit l inking these local and regional
inventories with this global approach because these
inventories are an important sources of information for
any global classification effort.
Related classification methods
The philosophy guiding the EcoVeg methodology
reflects, to varying degrees, the physiognomic-floristic-
ecological systems noted in the Introduction (e.g., see
Mueller-Dombois and Ellenberg [1974] for a general
presentation of such systems). Those systems have
typically not developed the kind of formal principles
and methods presented here, so comparisons are difficult
to make (but see the basic postulates of vegetation
classification and mapping for southern Africa presented
in Mucina and Rutherford [2006]). One system deserves
special comparison: the Braun-Blanquet approach, as
the HRWG learned much from it. Excellent summaries
of that approach are available in Westhoff and van der
Maarel (1973) and Dengler et al. (2008). First, both
approaches share many aspects of principles 1–6. In
particular, it is worth noting that principle 4, collecting
plot-based (releve
´
) vegetation data for classification
purposes, a core feature of the Braun-Blanquet ap-
proach, has become embedded in most formal vegeta-
tion classification approaches, including this one. But
there also some noteworthy differences. The Braun-
Blanquet approach strongly emphasizes floristic-diag-
nostic features at all levels (cf. EcoVeg principle 2),
restricts the scope of vegetation to natural and
seminatural or spontaneous (not planted) vegetation
(cf. EcoVeg principle 3), arranges the hierarchy based
strongly on floristic-ecological criteria, with less empha-
November 2014 537ECOLOGICAL VEGETATION CLASSIFICATION
REVIEWS
