You are on page 1of 11

Journal of Biogeography (J. Biogeogr.

) (2016) 43, 863–873

PERSPECTIVE Revising the biome concept for


understanding and predicting global
change impacts
Glenn R. Moncrieff1,2*, William J. Bond1,3 and Steven I. Higgins4,5

1
Fynbos Node, South African Environmental ABSTRACT
Observation Network (SAEON), Centre for
Biomes are globally distributed, structurally and functionally similar vegetation
Biodiversity Conservation, Kirstenbosch
Gardens, Private Bag X7, Rhodes Drive,
units defined without reference to plant species composition. The boundaries
Claremont 7735 Cape Town, South Africa, between biomes are presumed to correspond with species turnover and changes
2
Department of Botany and Zoology, in biogeochemical cycling. Determining the controls of biome distributions is
Stellenbosch University, Matieland 7602, thus critical for anticipating global change impacts. Historically, climate and
South Africa, 3Department of Biological soils have been understood to adequately explain the global distribution of
Sciences, University of Cape Town, biomes. Convergent evolution and environmental filtering are assumed to be
Rondebosch 7701, South Africa, 4Department pervasive, ultimately resulting in deterministic patterns of vegetation structure
of Botany, University of Otago, PO Box 56, and function in relation to prevailing environmental conditions. Recent studies
Dunedin 9054, New Zealand, 5Biodiversity have highlighted significant problems with this view of biomes. Firstly, system-
and Climate Research Centre (BiK-F), atic structural and functional divergence within biomes has been identified
Senckenberg Gesellschaft f€
ur Naturforschung, when comparing environmentally similar, yet floristically distinct regions. Sec-
Senckenberganlage 25, 60325 Frankfurt am ondly, climatic determinism is being further undermined by evidence suggest-
Main, Germany ing multiple stable biome states are possible for some combinations of climatic
drivers. We argue that biomes remain useful and necessary constructs for orga-
nizing our knowledge of how ecosystems function and for predicting how they
might respond to change. However, biome concepts should acknowledge the
limits to predictability from environmental conditions alone and the influence
of historical processes on modern vegetation patterns. We discuss how direct
mapping of plant structure and function, the incorporation of insights into
biome evolution and a new generation of vegetation models will lead to
*Correspondence: Glenn R. Moncrieff. Fynbos improvements in the concept of what biomes are, where they occur, and efforts
Node, South African Environmental to predict their distribution.
Observation Network (SAEON), Centre for
Biodiversity Conservation, Kirstenbosch Keywords
Gardens, Private Bag X7, Rhodes Drive,
Claremont 7735, Cape Town, South Africa.
alternate stable state, biome, DGVM, functional trait, global land cover, plant
E-mail: glenn.moncrieff@gmail.com functional type, potential natural vegetation, niche conservatism

Traditionally, it has been assumed that the occurrence of


INTRODUCTION
similar vegetation in areas distant both geographically and
There is complex large-scale variation in vegetation across evolutionarily is driven by two processes: environmental fil-
the surface of the Earth, with important consequences for tering of organisms not well suited to the prevailing climate
the resources available to support human wellbeing, and soils in the process of community assembly, and selec-
biodiversity and biogeochemical cycling (Cramer et al., 2001; tion for strategies and forms that optimally use available
Olson et al., 2001). Biomes are the basal unit used to orga- resources over evolutionary time resulting in convergence of
nize and describe this variation. They refer to globally distantly related organisms (Schimper, 1903; Woodward
convergent vegetation formations similar in structure and et al., 2004). Convergent evolution has been noted across a
function, explicitly ignoring floristic differences. The biome range of plant and animal taxa. Striking examples include
concept is attractive because it can be applied globally with- the parallel forms of marsupials and placental mammals,
out reference to plant species composition, allowing general- succulence and similar leaf morphology in distantly related
ity to be sought. desert plants or alpine cushion plants (Aubert et al., 2014)

ª 2016 John Wiley & Sons Ltd http://wileyonlinelibrary.com/journal/jbi 863


doi:10.1111/jbi.12701
G. R. Moncrieff et al.

defined plant functional types (PFTs) or vegetation units


BOX 1: DYNAMIC GLOBAL VEGETATION MODELS respond uniformly to environmental drivers, producing
Dynamic Global Vegetation Models (DGVMs) integrate biomes with consistent properties across continents.
plant physiological models into models of vegetation The fundamental assumptions of interregional convergence
biogeography and have allowed simulation of the temporal and the predictability of global vegetation patterns underly-
dimension of vegetation change and scaling of leaf-level ing the biome concept and efforts to model global biome
processes to whole canopies and the globe (Prentice et al., distribution have been increasingly undermined. Here we
2007). Gross primary productivity, respiration losses, discuss how new directions from community ecology, vegeta-
reproduction, mortality and interactions among plants are tion modelling and phylogenetics motivate for an update of
all incorporated into models that simulate vegetation how we define, map and understand the distribution of
dynamics in an effort to produce projections of the spatial global biomes. We explicitly ignore the substantial influence
and temporal trajectory of vegetation change, the carbon of human activity and focus only on potential natural vegeta-
cycle and feedbacks to the climate system. Uncertainties in tion. We start by synthesizing the evidence that constraints
the representation of various processes are large, as are imposed by evolutionary and biogeographic history limit the
parameter estimates, but without this complexity projec- functional convergence observed within biomes. Next we
tions of vegetation response to global change will be biased. highlight how climate feedbacks, disturbance and ecosystem
DGVMs descend from plant geography models which engineering by plants alter conditions relevant for plant
initially attempted to predict the global distribution of functioning. These feedbacks result in non-deterministic
biomes based on climate correlates. Although no longer biome distribution patterns in which multiple biomes can
purely correlative, the biome concept still underlies the exist at the same location under the same macroclimatic and
manner in which the complex reality of vegetation patterns geological template. In such instances this environmental
is simplified. Plant diversity in DGVMs is usually repre- template will not necessarily prescribe the dominant biome.
sented using Plant Functional Types (PFTs). PFTs in We then look at new directions in mapping and modelling
DGVMs are defined based on similarity in photosynthetic spatial variation in plant form and function through func-
function (e.g. C3 vs. C4), response to disturbance, phenol- tional traits and recent analyses that show how historical
ogy and physiognomy (Lavorel et al., 1997). Biomes are processes over evolutionary and ecological time-scales shape
delimited based on the relative dominance and combina- modern biome distributions. We end by discussing how the
tions of different PFTs (Bonan et al., 2002). PFTs provide a concept of biomes and their use to organize our knowledge
useful conceptual link from biomes, through the plant types of global vegetation patterns and biogeochemistry can be
of which they are composed, to ecosystem function, while revised through the incorporation of these new directions.
not necessitating the inclusion of species-specific informa-
tion. The traits that define PFTs can be related to specific CHALLENGES TO BIOMES
functions, allowing the mechanisms that control biome
distribution to be examined rather than merely producing Interregional divergence in form and function
correlative models.
Biomes are defined and delimited using a variety of schemes
based on similarity in structure, function or climatic condi-
tions, but usually (at least when defining units at the global
and the independent evolution of C4 and CAM photosynthe- scale) plant species composition and geographical location is
sis in multiple plant families (Keeley & Rundel, 2003). intentionally ignored (Box 2: Biome concepts). A comple-
The upshot of environmental filtering during community mentary view of vegetation recognizes biogeographical realms
assembly and convergent evolution in similar environments based on compositional similarity and shared evolutionary
is that the distribution of the plant forms that define biomes history (Holt et al., 2013). Mapped boundaries between
should be predictable from the environment and not depend realms represent significant barriers to current and historical
on any regional or historical context. This assumption has gene flow. Despite an important role for intercontinental and
been at the foundation of efforts to understand plant bio- trans-oceanic dispersal in the assembly of modern floras
geography at large scales. The equilibrium distribution of (Pennington et al., 2004), past speciation, extinction and
biomes is often understood to be explained by only a few vicariance have left an imprint on modern floras with geo-
variables that roughly describe the conditions relevant to graphically distant or isolated regions harbouring different
plants (Whittaker, 1975). The inclusion of additional vari- species assemblages. Historical differences in these processes
ables and information on local topo-edaphic conditions can among regions are reflected in differences in contemporary
improve predictions (Holdridge, 1947; Walter, 1973). This phylogenetic structure. Indeed it is possible for the same
thinking underpins most attempts to project the current, biome to be composed of entirely different and distantly
past and future distribution of biomes using only climatic, related species in two different realms.
edaphic and atmospheric forcing variables as input (Box 1: The evolutionary distinctiveness of different floras is likely
DGVMs). In a typical global vegetation model, globally to have ecological implications. Closely related taxa are often

864 Journal of Biogeography 43, 863–873


ª 2016 John Wiley & Sons Ltd
Biomes for understanding global change impacts

BOX 2: BIOME CONCEPTS


The occurrence of analogous vegetation formations in geographically disjunct although climatically similar regions was described
over 200 years ago by von Humboldt (1807). Schimper (1903) first defined and named the world’s biomes in an essentially
modern way. He argued that the dominant life-forms in each biome had convergent physiological attributes as a result of
selection to cope with the prevailing climate. Common to many early biome concepts is the explicit incorporation of climate
parameters into the definition of a biome (Table 1). Holdridge (1947) produced a systematic scheme of Life Zones describing
large vegetation formations that map onto axes of precipitation, temperature and potential evapotranspiration. Walter (1973)
similarly defined biomes in terms of the climate zone they occupy (Zonobiomes), with modifications caused by soils and
orographic features. The inclusion of climate to delimit a biome serves as a proxy for functional characteristics that are difficult to
quantify or map globally. For example, ‘tropical rain forests’ are, by definition, composed of plants sensitive to cold or frost and
non-deciduous. Remote sensing of land cover with satellites has allowed new global biome distribution maps to be created where
boundaries between vegetation units are ostensibly objectively defined (e.g. Arino et al., 2007). To allow the mapping of biomes
from remotely sensed data, most modern biome classification schemes explicitly ignore geographical context and climate when
defining biomes, rather emphasizing shared phenology and structure, and consistencies in the combination and relative
dominance of PFTs (Woodward et al., 2004). Focussing on characteristics observable from space can, however, lead to important
characteristics being ignored, such as whether plants are drought- or cold-deciduous and boundaries delimited based on arbitrary
cut-offs in order to produce agreement with training data. Both early and contemporary biome concepts and maps recognize
lower hierarchical levels of vegetation formations below global biomes which are defined with reference to geographical context
(e.g. Ecozones (Olson et al., 2001) or Biomes sensu strictu (Walter, 1973)). While these lower hierarchical levels are useful for
purposes such as conservation planning and regional analyses, we concern ourselves with biomes defined globally without
reference to geographical context or environmental conditions. This approach allows the distribution of PFTs and biomes in
relation to the environment to be analysed without circularity, and this biome concept is implicit in global vegetation models and
meta-analyses of global biogeochemical cycling or species richness.

Table 1 An overview of some well known biome concepts used for vegetation modelling, mapping and understanding the
determinants of global vegetation patterns.

Actual or
Biome potential
concept Term Defined based on Mapping variables vegetation Hierarchical Reference

Walter Zonobiome Climate Climate Potential Yes Walter, 1973


Schultz Ecozone Climate Climate Potential Yes Schultz, 2002
WWF Biome Climate, Structure, Climate, Soils, Expert Potential Yes Olson et al., 2001
Orography knowledge, Structure
MODIS- Land cover Structure, Function, Spectral reflectance, Actual No Friedl et al., 2002
IGBP Human influence Temperature, Topography,
Snow/Ice
DGVMs Biome Structure, Function Structure, Function Actual and No Sitch et al., 2003;
(Combination of PFTs) Potential Prentice et al., 2007

more similar ecologically than would be expected if traits convergence in plants and animals exist, when systematically
evolved randomly, resulting from the tendency of species to comparing ostensibly convergent plant groups among realms
retain their ancestral ecology (Wiens et al., 2010). This phy- functional differences masked by superficial structural simi-
logenetic conservatism is evident in species traits (Chazdon larity are often revealed (Alvarado-Cardenas et al., 2013).
et al., 2003), environmental tolerances (Petitpierre et al., When using PFTs to model biome distributions, optimal-
2012) and the biomes in which species occur (Crisp et al., ity assumptions are used that imply convergence (Fisher,
2009). If regional floras are phylogenetically structured, and 2013). For example, the perfect plasticity assumption (PPA)
phylogeny constrains trait evolution, it follows that there – used when simulating tree canopy interactions and light
ought to be divergence in plants traits among floristically availability – assumes that tree crowns perfectly explore and
distinct regions, potentially affecting emergent ecosystem fill available canopy space (Purves et al., 2007). This greatly
properties. In contradiction, convergence of floras to similar simplifies the task of modelling light competition among
vegetation structure and function under similar environmen- trees and can improve model performance (Purves et al.,
tal conditions is a key assumption implicit in the concept 2008). Representing the wide array of plant architecture
of global biomes. While many convincing examples of observed naturally is another major challenge for modelling

Journal of Biogeography 43, 863–873 865


ª 2016 John Wiley & Sons Ltd
G. R. Moncrieff et al.

vegetation structure. However, rather than quantifying and America which diverge from those from Africa (Moncrieff
modelling the entire range of architectures observed across et al., 2014a,b; Fig. 1). Similarly, the dipterocarp-dominated
biomes and environments, the assumption that plant forests of south-east Asia are composed of trees relatively tal-
resource supply networks are designed to optimize the trans- ler and thinner than those found in other tropical forests
port of nutrients and water has been used to estimate plant (Banin et al., 2012). In both examples cited above, the lack
allometries that are ostensibly globally applicable (Enquist & of systematic divergence among all species in the regions
Niklas, 2002; Price et al., 2007). compared suggests that observed differences are not the
Intercontinental comparisons are beginning to reveal bio- result of any unaccounted-for systematic abiotic or biotic
geographical biases across scales of plant form and function factor, but rather the product of the idiosyncrasies of a few
that suggest convergence upon optimal strategies is seldom dominant groups. Scaling upwards from species-level trait
realized. In forests and savannas, plant architectural traits are divergence to emergent ecosystem patterns, the structural
starkly different among continents (Banin et al., 2012; Dan- responses of savannas to variation in climate, soils and dis-
tas & Pausas, 2013; Moncrieff et al., 2014a). These differ- turbance also vary among continents, leading to differences
ences remain even after controlling for environmental in the environmental limits of savannas (Lehmann et al.,
differences, and are explained by the divergent architecture 2011, 2014). Although not explicitly linked, it is likely that
of a few regionally dominant taxa. The remarkably tall, thin- this pattern is partially underpinned by the trait divergence
stemmed and narrow canopied eucalypts of Australian savan- described by Moncrieff et al. (2014a) and Dantas & Pausas
nas diverge from the architectures of savanna trees of South (2013)

Figure 1 Comparative architecture of savanna trees from three continents. Canopy diameter, tree height and bark thickness are shown
at 20 cm stem diameter summarizing data from Moncrieff et al. (2014a), Dantas & Pausas (2013) and Lawes et al. (2011). Allometries
are drawn to scale except bark thickness, which is exaggerated by a factor of 2 relative to stem diameter in the diagram. The extensive
savannas now found in South America, Africa and Australia evolved in the late Miocene (c. 5–8 Ma, Beerling & Osborne, 2006) long
after Gondwana began to break-up; circa 184 Ma. The tree flora of each continent has thus been independently assembled. Savanna
trees evolved in situ on each continent from ancestors in closed-canopy forests and sclerophyll biomes (Simon et al., 2009; Bouchenak-
Khelladi et al., 2010; Crisp et al., 2011; Maurin et al., 2014). Thus, different clades characterize different continents, with Mimosoideae,
Melastomataceae and Malpighiaceae and the genus Qualea common in South American Cerrado, Myrtaceae (particularly Eucalyptus and
Corymbia) often dominant in Australian savannas and Mimosoideae, Caesalpinioideae and Combretaceae common throughout African
savannas.

866 Journal of Biogeography 43, 863–873


ª 2016 John Wiley & Sons Ltd
Biomes for understanding global change impacts

Divergence in structure and function among multiple predictable based on the climate. If multiple biomes are truly
instances of a single biome has important consequences for possible within a set of climatic constraints, after a transition
ecological forecasting. Although savannas share multiple occurs, vegetation (and the fire regime) must not return to
characteristics (including a discontinuous tree layer, continu- the previous state once the external forcing is removed. Mul-
ous understory of C4 grass, frequent fire), instances of the tiple stable biome states have been suggested over vast areas
savanna biome differ enough in structure and function that of the tropics and subtropics as a result of feedbacks among
each is expected to respond differently to global change fire, flammable, shade-intolerant grasses and fire-sensitive
(Lehmann et al., 2014). This observation makes modelling all trees (Staver et al., 2011a). Through promoting recurring
savannas as a single cohesive entity indefensible. When it natural fires, grasses can prevent the establishment of
comes to tropical forests there is an imperative to provide fire-sensitive forest trees and maintain a grassland or savanna
reliable ecological forecasting, as they are critical carbon ecosystem. Alternately, low light availability below forest tree
reservoirs and are extremely biodiverse. But is the concept of canopies reduces C4 grass growth and fuel bed continuity,
a single ‘tropical forest’ biome helpful when biodiversity, ultimately reducing the occurrence of fires and maintaining
architecture and biogeochemistry vary to the extent that an ecosystem with an entirely different suite of species and
insights from one region are not relevant to another (Corlett drastically altered structure (Scheiter & Higgins, 2007; Hoff-
& Primack, 2006; Banin et al., 2014)? mann et al., 2012; Parr et al., 2012). These alternative states
can be internally stable – forests maintaining a fire-free state
without the need for external forcing and resistant to the
The potential for multiple stable states
incursion of fires and flammable species; savannas persisting
Despite the widespread notion that biome distributions are cli- in a frequently burned state unless perturbed by anthro-
matically predetermined, it is well established that there exist pogenic action or a rare climatic event (Staver et al., 2012).
regions of climate space in which the dominant biome cannot Herbivores can also play an important role along with fire in
be accurately predicted (Whittaker, 1975; Bond, 2005). Vegeta- facilitating stabilizing feedbacks and transitions between
tion is not simply a product of the environmental template but alternative stable states (Dublin et al., 1990).
simultaneously alters that template across scales, from local to Perceptions of the factors controlling vegetation distribu-
global. At large scales, the earth’s vegetation feeds back to cli- tion are built upon the assumption that universal bioclimatic
mate by altering surface albedo, hydrology and the partitioning constraints acting on convergent plant functional types pro-
of latent and sensible heat (Bonan, 2008). Furthermore, vege- duce predictable patterns of biome distribution (Whittaker,
tation affects both directly and indirectly the fluxes and storage 1975). This thinking is reflected in the design of the current
of a range of greenhouses gases (IPCC, 2013). At smaller scales, generation of global vegetation models (Prentice et al.,
the presence of plants creates local micro-climates, altering 2007), and in maps of potential natural vegetation that give
wind, temperature and moisture availability (Hoffmann et al., the impression of deterministic biome distribution patterns
2012). Either actively or passively, plants change the physical, (Olson et al., 2001). These maps are used for benchmarking
chemical and biological properties of soils (Hawkes et al., model performance, and hence further select for models that
2005). Beyond impacting climate and soils, vegetation itself represent biomes as deterministic constructs. Some models
constructs ecosystem attributes relevant to plant functioning. now incorporate the processes responsible for non-determi-
The amount of light that passes through canopies and is avail- nistic biome distribution patterns (Higgins & Scheiter, 2012;
able to the lowest vegetation layers is determined by the den- Moncrieff et al., 2014b). If a model allows for multiple
sity and arrangement of leaves. The ignition and spread of potential biome states it is the initialization and maintenance
wildfires is a function of the arrangement, production and of feedbacks that determines which of the suite of potential
moisture content of the plant growth forms fuelling the fires biomes is realized (Fig. 2). The legacy of past climates is
(Burgan & Rothermel, 1984). These feedbacks create the indeed visible in current species and biome distributions,
potential for alternate biome states with profoundly different some of which may not yet have reached equilibrium with
biodiversity, ecological function and evolutionary history prevailing conditions (Svenning & Skov, 2007; Moncrieff
given the same macroclimatic and geological templates that et al., 2014b). Unfortunately many DGVM analyses do not
are not simply transitional to a predictable climax vegetation routinely report on the extent to which predictions are influ-
state (sensu Clements (1916)) but stable over long time series enced by initial conditions, making it difficult to assess how
(Charles-Dominique et al., 2015). widespread this problem is. Even a perfect model could sim-
Fire is a good example of a non-climatic factor altering ulate incorrect vegetation patterns with a mis-specified ini-
vegetation at large scales. Many studies have shown that tialization. Separating these two sources of uncertainty
when fire regimes are artificially altered vegetation change (model versus initial condition) would require vegetation
often follows, sometimes leading to a biome switch (Bond maps used for validation to reflect not only observed vegeta-
et al., 2005; Higgins et al., 2007). However, this does not tion patterns, or a single climax community defined by cli-
imply that vegetation is not predetermined by climate. If cli- mate, but rather the full range of possible states given the
mate ultimately predetermines the prevailing fire regime environmental template used to force models (e.g. Staver
unless subject to external forcing, then vegetation will still be et al. (2011a), Fig. 2).

Journal of Biogeography 43, 863–873 867


ª 2016 John Wiley & Sons Ltd
G. R. Moncrieff et al.

(a) Historical biomes (b) Potential biomes (c) Predicted biomes


alternate past climates actual current climate actual current climate

Low tree cover


Both states possible
High tree cover

Figure 2 Differences in historical vegetation and climate can result in divergent contemporary biome distribution patterns even under a
fixed climate if multiple stable biomes are possible. In (a) two different hypothetical historical states over ecological time-scales (0.1–
10 ka) show a wetter Africa with more extensive forests (top panel) and a drier Africa with more extensive deserts, shrublands,
grasslands and savannas (bottom panel). In (b) the relationship between climate and vegetation is shown in climate space (bottom
panel) and projected in geographical space (top panel) based on contemporary climate data from New et al. (2000). The plotted
relationship between climate and biome state shows deterministically high tree cover in wet environments with aseasonal rainfall,
deterministically low tree cover when mean annual precipitation is low and the potential for multiple stable biomes states with
intermediate, seasonal rainfall (Staver et al., 2011b). Contemporary biome distributions (c) will depend on historical conditions in areas
where multiple stable biome are possible and not be predictable without the inclusion of accurate information on initial conditions.

trade-offs between species traits these analyses suggest traits


NEW DIRECTIONS
that have a globally consistent interpretation (Wright et al.,
2004). The increasing availability of global data sets on plant
Mapping and modelling functional traits
traits means that definitions of biomes that are driven by
Once we accept that history may often overwhelm conver- data on plant functional traits is an realizable prospect
gence as an evolutionary force it becomes tempting to aban- (Kattge et al., 2011).
don global plant biomes for more regional vegetation Rather than producing biome classifications and delimiting
formations as the highest unit of vegetation organization boundaries based on artificially imposed groupings and dis-
(e.g. ecozones sensu Olson et al. (2001), or biomes crossed junctions that create the illusion of sharp boundaries or
with biogeographical realms e.g. Neotropical savannas). It mask intercontinental differences, new maps based on multi-
may be beneficial to avoid extrapolation of findings among variate analyses of trait distributions reflect where change in
other regions when studies are concerned with only a single vegetation structure is gradual or abrupt. Such trait maps
continent or region. However, we do not deem this necessary would not need to group distinct vegetation types in differ-
or prudent as a general rule. Biomes are widely used in ent regions based only on coarse structural similarity. Maps
global conservation planning, meta-analyses, and Earth sys- obtained from direct measurements exist for well studied
tem modelling. This necessitates a construct for organizing regions (Dahlin et al., 2013). Van Bodegom et al. (2014)
our knowledge and understanding of the world’s major produced global maps for three plant functional traits (leaf
ecosystem types. Recent developments have linked plant mass per area, stem-specific density and seed mass) by pro-
traits and ecosystem functions such as nitrogen cycling or jecting trait-environment relationships geographically. An
carbon sequestration, thereby implying new ways of defining advantage of mapping plant traits and defining biomes based
global ecosystems (Reich et al., 1997). By revealing universal on trait clustering is the necessity to simultaneously map

868 Journal of Biogeography 43, 863–873


ª 2016 John Wiley & Sons Ltd
Biomes for understanding global change impacts

trait variability at any location and the uncertainty in vegeta- past changes in the earth system and evolutionary innova-
tion classification. tion. We believe elucidating the historical assembly of mod-
Another approach is to directly examine and map the ern biomes can help identify convergence and divergence
functional response of vegetation as revealed by remotely among floristically distinct regions and move explanations of
sensed metrics of vegetation function. For example, Buiten- modern difference in traits and vegetation patterns beyond
werf et al. (2015) examined NDVI time series, calculated the simple invocation of ‘biogeographic idiosyncrasies’. For
important phenological parameters for each pixel, and classi- example, the divergence already described in the savanna tree
fied them into ‘phenomes’. This differs from the approach flora (Fig. 1) contrasts with the relative similarity of the
previously used to map biomes or land cover types from savanna grass flora. While the globally widespread savanna
remotely sensed reflectance data – where regression or biome and endemic savanna trees are relatively young
machine learning approaches are used to train classifiers – (c. 5–8 Ma), C4 savanna grasses have existed for over 30 Ma
because biome boundaries reflect interpretable functional (Beerling & Osborne, 2006). Does the difference in age of
turnover. Biomes defined and delimited based on such func- savanna trees and grasses explain why there is divergence in
tional similarities are more likely to represent units that will the properties of the tree layer but not the grass layer among
respond coherently to global change than units delimited Africa, Australia and South America? Relative convergence/
using artificial thresholds chosen to match training data. divergence as a function of age of a biome is becoming testa-
Moving away from a PFT-based approach to dynamic veg- ble across biomes as information on historical biome assem-
etation modelling in favour of trait-based models is already a bly becomes available.
research theme. Much criticism has been levelled at the con- Another useful example is provided by fire-prone biomes
cept of PFTs, and the potential biases they introduce when of Australia and the eucalypt trees (Eucalyptus, Corymbia,
attempting to summarize all plant variation into these crude Angophora), which often dominate these vegetation types.
groupings (Scheiter et al., 2013). New models of global vege- Australian biome distribution patterns and the variation of
tation patterns base projections on the simulated distribution basal area in relation to fire and climate differ markedly
and functioning of plant types or individuals with continu- from other continents (Lehmann et al., 2011; Murphy &
ous variation in the values of multiple traits (Pavlick et al., Bowman, 2012; Moncrieff et al., 2015). While the non-euca-
2013; Scheiter et al., 2013; Sakschewski et al., 2015). Plant lypt tree component of these biomes responds to fire accord-
performance in such a trait-based model is the outcome of ing to expectations derived from studies on the fire ecology
traits influencing plant physiology and ecological interac- of tree species from other continents, eucalypts appear to be
tions, and the integration of numerous trade-offs which con- remarkably insensitive to variation in fire regimes (Bond
strain the possible realm of trait combinations. Yet even if et al., 2012; Murphy et al., 2015). A unique suite of traits in
the grand task of linking plant performance to traits and eucalypts related to growth rates and the abundance and dis-
defining trade-off surfaces is achieved there is no guarantee tribution of bud-forming structures in stems and branches
that trait distributions will be predictable based on environ- enables fire escape and post-fire epicormic resprouting in
mental conditions alone. Weak correlations between environ- stems otherwise vulnerable to the total loss of aboveground
mental factors and community-weighted traits and biomass after intense fires (Burrows, 2013). Modern euca-
differences among separate evolutionary lineages and floras lypts evolved from ancestors in fire-prone sclerophyllous
suggest that trait convergence under similar environmental biomes (Crisp et al., 2011). The ancestral condition of fire-
conditions is unlikely (Heberling & Fridley, 2012; Mitchell adapted traits in eucalypts (c. 60 Ma) may explain their
et al., 2015). In addition, the presence and intensity of top- remarkable success in fire-prone areas of Australia now
down controls on vegetation – such as fire and herbivores – (Burrows, 2013). In contrast, the savanna tree flora of the
will profoundly alter traits (Staver et al., 2012; Wigley et al., Brazilian Cerrado evolved from wet forest ancestors and cor-
2015), but may not be predictable from climate and soils. respondingly these species differ from eucalypts in their fire
Constraining the potential trait space available to modelled adaptations (Simon et al., 2009; Fig. 1). These differences
plants according to regional trait pools, and adjusting the may partially explain why the distribution of South Ameri-
parameterisation of trade-offs for different regions may can savannas is restricted in environmental space relative to
increase agreement between simulated and observed trait dis- African and Australian savannas (Lehmann et al., 2011,
tributions. These constraints could be viewed as a means of 2014). However, other factors such as differences in
optimizing traits and trade-offs to match observations. We megafaunal associations and their extinction rates may also
view them rather as the necessary acknowledgement of the be involved (Doughty et al., 2015). Regardless, it is evident
importance of the unique evolutionary context of different that the historical context from which the fire-adapted flora
regions for ecosystem functioning. of Australia and South America evolved underpins modern
biome-level divergence.
We are also beginning to appreciate the importance of his-
Insights from evolutionary and ecological history
tory over shorter (ecological) time spans for predicting the dis-
Modern biomes have been assembled over millions of years tribution of global plant biomes. The importance of boundary
with their distributions constantly changing in response to conditions for modelling climate and weather has long been

Journal of Biogeography 43, 863–873 869


ª 2016 John Wiley & Sons Ltd
G. R. Moncrieff et al.

recognized. This behaviour is not viewed as an artefact of to predict how global change will reorganize vegetation pat-
model structure, but a realistic representation of the chaotic terns at large scales.
dynamics of atmospheric circulation (Lorenz, 1963). In global
vegetation models there is seldom direct lateral transfer of
Moving forward
mass or energy among pixels. Thus, the only relevant bound-
ary conditions relate to historical vegetation states. Progress in At the boundaries between biomes there is often rapid turn-
modelling global vegetation depends on the acknowledgement over of species, nutrient cycling and ecosystem services (Pay-
that vegetation patterns are not deterministic in relation to cli- ette et al., 2001; Hoffmann et al., 2012). Biomes themselves
mate, the inclusion of processes that produce the feedbacks are the theatre of evolution, shaping contemporary patterns
responsible for multiple stable equilibria and explicit explo- of diversity, leaving an imprint on species evolutionary tra-
ration of vegetation history – especially after the end of the last jectories (Pennington et al., 2004; Crisp et al., 2009). They
glacial. DGVM initializations should consider realistic histori- are thus not fictional constructs, and can indeed represent
cal vegetation states rather than uniform initial states. When important ecological and evolutionary units (Crisp et al.,
data on vegetation history are unavailable multiple initializa- 2009). We should therefore not do away with the biome
tions should be explored (Moncrieff et al., 2014b). concept, but rather recognize that it requires updating from
its original conception over a century ago. Functional simi-
larities, which can now be observed globally – either directly
CONCLUSIONS
across the entire globe using satellites or as plant traits syn-
thesized in databases that are increasingly formidable – form
What we thought
a better basis for defining and delimiting vegetation units.
The biome concept as developed by Schimper (1903) is Indicating uncertainty and multiple potential biomes on dis-
based on the idea that similar climates will select for similar tribution maps will help to move away from the misleading
plant forms independent of differences in history. Underlying notion of biomes as climatically predetermined. Including
this is the assumption that given the constraints common to insights into historical biome assembly and historic vegeta-
all plants optimal solutions to environmental conditions have tion states will help explain why differences in the behaviour
evolved. Deviations from global vegetation-climate patterns of different instances of a biome exist for no apparent envi-
at local scales have always been acknowledged (Walter, ronmental reason. The result of this update will be an
1973). Many landscapes are a complex mosaic of vegetation improved understanding of the drivers of spatial and tempo-
types often very different from the macroclimate-predicted ral vegetation change and an improved ability to predict how
biome. At regional and continental scales, however, it was global vegetation patterns will respond to global environ-
presumed that aggregate properties would converge to the mental change.
macroclimate-derived expectation as fundamental biophysical
constraints limit the domain of possibilities and evolutionary ACKNOWLEDGEMENTS
change drives global convergence. Accordingly, biome distri-
bution patterns have been thought of as deterministic, and We benefitted from discussions with Jasper Slingsby, Tho-
thus predictable given knowledge of the prevailing climate mas Hickler and Robert Buitenwerf. Funding was provided
and soils. Hence most efforts to model current and future by the South African Environmental Observation Network
vegetation patterns work on the implicit assumption that and the National Research Foundation of South Africa. We
only one biome should be possible given prescribed environ- are grateful for the helpful feedback from three anonymous
mental conditions. referees.

What we have learnt REFERENCES

Biomes defined by coarse structural similarity do not occur Alvarado-Cardenas, L.O., Martınez-Meyer, E., Feria, T.P.,
within a consistent region of environmental space even at Eguiarte, L.E., Hernandez, H.M., Midgley, G. & Olson,
the largest scales nor will they respond uniformly to global M.E. (2013) To converge or not to converge in environ-
change. The legacy of evolutionary history constrains con- mental space: testing for similar environments between
temporary plant functioning, resulting in divergence among analogous succulent plants of North America and Africa.
floristically distinct regions. Furthermore, many regions can Annals of Botany, 111, 1125–1138.
support multiple stable biome states rather than having Arino, O., Gross, D., Ranera, F., Bourg, L., Leroy, M.,
deterministic climate-vegetation associations. In these Bicheron, P., Latham, J., Di Gregorio, A., Brockman, C.,
regions, present-day vegetation patterns will depend on his- Witt, R., Defourny, P., Vancutsem, C., Herold, M., Sam-
torical conditions and maintenance of feedbacks that create bale, J., Achard, F., Durieux, L., Plummer, S. & Weber, J.-
the potential for alternate stable states. A biome concept that L. (2007) GlobCover: ESA service for global land cover
does not account for role of evolutionary history and multi- from MERIS. Geoscience and Remote Sensing Symposium,
ple stable biomes states will hinder, rather than help, efforts 2007. IGARSS 2007. IEEE International, 2412–2415.

870 Journal of Biogeography 43, 863–873


ª 2016 John Wiley & Sons Ltd
Biomes for understanding global change impacts

Aubert, S., Boucher, F., Lavergne, S., Renaud, J. & Choler, P. traits of woody species in wet tropical forests. Ecological
(2014) 1914–2014: A revised worldwide catalogue of cush- Monographs, 73, 331–348.
ion plants 100 years after Hauri and Schr€ oter. Alpine Bot- Clements, F.E. (1916) Plant succession: an analysis of the
any, 124, 59–70. development of vegetation. Carnegie Institution of Washing-
Banin, L., Feldpausch, T.R., Phillips, O.L., Baker, T.R., Lloyd, ton, Washington DC.
J., Affum-Baffoe, K., Arets, E., Berry, N.J., Bradford, M. & Corlett, R.T. & Primack, R.B. (2006) Tropical rainforests and
Brienen, R.J.W. (2012) What controls tropical forest the need for cross-continental comparisons. Trends in
architecture? Testing environmental, structural and floris- Ecology and Evolution, 21, 104–110.
tic drivers. Global Ecology and Biogeography, 21, 1179– Cramer, W., Bondeau, A., Woodward, F.I., Prentice, I.C., Betts,
1190. R.A., Brovkin, V., Cox, P.M., Fisher, V., Foley, J.A., Friend,
Banin, L., Lewis, S.L., Lopez-Gonzalez, G., Baker, T.R., A.D., Kucharik, C., Lomas, M.R., Ramankutty, N., Sitch, S.,
Quesada, C.A., Chao, K.-J., Burslem, D.F.R.P., Nilus, R., Abu Smith, B., White, A. & Young-Molling, C. (2001) Global
Salim, K., Keeling, H.C., Tan, S., Davies, S.J., Monteagudo, response of terrestrial ecosystem structure and function to
A., Vasquez, R., Lloyd, J., Neill, D., Pitman, N. & Phillips, CO2 and climate change: results from six dynamic global
O.L. (2014) Tropical forest wood production: a cross-conti- vegetation models. Global Change Biology, 7, 357–373.
nental comparison. Journal of Ecology, 102, 1025–1037. Crisp, M.D., Arroyo, M.T., Cook, L.G., Gandolfo, M.A., Jor-
Beerling, D.J. & Osborne, C.P. (2006) The origin of the dan, G.J., McGlone, M.S., Weston, P.H., Westoby, M.,
savanna biome. Global Change Biology, 12, 2023–2031. Wilf, P. & Linder, H.P. (2009) Phylogenetic biome conser-
Bonan, G.B. (2008) Forests and climate change: forcings, vatism on a global scale. Nature, 458, 754–756.
feedbacks, and the climate benefits of forests. Science, 320, Crisp, M.D., Burrows, G.E., Cook, L.G., Thornhill, A.H. &
1444–1449. Bowman, D.M.J.S. (2011) Flammable biomes dominated
Bonan, G.B., Oleson, K.W., Vertenstein, M., Levis, S., Zeng, by eucalypts originated at the Cretaceous-Palaeogene
X., Dai, Y., Dickinson, R.E. & Yang, Z.-L. (2002) The land boundary. Nature Communications, 2, 193.
surface climatology of the community land model coupled Dahlin, K.M., Asner, G.P. & Field, C.B. (2013) Environmen-
to the NCAR community climate model. Journal of Cli- tal and community controls on plant canopy chemistry in
mate, 15, 3123–3149. a Mediterranean-type ecosystem. Proceedings of the
Bond, W.J. (2005) Large parts of the world are brown or National Academy of Sciences USA, 110, 6895–6900.
black: a different view on the “Green World” hypothesis. Dantas, V. de L. & Pausas, J.G. (2013) The lanky and the
Journal of Vegetation Science, 16, 261–266. corky: fire-escape strategies in savanna woody species.
Bond, W.J., Woodward, F.I. & Midgley, G.F. (2005) The glo- Journal of Ecology, 101, 1265–1272.
bal distribution of ecosystems in a world without fire. Doughty, C.E., Faurby, S. & Svenning, J.C. (2015) The impact
New Phytologist, 165, 525–538. of the megafauna extinctions on savanna woody cover in
Bond, W.J., Cook, G.D. & Williams, R.J. (2012) Which trees South America. Ecography, doi:10.1111/ecog.01593.
dominate in savannas? The escape hypothesis and Dublin, H.T.Sinclair, A.R.E. & McGlade, J. (1990) Elephants
eucalypts in northern Australia. Austral Ecology, 37, and fire as causes of multiple stable states in the Serengeti-
678–685. Mara woodlands. Journal of Animal Ecology, 59, 1147–1164.
Bouchenak-Khelladi, Y., Maurin, O., Hurter, J. & Van der Enquist, B.J. & Niklas, K.J. (2002) Global allocation rules for
Bank, M. (2010) The evolutionary history and biogeogra- patterns of biomass partitioning in seed plants. Science,
phy of Mimosoideae (Leguminosae): an emphasis on Afri- 295, 1517–1520.
can acacias. Molecular Phylogenetics and Evolution, 57, Fisher, R.A. (2013) Modelling plant ecology. Environmental
495–508. Modelling (ed. by J. Wainwright and M. Mulligan), pp.
Buitenwerf, R., Rose, L. & Higgins, S.I. (2015) Three decades 207–220. John Wiley & Sons, Ltd, Chichester, UK.
of multi-dimensional change in global leaf phenology. Friedl, M.A., McIver, D.K., Hodges, J.C.F., Zhang, X.Y.,
Nature Climate Change, 5, 364–368. Muchoney, D., Strahler, A.H., Woodcock, C.E., Gopal, S.,
Burgan, R.E. & Rothermel, R.C. (1984) BEHAVE: fire behav- Schneider, A. & Cooper, A. (2002) Global land cover map-
ior prediction and fuel modeling system–FUEL subsystem. ping from MODIS: algorithms and early results. Remote
General technical report/Intermountain Forest and Range Sensing of Environment, 83, 287–302.
Experiment Station. USDA (no. INT-167). Hawkes, C.V., Wren, I.F., Herman, D.J. & Firestone, M.K.
Burrows, G.E. (2013) Buds, bushfires and resprouting in the (2005) Plant invasion alters nitrogen cycling by modifying
eucalypts. Australian Journal of Botany, 61, 331–349. the soil nitrifying community. Ecology Letters, 8, 976–985.
Charles-Dominique, T., Staver, A.C., Midgley, G.F. & Bond, Heberling, J.M. & Fridley, J.D. (2012) Biogeographic con-
W.J. (2015) Functional differentiation of biomes in an straints on the world-wide leaf economics spectrum. Glo-
African savanna/forest mosaic. South African Journal of bal Ecology and Biogeography, 21, 1137–1146.
Botany 101, 82–90. doi:10.1016/j.sajb.2015.05.005. Higgins, S.I. & Scheiter, S. (2012) Atmospheric CO2 forces
Chazdon, R.L., Careaga, S., Webb, C. & Vargas, O. (2003) abrupt vegetation shifts locally, but not globally. Nature,
Community and phylogenetic structure of reproductive 488, 209–212.

Journal of Biogeography 43, 863–873 871


ª 2016 John Wiley & Sons Ltd
G. R. Moncrieff et al.

Higgins, S.I., Bond, W.J., February, E.C., Bronn, A., Euston- & Higgins, S.I. (2014a) Contrasting architecture of key
Brown, D.I., Enslin, B., Govender, N., Rademan, L., African and Australian savanna tree taxa drives interconti-
O’Regan, S. & Potgieter, A.L. (2007) Effects of four dec- nental structural divergence. Global Ecology and Biogeogra-
ades of fire manipulation on woody vegetation structure phy, 23, 1235–1244.
in savanna. Ecology, 88, 1119–1125. Moncrieff, G.R., Scheiter, S., Bond, W.J. & Higgins, S.I.
Hoffmann, W.A., Geiger, E.L., Gotsch, S.G., Rossatto, D.R., (2014b) Increasing atmospheric CO2 overrides the histori-
Silva, L.C., Lau, O.L., Haridasan, M. & Franco, A.C. cal legacy of multiple stable biome states in Africa. New
(2012) Ecological thresholds at the savanna-forest bound- Phytologist, 201, 908–915.
ary: how plant traits, resources and fire govern the distri- Moncrieff, G.R., Hickler, T. & Higgins, S.I. (2015) Intercon-
bution of tropical biomes. Ecology Letters, 15, 759–768. tinental divergence in the climate envelope of major plant
Holdridge, L.R. (1947) Determination of world plant forma- biomes. Global Ecology and Biogeography, 24, 324–334.
tions from simple climatic data. Science, 105, 267–268. Murphy, B.P. & Bowman, D.M. (2012) What controls the
Holt, B.G., Lessard, J.-P., Borregaard, M.K., Fritz, S.A., Ara
ujo, distribution of tropical forest and savanna? Ecology Letters,
M.B., Dimitrov, D., Fabre, P.-H., Graham, C.H., Graves, 15, 748–758.
G.R. & Jønsson, K.A. (2013) An update of Wallace’s zoogeo- Murphy, B.P., Liedloff, A.C. & Cook, G.D. (2015) Does fire
graphic regions of the world. Science, 339, 74–78. limit tree biomass in Australian savannas? International
IPCC (2013) Climate change 2013: The physical science basis. Journal of Wildland Fire, 24, 1–15.
Working group I contribution to the fifth assessment report New, M., Hulme, M. & Jones, P. (2000) Representing twenti-
of the Intergovernmental Panel on Climate Change. Cam- eth-century space-time climate variability. Part II: Devel-
bridge University Press, Cambridge, UK. opment of 1901–96 monthly grids of terrestrial surface
Kattge, J., Diaz, S., Lavorel, S., Prentice, I.C., Leadley, P., climate. Journal of Climate, 13, 2217–2238.
B€onisch, G., Garnier, E., Westoby, M., Reich, P.B. & Olson, D.M., Dinerstein, E., Wikramanayake, E.D., Burgess,
Wright, I.J. et al. (2011) TRY–a global database of plant N.D., Powell, G.V., Underwood, E.C., D’amico, J.A.,
traits. Global Change Biology, 17, 2905–2935. Itoua, I., Strand, H.E. & Morrison, J.C. (2001) Terrestrial
Keeley, J.E. & Rundel, P.W. (2003) Evolution of CAM and ecoregions of the world: A new map of life on earth: A
C4 carbon-concentrating mechanisms. International Jour- new global map of terrestrial ecoregions provides an
nal of Plant Sciences, 164, S55–S77. innovative tool for conserving biodiversity. BioScience, 51,
Lavorel, S., McIntyre, S., Landsberg, J. & Forbes, T.D.A. 933–938.
(1997) Plant functional classifications: from general groups Parr, C.L., Gray, E.F. & Bond, W.J. (2012) Cascading
to specific groups based on response to disturbance. biodiversity and functional consequences of a global
Trends in Ecology and Evolution, 12, 474–478. change–induced biome switch. Diversity and Distributions,
Lawes, M.J., Adie, H., Russell-Smith, J., Murphy, B.P. & 18, 493–503.
Midgley, J.J. (2011) How do small savanna trees avoid Pavlick, R., Drewry, D.T., Bohn, K., Reu, B. & Kleidon, A.
stem mortality by fire? The roles of stem diameter, height (2013) The Jena Diversity-Dynamic Global Vegetation
and bark thickness. Ecosphere, 2, art42. Model (JeDi-DGVM): a diverse approach to representing
Lehmann, C.E.R., Archibald, S., Hoffmann, W.A. & Bond, terrestrial biogeography and biogeochemistry based on
W.J. (2011) Deciphering the distribution of the savanna plant functional trade-offs. Biogeosciences, 10, 4137–4177.
biome. New Phytologist, 191, 197–209. Payette, S., Fortin, M.-J. & Gamache, I. (2001) The Subarctic
Lehmann, C.E.R., Anderson, T.M., Sankaran, M. et al. Forest–Tundra: The structure of a biome in a changing cli-
(2014) Savanna vegetation-fire-climate relationships differ mate. BioScience, 51, 709–718.
among continents. Science, 343, 548–552. Pennington, R.T., Cronk, Q.C. & Richardson, J.A. (2004)
Lorenz, E.N. (1963) Deterministic nonperiodic flow. Journal Introduction and synthesis: plant phylogeny and the origin
of the Atmospheric Sciences, 20, 130–141. of major biomes. Philosophical Transactions of the Royal
Maurin, O., Davies, T.J., Burrows, J.E., Daru, B.H., Yes- Society B: Biological Sciences, 359, 1455–1464.
soufou, K., Muasya, A.M., Bank, M. & Bond, W.J. (2014) Petitpierre, B., Kueffer, C., Broennimann, O., Randin, C.,
Savanna fire and the origins of the “underground forests” Daehler, C. & Guisan, A. (2012) Climatic niche shifts are rare
of Africa. New Phytologist, 204, 201–214. among terrestrial plant invaders. Science, 335, 1344–1348.
Mitchell, N., Moore, T.E., Mollmann, H.K., Carlson, J.E., Prentice, I.C., Bondeau, A., Cramer, W., Harrison, S.P.,
Mocko, K., Martinez-Cabrera, H., Adams, C., Silander, Hickler, T., Lucht, W., Sitch, S., Smith, B. & Sykes, M.T.
J.A., Jr, Jones, C.S., Schlichting, C.D. & Holsinger, K.E. (2007) Dynamic Global Vegetation Modeling: quantifying
(2015) Functional traits in parallel evolutionary radiations terrestrial ecosystem responses to large-scale environmental
and trait-environment associations in the Cape Floristic change. Terrestrial Ecosystems in a Changing World (ed. by
Region of South Africa. The American Naturalist, 185, J.G. Canadell, D.E. Pataki and L.F. Pitelka), pp. 175–192.
525–537. Springer, Berlin Heidelberg.
Moncrieff, G.R., Lehmann, C.E., Schnitzler, J., Gambiza, J., Price, C.A., Enquist, B.J. & Savage, V.M. (2007) A general
Hiernaux, P., Ryan, C.M., Shackleton, C.M., Williams, R.J. model for allometric covariation in botanical form and

872 Journal of Biogeography 43, 863–873


ª 2016 John Wiley & Sons Ltd
Biomes for understanding global change impacts

function. Proceedings of the National Academy of Sciences Staver, A.C., Archibald, S. & Levin, S.A. (2011a) The global
USA, 104, 13204–13209. extent and determinants of savanna and forest as alterna-
Purves, D.W., Lichstein, J.W. & Pacala, S.W. (2007) Crown tive biome states. Science, 334, 230–232.
plasticity and competition for canopy space: A new spa- Staver, A.C., Archibald, S. & Levin, S.A. (2011b) Tree cover in
tially implicit model parameterized for 250 North Ameri- sub-Saharan Africa: Rainfall and fire constrain forest and
can tree species. PLoS ONE, 2, e870. savanna as alternative stable states. Ecology, 92, 1063–1072.
Purves, D.W., Lichstein, J.W., Strigul, N. & Pacala, S.W. Staver, A.C., Bond, W.J., Cramer, M.D. & Wakeling, J.L.
(2008) Predicting and understanding forest dynamics (2012) Top-down determinants of niche structure and adap-
using a simple tractable model. Proceedings of the National tation among African Acacias. Ecology Letters, 15, 673–679.
Academy of Sciences USA, 105, 17018–17022. Svenning, J.C. & Skov, F. (2007) Could the tree diversity pat-
Reich, P.B., Walters, M.B. & Ellsworth, D.S. (1997) From tern in Europe be generated by postglacial dispersal limita-
tropics to tundra: Global convergence in plant functioning. tion? Ecology Letters, 10, 453–460.
Proceedings of the National Academy of Sciences USA, 94, van Bodegom, P.M., Douma, J.C. & Verheijen, L.M. (2014)
13730–13734. A fully traits-based approach to modeling global vegeta-
Sakschewski, B., Bloh, W., Boit, A., Rammig, A., Kattge, J., tion distribution. Proceedings of the National Academy of
Poorter, L., Pe~ nuelas, J. & Thonicke, K. (2015) Leaf and Sciences USA, 111, 13733–13738.
stem economics spectra drive diversity of functional plant von Humboldt, A. (1807) Voyage de Humbolt et Bonpland,
traits in a dynamic global vegetation model. Global Change Premiere Partie Physique Generate, et Relation Historique
Biology, 21, 2711–2725. du Voyage. Chez Fr. Schoell, Libraire and A. Tubingue,
Scheiter, S. & Higgins, S.I. (2007) Partitioning of root and Chez JG Cotta, Libraire, Paris.
shoot competition and the stability of savannas. The Amer- Walter, H. (1973) Vegetation of the earth in relation to climate
ican Naturalist, 170, 587–601. and the eco-physiological conditions. Springer, New York.
Scheiter, S., Langan, L. & Higgins, S.I. (2013) Next-genera- Whittaker, R.H. (1975) Community and ecosystems. McMil-
tion dynamic global vegetation models: learning from lan, New York.
community ecology. The New Phytologist, 198, 957–969. Wiens, J.J., Ackerly, D.D., Allen, A.P., Anacker, B.L., Buckley,
Schimper, A.F.W. (1903) Plant-geography upon a physiological L.B., Cornell, H.V., Damschen, E.I., Jonathan Davies, T.,
basis. Clarendon Press, Oxford. Grytnes, J.A., Harrison, S.P., Hawkins, B.A., Holt, R.D.,
Schultz, J. (2002) The ecozones of the world. The ecological McCain, C.M. & Stephens, P.R. (2010) Niche conservatism
division of the geosphere. Springer, Berlin. as an emerging principle in ecology and conservation biol-
Simon, M.F., Grether, R., de Queiroz, L.P., Skema, C., ogy. Ecology Letters, 13, 1310–1324.
Pennington, R.T. & Hughes, C.E. (2009) Recent assem- Wigley, B.J., Bond, W.J., Fritz, H. & Coetsee, C. (2015)
bly of the Cerrado, a neotropical plant diversity hotspot, Mammal browsers and rainfall affect acacia leaf nutrient
by in situ evolution of adaptations to fire. Proceedings of content, defense, and growth in South African savannas.
the National Academy of Sciences USA, 106, 20359– Biotropica, 47, 190–200.
20364. Woodward, F.I., Lomas, M.R. & Kelly, C.K. (2004) Global
Sitch, S., Smith, B., Prentice, I.C., Arneth, A., Bondeau, A., climate and the distribution of plant biomes. Philosophical
Cramer, W., Kaplan, J.O., Levis, S., Lucht, W., Sykes, Transactions of the Royal Society B: Biological Sciences, 359,
M.T. & others (2003) Evaluation of ecosystem dynamics, 1465–1476.
plant geography and terrestrial carbon cycling in the LPJ Wright, I.J., Reich, P.B., Westoby, M. et al. (2004)
dynamic global vegetation model. Global Change Biology, The worldwide leaf economics spectrum. Nature, 428,
9, 161–185. 821–827.

BIOSKETCH

Glenn Moncrieff is a postdoctoral researcher at the Fynbos Node of the South African Environmental Observation Network
(SAEON). He is interested in how evolutionary history affects present-day vegetation functioning and response to disturbance at
very large scales, and applies a range of tools from dynamic global vegetation models to phylogenetic methods to investigate
these processes.

Editor: Malte Ebach

Journal of Biogeography 43, 863–873 873


ª 2016 John Wiley & Sons Ltd

You might also like