ArticlePDF Available

Abstract and Figures

Aim To map and characterize anthropogenic transformation of the terrestrial biosphere before and during the Industrial Revolution, from 1700 to 2000.Location Global.Methods Anthropogenic biomes (anthromes) were mapped for 1700, 1800, 1900 and 2000 using a rule-based anthrome classification model applied to gridded global data for human population density and land use. Anthropogenic transformation of terrestrial biomes was then characterized by map comparisons at century intervals.Results In 1700, nearly half of the terrestrial biosphere was wild, without human settlements or substantial land use. Most of the remainder was in a seminatural state (45%) having only minor use for agriculture and settlements. By 2000, the opposite was true, with the majority of the biosphere in agricultural and settled anthromes, less than 20% seminatural and only a quarter left wild. Anthropogenic transformation of the biosphere during the Industrial Revolution resulted about equally from land-use expansion into wildlands and intensification of land use within seminatural anthromes. Transformation pathways differed strongly between biomes and regions, with some remaining mostly wild but with the majority almost completely transformed into rangelands, croplands and villages. In the process of transforming almost 39% of earth's total ice-free surface into agricultural land and settlements, an additional 37% of global land without such use has become embedded within agricultural and settled anthromes.Main conclusions Between 1700 and 2000, the terrestrial biosphere made the critical transition from mostly wild to mostly anthropogenic, passing the 50% mark early in the 20th century. At present, and ever more in the future, the form and process of terrestrial ecosystems in most biomes will be predominantly anthropogenic, the product of land use and other direct human interactions with ecosystems. Ecological research and conservation efforts in all but a few biomes would benefit from a primary focus on the novel remnant, recovering and managed ecosystems embedded within used lands.
Content may be subject to copyright.
RESEARCH
PAPER
Anthropogenic transformation of the
biomes, 1700 to 2000
Erle C. Ellis1*, Kees Klein Goldewijk2, Stefan Siebert3, Deborah Lightman4
and Navin Ramankutty5
1Department of Geography and Environmental
Systems, University of Maryland, Baltimore
County, Baltimore, MD, USA, 2Netherlands
Environmental Assessment Agency, Bilthoven,
The Netherlands, 3Institute of Crop Science
and Resource Conservation, University of
Bonn, Bonn, Germany, 4McGill School of the
Environment, McGill University, Montreal,
QC, Canada, 5Department of Geography and
Earth System Science Program, McGill
University, Montreal, QC, Canada
ABSTRACT
Aim To map and characterize anthropogenic transformation of the terrestrial
biosphere before and during the Industrial Revolution, from 1700 to 2000.
Location Global.
Methods Anthropogenic biomes (anthromes) were mapped for 1700, 1800, 1900
and 2000 using a rule-based anthrome classification model applied to gridded
global data for human population density and land use. Anthropogenic transfor-
mation of terrestrial biomes was then characterized by map comparisons at century
intervals.
Results In 1700, nearly half of the terrestrial biosphere was wild, without human
settlements or substantial land use. Most of the remainder was in a seminatural
state (45%) having only minor use for agriculture and settlements. By 2000, the
opposite was true, with the majority of the biosphere in agricultural and settled
anthromes, less than 20% seminatural and only a quarter left wild. Anthropogenic
transformation of the biosphere during the Industrial Revolution resulted about
equally from land-use expansion into wildlands and intensification of land use
within seminatural anthromes. Transformation pathways differed strongly between
biomes and regions, with some remaining mostly wild but with the majority almost
completely transformed into rangelands, croplands and villages. In the process of
transforming almost 39% of earth’s total ice-free surface into agricultural land and
settlements, an additional 37% of global land without such use has become embed-
ded within agricultural and settled anthromes.
Main conclusions Between 1700 and 2000, the terrestrial biosphere made the
critical transition from mostly wild to mostly anthropogenic, passing the 50% mark
early in the 20th century. At present, and ever more in the future, the form and
process of terrestrial ecosystems in most biomes will be predominantly anthropo-
genic, the product of land use and other direct human interactions with ecosys-
tems. Ecological research and conservation efforts in all but a few biomes would
benefit from a primary focus on the novel remnant, recovering and managed
ecosystems embedded within used lands.
Keywords
Agriculture, anthromes, anthropogenic landscapes, conservation,
environmental history, global change, land-use change, novel ecosystems,
terrestrial ecosystems.
*Correspondence: Erle C. Ellis, Department of
Geography and Environmental Systems,
University of Maryland, Baltimore County, 1000
Hilltop Circle, Baltimore, MD 21250, USA.
E-mail: ece@umbc.edu
INTRODUCTION
For millennia, humans have reshaped the form and process of
ecosystems across the terrestrial biosphere, both intentionally
and unintentionally (Turner II et al., 1990; Redman, 1999;
Kirch, 2005; Dearing et al., 2006;). Starting with fairly transient
practices like hunting and gathering and building towards the
increasingly permanent use of land for agriculture and settle-
ments, the widespread and sustained presence of human popu-
lations has transformed ecosystems locally, regionally and
Global Ecology and Biogeography, (Global Ecol. Biogeogr.) (2010) 19, 589–606
© 2010 Blackwell Publishing Ltd DOI: 10.1111/j.1466-8238.2010.00540.x
www.blackwellpublishing.com/geb 589
globally. Human activities have facilitated species extinctions,
invasions, introductions and domestications, increased soil
erosion, altered fire frequency and hydrology, and incited pro-
found changes in primary productivity and other key bio-
geochemical and ecosystems processes (Turner II et al., 1990;
Vitousek et al., 1997; Defries et al., 2004; Foley et al., 2005;
Dearing et al., 2006; Hobbs et al., 2006; Ellis & Ramankutty,
2008; Hansen & Galetti, 2009).
Global estimates of direct transformation of ecosystems by
humans vary among studies, but there is growing consensus that
humans have now transformed ecosystem pattern and process
across most of the terrestrial biosphere (Sanderson et al., 2002;
Kareiva et al., 2007; Ellis & Ramankutty,2008). As a result, global
patterns of the form and function of terrestrial ecosystems are
no longer accurately depicted by the now classic approach to
mapping and modelling biomes as a function of climatic and
physiographic variables (Shelford, 1932; Holdridge, 1947;
Küchler, 1949; Dansereau, 1957; Whittaker, 1970; Prentice et al.,
1992).
While there is no doubt that global patterns of ecosystem
form and process will continue to be influenced and con-
strained by climate and other geophysical and biotic factors,
wherever human populations and activities are present, the
realized form and dynamics of terrestrial ecosystems, including
the presence of trees and their successional state, are deter-
mined largely by the type, intensity and duration of human
interactions with ecosystems (Hobbs et al., 2006; Ellis &
Ramankutty, 2008). To characterize and understand these
interactions and the global ecological patterns produced by
them, Ellis & Ramankutty (2008) introduced the concept of
anthropogenic biomes, or anthromes, and developed a global
classification and map of these as a new framework for global
ecology and earth science (Alessa & Chapin, 2008). Here we
build on this framework by assessing and mapping the anthro-
pogenic transformation of climate-based ‘potential natural
vegetation’ biomes (also termed ‘classic biomes’ or ‘potential
vegetation’) into anthromes beginning in 1700, an entirely pre-
industrial time period, and proceeding through the Industrial
Revolution to 2000, with the goal of understanding the history
and current state of anthropogenic transformation of the ter-
restrial biosphere.
Ellis & Ramankutty (2008) classified anthromes empirically
using a cluster analysis algorithm that identified globally signifi-
cant patterns in land use and human population density; key
variables characterizing the type and intensity of direct human
interactions with ecosystems (Ellis & Ramankutty, 2008).
However, this a posteriori approach fits anthrome classes and
their statistical signatures to the unique statistical patterns
within a specific dataset, thereby producing optimal, but differ-
ent, classifications when input datasets differ by time period or
estimation method. To investigate changes in anthromes across
time periods, we therefore developed an a priori anthrome clas-
sification procedure to facilitate consistent identification of
anthrome classes similar to those of Ellis & Ramankutty (2008)
across time periods based on the same land-use and human
population variables.
Previous studies have analysed changes in the extent and
composition of the classic biomes caused by land-use changes
over the past 300 years (e.g. Ramankutty & Foley, 1999; Klein
Goldewijk, 2001; Hurtt et al., 2006). In this study, we investigate
the transformation of classic biomes into anthromes with the
goal of understanding the context and history of the novel
anthropogenic ecosystems created by humans over the long
term (Hobbs et al., 2006). Towards this end, we investigate both
the intentional use of land for agriculture and settlements and
the major unintended consequence of this use: the embedding
of remnant, recovering and other non-agricultural and non-
urban lands within the complex anthropogenic landscape
mosaics created by human use of land (Ellis & Ramankutty,
2008).
MATERIALS AND METHODS
Global patterns of anthropogenic transformation of terrestrial
biomes were assessed at 5resolution by comparing potential
natural vegetation maps with anthrome maps at century inter-
vals from 1700 to 2000 using overlay analysis and other analyti-
cal geographic information system (GIS) software tools.
Anthrome classification and mapping was achieved using a
newly developed rule-based anthrome classification model
applied to existing global data for human population density
and agricultural and urban land use. All spatial data, including
inputs and results, are available for download at http://
ecotope.org/anthromes/v2.
Potential natural vegetation – the classic biomes
Ramankutty & Foley’s (1999) ‘potential natural vegetation’
dataset served as the primary biome system used in this study, as
its native 5resolution and derivation from both remote sensing
and ground-based maps makes it an especially reliable base
dataset. Pixels missing from the original dataset, constrained by
a relatively restricted land mask (mostly missing some island
and coastal areas), were filled using a nearest neighbour algo-
rithm. Separate broadleaf and needleleaf classes were combined,
as were ‘desert’ and‘polar desert’ to produce a simplified 12-class
‘potential vegetation’ biome dataset from the 15 original classes.
For an alternative ‘classic’ view of the biosphere, we also con-
ducted analyses using the ‘Olson biomes’ of Olson et al. (2001),
the most commonly used current biome dataset, as described in
Appendix S1 in Supporting Information.
Anthromes
Data sources
Data inputs for anthrome classification were obtained for years
1700, 1800, 1900 and 2000 from existing and newly produced
global 5datasets, to match those of Ellis & Ramankutty (2008).
Global data for human population density and percentage cover
by urban, crop and pasture lands at 5resolution were obtained
using the HYDE data model (Klein Goldewijk & van Drecht,
E. C. Ellis et al.
Global Ecology and Biogeography,19, 589–606, © 2010 Blackwell Publishing Ltd590
2006), a widely used standard in global investigations of land-
use change and its effects (Feddema et al., 2005; Hurtt et al.,
2006). Data used here were obtained using an updated version
of the HYDE 3.1 data model, based on Klein Goldewijk & van
Drecht (2006) in three configurations: an initial data model
very similar to the original published HYDE 3.1 dataset
(http://www.pbl.nl/hyde), a revised ‘best’ data model (the stan-
dard used in this analysis) and lower and upper bounds
datasets designed to highlight uncertainties in the HYDE
dataset. HYDE population data for the year 2000 were
obtained by spatial aggregation of Landscan population data
(Oak Ridge National Laboratory, 2006), and 2000 croplands
and pastures combine remotely sensed land cover with and
agricultural census data. HYDE historical population and
land-use estimates were produced from historical population
and land-use data obtained at several administrative levels (the
largest units are country level) by using these to constrain
spatial allocation models that distribute historical population
densities and agricultural land areas based on their proximity
to urban settlements, climate constraints, soil suitability, dis-
tance to rivers and terrain (Klein Goldewijk & van Drecht,
2006). Data for the percentage of irrigated land were produced
by combining the global map of irrigation areas for 2000
(Siebert et al., 2007) with a set of historical irrigation statistics
and adjusting these to match crop area data in the ‘best’ data
model (see Appendix S2).
Global 5data for rice cover in 2000 were obtained from
Monfreda et al. (2008). Regions with substantial rice cover
(>20%) in previous centuries were then mapped from this by
assuming that, in 1700, rice areas within the densely settled
regionsoftheworld(>100 persons km-2or urban cover
20%) would have been similar to 2000, but for the conversion
of some rice lands into settlements and other non-agricultural
uses. While this approach does not allow the calculation of his-
torical rice areas, it did allow ‘rice villages’ to be identified in
historical periods using the areas of substantial rice cover
mapped for each century.Rice villages were determined by com-
bining substantial rice-cover cells with the densely settled cells
located within 1 geographic degree of these in 2000 to create a
‘maximum rice’ layer, and then removing the densely settled
cells present in each century.
Classification system
To classify anthromes consistently across time periods, a new
a priori anthrome classification algorithm was developed to
emulate the basic form of the a posteriori anthrome classification
of Ellis and Ramankutty (‘Anthromes 1’;2008),using a relatively
simple and transparent a priori classification model built on
standardized thresholds for classifying the same variables
(‘Anthromes 2’; Table 1; see Appendix S3). First, the initial stage
of the Anthromes 1 classification was replicated by stratifying 5
Table 1 Description of anthrome classes.
Level Class Description
Dense settlements Urban and other dense settlements
11 Urban Dense built environments with very high populations
12 Mixed settlements Suburbs, towns and rural settlements with high but fragmented populations
Villages Dense agricultural settlements
21 Rice villages Villages dominated by paddy rice
22 Irrigated villages Villages dominated by irrigated crops
23 Rainfed villages Villages dominated by rainfed agriculture
24 Pastoral villages Villages dominated by rangeland
Croplands Lands used mainly for annual crops
31 Residential irrigated croplands Irrigated cropland with substantial human populations
32 Residential rainfed croplands Rainfed croplands with substantial human populations
33 Populated rainfed cropland Croplands with significant human populations, a mix of irrigated and rainfed crops
35 Remote croplands Croplands without significant populations
Rangeland Lands used mainly for livestock grazing and pasture
41 Residential rangelands Rangelands with substantial human populations
42 Populated rangelands Rangelands with significant human populations
43 Remote rangelands Rangelands without significant human populations
Seminatural lands Inhabited lands with minor use for permanent agriculture and settlements
51 Residential woodlands Forest regions with minor land use and substantial populations
52 Populated woodlands Forest regions with minor land use and significant populations
53 Remote woodlands Forest regions with minor land use without significant populations
54 Inhabited treeless and barren lands Regions without natural tree cover having only minor land use and a range of populations
Wildlands Lands without human populations or substantial land use
61 Wild woodlands Forests and savanna
62 Wild treeless and barren lands Regions without natural tree cover (grasslands, shrublands, tundra, desert and barren lands)
For details of classification see Appendix S3.
Anthropogenic transformation of the biomes
Global Ecology and Biogeography,19, 589–606, © 2010 Blackwell Publishing Ltd 591
cells into six population density classes differing by orders of
magnitude (urban, >2500 persons km-2; dense, >100 persons
km-2; residential, 10–100 persons km-2; populated, 1–10 persons
km-2; remote; <1personkm
-2; wild, 0 persons km-2). Then,
anthromes were classified by applying a sequence of classifica-
tion thresholds to global gridded data for land area covered by
urban settlements, rice, irrigated and rainfed crops and pastures
as detailed in Appendix S3 and described below.
The Anthromes 2 classification used the same five basic
anthrome classification levels as Anthromes 1 (see Appendix S3;
anthrome ‘groups’ in Ellis & Ramankutty, 2008) while simplify-
ing the system to improve the consistency and interpretability of
anthrome classes. Theforested’ level of Anthromes 1 was broad-
ened to a ‘seminatural’ level incorporating both woodlands and
‘inhabited treeless and barren lands’, retaining the general
meaning of this level as lands with relatively low levels of agri-
culture and urban land use (Table 1; see Appendix S3). The most
distinctive aspects of Anthromes 1 classes were retained while
simplifying anthrome class identification and interpretation by
collapsing village classes from 6 to 4, croplands classes from 5 to
4, and wildlands classes from 3 to 2. Classification was further
simplified by standardizing to a single ‘dominant’ land-cover
threshold of 20% and using this to classify anthromes in declin-
ing order of their land-use intensity and population density,
starting with the most intensively used (urban >rice >irrigated
>cropped >pastured) and densely populated (urban >dense >
residential >populated >remote) anthromes and finishing with
wildlands at the end (see Appendix S3). The identity of village
anthromes was clarified by limiting these only to regions with
histories of intensive subsistence agriculture (areas outside of
North America, Australia and New Zealand; see Appendix S3).
To simplify interpretation, anthrome levels were aggregated into
three basic categories: ‘Used anthromes’ (a combination of the
dense settlements, village, cropland and rangeland anthrome
levels), ‘seminatural anthromes’, and wildlands. Though the
spatial configuration of earth’s remaining wildlands is partly the
result of human activity, their ecology is still considered distinct
from that of anthromes and they are therefore referred to as
‘wildlands’ and not as ‘wildlands anthromes’.
Classification sensitivity
To develop the Anthromes 2 classification procedure used here,
a variety of algorithms were explored towards the goal of repli-
cating the Anthromes 1 classes as closely as possible using a
simple procedure, both from the original Anthromes 1 input
dataset and from the year 2000 data of our historical dataset.
Similarities between maps with different class definitions and
numbers of classes were tested using Cramer’s Vstatistic (Rees,
2008), a dimensionless symmetric indicator of association cor-
rected for chance that is similar to Kappa, with 1 representing
identical maps and 0 representing no relationship between
maps, calculated from land-area-weighted cross-tabulations of
mapped grid cells (see Appendix S4). Values of Cramer’s V
above 0.4 and 0.6 indicate ‘relatively strong’ and ‘strong’ simi-
larities between datasets, respectively (Rea & Parker, 1997).
The strong similarity of Anthromes 1 and 2 classification
was demonstrated by high values of Cramer’s Vwhen the
Anthromes 2 model was applied to the Anthromes1 input dataset
(0.67; see Appendix S4 and Table 1). Similarity remained rela-
tively strong, even when the two different classification models
were applied to their two different native datasets (Cramer’s V=
0.53), especially when considering Anthromes 2 maps for 2000 in
comparison with 1900 (Cramer’s V=0.46) and the comparison
of potential vegetation biomes (Ramankutty & Foley,1999) with
Olson biomes (Olson et al., 2001; Cramer’s V=0.49). The sensi-
tivity of Anthromes 2 classification to choice of model threshold,
variations in input datasets, and the spatial resolution of analysis
was also tested relative to changes between time periods, and in
comparison with the Anthromes 1 map and potential vegetation
maps as detailed in Appendix S4. With only one exception, the
largest differences between maps were observed across time
periods (Cramer’s Vof 0.46 for 2000 compared with 1900,
declining to 0.33 in 1700; Appendix S4 and Table 1), with changes
in anthrome level areas of 82% to 135% relative to 2000 (Appen-
dix S4 and Table 2). Coarsening the spatial resolution of analysis
(0.5° grid and 7700 km2equal-area hexagons) had the next
largest effect (Cramer’s Vas low as 0.48) especially at the dense
settlements level, producing the only instance of a larger effect
than temporal change. Halving and doubling the anthrome clas-
sification thresholds for land and population classes also pro-
duced significant differences, but these were much lower than
those between time periods and spatial resolutions (Cramer’s V
as low as 0.59; anthrome level changes between 19% and 89%),
while the use of input datasets with different model assumptions
and with higher and lower population and land use had the
smallest effects, with Cramer’s Vs greater than 0.94 and anthrome
level changes between 4% and 51% of the standard year 2000
values.
Areas of anthromes, ‘used lands’ and ‘unused lands’
Anthrome areas represent total land areas within anthrome cells,
and therefore include both lands in use for agriculture and
settlements and those without such use. Areas of different land-
use classes may therefore be calculated within each anthrome
class, for example the area of pasture or crops land use within
rangelands anthromes, crops and urban lands within croplands
anthromes, or crops and urban lands within urban anthromes.
To facilitate global investigation of lands with and without use
for agriculture and settlements, we further define the category of
‘used lands’ as the sum of all crop, pasture and urban land-use
areas within each cell. Areas of ‘unused lands’ were then calcu-
lated from land areas remaining in cells after ‘used land’ areas
were subtracted. ‘Unused lands’ are thus defined as‘lands not in
use for agriculture or urban settlements’ and therefore still con-
taining land managed for uses other than crops, pasture and
urban settlements (e.g. forestry, mining, parks and non-urban
housing), together with terrestrial ecosystems either recovering
from some use, or never used directly by humans. Unless oth-
erwise noted, ‘unused lands’ include both unused lands embed-
ded within anthromes and wildland areas outside anthromes.
E. C. Ellis et al.
Global Ecology and Biogeography,19, 589–606, © 2010 Blackwell Publishing Ltd592
RESULTS
Global and regional changes in biomes and
anthromes, 1700–2000
Changes in global areas
Anthromes are mapped by century in Fig. 1 and characterized as
a percentage of ice-free land by biome and region in Fig. 2 (see
Appendix S5 for more comprehensive statistics; original
datasets and online maps are available at: http://ecotope.org/
anthromes/v2). Figure 3 illustrates the percentage of global
population and used and unused lands found within each
anthrome. In 1700, about 95% of earth’s ice-free land was in
wildlands and seminatural anthromes (Fig. 2a). By 2000, 55% of
earth’s ice-free land had been transformed into rangelands,
croplands, villages and densely settled anthromes, leaving less
than 45% of the terrestrial biosphere wild and seminatural
(Fig. 2a). Further, anthropogenic changes between 1700 and
1800 were far smaller than those of the following centuries, and
the rate of change increased over time. As a result, the 20th
century stands out not only as the most dynamic period of
anthropogenic ecosystem transformation of the past 300 years,
but also as the period during which the terrestrial biosphere
transitioned from a primarily wild and seminatural state to a
primarily used state (Fig. 2a).
Changes in biomes
Anthropogenic transformation of most biomes followed similar
trends, with some key exceptions (Fig. 2b). The colder and drier
biomes, including boreal and mixed woodlands, tundra, and
deserts, showed very little change in wild area over time. The
same was observed in the analogous Olson biomes (see Appen-
dix S1), except for the ‘deserts and xeric shrublands’ biome,
probably because it merges shrublands with deserts. Temperate
deciduous woodlands were already used fairly heavily in 1700
(28% in used anthromes), but most other woodlands, savannas
and grasslands were predominantly in use at lower, seminatural
levels. Over the next 300 years, most of these biomes were con-
verted from wild and seminatural lands to croplands, rangelands
and other more intensively used anthromes (Fig. 2b).
Differences in biome composition help explain some
regional differences in biome transformation patterns (Fig. 2c),
including the long-term maintenance of wildlands in the Near
East, Africa and Eurasia, which have large deserts and/or boreal
areas. But other regional differences are distinct from this,
including the high pre-industrial levels of wildlands in North
America, Australia and New Zealand, and in Latin America and
the Caribbean, and their very dramatic conversion to used
anthromes by 2000. These regions contrast with Africa, Asia
and Oceania, which were primarily seminatural in 1700 and
were then transformed into more heavily used cropland and
village anthromes. Europe, which was mostly used in 1700,
stayed that way.
Changes in human population
As human populations and their use of land expanded from
1700 to 2000, their distribution among anthromes also changed
(Fig. 3a). In 1700, nearly half of earth’s human population lived
in seminatural lands – thinly dispersed in relatively extensively
used landscapes, with the remainder dwelling about equally in
croplands and villages (Fig. 3a). By 2000, this had changed com-
pletely, with only 4% still living in seminatural anthromes and
more than half dwelling in villages (51%). Half of earth’s popu-
lation now lives in cities (UNFPA, 2007), of which about 60%
reside in urban anthromes (29% of global population) with the
other 40% of urban populations dwelling in the smaller cities
and towns embedded within villages and other anthromes.
Changes in land distribution within and among anthromes
Changes in the distribution of ice-free land among anthromes
followed similar trends as population, with a dramatic shift away
from seminatural anthromes and wildlands towards the used
anthromes (Fig. 3a, b). Still, dense settlements and villages, the
most populous anthrome levels, even today account for less than
8% of global ice-free land. Change trends in unused lands
(Fig. 3c) continued to resemble those of total land (Fig. 3b) even
while their total area declined by 34% globally,from nearly 95%
of earth’s ice-free land in 1700 to just 61% in 2000. Over this
period, the portion of unused lands that were embedded within
the seminatural and used anthromes, and therefore outside of
wildlands, remained fairly stable, hovering near half of the
global total from 1700 to 1800 and increasing to about 60% in
1900 and staying there (Fig. 3c) However, the global proportion
of unused lands embedded within the used anthromes increased
tenfold from 1700 to 2000, from 3% to 30% of the global total,
with unused lands embedded in the seminatural anthromes
simultaneously declining from 45% to 29% of global land
remaining unused.
From 1700–2000, lands used for agriculture and urban settle-
ments increased from 5% to 39% of total ice-free land area,
while retaining a fairly constant proportion of dense settle-
ments, villages and croplands (Fig. 3d). However, the extent of
rangelands increased rapidly in every century, eventually replac-
ing seminatural as the dominant anthrome level between 1900
and 2000 (Fig. 3b, d). As a result, the foremost global land-use
change of the Industrial Revolution in terms of total area was the
expansion of pastures from 3% of ice-free land in 1700 to 26%
in 2000 (Fig. 3e). During this period, pastures shifted from
being predominantly a minor land use embedded within the
seminatural anthromes to becoming essentially an anthrome
unto itself, with nearly three quarters of all pastures located
within rangelands anthromes by 2000 (Fig. 3e). Similarly, crop
areas rose spectacularly during the Industrial Revolution, from
about 2% in 1700 to about 12% of global land area by 2000,
and like pasture lands, crops also became less a component of
seminatural anthromes than the defining component of crop-
land and village anthromes over this period (Fig. 3f). Irrigated
areas also increased very rapidly, yet were always concentrated
Anthropogenic transformation of the biomes
Global Ecology and Biogeography,19, 589–606, © 2010 Blackwell Publishing Ltd 593
within the most intensively used of the agricultural anthromes –
the villages and croplands – with their increase causing ‘rice
villages’, ‘irrigated villages’and‘residential irrigated croplands’ to
expand over time (Fig. 3g). Finally, the most intensively trans-
formed lands on earth, urban lands, also expanded in the most
dramatic fashion of all, changing by a factor of 40 from almost
insignificant in 1700 (0.01% of all land) to 0.4% of all land in
2000 (0.53 ¥106km2), shifting from a minor component of
2500 km
2000
1900
Urban
Mixed settlements
Dense Settlements
Wild woodlands
Wild treeless & barren lands
Wildlands
Inhabited treeless & barren lands
Residential woodlands
Populated woodlands
Remote woodlands
Seminatural
Rangelands
Residential rangelands
Populated rangelands
Remote rangelands
Residential irrigated croplands
Residential rainfed croplands
Populated croplands
Remote croplands
Croplands
Rice villages
Irrigated villages
Rainfed villages
Pastoral villages
Villages
Anthromes
Wild
larutanimeSdesU
Figure 1 Anthropogenic biomes, 1700–2000 (anthromes; class descriptions in Table 1). Region boundaries (2000) are distinguished by
black lines; same regions as Ellis & Ramankutty (2008). Eckert IV projection.
E. C. Ellis et al.
Global Ecology and Biogeography,19, 589–606, © 2010 Blackwell Publishing Ltd594
villages and croplands to becoming the defining land use of
urban anthromes by 2000 (Fig. 3h).
Pathways and dynamics of biome transformation
The classic view of the biosphere, as composed of natural veg-
etation biomes, is illustrated in Fig. 4(a) together with maps
portraying the global extent to which these were transformed
into anthromes by 1700 (Fig. 4c) and 2000 (Fig. 4b). From these
figures it is clear that even in 1700, slightly more than half of the
terrestrial biosphere was already inhabited and used signifi-
cantly (Fig. 4c), albeit mostly at relatively low seminatural levels
and mostly in Europe, sub-Saharan Africa, South and East Asia
and Central America. By 2000, the overwhelming majority of
the terrestrial biosphere had been transformed into anthromes
(Figs 2a, 4b & 5a), the result of converting about half of both the
wildlands (Figs 4d & 5a) and the seminatural anthromes
(Figs 4e & 5a) of 1700 into used anthromes. The historical wild-
lands of 1700 became rangelands in most parts of the world and
croplands in North America and South Australia (Fig. 4d). The
seminatural lands of 1700 became a mix of rangelands and
croplands in most of the world, and villages in Asia (Fig. 4e).
The dynamics of anthropogenic transformation differed pro-
foundly among biomes (Fig. 5b). Boreal and mixed woodlands,
tundra and deserts changed little over the past 300 years, with
less than 20% of their wildlands transformed and most change
occurring in areas already transformed to seminatural
anthromes by 1700. Contrasting with this, grasslands, savannas
and shrublands showed the greatest changes over time, with all
of these experiencing >80% conversion to used anthromes from
1800
1700
Figure 1 Continued
Anthropogenic transformation of the biomes
Global Ecology and Biogeography,19, 589–606, © 2010 Blackwell Publishing Ltd 595
1700 to 2000. Most of this was the result of converting
both wildlands and seminatural anthromes to rangelands,
though the conversion to croplands was substantial in
grasslands (28%), savanna (23%) and dense shrublands
(19%).
Woodlands showed more moderate change. Over the course
of the Industrial Revolution, about one-third of earth’s tropical
evergreen woodlands were converted into used anthromes
(from wildlands and seminatural anthromes), and about 22% of
their wildlands were converted to seminatural and rangeland
anthromes. More than half of the area of the other woodlands
biomes were transformed into used anthromes between 1700
and 2000. Tropical deciduous woodlands were transformed
predominantly into rangelands (28%) and villages (24%), while
the temperate woodlands were transformed into croplands
(23–28%), villages (14–18%) and dense settlements (4–7%),
with most change occurring in areas previously covered by
seminatural anthromes.
Current state of the anthropogenic biosphere
The history and intensity of land transformation varies tremen-
dously across the surface of the earth, with some biomes and
Tundra
Potential
vegetation
Desert &
Barren
Open
Shrubland
Dense
Shrubland
Grassland
& Steppe
Savanna
Mixed
Woodland
Boreal
Woodland
Temperate
Deciduous
Woodland
Temperate
Evergreen
Woodland
Tropical
Deciduous
Woodland
1900
1800
1700
(a)
Tropical
Evergreen
Woodland
2000
1900
1800
1700
2000
1900
1800
1700
2000
Eurasia
Europe
Asia
&
Oceania
Near
East
Latin
America
&
Caribbean
Africa
North
America
Australia
&
New
Zealand
(c)
100%
0%
50%
(b)
Seminatural
Urban
Mixed settlements
Dense Settlements
Wild woodlands
Wild treeless
& barren lands
Inhabited treeless
& barren lands
Residential woodlands
Populated woodlands
Remote woodlands
Rangelands
Residential
Populated
Remote
Residential irrigated
Residential rainfed
Populated
Remote
Croplands
Rice
Irrigated
Rainfed
Pastoral
Villages
Wild
Used
Wildlands
Seminatural
Anthromes
(%ice-
free
area)
Global
land
100%
0%
100%
0%
Potential
vegetation
Regions
Figure 2 Global changes in anthromes, 1700–2000, expressed as a percentage of global ice-free land area (a), as a percentage of land area
within potential natural vegetation biomes (b; Ramankutty & Foley, 1999) and as a percentage of global region area (c; regions outlined in
Fig. 1). Columns in (a), (b) and (c) sum to 100% global ice-free land area. Trends in the combined areas of ‘used’ anthromes are
highlighted by red shading and wildlands by green shading; seminatural anthromes left blank. Anthrome changes within Olson biome
classes (Olson et al., 2001) are in Appendix S1.
E. C. Ellis et al.
Global Ecology and Biogeography,19, 589–606, © 2010 Blackwell Publishing Ltd596
regions almost entirely transformed and others almost uninflu-
enced by direct human activity (Figs 2 & 6). The maps in Fig. 6
summarize and illustrate these differences and their dynamics in
a general way. In 1700, most of earth’s land was already moder-
ately transformed by human populations and land use (Fig. 6a).
In subsequent centuries, land use intensified, accelerated and
spread in highly dynamic and often contradictory patterns
(Fig. 6b, c, d), such that even while most regions were being
transformed at their most rapid rates in history, others, such as
eastern North America and the northern fringes of the former
Soviet Union experienced attenuation of human influence, espe-
cially in the 20th century (Fig. 6d). And while regions experi-
encing the most intensive transformation, to villages and
croplands (Fig. 6f ), tended also to have the longest periods of
human habitation and land use (Fig. 6g), this was not always the
case, most notably, in the grasslands of central North America,
which experienced major transformation (Fig. 6f) but were
mostly unused for agriculture and urban settlements prior to
1800 (Fig. 6a, b).
Even after 300 years of extensive anthropogenic transforma-
tion, more than 60% of the terrestrial biosphere remains
unused directly for agriculture or urban settlements (Fig. 3c).
Of these unused lands still remaining on earth in 2000, only
about 40% are wildlands (see Appendix S5; Fig. 7). The other
60% are embedded within dense settlements (1%), villages
(3%), croplands (7%), rangelands (19%) and seminatural
anthromes (29%). Taken together, these embedded unused
lands (Fig. 6h) represent an extent of human-altered terrestrial
ecosystems that is substantially greater than that of all of
earth’s remaining wildlands combined, accounting for about
37% of all ice-free land (19% in used anthromes, 18%
seminatural). Still, the global extent, type, duration and inten-
sity of anthropogenic transformation of these embedded
unused lands is a challenge to determine. Estimating the extent
of older anthromes, with more than 300 years since their
transformation from wildlands (Fig. 6g), and with significant
areas (>20%) of their land remaining unused (Fig. 6i), reveals
that potentially ancient novel anthropogenic ecosystems now
100 100 100 100 100 100 100 100
Land
Used IrrigatedCropPasture
Population UrbanUnused
(%)
(b) (d) (h)(g)(f)(e)
6
(a) (c)
All
1700
2000
1800
1900
0
0
(106km2)
128 50
0.5
2.5
15
30
120
Urban
Mixedsettlements
Dense Settlements
Wild woodland
s
Wildlands
Inhabited treeless & barren
Residential woodlands
Populated woodlands
Remote woodlands
Rangelands
Residential rangelands
Populated rangelands
Remoterangelands
Residential irrigated croplands
Residential rainfed croplands
Populated croplands
Remote croplands
Croplands
Rice villages
Irrigated villages
Rainfed villages
Pastoral villa
g
es
Villages Seminatural
Wild treeless
& barren
Global
area
(%)
Global
(109 persons)
Global
land
Global
00 0 0 0 0 0
0 0 0 0 0 0 0
Figure 3 Global changes in anthromes, 1700–2000, relative to global population (a) and land use (b–h). Global percentages for each time
period are at the top, absolute values at the bottom (scales differ for each variable, as marked), for global population (a), all ice-free land
(b), unused lands (c; lands not in use for agriculture or urban settlements), used lands (d; crops +pasture +urban), pasture land (e),
cropped land (f), irrigated land (g) and urban land (h).
Anthropogenic transformation of the biomes
Global Ecology and Biogeography,19, 589–606, © 2010 Blackwell Publishing Ltd 597
(b)
(c)
(e)
(d)
2500 km
Dense settlements
Villages
Croplands
Anthrome Level
Rangelands
Seminatural
Wildlands
(a)
Tropical Evergreen Woodland
Tropical Deciduous Woodland
Temperate Evergreen Woodland
Temperate Deciduous Woodland
Boreal Woodland
Mixed Woodland
Grassland & Steppe
Deserts & Barren
Tundra
Savanna
Open Shrubland
Dense Shrubland
Potential Vegetation
2000
1700
Anthrome
Level
Potential
Vegetation
2000
Wildlands
of 1700
2000
Seminatural
of 1700
Figure 4 Potential natural vegetation biomes (a; Ramankutty & Foley, 1999) and their anthropogenic transformation from 1700 to 2000
(b–d). Levels of anthrome transformation of the terrestrial biosphere in 2000 (b) and in 1700 (c) are illustrated, along with the year
2000 anthrome levels of 1700s-era wildlands (d) and seminatural anthromes (e). White spaces in (d) and (e) are non-wild and
non-seminatural areas, respectively, in 1700. Eckert IV projection.
E. C. Ellis et al.
Global Ecology and Biogeography,19, 589–606, © 2010 Blackwell Publishing Ltd598
cover more than 19% of all ice-free land (11% in used
anthromes, 8% in seminatural). The degree to which earth’s
remaining unused lands are present in wildlands versus
embedded within different anthromes is illustrated for differ-
ent biomes and regions in Fig. 7(a) and (b), respectively. From
this, it is clear that only the cold and dry biomes (boreal, shru-
blands, deserts), and the global regions with large extents of
these (North America, Australia and New Zealand, the Near
East and Eurasia) still have large extents of wildlands. Most
of Earth’s unused lands are now embedded within the
agricultural and settled landscapes of seminatural, rangeland,
cropland and village anthromes.
DISCUSSION
Anthromes as descriptors of the biosphere
Limits to anthrome classification
Anthromes, like biomes, are generalizations useful for
understanding global patterns of ecosystem form and process.
Tundra Desert &
Barren
Open
Shrubland
Dense
Shrubland
Grassland
& Steppe
Savanna
Mixed
Boreal
Temperate
Deciduous
Temperate Evergreen
Tropical
Deciduous
Tropical
Evergreen
Wildlands
of 1700
Seminatural
anthromes
of 1700
Used
anthromes
of 1700
Woodlands
1700 2000
Wildlands
of 1700
Seminatural
anthromes
of 1700
Used
anthromes
of 1700
All Land
All
Land
Potential
Vegetation
100%0%
0%
100%
0%
100%
(b)
(a)
(%1700 area) (%1700 area) (%1700 area)
(% global area)
% potential
vegetation
class area
% potential
vegetation
class area
0%
100%
0%
100%
0%
100%
(% ice-free land)
Terrestrial Biosphere
Dense settlements
Villages
Croplands
Rangelands
Seminatural lands
Wildlands
Anthrome Level
Anthromes
1700
1700 2000
0%
100%
Anthromes
1700
0% 100%
Figure 5 Transformation of the terrestrial biosphere (a) and potential natural vegetation biomes (b; Ramankutty & Foley, 1999) from 1700
to 2000. Year 1700 anthrome levels across the entire terrestrial biosphere (a) and within each potential vegetation biome (b) are depicted using
horizontal bars at the top of (a) and (b); bars sum to 100% of global ice-free land area. Anthrome-level changes from 1700 to 2000 across the
terrestrial biosphere (a) and within each potential vegetation biome (b) are illustrated by vertical area charts representing all land (all land; a
on left; b in top row), and the 1700s areas of wildlands, seminatural anthromes and used anthromes (on right in a; lower rows in b).
Anthropogenic transformation of the biomes
Global Ecology and Biogeography,19, 589–606, © 2010 Blackwell Publishing Ltd 599
Figure 6 Global patterns of anthropogenic transformation and novel ecosystem development, 1700–2000. Anthropogenic land
transformations are highlighted using an index of transformation calculated by subtracting anthrome classes between time periods (legend
at upper right), for all change up to 1700 (a), between 1700 and 1800 (b), 1800 and 1900 (c), 1900 and 2000 (d), 1700 and 2000 (e) and for
all change up to 2000 (f). Time since conversion to anthromes (g), percentage of anthrome area consisting of embedded unused lands (h;
lands not used for agriculture or settlements, not including wildlands), and the anthrome level of all cells with >20% cover by unused
lands with at least 300 years elapsed since their conversion from wild biomes (i). Eckert IV projection.
E. C. Ellis et al.
Global Ecology and Biogeography,19, 589–606, © 2010 Blackwell Publishing Ltd600
Moreover, anthromes provide a simple framework for assessing
and modelling both past and future global biotic and ecological
patterns in the light of the extent, intensity and duration of their
modification by humans. Still, anthromes, like biomes or any
other global classification system, are built on subjective trade-
offs between detail and simplicity and usually require a variety
of practical compromises to make their mapping possible using
available data. One example of such a trade-off is our use of a
priori classification thresholds, which force the division of lands
into different categories at a given value of a variable, even when
Figure 7 Changes in unused land areas and their distribution among anthromes from 1700 to 2000 within potential natural vegetation
biomes (a) and regions (b). Unused land areas (lands not used for agriculture or settlements) within potential vegetation biomes (a) and
regions (b) are depicted as a percentage of global ice-free land at the left in (a) and (b) The relative areas of potential vegetation biomes are
highlighted in (b) using the same colours as in (a); columns sum to 100% of global land. Changes in the distribution of unused lands
among anthromes over time are illustrated at right within each potential vegetation biome (a) and region (b) as a percentage of the total
unused land area at each time (sum =100% of unused land area within each biome or region at each time). Trends in the combined areas
of ‘used anthromes’ are highlighted by red shading and wildlands by green shading; seminatural anthromes are left blank.
Anthropogenic transformation of the biomes
Global Ecology and Biogeography,19, 589–606, © 2010 Blackwell Publishing Ltd 601
they may differ only slightly. An important case in point is the
distinction between wildlands and the remote woodlands,
rangelands and croplands anthromes, which can differ only
because of very small differences in population density and land
use (see Appendix S3). It is therefore necessary that any
anthrome classification, including this one, be applied with full
knowledge of these limitations and with consideration of the
fact that other anthrome classifications are possible. Different
systems would undoubtedly yield different results, depending
both on differences in the classification model and on the data
used as input.
On the positive side, our current anthrome system proved to
have more in common with the original system of Ellis &
Ramankutty (2008) than the two classic biome systems we used
in this study had with each other (see Appendix S4). Moreover,
the degree of variation introduced by our choice of model, data
and the spatial resolution of analysis, while significant, was
quantifiable and fairly predictable, with the largest effects caused
by changing spatial resolution, and with all of these variations
being substantially less than those caused by changes over time
(see Appendix S4). For example, in comparison with our stan-
dard area estimate for the year 2000, the global wildlands esti-
mate varied by about 20% across datasets and models, but
was lowered by 79% at the coarsest resolution of analysis, with
our standard 1700 estimate being 95% lower than for 2000.
Interpreting the ecology of anthromes
The ecological properties of anthrome classes must also be inter-
preted with some caution. For example, while the pasture land
use that defines rangelands anthromes is usually indicative of
enhanced herbivory and disturbance by livestock populations,
there are also cases where the ecological effects of pasture land
use on the form and diversity of vegetation can be quite small,
such that rangelands may closely resemble wildlands within a
particular biome (Steinfeld et al., 2006). Further, unlike classic
biomes, which attempt to represent fairly homogeneous forms
of vegetation, anthromes represent complex mixtures of differ-
ent land uses and land covers that are far harder to characterize
in simple terms (Ellis & Ramankutty, 2008).Moreover,the same
statistical characteristics of human populations and land use
that define a particular anthrome can yield quite different eco-
systems in different biomes, such as remote rangelands classified
within woodlands versus shrublands or deserts.
The same anthrome class may also differ in important ways
during different periods in human history. A key example is
provided by seminatural woodlands – woodlands with relatively
low levels of land use and human populations (Table 1; see
Appendix S3). In 1700 and especially before, most seminatural
woodlands were probably managed by shifting cultivators,
leaving most of their forest vegetation in relatively early stages of
recovery from land clearing and brief use for agriculture (Hurtt
et al., 2006; Ruddiman & Ellis, 2009). By 2000, however, shifting
cultivation was far less common, leaving seminatural woodlands
in regions with heterogeneous terrain allowing only partial use
of land for agriculture, as in much of Southeast Asia (Fig. 1), and
in areas experiencing agricultural abandonment, such as the
eastern United States and northern Europe (Fig. 6d). In both
cases these anthromes probably support forests in much later
successional stages, and often with human settlements under-
neath the forest canopy (Rudel et al., 2005).
Scientific understanding of anthrome ecology
Difficulties in interpreting the global ecological patterns created
and sustained by direct human interactions with ecosystems
result from our very limited scientific understanding of coupled
human and ecological systems (Turner et al., 2007; Ellis &
Ramankutty, 2008). Certainly our understanding of these
systems is far inferior to our understanding of the global eco-
logical patterns produced by biophysical processes alone. While
observations from remote sensing have revolutionized our
ability to see the global patterns and dynamics of vegetation and
other land covers across the earth’s surface, the causes of these
patterns and their dynamics are not directly observable from
above, or even from the ground,without intensive local research
efforts aimed at understanding both ecological and human
systems (Rindfuss et al., 2004; Turner et al., 2007; Ellis et al.,
2009). Given that such efforts are extremely costly and time-
consuming, global strategies are required to allocate such local
observations effectively (Ellis et al., 2009). By stratifying the
global ecological patterns created by humans, anthrome classi-
fication may serve both as an aid in selecting global samples of
local land change processes for observation and in the synthesis
of these observations into theory, helping to build a global
ecology that incorporates humans as sustained shapers and
managers of local and global ecosystem form and function.
Anthropogenic transformation of the biosphere,
1700–2000
When did the terrestrial biosphere become anthropogenic?
Historical analysis of changes in anthrome extent and compo-
sition confirm that the terrestrial biosphere shifted from a pri-
marily wild to a primarily anthropogenic state between 1700
and 2000 and that rapid intensification of land use in the 20th
century finally pushed the biosphere into its present anthro-
pogenic state (Fig. 2a). Still, it is important to recognize that
while about half of this transition was caused by the anthro-
pogenic transformation of lands that were still wild in 1700
(Figs 4d & 5a) the other half was caused by intensification of
land use in the seminatural anthromes that already covered
nearly half of the terrestrial biosphere in 1700 (Figs 4e & 5a).
The ecological significance of more than 45% of earth’s ice-
free land being inhabited and used at lower levels in 1700
should not be underestimated, as about 60% of earth’s tropical
and temperate woodlands were included in this seminatural
area, and these were most likely inhabited primarily by shifting
cultivators who may have cleared almost this entire area, one
small patch at a time, at some point in history or pre-history
(Ruddiman & Ellis, 2009). Moreover, even areas without
E. C. Ellis et al.
Global Ecology and Biogeography,19, 589–606, © 2010 Blackwell Publishing Ltd602
significant human populations or use of land for crops or
pasture in 1700 and later, and therefore considered ‘wild’ in
this analysis, may still have been significantly altered ecologi-
cally by prior use of land and by intensive and systematic for-
aging by sparse human populations (Cronon, 1983). It is also
important to note that the lion’s share of global land transfor-
mation since 1700 was the result of increasing pasture areas
and the consequent rise of rangelands anthromes (Fig. 2a).
This may have had relatively light impacts on ecosystem form
and process even in comparison with the conversion of wild-
lands to seminatural remote woodlands, if these were managed
by shifting cultivators. Nevertheless, by almost any standard,
the extent of human populations and their use of land in the
20th century supports the conclusion that, by 2000, most of
the terrestrial biosphere was transformed into predominantly
anthropogenic ecological patterns combining lands used for
agriculture and urban settlements and their legacy; the
remnant, recovering and other managed novel ecosystems
embedded within anthromes.
How did the terrestrial biosphere become anthropogenic?
The largest global change in land use over the past 300 years
was the near sixfold increase in the global extent of pastures
from 1800 to 2000 (Figs 2a & 3e). This vast increase in pastures
drove the emergence of new rangelands anthromes across the
wild and mostly dryer biomes of the Americas, Australia,
Central Asia and southern Africa (Figs 1 & 4d) and in the semi-
natural anthromes of the mostly moister wooded biomes of
sub-Saharan Africa, Central America and Eurasia (Fig. 4e). The
other major driver of biospheric transformation was the rapid
expansion of crops and croplands into the wild grasslands of
North and South America, the shrublands and wild woodlands
of southern Australia (Fig. 4d) and the seminatural grasslands
of Eurasia (Fig. 4e).
While these tremendous expansions in land use might seem
the most important anthropogenic changes in the terrestrial
biosphere during the Industrial Revolution, given that together
they covered nearly 50% of global land, it should not be for-
gotten that over the same period these were combined with
dramatic intensifications in land use. Land-use intensification
caused the most intensively used anthromes to expand in every
century, not only in terms of the increasing global extents of
villages and dense settlements, but also by increasing the areas
of the more densely populated anthrome classes within each
anthrome level (Figs 1, 2a & 6f). The precipitous decline in
both wildlands and seminatural anthromes in all but the
coldest and driest biomes must therefore be interpreted as the
combined result of both land-use expansion and land-use
intensification (Figs 2a & 5b). As a consequence, the terrestrial
biosphere is now used far more intensively than it ever has
been, though some attenuation of land use did occur during
the 20th century due to agricultural abandonment in northern
Eurasia, the eastern United States and parts of sub-Saharan
Africa (Fig. 6d).
Implications of an anthropogenic biosphere
Conserving nature in anthromes
At this point in history, about 40% of all ice-free land on earth
is in direct use for agriculture or urban settlements (Fig. 3d). An
additional 37% of ice-free land is not currently used for these
purposes, but is embedded within anthromes having these uses
(Figs 6i & 7). This leaves wildlands in the minority, a mere 22%
of global ice-free land area, with about 85% of these located only
in the cold and dry biomes of the world, a result confirming
earlier estimates (Sanderson et al., 2002).
As most of earth’s land not currently in use for agriculture or
urban settlements is now embedded within anthromes, it is the
ecology of these embedded ‘unused lands’ that should now
matter most in conserving the species and ecosystems we value.
The critical challenge, therefore, is in maintaining, enhancing
and restoring ecological functions in the remnant, recovering
and managed ecosystems formed by land use and its legacies
within the complex multifunctional anthropogenic landscape
mosaics that will be the predominant form of terrestrial ecosys-
tems today and into the future (Hobbs et al., 2006).
There is growing evidence that some lands used directly for
agricultural production can sustain high levels of biodiversity,
similar to those of lands unused for agriculture or settlements,
especially in ancient agricultural regions (Ranganathan et al.,
2008; Chazdon et al., 2009). Moreover, depending on how the
mosaic structure of landscapes is managed to enhance connec-
tivity and habitat values, it is possible to sustain high levels of
wild native biodiversity even in urban and village anthromes
where built-up lands and intensive cropping systems predomi-
nate (Ricketts, 2001; Fahrig, 2003; Lindenmayer et al., 2008;
Chazdon et al., 2009). Yet efforts to sustain and enhance biodi-
versity in anthromes can be challenged by trade-offs between
conservation values and the benefits of using land for agricul-
tural production and settlements (Chazdon et al., 2009). Despite
this challenge, anthromes composed entirely of agricultural and
settled lands are rare; landscape mosaics containing substantial
areas of unsettled agricultural land are the global norm
(Fig. 6h). As a result, global efforts to conserve, enhance and
restore biodiversity within anthromes may be possible without
directly challenging these land uses. Success in this effort will
require that novel anthropogenic ecosystems be the focus of
expanded research, monitoring and conservation efforts in most
terrestrial biomes, as their optimal management, landscape and
community structure, habitat connectivity, ecosystem processes
and dynamics remain poorly understood and cannot be reliably
predicted from past trends or historical environmental con-
straints (Hobbs et al., 2006; Lindenmayer et al., 2008; Chazdon
et al., 2009; Jones & Schmitz, 2009).
Global observing systems and models appropriate for an
anthropogenic biosphere
Existing global land-use and population data, vegetation
models, remote sensing platforms and other data acquisition
Anthropogenic transformation of the biomes
Global Ecology and Biogeography,19, 589–606, © 2010 Blackwell Publishing Ltd 603
systems and models are certainly useful for investigating
current, historical and future ecological patterns across the ter-
restrial biosphere. Indeed, the present study made use of these
to investigate current and historical patterns, and similar
methods may help push this investigation into pre-history. But
this remains a mere descriptive sketch. There remain tremen-
dous uncertainties in our understanding and ability to model
even current global patterns of ecosystem function and biodi-
versity across the anthropogenic biosphere.
Solid theoretical and predictive global models of coupled
human and ecological system dynamics are only now being
developed (e.g. Bouwman et al., 2006; Bondeau et al., 2007), and
most tend to focus on land-cover interactions with climate,
rather than ecosystems (e.g. Brovkin et al., 2006; Olofsson &
Hickler,2008).We need human systems models that are as theo-
retically strong, predictive and useful as the best current bio-
physical models of natural biospheric pattern, process and
dynamics, and we need these models to be coupled together to
produce useful predictions of global ecological patterns, pro-
cesses and dynamics.
The remedy is clear, but both expensive and logistically chal-
lenging: a human biosphere observing and modelling
system built on standardized global observations of coupled
human and ecological systems in the field. Global remote
sensing is a tremendous asset in this effort but it is
simply incapable of observing the causes of human and eco-
logical dynamics. We need standardized observations across
the global spectrum of anthropogenic ecosystems that inte-
grate ecological measurements and social surveys of human
practices at the relatively fine spatial scales at which these
interact (Alessa & Chapin, 2008; Ellis et al., 2009). Ultimately,
based on these observations, we can build strong theoretical
and applied models of anthropogenic ecosystem dynamics at
local, regional and global scales. Given that most of the terres-
trial biosphere is now anthropogenic, the future of all species,
including ours, will depend on understanding and modelling
the past, present and potential future ecology of our
anthropogenic biosphere as we continue to directly alter and
manage it.
ACKNOWLEDGEMENTS
Erle Ellis thanks the audience member at his 2007
American Geophysical Union presentation on anthromes
who asked the question: ‘So how did the biosphere become
anthropogenic anyway?’. Collaborations supporting this
project were initiated at the 2008 Global Land Use
Change Workshop in Vienna, Austria, supported by the Global
Land Project and the Netherlands Ministry of Environmental
Planning. Navin Ramankutty was supported by the Natural
Sciences and Engineering Research Council (NSERC) of
Canada. Thanks to Jonathan Dandois, Erica Antill, William
Ruddiman and Diann Prosser for helpful comments on
the manuscript. Skype software played a key role in project
collaboration.
REFERENCES
Alessa, L. & Chapin, F.S., III (2008) Anthropogenic biomes: a key
contribution to earth-system science. Trends in Ecology and
Evolution,23, 529–531.
Bondeau, A., Smith, P.C., Zaehle, S., Schaphoff, S., Lucht, W.,
Cramer, W., Gerten, D., Lotze-Campen, H., Müller, C., Reich-
stein, M. & Smith, B. (2007) Modelling the role of agriculture
for the 20th century global terrestrial carbon balance. Global
Change Biology,13, 679–706.
Bouwman, A.F., Kram, T. & Klein Goldewijk, K. (2006) Inte-
grated modelling of global environmental change: an overview of
Image 2.4. Netherlands Environmental Assessment Agency,
Bilthoven, The Netherlands.
Brovkin, V., Claussen, M., Driesschaert, E., Fichefet, T.,
Kicklighter, D., Loutre, M.F., Matthews, H.D., Ramankutty,
N., Schaeffer, M. & Sokolov, A. (2006) Biogeophysical effects
of historical land cover changes simulated by six earth system
models of intermediate complexity. Climate Dynamics,26,
587–600.
Chazdon, R.L., Harvey, C.A., Komar, O., Griffith, D.M.,
Ferguson, B.G., Martínez-Ramos, M., Morales, H., Nigh, R.,
Soto-Pinto, L., Breugel, M.V. & Philpott, S.M. (2009) Beyond
reserves: a research agenda for conserving biodiversity in
human-modified tropical landscapes. Biotropica,41, 142–
153.
Cronon, W. (1983) Changes in the land: Indians, colonists, and
the ecology of New England. Hill and Wang, New York.
Dansereau, P.M. (1957) Biogeography: an ecological perspective.
Ronald Press Company, New York.
Dearing, J., Battarbee, R., Dikau, R., Larocque, I. & Oldfield, F.
(2006) Human–environment interactions: learning from the
past. Regional Environmental Change,6, 1–16.
DeFries, R.S., Asner, G.P. & Houghton, R.A. (2004) Ecosystems
and land use change. Geophysical Monograph Series. Ameri-
can Geophysical Union, Washington, DC.
Ellis, E.C. & Ramankutty, N. (2008) Putting people in the map:
anthropogenic biomes of the world. Frontiers in Ecology and
the Environment,6, 439–447.
Ellis, E.C., Neerchal, N., Peng, K., Xiao, H.S., Wang, H., Yan, Z.,
Li, S.C., Wu,J.X., Jiao, J.G., Ouyang, H., Cheng, X. & Yang,L.Z.
(2009) Estimating long-term changes in China’s village land-
scapes. Ecosystems,12, 279–297.
Fahrig, L. (2003) Effects of habitat fragmentation on biodiver-
sity. Annual Review of Ecology Evolution and Systematics,34,
487–515.
Feddema, J.J., Oleson, K.W., Bonan, G.B., Mearns, L.O., Buja,
L.E., Meehl, G.A. & Washington, W.M. (2005) The impor-
tance of land-cover change in simulating future climates.
Science,310, 1674–1678.
Foley, J.A., DeFries, R., Asner, G.P., Barford, C., Bonan, G., Car-
penter, S.R., Chapin, F.S., III, Coe, M.T., Daily, G.C., Gibbs,
H.K., Helkowski, J.H., Holloway, T., Howard, E.A., Kucharik,
C.J., Monfreda, C., Patz, J.A., Prentice, I.C., Ramankutty, N. &
Snyder, P.K. (2005) Global consequences of land use. Science,
309, 570–574.
E. C. Ellis et al.
Global Ecology and Biogeography,19, 589–606, © 2010 Blackwell Publishing Ltd604
Hansen, D.M. & Galetti, M. (2009) Ecology: the forgotten mega-
fauna. Science,324, 42–43.
Hobbs, R.J., Arico, S., Aronson, J., Baron, J.S., Bridgewater, P.,
Cramer, V.A., Epstein, P.R., Ewel, J.J., Klink, C.A., Lugo, A.E.,
Norton, D., Ojima, D., Richardson, D.M., Sanderson, E.W.,
Valladares, F., Vila, M., Zamora, R. & Zobel, M. (2006) Novel
ecosystems: theoretical and management aspects of the new
ecological world order. Global Ecology and Biogeography,15,
1–7.
Holdridge, L.R. (1947) Determination of world plant
formations from simple climatic data. Science,105, 367–
368.
Hurtt, G.C., Frolking, S., Fearon, M.G., Moore, B., Shevliakova,
E., Malyshev, S., Pacala, S.W. & Houghton, R.A. (2006) The
underpinnings of land-use history: three centuries of global
gridded land-use transitions, wood-harvest activity, and
resulting secondary lands. Global Change Biology,12, 1208–
1229.
Jones, H. & Schmitz, O. (2009) Rapid recovery of damaged
ecosystems. PLoS ONE,4, e5653.
Kareiva, P., Watts, S., McDonald, R. & Boucher, T. (2007)
Domesticated nature: shaping landscapes and ecosystems for
human welfare. Science,316, 1866–1869.
Kirch, P.V. (2005) Archaeology and global change: the Holocene
record. Annual Review of Environment and Resources,30, 409–
440.
Klein Goldewijk, K. (2001) Estimating global land use change
over the past 300 years: the HYDE database. Global Bio-
geochemical Cycles,15, 417–433.
Klein Goldewijk, K. & van Drecht, G. (2006) HYDE 3: current
and historical population and land cover. Integrated modelling
of global environmental change. An overview of IMAGE 2.4 (ed.
by A.F. Bouwman, T. Kram and K. Klein Goldewijk), pp.
93–111. Netherlands Environmental Assessment Agency
(MNP), Bilthoven, The Netherlands.
Küchler, A.W. (1949) A physiognomic classification of vegeta-
tion. Annals of the Association of American Geographers,39,
201–210.
Lindenmayer, D., Hobbs, R.J., Montague-Drake, R. et al. (2008)
A checklist for ecological management of landscapes for con-
servation. Ecology Letters,11, 78–91.
Monfreda, C., Ramankutty, N. & Foley, J.A. (2008) Farming the
planet: 2. Geographic distribution of crop areas, yields, physi-
ological types, and net primary production in the year 2000.
Global Biogeochemical Cycles,22, GB1022.
Oak Ridge National Laboratory (2006) LandScan global popula-
tion database. Oak Ridge National Laboratory, Oak Ridge, TN.
Available at: http://www.ornl.gov/landscan/ (accessed 28 July
2008).
Olofsson, J. & Hickler, T. (2008) Effects of human land-use on
the global carbon cycle during the last 6,000 years. Vegetation
History and Archaeobotany,17, 605–615.
Olson, D.M., Dinerstein, E., Wikramanayake, E.D., Burgess,
N.D., Powell, G.V.N., Underwood, E.C., D’Amico, J.A., Itoua,
I., Strand, H.E., Morrison, J.C., Loucks, C.J., Allnutt, T.F.,
Ricketts, T.H., Kura, Y., Lamoreux, J.F., Wettengel, W.W.,
Hedao, P. & Kassem, K.R. (2001) Terrestrial ecoregions of the
world: a new map of life on Earth. BioScience,51, 933–938.
Prentice, I.C., Cramer, W., Harrison, S.P., Leemans, R., Monse-
rud, R.A. & Solomon, A.M. (1992) A global biome model
based on plant physiology and dominance, soil properties and
climate. Journal of Biogeography,19, 117–134.
Ramankutty, N. & Foley, J.A. (1999) Estimating historical
changes in global land cover: croplands from 1700 to 1992.
Global Biogeochemical Cycles,13, 997–1027.
Ranganathan, J., Daniels, R.J.R., Chandran, M.D.S., Ehrlich, P.R.
& Daily, G.C. (2008) Sustaining biodiversity in ancient tropi-
cal countryside. Proceedings of the National Academy of Sci-
ences USA,105, 17852–17854.
Rea, L.M. & Parker, R.A. (1997) Designing and conducting survey
research: a comprehensive guide. Jossey-Bass, San Francisco.
Redman, C.L. (1999) Human impact on ancient environments.
University of Arizona Press, Tucson.
Rees, W.G. (2008) Comparing the spatial content of thematic
maps. International Journal of Remote Sensing,29, 3833–
3844.
Ricketts, T.H. (2001) The matrix matters: effective isolation in
fragmented landscapes. The American Naturalist,158, 87–99.
Rindfuss, R.R., Walsh, S.J., Turner, B.L., II, Fox, J. & Mishra, V.
(2004) Developing a science of land change: challenges and
methodological issues. Proceedings of the National Academy of
Sciences USA,101, 13976–13981.
Ruddiman, W.F. & Ellis,E.C. (2009) Effect of per-capita land use
changes on Holocene forest clearance and CO2emissions.
Quaternary Science Reviews,28, 3011–3015.
Rudel, T.K., Coomes, O.T., Moran, E., Achard, F., Angelsen, A.,
Xu, J.C. & Lambin, E. (2005) Forest transitions: towards a
global understanding of land use change. Global Environmen-
tal Change – Human and Policy Dimensions,15, 23–31.
Sanderson, E.W., Jaiteh, M., Levy, M.A., Redford, K.H.,
Wannebo, A.V. & Woolmer, G. (2002) The human footprint
and the last of the wild. BioScience,52, 891–904.
Shelford, V.E. (1932) Basic principles of the classification of
communities and habitats and the use of terms. Ecology,13,
105–120.
Siebert, S., Döll, P., Feick, S., Hoogeveen, J. & Frenken, K. (2007)
Global map of irrigation areas version 4.0.1. Johann Wolfgang
Goethe University, Frankfurt am Main, Germany/Food and
Agriculture Organization of the United Nations, Rome, Italy.
Available at: http://www.fao.org/nr/water/aquastat/irriga
tionmap/index.stm (accessed 22 September 2008).
Steinfeld, H., Gerber, P., Wassenaar, T., Castel, V., Rosales, M. &
De Haan, C. (2006) Livestock’s long shadow: environmental
issues and options. Food and Agriculture Organization of the
United Nations, Rome.
Turner, B.L., II, Clark, W.C., Kates, R.W., Richards, J.F.,Mathews,
J.T. & Meyer, W.B. (1990) The earth as transformed by human
action: global and regional changes in the biosphere over the past
300 years. Cambridge University Press with Clark University,
Cambridge and New York.
Turner, B.L., II, Lambin, E.F. & Reenberg, A. (2007) The emer-
gence of land change science for global environmental change
Anthropogenic transformation of the biomes
Global Ecology and Biogeography,19, 589–606, © 2010 Blackwell Publishing Ltd 605
and sustainability. Proceedings of the National Academy of Sci-
ences USA,104, 20666–20671.
UNFPA (2007) State of world population 2007: unleashing the
potential of urban growth. United Nations Population Fund,
New York.
Vitousek, P.M., Mooney, H.A., Lubchenco, J. & Melillo, J.M.
(1997) Human domination of earth’s ecosystems. Science,
277, 494–499.
Whittaker, R.H. (1970) Communities and ecosystems.Mac-
millan, New York.
SUPPORTING INFORMATION
Additional Supporting Information may be found in the online
version of this article:
Appendix S1. Analyses based on Olson biomes (Olson et al.,
2001; Acrobat file).
Appendix S2. Irrigated areas 1700–2000: methodology and data
sources (Acrobat file).
Appendix S3. Anthrome classification algorithm (Acrobat file).
Appendix S4. Anthrome classification sensitivity to data inputs,
model thresholds and spatial resolution (Acrobat file).
Appendix S5. Statistical data for anthromes and biomes and
their transformations (Excel file).
As a service to our authors and readers, this journal provides
supporting information supplied by the authors. Such materials
are peer-reviewed and may be reorganized for online delivery,
but are not copy-edited or typeset. Technical support issues
arising from supporting information (other than missing files)
should be addressed to the authors.
BIOSKETCH
This research was conducted by members of the
anthromes working group, an informal international
collaboration of researchers dedicated to investigating,
understanding and modelling human transformation
and management of the terrestrial biosphere past, present
and future.
Author contributions: E.C.E conceived the initial idea,
conducted the anthrome analysis and led the writing;
K.K.G. prepared the land-use and population datasets;
S.S. produced the irrigation datasets; D.L. and N.R. devel-
oped early versions of the anthrome classification system;
N.R. developed an initial version of Appendix S3. All
authors contributed to the analysis and to revising the
manuscript for publication.
Web link: http://ecotope.org/anthromes/group/
Editor: Erica Fleishman
E. C. Ellis et al.
Global Ecology and Biogeography,19, 589–606, © 2010 Blackwell Publishing Ltd606
... In North America, grassland birds have experienced steeper long-term declines than any other avian guild during the past 50 years (Rosenberg et al., 2019;Sauer, 2020) and more than 79% of grasslands have been lost since the early 1800s (Samson and Knopf, 1994;White, 2000). While factors such as fire suppression, overgrazing, desertification, and the introduction of non-native plant species have contributed to degradation of native prairies, conversion to large-scale, row-crop agriculture remains one of the greatest threats to grassland ecosystems (Knapp et al., 1999;Blann, 2006;Ellis et al., 2010;Wright and Wimberly, 2013). Indeed, intensification of agricultural practices is considered a leading driver of grassland bird population declines worldwide (Wilson et al., 2005;Quinn et al., 2017;Davis et al., 2020). ...
... The loss of native grasslands through conversion to cropland is a primary driver of grassland bird population declines, including thick-billed longspurs (Samson et al., 2004;Blann, 2006;Ellis et al., 2010;Wright and Wimberly, 2013). Nevertheless, we show that longspurs breeding in a region of relatively intact grasslands in northeastern Montana can successfully nest in nearby crop fields. ...
Preprint
Conversion of the North American prairies to cropland remains a prominent threat to grassland bird populations. Yet, a few species nest in these vastly modified systems. The thick-billed longspur (Rhynchophanes mccownii) is an obligate grassland bird whose populations have declined 4% annually during the past 50 years. Thick-billed longspurs historically nested in recently disturbed or sparsely vegetated patches within native mixed-grass prairie, but observations of longspurs in crop fields during the breeding season suggest such fields also provide cues for habitat selection. Maladaptive selection for poor quality habitat may contribute to ongoing declines in longspur populations, but information on thick-billed longspur breeding ecology in crop fields is lacking. We hypothesized that crop fields may function as ecological traps; specifically, we expected that crop fields may provide cues for territory selection but frequent human disturbance and increased exposure to weather and predators would have negative consequences for reproduction. To address this hypothesis, we compared measures of habitat selection (settlement patterns and trends in abundance) and productivity (nest density, nest survival, and number of young fledged) between crop fields and native sites in northeastern Montana, USA. Settlement patterns were similar across site types and occupancy ranged from 0.52 ± 0.17SE to 0.99 ± 0.01 on April 7 and 30, respectively. Early season abundance differed by year and changes in abundance during the breeding season were associated with precipitation-driven vegetation conditions, rather than habitat type. Standardized nest density (0.19 ± 0.27SD nests/plot/hour), the number of young fledged per successful nest (2.9 ± 0.18SE), and nest survival (0.24 ± 0.03 SE; n=222 nests) were similar for crop and native sites. Collectively, the data did not support our hypothesis that crop fields are ecological traps: longspurs did not exhibit a clear preference for cropland and reproductive output was not significantly reduced. Our results indicate that crop fields provide alternative breeding habitat within a human-dominated landscape.
...  The safeguarding the integrity of this interaction is vital to protecting and sustaining the area and its associated nature conservation and other values. However, the continent is not only a mosaic of landscapes shaped by natural conditions but also the result of intense human interventions over many centuries (Ellis et al., 2010 ...
... As early as 12,000 years ago, large parts of Earth's land surface were colonized by human societies that used transformative land-use practices (Ellis et al., 2021). The period of industrialization and globalization in the last two centuries has accelerated land use towards more intensive cultivated and homogeneous cultural landscapes, resulting in ecologically simplified habitats, biotic homogenization and species translocations around the world (Ellis et al., 2010(Ellis et al., , 2021. Beyond historical landscape transformations, today's land cover, which provides habitat for living organisms, can be geomorphometrically characterized as a continuum from natural to semi-natural to highly human-influenced (Cao et al., 2020). ...
Thesis
Protected areas are among the key elements of global biodiversity conservation strategies and aim to conserve native species, habitats and ecosystems. Globalisation has led to increased introductions of species outside their natural range. In their new environment, some of these non-native species have the potential to affect ecosystems and compete with or threaten native species. The environment in close proximity to protected areas is likely to be the stepping stone for non-native species to become established in protected areas. However, little is known about the role that protected area surroundings play in the permeability of protected areas to non-native species. In this thesis, I focused on protected areas and their surrounding belts to address the issue of permeability to non-native species. Examining protected areas in Norway, I showed that non-native species surrounding protected areas have a qualitative impact on the community of non-native species in protected areas. Moreover, the proportion of invasive species was higher in protected areas (40 %) compared to their belts (12 %). The number of non-native species in the surrounding areas also significantly determined the number of non-native species in protected areas. I have also highlighted the dynamics of colonization from the belts to the protected areas by showing that non-native species were detected in the protected areas on average several years after they were recorded in the belts. In addition, I showed in four European countries that the type of land use and land cover in the proximity and within protected areas plays a central role in the establishment of non-native species in protected areas. Anthropogenic land use and land cover around protected areas promoted the establishment of non-native species inside protected areas, regardless of the land use and land cover present in them. Finally, I investigated the colonization dynamics of Acacia dealbata, an invasive t ree species, in and around protected areas in central Portugal over the last twenty years. I showed that disturbances by fires and the loss of tree cover had a significant positive effect on the presence of the species. This thesis highlights the importance of the protected area surroundings for the colonization of non-native species. This is particularly relevant for future management strategies for non-native species in protected areas.
... This conversion has turned the peatland carbon sink into a huge carbon source (Leifeld and Menichetti, 2018). The transformation of landscapes, from being shaped predominantly by natural processes to anthropogenic drivers, is one of the key indicators used to define the existence and onset of the "Anthropocene" (Corlett, 2015;Ellis et al., 2010;Lewis and Maslin, 2015;Smith and Zeder, 2013), with associated debates on how to manage the emerging novel ecosystems (Clement and Standish, 2018;Folke et al., 2021;Hobbs et al., 2014). To contextualise contemporary changes, understanding how human interactions with these ecosystems have unravelled in the past is important. ...
Article
Tropical peatlands are a globally important carbon store. They host significant biodiversity and provide a range of other important ecosystem services, including food and medicines for local communities. Tropical peatlands are increasingly modified by humans in the rapid and transformative way typical of the “Anthropocene,” with the most significant human—driven changes to date occurring in Southeast Asia. This review synthesizes the dominant changes observed in human interactions with tropical peatlands in the last 200 years, focusing on the tropical lowland peatlands of Southeast Asia. We identify the beginning of transformative anthropogenic processes in these carbon-rich ecosystems, chart the intensification of these processes in the 20th and early 21st centuries, and assess their impacts on key ecosystem services in the present. Where data exist, we compare the tropical peatlands of Central Africa and Amazonia, which have experienced very different scales of disturbance in the recent past. We explore their global importance and how environmental pressures may affect them in the future. Finally, looking to the future, we identify ongoing efforts in peatland conservation, management, restoration, and socio-economic development, as well as areas of fruitful research toward sustainability of tropical peatlands.
... plantations, and settlements (urban and rural). Agriculture and urban settlements cover almost 40% of Earth's ice-free terrestrial land, with an additional 37% being rangelands and semi-natural habitats that are embedded within agricultural or settled landscapes (Ellis et al. 2010). How much habitat is needed and how it should be distributed within agricultural landscapes is not yet known Brosi et al. 2008). ...
Article
Insect pollination improves the yield of most crop species and contributes to one-third of global crop production. The importance of this ecosystem service in improving agricultural production has largely been overlooked, however, in favour of practices that improve soil conditions such as fertiliser use and supplementary irrigation. Using economic modelling, this study estimates the value of insect pollination under different land-use types in Kenya. Our model assumes that a combination of land-use type and the foraging distance of insect pollinators influences the intensity of pollination and the value of agricultural output. To demonstrate the hypothesised relationships, areas under different land-use types, e.g. forest, grassland and cropland, and their distances from households were used as proxies for insect pollination. Concentric buffer zones representing foraging distances of pollinators from the land-use types were drawn at 250 m, 500 m, 1 000 m, 2 000 m and 3 000 m from the farms, and areas under each land use in the buffer zones were estimated for the years 2004, 2007 and 2010. Using the random-effects model and an output distance-function stochastic frontier model, the land-use areas, other factors of production and climate variables were regressed on the value of agricultural output in each buffer zone to determine their contribution to agricultural output resulting from insect pollination. The results indicate higher crop productivity on farms bordering forests and grasslands. This implies that insect pollinators are important for crop production, and increasing the number of pollinator habitats closer to the farms will increase food production in the tropics.
Article
Tropical forest tree communities are structured by a range of large‐scale drivers including elevation, certain high‐impact anthropogenic activities (e.g., deforestation), and fires. However, low‐impact human activities such as foraging may also be subtly but notably altering the composition of tropical forest tree communities. The study assessed the (i) differences in species diversity, patterns of relative abundance, and pairwise beta diversity between trees with edible and inedible fruits and seeds along an elevation gradient, and (ii) impact of human foraging on the forest tree communities in Oban Division of Cross River National Park, Nigeria. Fifteen permanent 40 by 40 m plots were established along an elevational gradient (120–460 m above mean sea level). All trees of 0.1 m diameter at breast height (dbh) and above were measured, identified, and, with the aid of structured questionnaires, classified into those with edible and inedible fruits/seeds. A total of 35 edible species with density of 128 stems/hectare and basal area of 11.99 m2/hectare, and 109 inedible species with density of 364 stems/hectare and basal area of 22.42 m2/hectare were sampled. However, the evenness of edible and inedible species was similar at pooled and plot levels. For inedible species, there was a positive relationship between pairwise beta diversity and elevation, and this was driven mainly by turnover. In contrast, edible species exhibited a non‐significant trend between elevation and beta diversity. Thus, the study showed that human foraging of edible fruits may have subtly influenced patterns of species diversity and community structure in this tropical forest. Tropical forest tree communities are structured by a range of large‐scale drivers including elevation, certain high‐impact anthropogenic activities (e.g., deforestation), and fires. However, low‐impact human activities such as foraging may also be altering the composition and distribution of tropical forest trees species. The study assessed the (i) differences in species diversity, patterns of relative abundance, and pairwise beta diversity between trees with edible and inedible fruits and seeds along an elevation gradient, and (ii) impact of human foraging on the forest tree communities in Oban Division of Cross River National Park, Nigeria. The study showed that human foraging of edible fruits may have subtly influenced patterns of species diversity and community structure in this tropical forest.
Article
Shifts in protein production methods are an emerging challenge toward realizing a sustainable society. This paper aims to examine preferences among Japanese consumers regarding attributes of beef mince and its substitutes, to develop consumer segments based on these preferences, and to explore the segment with higher acceptance of replacement from conventional products. This paper also aims to explain intersegment differences from consumer heterogeneity in human values, scientific literacy, and sociodemographic viewpoints for a deeper understanding of consumer behavior in each segment. The results of an online choice experiment involving 4421 consumers in Japan, using food labels on mince showed that Japanese-origin organic beef was associated with the highest utility among the five production methods mentioned. Five consumer segments were identified with latent class analysis: novelty accepters, generous customers, attribute-economy balancers, price–conscious, and conservatives, which vary in preference in choice behavior, sociodemographic, human values, and scientific literacy.
Article
Full-text available
Humans are promoting drastic changes in biological communities that result in ‘winner-loser’ species replacements across multiple spatial scales. In tropical regions, such replacements can be particularly driven by deforestation, especially in landscapes devoted to free-ranging livestock production in which mixing native and exotic species can create species-poor and homogenized communities. We tested this hypothesis assessing medium- and large-bodied non-volant mammals in four 16-km2 landscapes with varying deforestation levels (5%, 30%, 70%, and 95% forest cover), where exotic mammals (e.g., cows and goats) have free access to the remaining Caatinga dry forest. Using camera traps, we obtained 2808 independent records of 17 species, most of them (2054 records, 73%) corresponding to seven exotic species. Native Cerdocyon thous (crab-eating fox) and exotic Capra hircus (goat) accounted for almost half of the records and 60-80% of the records in the two most deforested landscapes. Alpha diversity did not differ significantly among landscapes, but the two more forested landscapes tended to have more native species than exotic ones. Beta diversity patterns among and within landscapes were relatively low regardless of species abundance, indicating biotic homogenization at multiple spatial scales. We conclude that a novel mammal community full of exotic mammals and a few generalist natives has been established in the study region. To promote environmental-friendly livestock farming in the largest tropical dry forest of South America, we should avoid deforestation, especially in sites that concentrate native species, segregate lands for livestock and conservation, and boost inspection against illegal hunting.
Article
Pathways to eradicate global hunger while bending the curve of biodiversity loss unanimously suggest changing to less energy-rich diets, closing yield gaps through agroecological principles, adopting modern breeding technologies to foster stress resilience and yields, as well as minimizing harvest losses and food waste. Against the background of a brief history of global agriculture, we review the available evidence on how the global food system might look given a global temperature increase by 3°. We show that a moderate gain in the area suitable for agriculture is confronted with substantial yield losses through strains on crop physiology, multitrophic interactions, and more frequent extreme events. Self-amplifying feedbacks are unresolved and might lead to further losses. In light of these uncertainties, we see that complexity is underestimated and more systemic research is needed. Efficiency gains in agriculture, albeit indispensable, will not be enough to achieve food security under severe climate change.
Article
Agricultural policies without explicit environmental goals can indirectly affect the natural environment through its effect on farmer input use behavior. For example, the highly-subsidized crop insurance program in the United States (US), while developed to protect farmers against yield and revenue risks, also has the potential to influence fertilizer and land use decisions, which can then impact the extent of excess nitrogen and phosphorus that can run-off and pollute nearby water bodies. This study utilizes county-level panel data from 1989-2015 to directly evaluate the impact of crop insurance participation on nitrogen and phosphorus concentration in waterways. Results from linear panel fixed effects (FE) models suggest that counties with higher crop insurance participation tend to have lower nitrogen concentrations in its water bodies, but the effects are small. In contrast, we do not find a consistent statistically significant crop insurance effect on phosphorus concentrations. Findings based on alternative estimation techniques and other empirical specifications generally support our baseline FE model results. We posit that the modest crop insurance effects may be due to two competing mechanisms — the moral hazard effect of crop insurance (i.e., reducing fertilizer use), being counteracted by the incentive to bring in riskier crops or marginal land to production (i.e., increasing fertilizer use).
Article
Full-text available
Humans have fundamentally altered global patterns of biodiversity and ecosystem processes. Surprisingly, existing systems for representing these global patterns, including biome classifications, either ignore humans altogether or simplify human influence into, at most, four categories. Here, we present the first characterization of terrestrial biomes based on global patterns of sustained, direct human interaction with ecosystems. Eighteen "anthropogenic biomes" were identified through empirical analysis of global population, land use, and land cover. More than 75% of Earth's ice-free land showed evidence of alteration as a result of human residence and land use, with less than a quarter remaining as wildlands, supporting just 11% of terrestrial net primary production. Anthropogenic biomes offer a new way forward by acknowledging human influence on global ecosystems and moving us toward models and investigations of the terrestrial biosphere that integrate human and ecological systems.
Article
Full-text available
Human alteration of Earth is substantial and growing. Between one-third and one-half of the land surface has been transformed by human action; the carbon dioxide concentration in the atmosphere has increased by nearly 30 percent since the beginning of the Industrial Revolution; more atmospheric nitrogen is fixed by humanity than by all natural terrestrial sources combined; more than half of all accessible surface fresh water is put to use by humanity; and about one-quarter of the bird species on Earth have been driven to extinction. By these and other standards, it is clear that we live on a human-dominated planet.
Article
Contrasts the precolonial ecosystems with those that existed in the early 19th century, then compares the ecological relationships of precolonial Indian communities with those of the arriving Europeans. Against this background, the 2nd half of the book describes the processes by which ecological changes were instigated by the European settlers. -J.Sheail
Article
Attempts to document recent changes in the biosphere, contrast global and regional scale patterns of change, and explain the main human forces which have driven such changes. The book presents studies from around the world of societal and environmental alterations brought about by the development and civilization of mankind. The first section deals with the major human forces of the last 300 yr, whilst the second presents detailed accounts of transformations to the global environment brought about by human action. The third section uses regional case studies of such transformations to describe the multivariable interactions of environmental change, and a brief final section examines a range of perspectives and theories purporting to explain human actions in relation to the biosphere. The great majority of papers from the first three sections are abstracted separately. -after Editors
Article
Land use is at the center of one of the most vexing challenges for the coming decades: to provide enough food, fiber and shelter for the world's population; raise the standard of living for the billion people currently below the poverty line; and simultaneously sustain the world's ecosystems for use by humans and other species. The intended consequence of cropland expansion, urban growth, and other land use changes is to satisfy demands from the increasing appetite of the world's population. Unintended consequences, however, can alter ecological processes and have far-reaching and long-term effects that potentially compromise the basic functioning of ecosystems. Recently, the scientific community has begun to confront such issues. Several national and international programs have been at the forefront of scientific enquiry on the causes and consequences of land use change, including: the Land Use and Land Cover Change Program of the National Aeronautics and Space Administration, the Land Use program element in the interagency U.S. Climate Change Science Program, and the International Geosphere-Biosphere's Land Use and Cover Change (LUCC) core project. The result has been significant advances in understanding the complex socioeconomic, technological, and biophysical factors that drive land use change worldwide.
Article
Measures of categorical association can be used to assess the degree of similarity between two thematic maps, having regard to the spatial distribution of map classes but not to the identity of the classes. Such measures can be used, for example, to compare maps that use different classification schemes, or to identify misregistration between maps. The theory of these measures and their relationship to measures of categorical agreement are discussed. Three representative measures of categorical association are investigated quantitatively using various categorical maps, including maps generated by classification of satellite images, of a study area in the Russian north. On the basis of this investigation, it is suggested that Cramer's V statistic provides a simple but effective measure of the similarity in the spatial content of maps.