1 CONCEPTS AND QUESTIONS Putting people in the map: anthropogenic biomes of the worl 439 Erle C Ellis 1* an Navin Ramankutty 2 Humans have funamentally altere global patterns of bioiversity an ecosystem processes. Surprisingly, existing systems for representing these global patterns, incluing biome classifications, either ignore humans altogether or simplify human influence into, at most, four categories. Here, we present the first characterization of terrestrial biomes base on global patterns of sustaine, irect human interaction with ecosystems. Eighteen anthropogenic biomes were ientifie through empirical analysis of global population, lan use, an lan cover. More than 75% of Earth s ice-free lan showe evience of alteration as a result of human resience an lan use, with less than a quarter remaining as willans, supporting just 11% of terrestrial net primary prouction. Anthropogenic biomes offer a new way forwar by acknowleging human influence on global ecosystems an moving us towar moels an investigations of the terrestrial biosphere that integrate human an ecological systems. Front Ecol Environ 2008; 6(8): , oi: / Humans have long istinguishe themselves from other species by shaping ecosystem form an process using tools an technologies, such as fire, that are beyon the capacity of other organisms (Smith 2007). This exceptional ability for ecosystem engineering has helpe to sustain unpreceente human population growth over the past half century, to such an extent that humans now consume about one-thir of all terrestrial net primary prouction (NPP; Vitousek et al. 1986; Imhoff et al. 2004) an move more earth an prouce more reactive nitrogen than all other terrestrial processes combine (Galloway 2005; Wilkinson an McElroy 2007). Humans are also causing global extinctions (Novacek an Clelan 2001) an changes in climate that are comparable to any observe in the natural recor (Ruiman 2003; IPCC 2007). Clearly, Homo sapiens has emerge as a force of nature rivaling climatic In a nutshell: Anthropogenic biomes offer a new view of the terrestrial biosphere in its contemporary, human-altere form Most of the terrestrial biosphere has been altere by human resience an agriculture Less than a quarter of Earth s ice-free lan is wil; only 20% of this is forests an > 36% is barren More than 80% of all people live in ensely populate urban an village biomes Agricultural villages are the most extensive of all ensely populate biomes an one in four people lives in them 1 Department of Geography an Environmental Systems, University of Marylan, Baltimore, MD * 2 Department of Geography an Earth System Science Program, McGill University, Montreal, QC, Canaa an geologic forces in shaping the terrestrial biosphere an its processes. Biomes are the most basic units that ecologists use to escribe global patterns of ecosystem form, process, an bioiversity. Historically, biomes have been ientifie an mappe base on general ifferences in vegetation type associate with regional variations in climate (Uvary 1975; Matthews 1983; Prentice et al. 1992; Olson et al. 2001; Bailey 2004). Now that humans have restructure the terrestrial biosphere for agriculture, forestry, an other uses, global patterns of species composition an abunance, primary prouctivity, lan-surface hyrology, an the biogeochemical cycles of carbon, nitrogen, an phosphorus, have all been substantially altere (Matson et al. 1997; Vitousek et al. 1997; Foley et al. 2005). Inee, recent stuies inicate that human-ominate ecosystems now cover more of Earth s lan surface than o wil ecosystems (McCloskey an Spaling 1989; Vitousek et al. 1997; Sanerson et al. 2002, Mittermeier et al. 2003; Foley et al. 2005). It is therefore surprising that existing escriptions of biome systems either ignore human influence altogether or escribe it using at most four anthropogenic ecosystem classes (urban/built-up, croplan, an one or two croplan/natural vegetation mosaic(s); classification systems inclue IGBP, Lovelan et al. 2000; Olson Biomes, Olson et al. 2001; GLC 2000, Bartholome an Belwar 2005; an GLOBCOVER, Defourny et al. 2006). Here, we present an alternate view of the terrestrial biosphere, base on an empirical analysis of global patterns of sustaine irect human interaction with ecosystems, yieling a global map of anthropogenic biomes. We then examine the potential of anthropogenic biomes to serve as a new global framework for ecology, complete with
2 Anthropogenic biomes of the worl EC Ellis an N Ramankutty 440 testable hypotheses, that can avance research, eucation, an conservation of the terrestrial biosphere as it exists toay the prouct of intensive reshaping by irect interactions with humans. Human interactions with ecosystems Human interactions with ecosystems are inherently ynamic an complex (Folke et al. 1996; DeFries et al. 2004; Rinfuss et al. 2004); any categorization of these is a gross oversimplification. Yet there is little hope of unerstaning an moeling these interactions at a global scale without such simplification. Most global moels of primary prouctivity, species iversity, an even climate epen on stratifying the terrestrial surface into a limite number of functional types, lan-cover types, biomes, or vegetation classes (Haxeltine an Prentice 1996; Thomas et al. 2004; Feema et al. 2005). Human interactions with ecosystems range from the relatively light impacts of mobile bans of hunter-gatherers to the complete replacement of pre-existing ecosystems with built structures (Smil 1991). Population ensity is a useful inicator of the form an intensity of these interactions, as increasing populations have long been consiere both a cause an a consequence of ecosystem moification to prouce foo an other necessities (Boserup 1965, 1981; Smil 1991; Netting 1993). Inee, most basic historical forms of human ecosystem interaction are associate with major ifferences in population ensity, incluing foraging (< 1 person km 2 ), shifting (> 10 persons km 2 ), an continuous cultivation (> 100 persons km 2 ); populations enser than 2500 persons km 2 are believe to be unsupportable by traitional subsistence agriculture (Smil 1991; Netting 1993). In recent ecaes, inustrial agriculture an moern transportation have create new forms of human ecosystem interaction across the full range of population ensities, from low-ensity exurban evelopments to vast conurbations that combine high-ensity cities, low-ensity suburbs, agriculture, an even foreste areas (Smil 1991; Qaeer 2000; Theobal 2004). Nevertheless, population ensity can still serve as a useful inicator of the form an intensity of human ecosystem interactions within a specific locale, especially when populations iffer by an orer of magnitue or more. Such major ifferences in population ensity help to istinguish situations in which humans may be consiere merely agents of ecosystem transformation (ecosystem engineers), from situations in which human populations have grown ense enough that their local resource consumption an waste prouction form a substantial component of local biogeochemical cycles an other ecosystem processes. To begin our analysis, we therefore categorize human ecosystem interactions into four classes, base on major ifferences in population ensity: high population intensity ( ense, >100 persons km 2 ), substantial population intensity ( resiential, 10 to 100 persons km 2 ), minor population ( populate, 1 to 10 persons km 2 ), an inconsequential population ( remote, < 1 person km 2 ). Population class names are efine only in the context of this stuy. Ientifying anthropogenic biomes: an empirical approach We ientifie an mappe anthropogenic biomes using the multi-stage empirical proceure etaile in WebPanel 1 an outline below, base on global ata for population (urban, non-urban), lan use (percent area of pasture, crops, irrigation, rice, urban lan), an lan cover (percent area of trees an bare earth); ata for NPP, IGBP lan cover, an Olson biomes were obtaine for later analysis (WebPanel 1 inclues references for all ata sources). Biome analysis was conucte at 5 arc minute resolution (5 gri cells cover ~ 86 km 2 at the equator), a spatial resolution selecte as the finest allowing irect use of high-quality lan-use area estimates. First, anthropogenic 5 cells were separate from wil cells, base on the presence of human populations, crops, or pastures. Anthropogenic cells were then stratifie into the population ensity classes escribe above ( ense, resiential, populate, an remote ), base on the ensity of their non-urban population. We then use cluster analysis, a statistical proceure esigne to ientify an optimal number of istinct natural groupings (clusters) within a ataset (using SPSS 15.01), to ientify natural groupings within the cells of each population ensity class an within the wil class, base on non-urban population ensity an percent urban area, pasture, crops, irrigate, rice, trees, an bare earth. Finally, the strata erive above were escribe, labele, an organize into broa logical groupings, base on their populations, lan-use an lan-cover characteristics, an their regional istribution, yieling the 18 anthropogenic biome classes an three wil biome classes illustrate in Figure 1 an escribe in Table 1. (WebTables 1 an 2 provie more etaile statistics; WebPanel 2 provies maps viewable in Google Earth, Google Maps, an Microsoft Virtual Earth, a printable wall map, an map ata in GIS format.) A tour of the anthropogenic biomes When viewe globally, anthropogenic biomes clearly ominate the terrestrial biosphere, covering more than threequarters of Earth s ice-free lan an incorporating nearly 90% of terrestrial NPP an 80% of global tree cover (Figures 1 an 2a; WebTable 2). About half of terrestrial NPP an lan were present in the foreste an rangelan biomes, which have relatively low population ensities an potentially low impacts from lan use (excluing resiential rangelans; Figures 1 an 2a). However, one-thir of Earth s ice-free lan an about 45% of terrestrial NPP occurre within cultivate an substantially populate biomes (ense settlements, villages, croplans, an resiential rangelans; Figures 1 an 2a).
3 EC Ellis an N Ramankutty Anthropogenic biomes of the worl 441 Anthropogenic biomes: % worl regions Anthropogenic biomes: legen Dense settlements 11 Urban 12 Dense settlements Villages 21 Rice villages 22 Irrigate villages 23 Croppe an pastoral villages 24 Pastoral villages 25 Rainfe villages 26 Rainfe mosaic villages Croplans 31 Resiential irrigate croplan 32 Resiential rainfe mosaic 33 Populate irrigate croplan 34 Populate rainfe croplan 35 Remote croplans 100% Rangelans 41 Resiential rangelans 42 Populate rangelans 43 Remote rangelans Foreste 51 Populate forests 52 Remote forests 50% Willans 61 Wil forests 62 Sparse trees 63 Barren Region bounary 0% Worl N. America, Europe, Austr., NZ evelope Asia, Oceania Eurasia eveloping Near East Latin America, Caribbean Africa F i g ur e 1. Anthropogenic biomes: worl map an regional areas. Biomes are organize into groups (Table 1), an sorte in orer of population ensity. Map scale = 1: , Plate Carrée projection (geographic), 5 arc minute resolution (5 = ). Regional biome areas are etaile in WebTable 3; WebPanel 2 provies interactive versions of this map. Of Earth s 6.4 billion human inhabitants, 40% live in ense settlements biomes (82% urban population), 40% live in village biomes (38% urban), 15% live in croplan biomes (7% urban), an 5% live in rangelan biomes (5% urban; foreste biomes ha 0.6% of global population; Figure 2a). Though most people live in ense settlements an villages, these cover just 7% of Earth s ice-free lan, an about 60% of this population is urban, living in the cities an towns embee within these biomes, which also inclue almost all of the lan we have classifie as urban (94% of 0.5 million km2, although this is probably a substantial unerestimate; Salvatore et al. 2005; Figure 2a). Village biomes, representing ense agricultural populations, were by far the most extensive of the ensely populate biomes, covering 7.7 million km2, compare with 1.5 million km2 for the urban an ense settlements biomes. Moreover, village biomes house about one-half of the worl s non-urban population (1.6 of ~ 3.2 billion persons). Though about one-thir of global urban area is also embee within these biomes, urban areas accounte for just 2% of their total extent, while agricultural lan (crops an pasture) average > 60% of their area. More than 39% of ensely populate biomes were locate in Asia, which also incorporate more than 60% of that continent s total global area, even though this region was the fifth largest of seven regions (Figure 1; WebTable 3). Village biomes were most common in Asia, where they covere more than a quarter of all lan. Africa was secon, with 13% of village biome area, though these covere just 6% of Africa s lan. The most intensive lan-use practices were also isproportionately locate in the village biomes, incluing about half the worl s irrigate lan (1.4 of 2.7 million km2) an two-thirs of global rice lan (1.1 of 1.7 million km2; Figure 2a). After rangelans, croplan biomes were the secon most extensive of the anthropogenic biomes, covering about 20% of Earth s ice-free lan. Far from being simple, crop-covere lanscapes, croplan biomes were mostly mosaics of cultivate lan mixe with trees an pastures (Figure 3c). As a result, croplan biomes constitute only slightly more than half of the worl s total crop-covere g
4 Anthropogenic biomes of the worl EC Ellis an N Ramankutty 442 Table 1. Anthropogenic biome escriptions Group Biome Description Dense settlements Dense settlements with substantial urban area 11 Urban Dense built environments with very high populations 12 Dense settlements Dense mix of rural an urban populations, incluing both suburbs an villages Villages Dense agricultural settlements 21 Rice villages Villages ominate by pay rice 22 Irrigate villages Villages ominate by irrigate crops 23 Croppe an pastoral Villages with a mix of crops an pasture villages 24 Pastoral villages Villages ominate by rangelan 25 Rainfe villages Villages ominate by rainfe agriculture 26 Rainfe mosaic villages Villages with a mix of trees an crops Croplans Annual crops mixe with other lan uses an lan covers 31 Resiential irrigate Irrigate croplan with substantial human populations croplan 32 Resiential rainfe mosaic Mix of trees an rainfe croplan with substantial human populations 33 Populate irrigate croplan Irrigate croplan with minor human populations 34 Populate rainfe croplan Rainfe croplan with minor human populations 35 Remote croplans Croplan with inconsequential human populations Rangelan Livestock grazing; minimal crops an forests 41 Resiential rangelans Rangelans with substantial human populations 42 Populate rangelans Rangelans with minor human populations 43 Remote rangelans Rangelans with inconsequential human populations Foreste Forests with human populations an agriculture 51 Populate forests Forests with minor human populations 52 Remote forests Forests with inconsequential human populations Willans Lan without human populations or agriculture 61 Wil forests High tree cover, mostly boreal an tropical forests 62 Sparse trees Low tree cover, mostly col an ari lans 63 Barren No tree cover, mostly eserts an frozen lan area (8 of 15 million km 2 ), with village biomes hosting nearly a quarter an rangelan biomes about 16%. The croplan biomes also inclue 17% of the worl s pasture lan, along with a quarter of global tree cover an nearly a thir of terrestrial NPP. Most abunant in Africa an Asia, resiential, rainfe mosaic was by far the most extensive croplan biome an the secon most abunant biome overall (16.7 million km 2 ), proviing a home to nearly 600 million people, 4 million km 2 of crops, an about 20% of the worl s tree cover an NPP a greater share than the entire wil forests biome. Rangelan biomes were the most extensive, covering nearly a thir of global ice-free lan an incorporating 73% of global pasture (28 million km 2 ), but these were foun primarily in ari an other low prouctivity regions with a high percentage of bare earth cover (aroun 50%; Figure 3c). As a result, rangelans accounte for less than 15% of terrestrial NPP, 6% of global tree cover, an 5% of global population. Foreste biomes covere an area similar to the croplan biomes (25 million km 2 versus 27 million km 2 for croplans), but incorporate a much greater tree-covere area (45% versus 25% of their global area). It is therefore surprising that the total NPP of the foreste biomes was nearly the same as that of the croplan biomes (16.4 versus 16.0 billion tons per year). This may be explaine by the lower prouctivity of boreal forests, which preominate in the foreste biomes, while croplan biomes were locate in some of the worl s most prouctive climates an soils. Willans without evience of human occupation or lan use occupie just 22% of Earth s icefree lan in this analysis. In general, these were locate in the least prouctive regions of the worl; more than two-thirs of their area occurre in barren an sparsely tree-covere regions. As a result, even though willans containe about 20% cover by wil forests (a mix of boreal an tropical forests; Figure 2c), willans as a whole contribute only about 11% of total terrestrial NPP. Anthropogenic biomes are mosaics It is clear from the biome escriptions above, from the lan-use an lan-cover patterns in Figure 3c, an most of all, by comparing our biome map against high-resolution satellite imagery (WebPanel 2), that anthropogenic biomes are best characterize as heterogeneous lanscape mosaics, combining a variety of ifferent lan uses an lan covers. Urban areas are embee within agricultural areas, trees are intersperse with croplans an housing, an manage vegetation is mixe with semi-natural vegetation (eg croplans are embee within rangelans an forests). Though some of this heterogeneity might be explaine by the relatively coarse resolution of our analysis, we suggest a more basic explanation: that irect interactions between humans an ecosystems generally take place within heterogeneous lanscape mosaics (Pickett an Caenasso 1995; Daily 1999). Further, we propose that this heterogeneity has three causes, two of which are anthropogenic an all of which are fractal in nature (Levin 1992), proucing similar patterns across spatial scales ranging from the lan holings of iniviual househols to the global patterning of the anthropogenic biomes. We hypothesize that even in the most ensely populate biomes, most lanscape heterogeneity is cause by natural variation in terrain, hyrology, soils, isturbance regimes (eg fire), an climate, as escribe by conventional moels of ecosystems an the terrestrial biosphere (eg Levin 1992; Haxeltine an Prentice 1996; Olson et
5 EC Ellis an N Ramankutty Anthropogenic biomes of the worl al. 2001). Anthropogenic enhancement of natural lanscape heterogeneity represents a seconary cause of heterogeneity within anthropogenic biomes, explaine in part by the human tenency to seek out an use the most prouctive lans first an to work an populate these lans most intensively (Huston 1993). At a global scale, this process may explain why willans are most common in those parts of the biosphere with the least potential for agriculture (ie polar regions, mountains, low fertility tropical soils; Figure 1) an why, at a given percentage of tree cover, NPP appears higher in anthropogenic biomes with higher population ensities (compare NPP with tree cover, especially in wil forests versus foreste biomes; Figure 3c). It may also explain why most human populations, both urban an rural, appear to be associate with intensive agriculture (irrigate crops, rice), an not with pasture, forests, or other, less intensive lan uses (Figure 3c). Finally, this hypothesis explains why most fertile valleys an flooplains in favorable climates are alreay in use as croplans, while neighboring hillslopes an mountains are often islans of semi-natural vegetation, left virtually unisturbe by local populations (Huston 1993; Daily 1999). The thir cause of lanscape heterogeneity in anthropogenic biomes is entirely anthropogenic: humans create lanscape heterogeneity irectly, as exemplifie by the construction of settlements an transportation systems in patterns riven as much by cultural as by environmental constraints (Pickett an Caenasso 1995). All three of these rivers of heterogeneity unoubtely interact in patterning the terrestrial biosphere, but their relative roles at global scales have yet to be stuie an surely merit further investigation, consiering the impacts of lanscape fragmentation on bioiversity (Vitousek et al. 1997; Sanerson et al. 2002). A conceptual moel for anthropogenic biomes Given that anthropogenic biomes are mosaics mixtures of settlements, agriculture, forests an other lan uses an lan covers how o we procee to a general ecological unerstaning of human ecosystem interactions within an across anthropogenic biomes? Before eveloping (a) 100 Worl total % (b) Biome % 0 Population Lan NPP Trees Bare Urban Rice Irrigate Crops Pasture All lan IGBP classes Snow an ice Barren or sparsely vegetate Deciuous neeleleaf forest (c) Biome % All lan Olson Biomes Tunra Lan cover Boreal forests Open shrublans Evergreen neeleleaf forest Deserts an xeric shrublans Lan use Temperate coniferous forests Flooe grasslans an savannas Montane grasslans an shrublans Meiterranean forests, woolans, an shrublans Mixe forests Evergreen broaleaf forest Grasslans Savannas Wooy savannas Permanent wetlans Deciuous broaleaf forest Close shrublans Tropical an Tropical an subtropical subtropical moist grasslans, savannas, broaleaf forests an shrublans Temperate broaleaf an mixe forests Tropical an subtropical ry broaleaf forests Temperate grasslans, savannas, an shrublans Urban Dense settlements Rice villages Irrigate villages Croppe an pastoral villages Pastoral villages Rainfe villages Rainfe mosaic villages Resiential rainfe mosaic Populate irrigate croplan Populate rainfe croplan Remote croplans Resiential rangelans Populate rangelans Remote rangelans Populate forests Remote forests Wil forests Sparse trees Barren Croplans Urban an built-up Croplan/natural vegetation mosaic Mangroves Tropical an subtropical coniferous forests Figure 2. Anthropogenic biomes expresse as a percentage of (a) global population, ice-free lan, NPP, lan cover, an lan use (WebTable 3), (b) IGBP lan-cover classes (Friel et al. 2002; WebTable 4), an (c) Olson biomes (Olson et al. 2001; WebTable 5). In (b) an (c), left columns show the anthropogenic biomes as a percentage of global ice-free lan, horizontal bars show (b) IGBP lan cover an (c) Olson biomes as a percentage of ice-free lan, an columns in center illustrate the percent area of each anthropogenic biome within each IGBP an Olson class, sorte in orer of ecreasing total wil biome area, left to right. Color an orer of anthropogenic biome classes are the same as in Figure
6 Anthropogenic biomes of the worl Willans (a) Foreste Rangelans Croplans pasture s crop rainfe Villages Dense settlements Population ensity Lan use forestry builtup irrigate ornamental (b) Lan cover bare ous ace trees herb NPP Carbon emissions + Reactive nitrogen Bioiversity native introuce (c) 104 Population ensity (persons km 2) urban Lan use rainfe crops (% area) pasture irrigate bare Lan cover (% area) 0 herbaceous trees 800 NPP (g m 2 year 1) 0 Sp Ba ar rren se tr W il ee fo s re s Re ts m Po ot pu e la te Re m Po ot pu e Re late si e nt Po Re ial pu m Po lat pu e ote ra la Re te infe si ir Re entia riga l t si r e ainfe e nt m ia o Ra l irr saic ig in at fe e m os a Ra ic Cr in op fe pe Pas an to r pa al st o Irr ral ig D at en e se se Ri c ttl em e en ts U rb an 444 EC Ellis an N Ramankutty F i g u re 3. Conceptual moel of anthropogenic biomes compare with ata. (a) Anthropogenic biomes structure by population ensity (logarithmic scale) an lan use (percent lan area), forming patterns of (b) ecosystem structure (percent lan cover), process (NPP, carbon balance; re = emissions, reactive nitrogen), an bioiversity (native versus non-native + omestic bioiversity; inicate relative to pre-existing bioiversity; white space inicates net reuction of bioiversity) within broa groups of anthropogenic biomes. (c) Mean population ensity, lan use, lan cover, an NPP observe within anthropogenic biomes (Figure 1; WebTable 1). Biome labels at bottom omit names of broa groups, at top. a set of hypotheses an a strategy for testing them, we first summarize our current unerstaning of how these interactions pattern terrestrial ecosystem processes at a global scale using a simple equation: Ecosystem processes = f(population ensity, lan use, biota, climate, terrain, geology) ontiersinecology.or g Those familiar with conventional ecosystem-process moels will recognize that ours is merely an expansion of these, aing human population ensity an lan use as parameters to explain global patterns of ecosystem processes an their changes. With some moification, conventional lan-use an ecosystem-process moels shoul therefore be capable of moeling ecological
7 EC Ellis an N Ramankutty Anthropogenic biomes of the worl changes within an across anthropogenic biomes (Turner et al. 1995; DeFries et al. 2004; Foley et al. 2005). We inclue population ensity as a separate river of ecosystem processes, base on the principle that increasing population ensities can rive greater intensity of lan use (Boserup 1965, 1981) an can also increase the irect contribution of humans to local ecosystem processes (eg resource consumption, combustion, excretion; Imhoff et al. 2004). For example, uner the same environmental conitions, our moel woul preict greater fertilizer an water inputs to agricultural lan in areas with higher population ensities, together with greater emissions from the combustion of biomass an fossil fuel. Some hypotheses an their tests Base on our conceptual moel of anthropogenic biomes, we propose some basic hypotheses concerning their utility as a moel of the terrestrial biosphere. First, we hypothesize that anthropogenic biomes will iffer substantially in terms of basic ecosystem processes (eg NPP, carbon emissions, reactive nitrogen; Figure 3b) an bioiversity (total, native) when measure across each biome in the fiel, an that these ifferences will be at least as great as those between the conventional biomes when observe using equivalent methos at the same spatial scale. Further, we hypothesize that these ifferences will be riven by ifferences in population ensity an lan use between the biomes (Figure 3a), a tren alreay evient in the general tenency towar increasing croppe area, irrigation, an rice prouction with increasing population ensity (Figure 3c). Finally, we hypothesize that the egree to which anthropogenic biomes explain global patterns of ecosystem processes an bioiversity will increase over time, in tanem with anticipate future increases in human influence on ecosystems. The testing of these an other hypotheses awaits improve ata on human ecosystem interactions obtaine by observations mae within an across the full range of anthropogenic lanscapes. Observations within anthropogenic lanscapes capable of resolving iniviually manage lan-use features an built structures are critical, because this is the scale at which humans interact irectly with ecosystems an is also the optimal scale for precise measurements of ecosystem parameters an their controls (Ellis et al. 2006). Given the consierable effort involve in making etaile measurements of ecological an human systems across heterogeneous anthropogenic lanscapes, this will require evelopment of statistically robust stratifiesampling esigns that can support regional an global estimates base on relatively small lanscape samples within an across anthropogenic biomes (eg Ellis 2004). This, in turn, will require improve global ata, especially for human populations an lan-use practices. Fortunately, evelopment of these atasets woul also pave the way towar a system of anthropogenic ecoregions capable of serving the ecological monitoring nees of regional an local stakeholers, a role currently occupie by conventional ecoregion mapping an classification systems (Olson et al. 2001). Are conventional biome systems obsolete? We have portraye the terrestrial biosphere as compose of anthropogenic biomes, which might also be terme anthromes or human biomes to istinguish them from conventional biome systems. This begs the question: are conventional biome systems obsolete? The answer is certainly no. Although we have propose a basic moel of ecological processes within an across anthropogenic biomes, our moel remains conceptual, while existing moels of the terrestrial biomes, base on climate, terrain, an geology, are fully operational an are useful for preicting the future state of the biosphere in response to climate change (Melillo et al. 1993; Cox et al. 2000; Cramer et al. 2001). On the other han, anthropogenic biomes are in many ways a more accurate escription of broa ecological patterns within the current terrestrial biosphere than are conventional biome systems that escribe vegetation patterns base on variations in climate an geology. It is rare to fin extensive areas of any of the basic vegetation forms epicte in conventional biome moels outsie of the areas we have efine as wil biomes. This is because most of the worl s natural ecosystems are embee within lans altere by lan use an human populations, as is apparent when viewing the istribution of IGBP an Olson biomes within the anthropogenic biomes (Figure 2 b,c). Ecologists go home! Anthropogenic biomes point to a necessary turnaroun in ecological science an eucation, especially for North Americans. Beginning with the first mention of ecology in school, the biosphere has long been epicte as being compose of natural biomes, perpetuating an outate view of the worl as natural ecosystems with humans isturbing them. Although this moel has long been challenge by ecologists (Oum 1969), especially in Europe an Asia (Golley 1993), an by those in other isciplines (Cronon 1983), it remains the mainstream view. Anthropogenic biomes tell a completely ifferent story, one of human systems, with natural ecosystems embee within them. This is no minor change in the story we tell our chilren an each other. Yet it is necessary for sustainable management of the biosphere in the 21st century. Anthropogenic biomes clearly show the inextricable intermingling of human an natural systems almost everywhere on Earth s terrestrial surface, emonstrating that interactions between these systems can no longer be avoie in any substantial way. Moreover, human interactions with ecosystems meiate through the atmosphere (eg climate change) are even more pervasive an are is- 445
8 Anthropogenic biomes of the worl EC Ellis an N Ramankutty 446 proportionately altering the areas least impacte by humans irectly (polar an ari lans; IPCC 2007; Figure 1). Sustainable ecosystem management must therefore be irecte towar eveloping an maintaining beneficial interactions between manage an natural systems, because avoiing these interactions is no longer a practical option (DeFries et al. 2004; Foley et al. 2005). Most importantly, though still at an early stage of evelopment, anthropogenic biomes offer a framework for incorporating humans irectly into global ecosystem moels, a capability that is both urgently neee an as yet unavailable (Carpenter et al. 2006). Ecologists have long been known as the scientists who travel to uninhabite lans to o their work. As a result, our unerstaning of anthropogenic ecosystems remains poor when compare with the rich literature on natural ecosystems. Though much recent effort has focuse on integrating humans into ecological research (Pickett et al. 2001; Rinfuss et al. 2004; WebPanel 3 inclues more citations) an support for this is increasingly available from the US National Science Founation (www.nsf.gov; eg HERO, CNH, HSD programs), ecologists can an shoul o more to come home an work where most humans live. Builing ecological science an eucation on a founation of anthropogenic biomes will help scientists an society take ownership of a biosphere that we have alreay altere irreversibly, an moves us towar unerstaning how best to manage the anthropogenic biosphere we live in. Conclusions Human influence on the terrestrial biosphere is now pervasive. While climate an geology have shape ecosystems an evolution in the past, our work contributes to the growing boy of evience emonstrating that human forces may now outweigh these across most of Earth s lan surface toay. Inee, willans now constitute only a small fraction of Earth s lan. For the foreseeable future, the fate of terrestrial ecosystems an the species they support will be intertwine with human systems: most of nature is now embee within anthropogenic mosaics of lan use an lan cover. While not intene to replace existing biome systems base on climate, terrain, an geology, we hope that wie availability of an anthropogenic biome system will encourage a richer view of human ecosystem interactions across the terrestrial biosphere, an that this will, in turn, guie our investigation, unerstaning, an management of ecosystem processes an their changes at global an regional scales. Acknowlegements ECE thanks S Gliessman of the Department of Environmental Stuies at the University of California, Santa Cruz, an C Fiel of the Department of Global Ecology, Carnegie Institute of Washington at Stanfor, for graciously hosting his sabbatical. P Vitousek an his group, G Asner, J Foley, A Wolf, an A e Bremon provie helpful input. T Rabenhorst provie much-neee help with cartography. Many thanks to the Global Lan Cover Facility (www.lancover.org) for proviing global lan-cover ata an to C Monfrea for rice ata. References Bartholome E an Belwar AS GLC2000: a new approach to global lan cover mapping from Earth observation ata. Int J Remote Sens 26: Boserup E The conitions of agricultural growth: the economics of agrarian change uner population pressure. Lonon, UK: Allen an Unwin. Boserup E Population an technological change: a stuy of long term trens. Chicago, IL: University of Chicago Press. Carpenter SR, DeFries R, Dietz T, et al Millennium Ecosystem Assessment: research nees. Science 314: Cox PM, Betts RA, Jones CD, et al Acceleration of global warming ue to carbon-cycle feebacks in a couple climate moel. Nature 408: Cramer W, Boneau A, Woowar FI, et al Global response of terrestrial ecosystem structure an function to CO 2 an climate change: results from six ynamic global vegetation moels. Global Change Biol 7: Cronon W Changes in the lan: Inians, colonists, an the ecology of New Englan. New York, NY: Hill an Wang. Daily GC Developing a scientific basis for managing Earth s life support systems. Conserv Ecol 3: 14. DeFries RS, Foley JA, an Asner GP Lan-use choices: balancing human nees an ecosystem function. Front Ecol Environ 2: Defourny P, Vancutsem C, Bicheron P, et al GLOBCOVER: a 300 m global lan cover prouct for 2005 using Envisat MERIS time series. In: Proceeings of the ISPRS Commission VII mi-term symposium, Remote sensing: from pixels to processes; 2006 May 8 11; Enschee, Netherlans. Ellis EC Long-term ecological changes in the ensely populate rural lanscapes of China. In: DeFries RS, Asner GP, an Houghton RA (Es). Ecosystems an lan-use change. Washington, DC: American Geophysical Union. Ellis EC, Wang H, Xiao HS, et al Measuring long-term ecological changes in ensely populate lanscapes using current an historical high resolution imagery. Remote Sens Environ 100: Feema JJ, Oleson KW, Bonan GB, et al The importance of lan-cover change in simulating future climates. Science 310: Folke C, Holling CS, an Perrings C Biological iversity, ecosystems, an the human scale. Ecol Appl 6: Foley JA, DeFries R, Asner GP, et al Global consequences of lan use. Science 309: Friel MA, McIver DK, Hoges JCF, et al Global lan cover mapping from MODIS: algorithms an early results. Remote Sens Environ 83: Galloway JN The global nitrogen cycle. In: Schlesinger WH (E). Treatise on geochemistry. Oxfor, UK: Pergamon. Golley FB A history of the ecosystem concept in ecology: more than the sum of the parts. New Haven, CT: Yale University Press. Haxeltine A an Prentice IC BIOME3: an equilibrium terrestrial biosphere moel base on ecophysiological constraints, resource availability, an competition among plant functional types. Global Biogeochem Cy 10: Huston M Biological iversity, soils, an economics. Science 262:
9 EC Ellis an N Ramankutty Imhoff ML, Bounoua L, Ricketts T, et al Global patterns in human consumption of net primary prouction. Nature 429: 870. IPCC (Intergovernmental Panel on Climate Change) Climate change 2007: the physical science basis. Summary for policy makers. A report of Working Group I of the Intergovernmental Panel on Climate Change. Geneva, Switzerlan: IPCC. Levin SA The problem of pattern an scale in ecology. Ecology 73: Lovelan TR, Ree BC, Brown JF, et al Development of a global lan-cover characteristics atabase an IGBP DISCover from 1 km AVHRR ata. Int J Remote Sens 21: Matson PA, Parton WJ, Power AG, an Swift MJ Agricultural intensification an ecosystem properties. Science 277: Matthews E Global vegetation an lan use: new high-resolution atabases for climate stuies. J Clim Appl Meteorol 22: McCloskey JM an Spaling H A reconnaissance level inventory of the amount of wilerness remaining in the worl. Ambio 18: Melillo JM, McGuire AD, Kicklighter DW, et al Global climate change an terrestrial net primary prouction. Nature 363: Mittermeier RA, Mittermeier CG, Brooks TM, et al Wilerness an bioiversity conservation. P Natl Aca Sci USA 100: Netting RM Smallholers, householers: farm families an the ecology of intensive sustainable agriculture. Stanfor, CA: Stanfor University Press. Novacek MJ an Clelan EE The current bioiversity extinction event: scenarios for mitigation an recovery. P Natl Aca Sci USA 98: Oum EP The strategy of ecosystem evelopment. Science 164: Olson DM, Dinerstein E, Wikramanayake ED, et al Terrestrial ecoregions of the worl: a new map of life on Earth. BioScience 51: Anthropogenic biomes of the worl Pickett STA an Caenasso ML Lanscape ecology: spatial heterogeneity in ecological systems. Science 269: Pickett STA, Caenasso ML, Grove JM, et al Urban ecological systems: linking terrestrial ecological, physical, an socioeconomic components of metropolitan areas. Annu Rev Ecol Syst 32: Qaeer MA Ruralopolises: the spatial organisation an resiential lan economy of high-ensity rural regions in South Asia. Urban Stu 37: Rinfuss RR, Walsh SJ, Turner II BL, et al Developing a science of lan change: challenges an methoological issues. P Natl Aca Sci USA 101: Ruiman WF The anthropogenic greenhouse era began thousans of years ago. Climatic Change 61: Salvatore M, Pozzi F, Ataman E, et al Mapping global urban an rural population istributions. Rome, Italy: UN Foo an Agriculture Organisation. Environment an Natural Resources Working Paper 24. Sanerson EW, Jaiteh M, Levy MA, et al The human footprint an the last of the wil. BioScience 52: Smil V General energetics: energy in the biosphere an civilization, 1st en. New York, NY: John Wiley & Sons. Smith BD The ultimate ecosystem engineers. Science 315: Theobal DM Placing exurban lan-use change in a human moification framework. Front Ecol Environ 2: Thomas CD, Cameron A, Green RE, et al Extinction risk from climate change. Nature 427: Turner II BL, Skole D, Sanerson S, et al Lan-use an lan cover change: science/research plan. Stockholm, Sween: International Geosphere Biosphere Ptrogramme. IGBP Report no 35. Vitousek PM, Mooney HA, Lubchenco J, an Melillo JM Human omination of Earth s ecosystems. Science 277: Wilkinson BH an McElroy BJ The impact of humans on continental erosion an seimentation. Geol Soc Am Bull 119: BRING ESA TO SCHOOL! STUDENT MEMBERS NEEDED! Enter the ESA Stuent Section Poster Contest, an you coul WIN one of three great prizes! It s simple: PRINT the flyer POST it on campus SUBMIT a photo an explanation to For flyers, rules, an eligibility, visit the Stuent Section website at Hurry! Contest ens December 15!