The Trouble with Savanna and Other Environmental Categories, Especially in Africa

  • Chris S. Duvall
  • Bilal Butt
  • Abigail Neely


Environmental categories are simplifications of reality meant to enable generalization, which is necessary to produce predictive physical geographic knowledge. We argue here that these categories are social constructions related to ideas shared broadly in society, including environmental deterministic explanations of human difference. The biophysical, philosophical, and sociocultural problems associated with environmental categories are exemplified by ‘savanna’ in Africa. Examining environmental categorization is an important point of engagement in critical physical geography because it is a social process, explicitly centered on simplification and generalization, and significant broadly across scientific practice and society.

The concept of environment is useful for understanding geography as the integration of biophysical and sociocultural, objective and subjective, material and nonmaterial processes (Gregory 2017; Radcliffe 2009). A formal definition of environment is the biophysical conditions, aesthetic values, and human activities that exist in a given location (Mayhew 2009: 159). Other terms can be approximately synonymous, including land cover, biome, region, and landscape. By categorizing environmental conditions, geographers can characterize and compare locations. Yet environmental categories pose inescapable problems of geographic philosophy—in short, how to identify and delimit meaningful features. These problems are resolved only through subjective sociocultural consensus rather than objective reasoning.

In this chapter, we argue that practitioners of Critical Physical Geography (CPG) must query the philosophical and sociocultural significance of environmental categories as much as their biophysical meaning. Environmental categories are simplifications meant to enable geographic generalization, which is necessary to develop predictive knowledge of biophysical reality. However, the environmental categories used in physical geography also provided bases for generalizing about human conditions. For instance, environmental determinism—the idea that environment controls human characteristics and behaviors—is founded upon ideas about biophysical conditions. In terms of transformative physical geographic scholarship (Tadaki et al. 2015), environmental categorization is an important point of engagement because it is: a social process, explicitly centered on simplification and generalization, and significant broadly across science and society (Gregory 2017; Simon 2016).

The remainder of this chapter comprises four sections. First, we outline geographic theory regarding the social nature of environmental categories. Second, we describe the history of the category of ‘savanna’ in Western geography. Third, we sketch the social meaning that savanna has had, particularly in Africa. In these sections, we analyze historical works from the perspective of postcolonial ecocriticism, which argues that environmental knowledge arises across whole societies and not separately in scientific and aesthetic contexts (Huggan and Tiffin 2010). Finally, we conclude that physical geographers must conscientiously avoid overgeneralizing the environmental context.

Geographic Philosophy and Society

Western environmental thought has long emphasized the relative abundance of trees versus grasses, and privileged trees over other flora (Dove 2004; Davis, this volume). Within this frame of reference, savanna represents a real but irresolvable biophysical condition. Globally, tree-grass abundance exhibits clinal variation associated primarily with water availability. There are locations where trees dominate, with humid climates and/or high soil moisture. At the other extreme, there are locations where vegetation is sparse and ephemeral because conditions are too dry for most plants. For economic reasons—particularly livestock husbandry—Europeans have long noticed the middle condition of tree-grass co-abundance. The extremes and the middle are the most distinct positions when viewed as independent samples along the continuum of tree-grass co-abundance. Nonetheless, “[a]ny definition of the limits of savanna on this continuum is unavoidably arbitrary” (Scholes 1997: 258).

The trouble with environmental categories begins with geographic ontology, or the fundamental properties of and relationships between geographic concepts (Smith 2001; Smith and Mark 2003). Features such as mountain, city, or lake seem unquestionably to exist; they are, indeed, basic concepts in the informal knowledge people use to navigate reality. Yet the existence of most physical geographic features is difficult to demonstrate formally. Few features have bona fide (natural) boundaries that correspond to objective discontinuities on the Earth’s surface, such as oceans, rivers, cliffs, or highways. Most features are some part of a field of continuous variation, such as surface elevation or plant density, so they have fiat (socially constructed) boundaries because they are not discrete objects, even if conceptualized as such (Smith 2001; Smith and Mark 2003; Smith and Varzi 2000; Varzi 2001). Any boundaries defined for geographic features within continuous fields of variation are inherently arbitrary (Bennett 2001).

Defined geographic features are thus merely one among many possibilities for differentiating portions of these continuous fields. As such, concepts like desert, moraine, or peneplain have historical ontologies: they emerged at particular moments and have changed within broader sociocultural change (Hacking 2002). Surprisingly, few physical geographic features are natural kinds, or entities recognized pan-culturally (Smith and Mark 2001, 2003). Rather, different sociocultural groups—including academic geographers—have unique geographic ontologies, featuring seemingly self-evident categorizations that are in fact specific to each group (Burenhult 2008; Mark et al. 2011; Blaut 1979).

Social consensus is prerequisite for constructing shared geographic knowledge, including environmental categories. Thus proposed geographic features, such as savannas or deserts, gain existence alongside shared theories of knowledge (that is, epistemologies) that make the underlying concepts observable in the real world (Inkpen 2005; Blaut 1979; Simon 2010, 2016). For instance, where does forest end and something else, such as savanna, begin? The range of answers to this question demonstrates that the physical arrangement of trees is secondary to social context in answering this question (Helms 2002; Körner 2007: 317).

A deep-seated epistemological debate in Western geographic thought is of generalization versus specificity (or nomothetic versus idiographic approaches) (Cresswell 2013: 84–88). Scholars debate how broadly or narrowly to define environmental categories, but all categories simplify reality. Placing multiple locations in a single category means that differences between locations must be overlooked to greater or lesser degrees, depending upon how generalizing or specifying the category’s definition might be.

As social constructs, physical geographic features relate to ideas shared widely in society. The concept of tropical rainforest, for example, arose within nineteenth-century European botany but reflected contemporaneous belief in social Darwinism, imperatives for colonial expansion, and imagined geographies of ‘the tropics’ (Stott 1999). These broadly shared ideas helped enable European powers to gain and exert political-economic authority worldwide, to the detriment of many non-European peoples. Environmental categories have had especially important social roles as bases for environmental determinism, which is an idea that has sustained racist explanations of human difference (Correia 2013; Raleigh et al. 2014), even if it is less controversial in explanations of non-human biogeographies (Rosindell et al. 2012). Environmental determinism subtly shapes resource management in many locations (Robbins 2001; Simon 2010, 2016). Most relevant for this chapter are examples from colonial and post-colonial contexts in the Global South (Bassett and Crummey 2003; Cinnamon 2003; Dove 1992, 1997, 2004; Ickowitz 2006). People and activities that are categorized as belonging to one environment are considered inappropriate, inauthentic, or incapable when in locations that are categorized differently. Socially constructed environmental mismatches have enabled authorities to claim rights or obligations to intervene in resource management. For instance, in colonial and post-colonial Guinea, ethnic groups categorized as “savanna people” were considered inimical to forest, and thus the became the targets of repressive policies intended to conserve forest vegetation (Fairhead and Leach 1996: 34).

It is important in CPG to recognize that environmental categories are useful in studying biophysical reality but also inescapably social constructions that enable thought about how external conditions affect biotic and abiotic objects. All categories are historically and geographically contingent concepts.

Historical Ontology of Savanna

By some accounts, savanna is deeply rooted in human experience (Domínguez-Rodrigo 2014). The savanna hypothesis in evolutionary psychology is that our East African origins make us prefer environments with “high resource-providing potential”, “distant views and low, grassy ground cover” and scattered trees (Orians and Heerwagen 1992: 559). Regardless of human paleontology, the category of savanna is neither timeless nor natural but a historically traceable concept.

Innumerable works claim that savanna simply is, presenting it as a self-evident natural feature and thus ignoring the social context that allows us to delineate part of a continuum as a discrete category. For example, “The savanna biome comprises a mix of trees and grasses. The trees do not form a continuous canopy, and lack of shade allows grass to grow. Savanna covers large areas of Africa, South Asia, South America, and Australia” (Rubenstein 2008: 27). Many works present savanna as a primary environmental category for African and tropical geography more broadly, composing with forest and desert the basic environmental structure for these expansive locations (Murray 1990: 14).

Ostensibly, savanna is identifiable through observations of features like trees and grasses, but scholars have long struggled to define the category in a way that consistently aligns with the range of conditions that various people have considered characteristic of savanna. By 1965, for instance, scholars had identified savanna in terms of “climate, soil, pedology, landforms, planation, hydrology, environment, landscape, fauna, agriculture, culture, economy, people, crops, [and] cattle”, not to mention vegetation (Hills 1965: 216). Repeatedly, since the 1920s, some have argued that no proposed criteria are sufficient to define savanna, whether globally or in particular locations (Cole 1963; Eiten 1986; Lawesson 1994; Pratt et al. 1966; Shantz and Marbut 1923: 7–8). Predictably, maps of savanna do not correspond (Fig. 6.1).
Fig. 6.1

Maps of African environmental geography. Hills and Archibold portray only savanna. Drude, Cole & de Blij, and Cunningham & Cunningham include scattered highland zones, Mediterranean areas at the northernmost and southernmost latitudes, and various other small patches, particularly temperate grassland (primarily South Africa and Ethiopia). Kromm includes tundra (sic; Kenya-Tanzania border) and various patches related to population density and agricultural productivity.

Despite the challenges of defining the category, savanna has been widely accepted because there is a spatial middle condition between global extremes of tree-grass abundance, as well as a conceptual middle between forest and desert. The absence of bona fide boundaries means that savanna is hardly constrained spatially. Similarly, its conceptual amorphousness—evident in the lack of a reliable definition—allows highly malleable fiat boundaries. Savanna might be any low-latitude, seasonally dry location; any place with lots of grass and some trees; areas where people exhibit certain behaviors; or places otherwise somehow between forest and desert.

In English, savanna dates to a 1555 translation of Peter Martyr’s Latin representation of Spanish borrowings of Native American words encountered before 1530 (Oxford English Dictionary 2006). Not surprisingly, the loanword’s meaning and origin are uncertain. Martyr identified savanna in two locations. In eastern Panama, he described an indigenous dominion that included “a playne of twelue leages in breadth and veary frutefull. This playne, they caule Zauana [Savanna]” (Arber 1885: 148). The relevant language was perhaps Kuna, but Martyr’s text is so imprecise that Zauana could refer to topography, political unit, or something else. The other location was Hispaniola, where the now-extinct Taino language was spoken. Hispaniola’s Zauana was a “lordshyp”, which encompassed forested mountains and “plaines […] withowt trees, whether the earth be with grasse or withowt” (Arber 1885: 169, 173, 212).

The concept taken into English came from Martyr’s table of contents, which described the Panamanian place as “the large and frutefull playne of Zauana” (Arber 1885: 395). In the following centuries, the concept of savanna appeared widely in the New World and subsequently elsewhere (Fig. 6.2).
Fig. 6.2

Local appearances of savanna in English, 1600–1800. Non-fictional uses in black boxes, fictional uses in gray

The pre-scientific category of savanna was not formally defined, but characteristically had grassy vegetation and planar topography, which enabled sweeping views and facilitated travel; its plants, soil, and livestock or game seemingly promised high productivity. Early uses compared savanna to pasture and meadow, economically significant concepts that by the fourteenth century were locations used for grazing livestock (Oxford English Dictionary 2006). Savanna began appearing in truly imagined contexts by 1700, showing that the concept was sufficiently generalized to become placeless. Most importantly, in the 1810s, The Swiss Family Robinson’s South Pacific island had three main environments: forests, a desert, and a grassy plain abounding with game and “interspersed at agreeable distances with little woods” (Wyss 2007 [1816]: 281). In the first English edition (1816) this area was unnamed; in the second edition (1848) it was “savanna” (Locke 1848: 198).

Savanna circulated via incompletely known social contexts (Domínguez-Rodrigo 2014). In English, into the 1800s it was recognized as a Spanish loanword, and often italicized and defined, suggesting novelty. Adoption varied spatially and between European language communities. For instance, in North America in 1799, English speakers widely said savanna while French speakers said prairie (Winterbotham 1799: 487), even though French savane was written decades before in Central America, the Caribbean, and Senegambia (Adanson 1757; Anonymous 1684: 40, 360, 417; Raveneau de Lussan 1690: 254).

About 1700, the concept of savanna began to enter scientific discourse. Its placelessness made savanna useful but imprecise, as English philosopher John Locke observed: savanna, like woodland, mountain, and plain, enabled “loose Description” of a country, but did not offer “the more useful Observations of the Soil, Plants, Animals and Inhabitants, with their several Sorts of Properties” (Locke 1706: 78). The generalizing power of savanna was attractive during the Age of Sail, when many locations outside Europe were seen briefly, with limited knowledge. Yet naturalists also sought to characterize place-specific conditions. They struggled to do so without having formally considered the epistemology of environmental categories, that is, how they might be defined and thus identified in the real world. Savanna posed particular problems, because, as a popular geography of North America recognized, “the dubious boundaries of the savannas, ris[e] imperceptibly toward the forests” (Pinkerton 1802: 584). Generalizing proved easier than specifying place-specific conditions. In Jamaica, for instance, Hans Sloane inconsistently defined savanna through comparisons and contrasts with other categories: he variously wrote “Savanna or Meadow”, “low Land Woods and Savanna’s”, “Savanna’s or Plains”, “Savanna Woods”, “woody parts of the Savanna’s, or Low-Lands”, “Low-Land or Savanna Woods”, “woody Savanna’s”, and “Savanna’s and Woods” (Sloane 1725: ix, 19, 24, 25, 26, 39, 131, 173, and elsewhere).

Alexander von Humboldt initiated formal discussion of environmental ontology and epistemology (Humboldt 1819: 148–154). His concern arose with regard to environments with few trees: “Europe has bruyères [moors], Asia steppes, Africa deserts, America savannas; but this distinction establishes contrasts that are founded neither in the nature of things, nor in the genius of language” (Humboldt 1819: 148). He considered vegetation the most meaningful criterion for environmental categories and divided environments with few trees into deserts (“bare lands, without a trace of plants”) and “savannas or steppes” (“lands covered with grasses”) (Humboldt 1819: 149). He also contrasted savanna and forest, at least in South America, where he consistently reduced environmental variation to a binary opposition as embodied by “savanna peoples” versus “forest peoples” in Brazil (Humboldt 1819: 609). Despite the implication that his categories were obvious and self-evident, Humboldt did not construct a consistent epistemology. He took local concepts as equivalent to global categories, challenged the applicability of general concepts in specific instances, and inserted criteria beyond his stated preference for vegetation. For example, he observed that “the puszta of Hungary are veritable savannas” but “the savannas of America, especially those in the temperate zone, [are often called] prairies; but this word seems to me hardly applicable to pasturage [that is] often very dry. Instead, environments like] the Llanos and Pampas of southern America are veritable [savannas or steppes]” (Humboldt 1819: 149). Humboldt’s savanna was not explicitly tropical but within his New World “equinoctial regions”, and he consistently contrasted low-latitude and mid-latitude environments with few trees.

Humboldt’s forest-savanna-desert model undergirded subsequent scholarship, though some preferred narrower categories. In particular, French geographer Conrad Malte-Brun argued that Humboldt’s broad, vegetation-based categories must be subdivided by soil, hydrology, and topography to be meaningful (Malte-Brun 1819: 215–220). Nonetheless, subsequent writers maintained savanna as a concept loosely defined only through juxtaposition with other categories. By 1860, science writers began writing explicitly of “tropical savanna” (Drude 1887; Grisebach 1872; Hartwig and Guernsey 1876; Mangin 1872; Müller 1857), despite its persistence in the southeastern United States. Economic and aesthetic values together shaped understandings of savanna, which commonly seemed to signify presumed productive potential as much as biophysical conditions. In 1889, for instance, forest occupied windward slopes in Fiji, while drier leeward slopes “offer[ed] only savannas […] where colonists find the most favorable lands, already suited for agriculture or raising livestock” (Reclus 1889: 874). Nineteenth-century scholars explicitly took aesthetic notions as a basis for environmental categorization (Warming 1909: 137), mixing beliefs about imagined geographic regions with Western preferences for trees over grasses, and verdant over senescent vegetation (Dove 2004; Stott 1999).

Vegetation structure was privileged in environmental geography because it was considered dependent upon climate (Grisebach 1872; Herbertson 1905; Köppen 1900; Schimper 1903).

The presence or absence of water in the Tropical World exerts an influence upon all forms of animal and vegetable life […]. Wherever water is absolutely wanting, the country is given over to barrenness. Wherever water is perpetual and abundant, the soil is clothed with lofty forests and a profusion of lush vegetation. Midway between these extremes are vast tracts dry at one season and wet at another. These regions […] we may call savannas […]. (Hartwig and Guernsey 1876: 499)

This epistemology—that vegetation indicated meaningful climatic differences—reflected early twentieth-century interest in environmental determinism. Tree-grass co-abundance became diagnostic of Köppen’s savanna climate zone. Nonetheless, named categories never corresponded precisely. Warming’s “savannah-vegetation” (1909) differed from Schimper’s “savanna forest” (1903) and Grisebach’s “tropical savanna” (1872). Differences reflected contrasting beliefs about vegetation ecology, and thus different interests to promote in scientific discourse (Shantz and Marbut 1923; Tansley 1920).

Generalizing concepts like savanna masked variation among environmental conditions in particular locations, which scholars tried to indicate with the plural form savannas or qualified forms like humid savanna or shrub savanna. Debates arose on the appropriateness of qualified terms versus alternatives like steppe, prairie, and grassland (Walter 1971: 238–239). Qualified and alternative forms represent competing ecological beliefs, as well as different social contexts (Domínguez-Rodrigo 2014; Eiten 1986; Gleave and White 1969; Hills 1965; Lawesson 1994). As the savanna lexicon expanded, prominent scholars unsuccessfully tried to impose specific epistemologies via standardized definitions (Forsberg 1967; Trochain 1957; UNESCO 1973; Walter 1971).

In the 1980s, leading scholars of savanna promulgated scale- and resolution-specific criteria for knowing savanna. This epistemology remains dominant, and makes savanna “those forms of vegetation that occur between the equatorial rain forests and the mid-latitude deserts and have a continuous grass stratum that is either treeless or [with] trees and shrubs [at] variable […] densit[ies]” (Cole 1986: 6). Thus, savanna is meaningful only for “general” vegetation description (White 1983: 18) because it refers to “a continuum of physiognomic [vegetation] types” (Menaut 1983: 110). This epistemology posits that ultimately (or at least theoretically) climate produces environmental geography; savanna represents precipitation seasonality due to the annual north-south shift of the subtropical high-pressure belt and inter-tropical convergence zone (Nix 1983).

This epistemology allows simultaneous generalization and specificity. Savanna is globally real, yet locally unobservable. For instance, one study states, “[we] call this collection of ecosystems ‘savannas,’ except when we need to distinguish particular habitats. We note that savannas range from well-wooded areas […], to open habitats with few trees” (Loarie et al. 2009: 3100). This epistemology acknowledges fine-scale environmental variability, yet contradictorily allows scholars to describe particular conditions as deviations from idealized conditions, rather than environments requiring particularized descriptions. Thus, one study described a site as including “tree savanna”, “wooded savanna”, “open shrub savanna”, “dense shrub savanna”, “disturbed savanna”, “grass savanna”, “savanna grassland”, “humid savanna grassland”, “humid savanna”, “dry savanna grassland”, and also “open forest”, without further differentiating these categories (Bassett and Koli Bi 2000).

The currently dominant epistemology does not resolve any objective geographic feature, but justifies socially grounded belief that savanna is real. “Although it may be difficult to define the term ‘savanna’ precisely,” one paper acknowledges, “the general concept of a tropical or subtropical mixed tree (or shrub)-grass community is widely accepted” (Jeltsch et al. 2000: 161). Savanna “is widely accepted”—has socially determined validity—because of its history in Western thought, and its social roles outside biophysical science.

Social Roles of African Savanna

The savanna concept enables generalizations about biophysical conditions but also sustains facile characterizations of human geography that are sustained broadly across society.

The concept of savanna first appeared in Africa—in Senegambia—in 1738 (Moore 1738), but many subsequent European travelers did not see it. Mungo Park (West Africa, 1790s–1800s) and David Livingstone (Southern Africa, 1850s–1860s) saw pasture, prairie, and meadow, while Richard Burton (Central and East Africa, 1860s–1870s) and Henry Morton Stanley (East, Southern, and Central Africa, 1860s–1870s) sometimes encountered savanna. The Portuguese generally saw campina, Germans Steppe, Afrikaaners veld, and the French savane, if not prairie. European scholars increasingly described African environments as savanna, beginning with Carl Ritter’s 1822 description of what is now Benin (Ritter 1822: 297). By 1880, savanna covered 21.3% of Africa, 37% by 1923, and 65% by 1995 (White 1892: 53; Archibold 1995: 61; Shantz and Marbut 1923: 57).

Humboldt’s forest-savanna-desert model, described above, has been the backbone for environmental geographies of Africa, garnished with other, marginal categories (Fig. 6.1).

By 1900, the forest-savanna-desert ontology supported deterministic portrayals of Africans (Richards 1996; Salazar 2009; Tilley 2011). Environmental stereotypes varied across the continent. Most importantly, in northern Africa, the Hamitic hypothesis flourished. This idea was that any vestiges of civilization were the heritage of Caucasian Hamites—descendants of Noah’s son Ham—who entered Africa from the north. This hypothesis became environmental because of the north-to-south progression perceived for the desert-savanna-forest categories. Thus, desert pastoralists surpassed savanna agriculturalists, who surpassed more southerly forest peoples (Mangin 1872: 181).

Savanna was, at least theoretically, observable in bodies: “Peoples of the Savannas are taller and, though often darker, are more mixed in [racial] origin than the negroes of the [forested] south” (Harrison Church et al. 1964: 213). Supposed linkages between environment, livelihood, and race generated corporeal consequences, particularly because the three environments were commonly portrayed as in conflict. In Rwanda and Burundi, for instance, the Tutsi identity arose under Belgian colonial rule, and linked pastoralism, savanna, and arrival from elsewhere (that is, a Hamitic past); this simplistic history was variously taken to show the benign superiority or domineering usurpation of Tutsi over Hutu, an identity linked to farming, indigeneity, and forest, which was considered to have declined as savanna expanded (Eltringham 2006; Hintjens 1999; Jefremovas 1997). Repeated conflicts since 1959, including genocide in 1972 and 1994, centered on these identities.

Environmental generalizations about people have contributed to conflicts elsewhere (Gruley and Duvall 2012; Richards 2001). Popular media disseminate such generalizations, including an educational website that tells us that: “The people of Africa’s vast savanna are united by their strong identity with the sprawling plains that surround them. [Amongst] these pastoral groups […] the Maasai have held the most tenaciously to their wanderlust. These tall, dark skinned herdsmen [now] share the plains with the Kikuyu, traditionally a nation of farmers […]” ( 2001).

Colonial ecological science portrayed savanna as if in existential struggle against forest and desert. Forest was nature’s aboriginal, tropical lushness; desert was barren nature (Stott 1999; Verstraete 1986). The potential bounty of savanna could be lost through “desertification” (see Davis and Sayre, this volume), although savanna itself could represent degradation—the “savannization” of forest (most notably: Aubréville 1949, 1962). Savanna people were believed to cause both. This narrative was especially prominent in West Africa, where it justified colonial and neo-colonial interventions (Bassett and Crummey 2003; Cinnamon 2003; Fairhead and Leach 1996; Ickowitz 2006). Ecologists widely found savanna productivity below their expectations, interpreting this as evidence that so-labeled savanna livelihood practices were maladaptive. Savanna located near forest was qualified as “derived”, to indicate belief in past (but unobserved) anthropogenic deforestation. An unnamed 1957 college textbook described by Rosenblum “claimed that one of the basic reasons for the underdevelopment of the savanna region was ‘the presence of the native population’” (Rosenblum 1963: 11). This blunt portrayal echoed earlier, subtler statements, and anticipated subsequent ones (e.g. Verstraete et al. 2009).

In Southern and East Africa, savanna was frequently understood as unpeopled, full of wildlife, and available to outsiders. Fictional portrayals of unbound “savanna Africa” stimulated tourism, and were prominent in White settler imaginations (Adams and McShane 1992; Akama 1996; Hughes 2011; Staples 2006). East African landscapes became the archetypal savanna, unchanged since the Pleistocene and the ultimate contrast to humanized Europe (Anderson and Grove 1987; Neumann 2011). Further, the (singular) African savanna became the archetypal wild environment (Preston-Whyte et al. 2006: 132), occupied only by wild animals and timeless people, like the Maasai stereotype above (Akama 2002; Galaty 2002; Norton 1996). Colonial and post-colonial authorities made this concept of savanna real by evicting people from protected areas (Neumann 1997), as African agriculturalists were considered inauthentic in or incompatible with the wild savanna.

Across the continent, socially constructed environmental “mismatches” allowed policymakers to claim and place blame for environmental changes, and to justify heavy-handed policies meant to either eliminate or promote savanna behaviors depending on context. Purportedly apolitical savanna imagery continues to justify resource dispossession (Robbins 2012: 12–13).

What/Where/Why Is Savanna?

It is neither necessary nor unnecessary that savanna exists as a category in Africa or elsewhere. Ever since Malte-Brun (1819), a small but persistent social group has argued against highly generalized environmental categories (Malte-Brun 1819). The specifying impulse might reflect desires to correct “geographer’s fictions” (Gleave and White 1969), beliefs about the characteristics and meaning of biophysical conditions (Domínguez-Rodrigo 2014), efforts to improve resource management (Raynaut 1997), or new approaches to data collection and analysis (Sayre et al. 2013). Emphatically, narrower environmental categories are just as social as broad ones, and thus can support non-biophysical generalizations and political-economic imperatives.

The generalizing versus specifying quandary in environmental thought has existed at least since the 1690s (Locke 1706: 78). Savanna represents a generalizing worldview: “Despite their […] differences, the savannas of the world are believed to share the same basic patterns of structure and function” (Scholes 1997: 259). In academic geography, debate about the value of general versus specific knowledge has been especially prominent since the 1950s, though not in physical geography (Cresswell 2013). Contrary to Miller and Goodchild (Miller and Goodchild 2015: 456), we would argue that it is not “perhaps wise” of physical geographers to avoid this debate, or other points of philosophical engagement. Geographic generalization is necessary and broad patterns and processes exist, but broad features arise through the spatial dependency and heterogeneity that exists in local contexts. Further, advances in complexity theory, data science, and biogeography underscore that interactions within particular environments produce emergent, system-wide behaviors that cannot be understood—or sometimes even observed—in examining either the local or the global alone (Graham and Shelton 2013; Gregory 2017; Miller and Goodchild 2015).

Instead of choosing either generalizations or specificities, we need both. The environmental categories with scientific lineages tracing to Humboldt (1819) are not suitable, and were not intended, for particular descriptions. As the spatial and conceptual center, savanna stabilizes the forest-savanna-desert ontology as fact without history, even though it was formalized in Europe in the 1810s: evident in The Swiss Family Robinson, it became scientific in Humboldt’s equinoctial Americas. The concept of savanna approximates the global condition of semi-arid climate and low-latitude location. Due to the term’s long history it is probably inescapable in labeling emergent behavior associated with this global condition. But importantly, spatial extent does not indicate complexity, globalness, or localness; forest, savanna, desert, and other highly generalizing categories should not be accepted as default descriptors of large areas. The specificities of locations grouped into the broad categories are important. Biogeographers have identified multivariate and multi-scalar patterns in semi-arid, low-latitude regions. Erstwhile savanna in Africa, for instance, can be meaningfully bifurcated based on various non-contiguous pairs of biophysical criteria: dystrophic versus eutrophic soils (Breman and Kessler 1995); bimodal versus unimodal precipitation seasonality (Ellis and Galvin 1994); Zambezian versus Sudanian floristic zones (White 1983); elevation greater or less than 3000 feet (Stock 2004); or clustered versus dispersed tree distribution (Moore 1996). Many studies described as relating to African savanna do not justify such generalization (e.g. Asner et al. 2009; Pellegrini et al. 2017). More globally, scholars call semi-arid, low-latitude locations categorically savanna, even if the purpose of generalization is, ironically, to characterize expansive areas as not uniform (Furley and Metcalfe 2007; Lehman et al. 2014). The need to generalize should not be privileged over the need to describe locally specific environmental conditions.

Finally, the usefulness of environmental description depends upon social context (Robbins 2001; Simon 2010, 2016). There is no inherently correct or incorrect way to categorize environments. Practicing CPG means actively inquiring why a particular environment might or might not be considered to exist in a location (see Davis, Sayre, and Simon, this volume). Environmental categories are useful to most physical geographers, though much more widely useful elsewhere in society. The forest-savanna-desert categories allow immediate, if facile, generalizations, and inevitably evoke notions about object-context relationships, including environmentally deterministic ideas about human identity, authenticity, and capability that can have profound environmental and ethical consequences for how landscapes are managed. All categories simplify reality. Nonetheless, physical geographers have social responsibilities to avoid overgeneralizing about environmental conditions and to embrace specificities as well, in order to weaken facile characterizations that can sustain unjust political-economic relationships.


  1. Adams, J.S., and T.O. McShane. 1992. The myth of wild Africa. New York: Norton.Google Scholar
  2. Adanson, M. 1757. Histoire naturelle du Sénégal. Paris: Bauche.Google Scholar
  3. Akama, J.S. 1996. Western environmental values and nature-based tourism in Kenya. Tourism Management 17 (8): 567–574.CrossRefGoogle Scholar
  4. ———. 2002. The creation of the Maasai image and tourism development in Kenya. In Cultural tourism in Africa, ed. J.S. Akama and P. Sterry, 43–54. Arnhem, The Netherlands: ATLE.Google Scholar
  5. Anderson, D., and R.H. Grove. 1987. Introduction. In Conservation in Africa, ed. D. Anderson and R.H. Grove, 1–12. Cambridge, UK: Cambridge University.Google Scholar
  6. Anonymous. 1684. Recueil de divers voyages faits en Afrique et en Amérique. Paris: Veuve Cellier.Google Scholar
  7. Arber, E. 1885. The first three English books on America. Birmingham, UK: [Publisher not named].Google Scholar
  8. Archibold, O.W. 1995. Ecology of world vegetation. London: Chapman & Hall.CrossRefGoogle Scholar
  9. Asner, G.P., et al. 2009. Large-scale impacts of herbivores on the structural diversity of African savannas. Proceedings of the National Academy of Sciences of the USA 106 (12): 4947–4952.CrossRefGoogle Scholar
  10. Aubréville, A. 1949. Climats, forêts, et désertification de l’Afrique tropicale. Paris: Société d’Éditions Géographiques, Maritimes, et Coloniales.Google Scholar
  11. ———. 1962. Savanisation tropicale et glaciations Quaternaires. Adansonia 2 (1): 16–84.Google Scholar
  12. Bassett, T.J., and D. Crummey, eds. 2003. African savannas. Oxford, UK: James Currey.Google Scholar
  13. Bassett, T.J., and Z. Koli Bi. 2000. Environmental discourses and the Ivorian savanna. Annals of the Association of American Geographers 90 (1): 67–95.CrossRefGoogle Scholar
  14. Bennett, B. 2001. What is a forest? On the vagueness of certain geographic concepts. Topoi 20: 189–201.CrossRefGoogle Scholar
  15. Blaut, J.M. 1979. Some principles of ethnogeography. In Philosophy in geography, ed. S. Gale and G. Olsson, 1–7. Dordrecht, The Netherlands: D. Reidel.Google Scholar
  16. Breman, H., and J.-J. Kessler. 1995. Woody plants in agro-ecosystems of semi-arid regions with an emphasis on the Sahelian countries. Berlin: Springer-Verlag.CrossRefGoogle Scholar
  17. Burenhult, N. (Ed.) 2008. Language and landscape: Geographical ontology in cross-linguistic perspective [Special Issue]. Language Sciences, 30 (2–3): 135–382.Google Scholar
  18. Cinnamon, J. 2003. Narrating equatorial African landscapes: Conservation, histories, and endangered forests in northern Gabon. Journal of Colonialism and Colonial History 4 (2).
  19. Cole, M.M. 1963. Vegetation nomenclature and classification, with particular reference to the savannas. South African Geographical Journal 45 (2): 3–14.CrossRefGoogle Scholar
  20. ———. 1986. The savannas: Biogeography and geobotany. London: Academic Press.Google Scholar
  21. Correia, D. 2013. F**k Jared Diamond. Capitalism Nature Socialism 24 (4): 1–6.CrossRefGoogle Scholar
  22. Cresswell, T. 2013. Geographic thought. Chichester: Wiley-Blackwell.Google Scholar
  23. Domínguez-Rodrigo, M. 2014. Is the “Savanna Hypothesis” a Dead Concept for Explaining the Emergence of the Earliest Hominins? Current Anthropology 55 (1): 59–81.CrossRefGoogle Scholar
  24. Dove, M.R. 1992. The dialectal history of “jungle” in Pakistan. Journal of Anthropological Research 48 (3): 231–253.CrossRefGoogle Scholar
  25. ———. 1997. The epistemology of Southeast Asia’s anthropogenic grasslands. Southeast Asian Studies (Kyoto) 35 (2): 223–239.Google Scholar
  26. ———. 2004. Anthropogenic grasslands in Southeast Asia. Agroforestry Systems 61: 423–435.Google Scholar
  27. Drude, O. 1887. Atlas der Pflanzenverbreitung. Gotha: Perthes.Google Scholar
  28. Eiten, G. 1986. The use of the term “savanna”. Tropical Ecology 27 (1): 10–23.Google Scholar
  29. Ellis, J., and K.A. Galvin. 1994. Climate patterns and land-use practices in the dry zones of Africa. BioScience 44 (5): 340–349.CrossRefGoogle Scholar
  30. Eltringham, N. 2006. The Hamitic hypothesis, race, and the Rwandan genocide. Social Identities 12 (4): 425–446.CrossRefGoogle Scholar
  31. Fairhead, J., and M. Leach. 1996. Misreading the African landscape. Cambridge: Cambridge University Press.CrossRefGoogle Scholar
  32. Forsberg, F.R. 1967. A classification of vegetation for general purposes. Tropical Ecology 2: 1–28.Google Scholar
  33. Furley, P.A., and S.E. Metcalfe. 2007. Dynamic changes in savanna and seasonally dry vegetation through time. Progress in Physical Geography 31 (6): 633–642.CrossRefGoogle Scholar
  34. Galaty, J. 2002. Narrative inventions of the pastoralists in East Africa. Visual Anthropology 15 (3): 347–367.CrossRefGoogle Scholar
  35. Gleave, M.B., and H.P. White. 1969. The West African Middle Belt: Environmental fact or geographer’s fiction? Geographical Review 59 (1): 123–139.CrossRefGoogle Scholar
  36. Graham, M., and T. Shelton. 2013. Geography and the future of big data, big data and the future of geography. Dialogues in Human Geography 3 (3): 255–261.CrossRefGoogle Scholar
  37. Gregory, K.J. 2017. Putting physical environments in their place: The next chapter? The Canadian Geographer / Le Géographe canadien 61(1): 11–18.Google Scholar
  38. Grisebach, A. 1872. Die Vegetation der Erde nach ihrer Klimatischen Anordnung, Band II. Leipzig: Engelmann.Google Scholar
  39. Gruley, J., and C.S. Duvall. 2012. The evolving narrative of the darfur conflict as represented in The New York Times and The Washington Post. GeoJournal 77 (1): 29–46.CrossRefGoogle Scholar
  40. Hacking, I. 2002. Historical ontology. Cambridge: Harvard University Press.CrossRefGoogle Scholar
  41. Harrison Church, R.J., et al. 1964. Africa and the islands. London: Longmans.Google Scholar
  42. Hartwig, G., and A.H. Guernsey. 1876. The polar and tropical worlds. Columbus, OH: Watrous.Google Scholar
  43. Helms, J.A. 2002. Forest, forestry, forester: What do these terms mean? Journal of Forestry 100 (8): 15–19.Google Scholar
  44. Herbertson, A.J. 1905. The major natural regions. Geographical Journal 25: 300–312.CrossRefGoogle Scholar
  45. Hills, T.L. 1965. Savannas: A review of a major research problem in tropical geography. Canadian Geographer 9 (4): 216–228.CrossRefGoogle Scholar
  46. Hintjens, H.M. 1999. Explaining the 1994 genocide in Rwanda. Journal of Modern African Studies 37 (2): 241–286.CrossRefGoogle Scholar
  47. Huggan, G., and H. Tiffin. 2010. Postcolonial ecocriticism. London: Routledge.Google Scholar
  48. Hughes, D.M. 2011. Whites lost and found: Immigration and imagination in savanna Africa. In Environment at the margins, ed. B. Caminero-Santangelo and G. Myers, 159–184. Athens: Ohio University Press.Google Scholar
  49. Humboldt, A. 1819. Voyage aux régions equinoxiales du nouveau continent, tome second. Paris: Maze.Google Scholar
  50. Ickowitz, A. 2006. Shifting cultivation and deforestation in tropical Africa. Development and Change 37 (3): 599–626.CrossRefGoogle Scholar
  51. Inkpen, R. 2005. Science, philosophy and physical geography. New York: Routledge.Google Scholar
  52. Jefremovas, V. 1997. Contested identities: Power and the fictions of ethnicity, ethnography and history in Rwanda. Anthropologica 39 (1–2): 91–104.CrossRefGoogle Scholar
  53. Jeltsch, F., G.E. Weber, and V. Grimm. 2000. Ecological buffering mechanisms in savannas: A unifying theory of long-term tree-grass coexistence. Plant Ecology 150 (1/2): 161–171.CrossRefGoogle Scholar
  54. Köppen, W. 1900. Versuch einer Klassifikation der Klimate, vorzugsweise nach ihren Beziehungen zur Pflanzenwelt. Geographische Zeitschrift 6 (11): 593–611.Google Scholar
  55. Körner, C. 2007. Climatic treelines: Conventions, global patterns, causes. Erdkunde 61 (4): 316–324.CrossRefGoogle Scholar
  56. Lawesson, J.E. 1994. Some comments on the classification of African vegetation. Journal of Vegetation Science 5: 441–444.CrossRefGoogle Scholar
  57. Lehman, C.E.R., et al. 2014. Savanna Vegetation-Fire-Climate relationships differ among continents. Science 343: 548–552.CrossRefGoogle Scholar
  58. Loarie, S., R. van Aarde, and S. Pimm. 2009. Elephant seasonal vegetation preferences across dry and wet savannas. Biological Conservation 142 (12): 3099–3107.CrossRefGoogle Scholar
  59. Locke, J. 1706. Of the conduct of the understanding. In The Posthumous Works of Mr. John Locke, 1–137. London: Churchill.Google Scholar
  60. Locke, J.C. 1848. The Swiss Family Robinson. second series, Vol. 1. New York: Harper & Brothers.Google Scholar
  61. Malte-Brun, C. 1819. Analyse critique: Voyage aux régions équinoxiales du nouveau continent. Nouvelle Annales des Voyages, de la Géographie et de l’Histoire 3: 202–222.Google Scholar
  62. Mangin, A. 1872. The desert world. London: Nelson and Sons.Google Scholar
  63. Mark, D.M., et al., eds. 2011. Landscape in language. Amsterdam: John Benjamins.Google Scholar
  64. Mayhew, S. 2009. Oxford dictionary of geography. 4th ed. Oxford: Oxford University Press.Google Scholar
  65. Menaut, J.C. 1983. The vegetation of African savannas. In Tropical savannas, ed. F. Bourlière, 109–150. Amsterdam: Elsevier.Google Scholar
  66. Miller, H.J., and M.F. Goodchild. 2015. Data-driven geography. GeoJournal 80: 449–461.CrossRefGoogle Scholar
  67. Moore, F. 1738. Travels into the inland parts of Africa. London: Cave.Google Scholar
  68. Moore, J.J. 1996. Savanna chimpanzees, referential models and the last common ancestor. In Great ape societies, ed. W.C. McGrew, L.F. Marchant, and T. Nishida, 275–292. Cambridge: Cambridge University.CrossRefGoogle Scholar
  69. Müller, K. 1857. Das Buch der Pflanzenwelt: Eine botanische Reise um die Welt, Band 1. Leipzig: Spamer.Google Scholar
  70. Murray, J., ed. 1990. The cultural atlas of the World: Africa. Alexandria: Stonehenge Press.Google Scholar
  71. Neumann, R.P. 1997. Imposing wilderness: Struggles over livelihood and nature preservation in Africa. Berkeley: University of California.Google Scholar
  72. ———. 2011. “Through the Pleistocene”: Nature and race in Theodore Roosevelt’s African Game Trails. In Environment at the Margins, ed. B. Caminero-Santangelo and G. Myers, 22–42. Athens: Ohio University Press.Google Scholar
  73. Nix, H.A. 1983. Climate of tropical savannas. In Tropical savannas, ed. F. Bourlière, 37–61. Amsterdam: Elsevier.Google Scholar
  74. Norton, A. 1996. Experiencing nature: The reproduction of environmental discourse through safari tourism in East Africa. Geoforum 27: 355–373.CrossRefGoogle Scholar
  75. Orians, G.H., and J.H. Heerwagen. 1992. Evolved responses to landscapes. In The adapted mind, ed. J.H. Barkow, L. Cosmides, and J. Tooby, 555–579. New York: Oxford University Press.Google Scholar
  76. Oxford English Dictionary. 2006. Oxford English dictionary online version. 3rd ed. Oxford: Oxford University Press.Google Scholar
  77. 2001. Africa: Explore the regions: Savanna. Accessed 21 Jan 2017.
  78. Pellegrini, A.F.A., et al. 2017. Woody plant biomass and carbon exchange depend on elephant-fire interactions across a productivity gradient in African savanna. Journal of Ecology 105 (1): 111–121.CrossRefGoogle Scholar
  79. Pinkerton, J. 1802. Modern geography. Vol. 2. London: Cadell and others.Google Scholar
  80. Pratt, D.J., P.J. Greenway, and M.D. Gwynne. 1966. A classification of East African rangelands, with an appendix on terminology. Journal of Applied Ecology 3 (2): 369–382.CrossRefGoogle Scholar
  81. Preston-Whyte, R., S. Brooks, and W. Ellery. 2006. Deserts and savannah regions. In Tourism and environmental change, ed. S. Gössling and C.M. Hall, 128–141. London: Taylor & Francis.Google Scholar
  82. Radcliffe, S.A. 2009. Environmentalist thinking and/in geography. Progress in Human Geography 33: 1–19.Google Scholar
  83. Raleigh, C., A. Linke, and J. O’Loughlin. 2014. Extreme temperatures and violence. Nature Climate Change 4 (2): 76–77.CrossRefGoogle Scholar
  84. Raveneau de Lussan. 1690. Journal du voyage fait à la mer de Sud. Paris: Coignard.Google Scholar
  85. Raynaut, C., ed. 1997. Sahels: Diversité et dynamiques des relations sociétés-nature. Paris: Karthala.Google Scholar
  86. Reclus, É. 1889. Nouvelle géographie universelle, tome 14: Océan et terres océaniques. Paris: Hachette.Google Scholar
  87. Richards, P. 1996. Forest indigenous peoples: Concept, critique and cases. Proceedings of the Royal Society of Edinburgh, Section B 104: 349–365.Google Scholar
  88. ———. 2001. Are “Forest” wars in Africa resource conflicts? In Violent environments, ed. N.L. Peluso and M. Watts, 65–82. Ithaca: Cornell University Press.Google Scholar
  89. Ritter, C. 1822. Die Erdkunde im Verhaltnis zur Natur und zur Geschichte des Menschen, Erster Theil: Afrika. Berlin: Reimer.Google Scholar
  90. Robbins, P. 2001. Fixed categories in a portable landscape: The causes and consequences of land-cover categorization. Environment and Planning A 33: 161–179.CrossRefGoogle Scholar
  91. ———. 2012. Political ecology. 2nd ed. Chichester: Wiley-Blackwell.Google Scholar
  92. Rosenblum, P. 1963. Africa as seen in American geography textbooks. Africa Today 10 (2): 4–5+11.Google Scholar
  93. Rosindell, J., et al. 2012. The case for ecological neutral theory. Trends in Ecology & Evolution 27 (4): 203–208.CrossRefGoogle Scholar
  94. Rubenstein, J.M. 2008. The cultural landscape. 9th ed. Upper Saddle River: Pearson.Google Scholar
  95. Salazar, N. 2009. Imaged or imagined? Cultural representations and the “tourismification” of peoples and places. Cahiers d’Études Africaines 49 (1–2): 49–71.Google Scholar
  96. Sayre, R., et al. 2013. A new map of standardized terrestrial ecosystems of Africa. Washington, DC: AAG.Google Scholar
  97. Schimper, A.F.W. 1903. Plant geography on a physiological basis. Oxford: Clarendon.CrossRefGoogle Scholar
  98. Scholes, R.J. 1997. Savanna. In Vegetation of Southern Africa, ed. R.M. Cowling, D.M. Richardson, and S.M. Pierce, 258–277. Cambridge: Cambridge University Press.Google Scholar
  99. Shantz, H.L., and C.F. Marbut. 1923. The vegetation and soils of Africa. New York: American Geographical Society.Google Scholar
  100. Simon, G.L. 2010. The 100th meridian, ecological boundaries, and the problem of reification. Society & Natural Resources 24 (1): 95–101.CrossRefGoogle Scholar
  101. ———. 2016. How regions do work, and the work we do: A constructive critique of regions in political ecology. Journal of Political Ecology 23: 197–203.CrossRefGoogle Scholar
  102. Sloane, H. 1725. A voyage to the Islands Madera, Barbados, Nieves, St. Christophers, and Jamaica, Vol. 2. London: Sloane.Google Scholar
  103. Smith, B. 2001. Fiat objects. Topoi 20 (2): 131–148.CrossRefGoogle Scholar
  104. Smith, B., and D.M. Mark. 2001. Geographic categories: An ontological investigation. International Journal of Geographic Information Science 15 (7): 591–612.CrossRefGoogle Scholar
  105. ———. 2003. Do mountains exist? Towards an ontology of landforms. Environment and Planning B 30 (3): 411–427.CrossRefGoogle Scholar
  106. Smith, B., and A.C. Varzi. 2000. Fiat and bona fide boundaries. Philosophy and Phenomenological Research 60: 401–420.CrossRefGoogle Scholar
  107. Staples, A. 2006. Safari adventure: Forgotten cinematic journeys in Africa. Film History 18 (4): 392–411.CrossRefGoogle Scholar
  108. Stock, R. 2004. Africa South of the Sahara. 2d ed. New York: Guilford.Google Scholar
  109. Stott, P. 1999. Tropical rainforest: A political ecology of hegemonic myth-making. London: IEA Environment Unit.Google Scholar
  110. Tadaki, M., et al. 2015. Cultivating critical practices in physical geography. The Geographical Journal 181 (2): 160–171.CrossRefGoogle Scholar
  111. Tansley, A.G. 1920. The classification of vegetation and the concept of development. Journal of Ecology 8 (2): 118–149.CrossRefGoogle Scholar
  112. Tilley, H. 2011. Africa as living laboratory: Empire, development and scientific knowledge, 1870–1950. Chicago: University of Chicago Press.CrossRefGoogle Scholar
  113. Trochain, J.-L. 1957. Accord interafricain sur la définition des types de végétation de l’Afrique tropicale. Bulletin de l’Institut des Etudes Centrafricaines 13–14: 55–94.Google Scholar
  114. UNESCO. 1973. International classification and mapping of vegetation. Paris: UNESCO.Google Scholar
  115. Varzi, A.C. 2001. Vagueness in geography. Philosophy & Geography 4 (1): 49–65.CrossRefGoogle Scholar
  116. Verstraete, M.M. 1986. Defining desertification. Climatic Change 9 (1): 5–18.CrossRefGoogle Scholar
  117. Verstraete, M.M., R.J. Scholes, and M. Stafford Smith. 2009. Climate and desertification: Looking at an old problem through new lenses. Frontiers in Ecology and the Environment 7 (8): 421–428.CrossRefGoogle Scholar
  118. Walter, H. 1971. Ecology of tropical and subtropical vegetation. Edinburgh: Oliver & Boyd.Google Scholar
  119. Warming, E. 1909. Oecology of plants. Oxford: Clarendon.Google Scholar
  120. White, A.S. 1892. The development of Africa. 2nd ed. London: George Philip.Google Scholar
  121. White, F.C. 1983. The vegetation of Africa. Paris: UNESCO/AETFAT/UNSO.Google Scholar
  122. Winterbotham, W. 1799. An historical, geographical, commercial, and philosophical view of the American United States, Vol. 2. 2nd ed. London: Printed for the Compiler.Google Scholar
  123. Wyss, J. 2007 [1816]. The Swiss Family Robinson. New York: Penguin Books.Google Scholar

Copyright information

© The Author(s) 2018

Authors and Affiliations

  • Chris S. Duvall
    • 1
  • Bilal Butt
    • 2
  • Abigail Neely
    • 3
  1. 1.Department of Geography and Environmental StudiesUniversity of New MexicoAlbuquerqueUSA
  2. 2.School for Environment and Sustainability, University of MichiganAnn ArborUSA
  3. 3.Department of GeographyDartmouth CollegeHanoverUSA

Personalised recommendations