In seeking to understand African historical geography, and to situate their research within those landscapes and their environmental changes, historians of Africa are likely to ask about the spatial distribution and interconnections of sites of interest, their associated natural resources/land use, and how these interrelate in space and time. Historians may be interested in the onset and duration of climatic changes reshaping biophysical and often social environments, in patterns of environmental resource modification by the intertwined effects of biophysical and sociocultural factors—their enhancement or decline, their husbanding, extraction, depletion, or ultimately unsustainable use. In seeking to know more about such changes, historians might wish to distinguish different drivers of environmental change (biophysical or anthropogenic), different forms of land/resource use, and the sequence and timings of such changes. They might want to know more about the nature and timing of environmental changes associated with (or driving changes in) these relations—for example, the introduction of new species (e.g., the arrival of zebu cattle, Bos indicus, from South Asia, and the implications for transformation of pastoralism) or technologies (e.g., the expansion of iron-working into the Great Lakes region of East Africa and the environmental transformation this wrought).1 Historians might want to look at regional impacts of climatic change, such as the spectacular shifts of Saharan climate and environment over the last 10,000 years, with major implications for population movements, political and economic interactions; or aridification curtailing agriculture on the Nile at the southernmost extent of Rameses II’s Egyptian empire, and the subsequent re-assertion of indigenous Nubian culture in the region.2 Historians need to understand the local impacts of environmental change—whether tectonic or climatic—creating or disrupting physical barriers to movement and/or trading relations between sites, such as the shifting patterns of flooding thought to have affected interactions between peoples of the Sudd or the grassland corridors in an otherwise forested landscape through which West African Bantu groups moved with their livestock into Southwest Africa.3
Answering these questions requires input from experts in a wide range of fields. This article looks at a subset of those fields that may be of use to historical research—at the sorts of information they yield, its potential and its pitfalls. It does not attempt to be exhaustive, nor to expound further upon those approaches that historians have long used—mainstream archaeology, ethnography, historical linguistics, oral histories, and time series landscape photography. Historians already commonly use ancient maps as texts in themselves and as a basis for reconstructing regional-scale changes in polities and movements and increasingly turn to indigenous accounts of landscapes and their changes to understand different place-based perspectives on accounts drawn and/or written by outsiders.4 Despite the rich and often transformative insights such methods have brought to the understanding of African historical geography, landscapes, and environmental change, this article necessarily focuses on a few recent and emerging technologies.5 Because historians need to know how and why we think we know what these lines of evidence tell us, it also addresses some of the ways interpretation of seemingly hard science approaches and findings may be colored by or based on environmental narratives with a political element at their heart.
Basic Methods and Sources
In order for historians to locate and formulate their questions in the context of geography, landscape, and environmental change, this article starts with a thumbnail sketch of African landscapes, the underlying factors determining them, and the environmental shifts they undergo as those drivers change both seasonally and in the longer term, before going on to the methods increasingly used to chart historical change in those environments.
African Environments
There is an underlying biophysical logic to the nature and distribution of African tropical and subtropical environments, stemming from basic relations of latitude, climate, and weather systems, modified by topography, soils, and drainage. From the tropical rainforests of West Central Equatorial Africa, present-day African landscapes grade northward and southward through successive latitudinal bands of subhumid wooded (“Guinea” or “Sudanian”) savanna, through semi-arid grasslands, to arid desert landscapes of the Sahara and, in the South, the Namib desert. Farther still north and south, Mediterranean climates or their southern analogues prevail. Tropical and subtropical ecoclimatic zones are to a great extent shaped and placed by the InterTropical Convergence Zone (ITCZ), where northeast and southeast monsoon or trade winds meet. The ITCZ is created by the heating effect of the overhead sun, causing warm humid air to rise, to cool, and its load of water vapor to condense and fall as rain. The ITCZ and its associated rainfall migrate to the northern tropic in the northern hemisphere summer and back across the equator to the southern tropic, lagging a month or so behind the overhead sun. This creates two rainfall peaks a year at the equator, supporting rain forest or woodland, but grades to a single short “summer” rainfall at each farthest tropic, enough perhaps to support only ephemeral grass growth in arid Sahelian or Zambesian zones. Though currently centered on the equator, the ITCZ has in the past shifted with changes in global climate, extent of the ice caps, and associated northern and southern hemisphere temperatures, driving vegetation belt shifts in its wake. During the Holocene, the ITCZ shifted far north, making the Sahara and Central and Eastern Sudan relatively wetter, more favorable, and more populated environments around 8000 BP. The return of the ITCZ to its present position around 4500 BP restored aridity to the Sahara and Egypt’s Western desert and bimodal rains to East Africa. Population and production system movements accompanying these shifts saw the first domesticated ruminants appear south of the present-day Sahara around 4500 BP.
From east to west, land mass and topography assert their influence over latitude in determining climate. The eastern Sahara merges into the deserts of Egypt, Sudan, and the Horn. The Great African Rift Valley and its associated faulted highlands and volcanics, and the proximity of the arid Arabian peninsula over which the northeast monsoon winds must pass, create a regional climate that is much drier in East than West and Central equatorial Africa. East African landscapes are dominated by arid or semi-arid savanna grasslands and deciduous woodland. More locally, individual highland areas such as the montane massifs of the Sahara, the Adamawa Plateau of Cameroon, and the volcanic mountains of East Africa create their own climatic effects, trapping rainfall that supports wooded or forest ecosystems and creating semi-arid rain shadow in their lee. Similarly, the great low-lying wetlands and lakes like the inland Delta of the Niger, the Sudd, and Lake Victoria collect surface and groundwater drainage, creating local weather and climate and influencing plant and animal and ultimately human production systems. The immense range of variation in local environment and vegetation stemming from these large-scale factors can be captured and integrated in a simple two-by-two matrix (see Table 1).
Table 1. Biophysical Determinants of African Vegetation Types
Open in new tab
Water Availability | |||
|---|---|---|---|
Soil Mineral Nutrients | High | Low | |
High | High biomass (plant and animal) E.g., East African Rift Valley Highlands | Plant growth: seasonal pulse Animals: temporary concentration E.g., Serengeti migration | |
Low | Plant production —low dietary quality —tall growth Low animal biomass Fire climax system E.g., Southeast Africa Miombo/mopane Woodland | Limited plant production Forage quality declines as rainfall increases Low animal biomass E.g., West African Sahel | |
Open in new tab
Locally, plant growth in Africa is primarily determined by plant-available moisture and by soil nutrients (in montane areas, temperature becomes limiting). When moisture and nutrients are both abundant, plant growth is vigorous and animal biomass high. These are production hotspots, perennially contested by different groups for different purposes, like the potentially forested highlands of the Great Rift Valley in East Africa. Where water is scarce but nutrients high, there is a temporary seasonal pulse of high nutrient-quality production during the rains (or, for major wetlands, in the drawdown period revealing waterlogged alluvium). Animals converge on temporarily rich resources, driving often large-scale migrations of wildlife and/or livestock herds, such as dominate the Serengeti ecosystem. Where both water and soil nutrient availability are low, forage quality is poor and plant growth and animal populations are sparse, such as across much of the West African Sahel. High water availability drives plant growth despite low soil nutrients, producing tall forage of poor dietary quality, supporting only sparse animal populations. Left ungrazed, the tall grasses dry in situ and trigger a fire regime characterized by fire-adapted trees and other species, such as constitute the miombo and mopane woodlands of Tanzania and Zambia/Zimbabwe. The simplified schema shown in Table 1 may help historians navigate some of the classic sources such as soil and vegetation surveys carried out by Trapnell during and after World War II and the Platt surveys resurrected by Veronica Berry.6
Methods
Having sketched the broad variety of landscapes and the biophysical logic that underpins them, this article goes on to look at the main techniques used, what historians need to know to understand and interpret these methods, and how and why each method is used. We group methods broadly into (1) macro approaches (aerial photography and remote sensing, geographic information system [GIS]; macro-species assemblages); (2) micro approaches (micro-species assemblages; chemical and isotope analyses for placing evidence in time, through dating techniques, and in space, by locating the origins of materials through their chemical and physical characteristics; genetic analyses tracing biological ancestry); and (3) modeling (climate models, interplay of land use, climate and land cover change).
In a way that must seem to historians curiously decontextualizing, most of these methods can be applied to data irrespective of time depths and timelines, depending only on having the technology and appropriate objects of study available: many offer an understanding of time depths and timelines, while being potentially applicable across a range that far exceeds human history and prehistory. The exception is remote sensing (and aerial or other photography), limited by the recent date of emergence of the technology able to capture images at a moment in time. The other approaches described depend equally on relatively recently developed technologies but focus on dimensions of objects—organic remains, artefacts—which can persist over long periods of time or at least are subject to change in ways and at rates that are understood and open to calibration. With this in mind, the text that follows indicates only briefly the time periods for which methods may be most suited.
Macro Methods in African Historical Geography, Landscapes, and Environmental Change
Remote Sensing and Aerial Photography
Since the mid-1970s, satellites circling the earth have recorded spectral data—different wavelengths of electromagnetic radiation (EMR) bouncing back off (or emitted from) the earth’s surface. Different ranges of EMR carry different types of information on temperature, water content, the extent to which any particular patch of the earth’s surface absorbs or reflects solar radiation, and atmospheric conditions. By combining different spectral bands—different ranges of wavelengths—remote sensing technology generates images of the earth’s surface displaying proxy measures of, for example, green biomass, surface water, and groundwater. Different satellite sensors deliver different types of data—different wavelengths captured at different resolutions and with differing periodicity. For example, real-time satellite data inform us of current and emerging weather conditions; near-real-time Google Earth images make it possible to “walk” through current landscapes of no more than a couple of weeks back. Processed data on wavelengths representing a proxy for green biomass (the normalized difference vegetation index [NDVI] and its more recent congener, the enhanced vegetation index [EVI], both proxies for available forage) are available monthly for most African locations at relatively coarse resolution. For example, a recent partly historical study of gendered forest use among the Loita Maasai of Kenya used fine (30 m) resolution images of global forest loss and gain for 2000–2012 using data freely available online alongside local rainfall records and national population census data. Detailed work on land cover change needs to use very fine-grained resolution images, but existing publicly accessible high-resolution imagery may not be enough to assemble a consistent time series for robust vegetation trend analysis. High-resolution images are less frequently captured and sometimes rendered less useful by large areas of cloud cover, but it is generally possible to access good-quality images representing wet or dry season conditions for the same location for say 5- or 10-year time slices to analyze successive detailed snapshots of change, while publicly available medium-resolution satellite imagery has captured observations of the same area at regular, short intervals making it possible to model and analyze change in vegetation through time.7 Images also show up large and small geomorphological/tectonic and topographic structures (current and fossil river systems, faults and the mountains and valleys that they engender, unstable zones, and salt pans) and human-made structures or the traces left by past such structures (settlements or their past traces, roads, walls, cleared and/or cultivated areas, and structures diverting and/or harnessing rivers like dams, canals, large-scale fish traps, and weirs). Aerial photography, effectively superseded by remote sensing, makes it possible to extend access to comparable data back in time through the 1960s for a subset of surface features, including topography, drainage, woody vegetation, man-made structures, and cleared areas.
Remotely sensed images are useful to historians in several ways. They can show surface features and proxy measures of surface conditions at particular moments in time since 1975 (the point at which such images first become publicly available). For historians operating with non- or poorly literate societies in the present day, remotely sensed images may help give a synoptic view of such landscape features at any given time over the last decades, with which local accounts can be triangulated. For example, remote sensing data records continent-wide and finer-grained regional or local fire occurrence, distribution, and spread on at least a monthly basis. Second, by compiling time series, historians of the present day can see how these landscapes have changed over the last decades. Third, historians can use Google Earth technology to summon up and “walk” through real-time landscapes with their key informants as an aid to gathering oral histories of land and resource use and environmental change alongside a better grasp of place-based perspectives on landscape.
Though remotely sensed images are particularly useful to researchers interested in relatively recent periods, historians interested in deeper time still find remotely sensed images useful in exploring the location, extent, and interrelations of past human-made structures and in visualizing the spatial relations between such structures and particular landscape features—water bodies, rivers, and their past channels, mountain ranges and passes, salt pans, hot springs—all useful context in terms of understanding spatial extent and interrelations of past structures in relation to the biophysical landscape. Just as historians may work with present-day occupants of those landscapes to elicit their histories, they have the potential to do the same with past travelers’ accounts, following their sources as they move through the landscape from one key feature or location to the next, working toward better comprehending what those people saw when they were there and what they may have missed.8 Historians now have at their fingertips an extraordinary technology for visualizing landscapes, sharing insights, and co-producing new understanding as well as new questions, both in grasping key features of the landscape now before them and in understanding how it might have changed in past times.
GIS
Whereas remote sensing and aerial photography are potentially very useful sources of spatial data, GIS represents a system for organizing and analyzing such data. It can be used for visualizing landscape features and inferring environmental change wherever such data contain spatially explicit historical information. Put simply, any data that are georeferenced (i.e., given a precise location) can be stored in layers in a data matrix, which then allows the researcher not only to produce spatially explicit maps but also to explore the spatial associations between different kinds of variables. For example, a researcher might build a GIS for a specified area of interest with separate spatially explicit layers representing topographic features (mountains, valleys, rivers, wetlands), physical infrastructure (roads, settlements, walls, mines), land use and land cover/vegetation type (farmed fields, forest, pastures), administrative boundaries as they currently are and as they were at one or more previous periods of interest (potentially a complete historical record of spatially and temporally explicit boundary changes), distribution of people and of the domestic species and wildlife alongside which they live, and distribution of certain kinds of artefact or materials associated with specific cultural groups or specific provenance. Separate layers can be created both for the present and for past periods wherever those data exist, and such systems can be created for sites of small or large scale. For example, archaeologists might use GIS to organize data on location and dating of the cultural artefacts and biological materials they recover within a particular site. GIS then allows the researcher to look for statistical associations between variables in separate layers, say, between paleo-ecological features and the distribution of Neolithic hearths and kill sites; between locations within an archaeological site and concentrations of particular finds associated with ritual, hearth, storage, or other designated spaces; or between administrative boundary changes and land use/land cover variables. Using GIS requires software and statistical skills, and inevitably the reliability and sophistication of the outputs are determined by the quality of the data and the skills of the analyst, but GIS becomes particularly helpful when handling large volumes of spatially referenced data on multiple related variables.
Dating and Species Assemblages
Anyone working with archaeologists will be familiar with the concept of dating finds, whether plant, animal, or human remains or artefacts or indeed the traces of past geophysical events—volcanic eruption, floods, extended periods of exceptionally warm/wet or cold/dry conditions, exceptionally high or low lake levels. Dating techniques span our macro/micro divide. This article does not deal with the most macro of geological dating techniques—stratigraphy of sedimentary or volcanic layers, whether in cores or in situ, generally used for dating periods spanning millions of years—nor does it deal in any detail with dating by growth rings in corals, potentially spanning time periods of hundreds of thousands of years, or in trees, potentially spanning centuries, but both evidently covering shorter periods than stratigraphic layers. Any durable material laid down through life by living organisms tends to display a layered structure reflecting (typically seasonal) variation in environmental conditions and therefore in growth rate. Corals, trees, and mammalian teeth have all been used not only to count years of deposition but also to chart the progression of seasons and of “good” and “bad” years. These materials have both the organic material making carbon dating possible and the layered structure allowing seasons and years to be counted, potentially offering detailed dating and additional environmental information. Coral reefs, for example, embody growth rings, making it possible to combine oxygen isotope measures (indicative of environmental conditions) with precise dating. These methods are well described, their potential and limitations well understood, and we do not deal with them further here.
Historians will also be familiar with the idea of species assemblages in archaeological remains with, for example, hearths and middens giving evidence of hunted, fished, gathered, or domesticated species in common usage and both the potential of such finds as evidence of particular ways of life and also their limitations, with biophysical, taphonomic (processes by which dead organisms and other objects are deposited, sometimes moved around by wind, water, or soil disturbance, decay, and/or fossilization), or social processes (trade) potentially complicating the picture. Such lines of evidence are commonly used to reconstruct plant and animal assemblages—vegetation formations and animal species’ guilds and communities—with implications for understanding ecological succession and relating these to both bioclimatic and anthropogenic effects in a given area. Species identifications and their presence/absence may give evidence of specific environmental conditions of temperature, vegetation, fire and flood regimes etc.; dating their appearance may give evidence of the introduction of species indicating, for example, communications and/or trading relations with their areas of origin. Such analyses can span periods from the recent past to tens of thousands of years ago. They have been widely discussed in both specialist and popular literature and no further detail is presented here on macro-species assemblages. There is, however, a proliferation of micro methods of dating and of micro-species assemblages with potential applications for historians, which we consider in more detail here.
Micro Methods: Dating, Species Assemblages, Biochemical, and Genetic Approaches
Micro methods in dating estimate time since deposition by measuring the extent to which unstable isotopes or other materials have undergone decay since the object of which they form part was “fixed”—by the death of the plant or animal or by the submergence or burial of the artefact in question. We look briefly at isotopic decay and thermoluminescence techniques here.
Carbon Isotope and Thermoluminescence Dating
Carbon has an unstable, radioactive isotope 14C, formed in the upper atmosphere from nitrogen under the impact of cosmic rays (and nuclear testing). Of all the carbon worldwide, 99 percent is stable isotope 12C, and 1 percent is stable isotope 13C. 14C occurs in only very tiny trace amounts. It decays over time at a known rate, with a half-life of 5,730 ± 40 years. In other words, if you were to take a sample of 14C and revisit it after 5,730 years, you would find half of the radioactive 14C would have decayed (reverted to a stable isotope of nitrogen). This known rate of decay makes it possible to date materials up to around 50,000 years before present. Organisms are continually exchanging carbon (and other elements) with the environment, but at death their tissues cease this turnover and instead “fix” carbon isotopes in proportions reflecting those prevailing in their environment. Evaluating how much of the original 14C has decayed in organic remains makes it possible to date those remains.
A comparable technique, thermoluminescence, uses decay of mineral rather than organic residues to date objects or deposits over periods comparable to, or reaching further back than, 14C dating. Sedimentary deposits contain radioactive elements that decay over time. Grains of quartz, potassium feldspar, or other mineral grains will have been fully bleached, by even short exposure to daylight, prior to being buried in sediment. These mineral grains absorb radiation from the materials in soil or sediment surrounding them. Stimulating these mineral grains using either light of specific wavelengths (optically stimulated thermoluminescence) or heat (thermoluminescence) causes them to emit a light signal as the stored energy is released. The intensity of the light signal emitted varies depending on the amount of radiation absorbed during burial (and hence on time since burial) as well as on specific properties of individual minerals. This technique has been used to date, for example, fragments of fallen rock art buried in the sediment below a rock wall.9
Pollen, Diatoms, and Other Micro-Particles
Micro-dating techniques overlap and dovetail with techniques indicating environmental conditions at the time of deposition. For example, the species assemblages of microscopic plants and animals found in lake or marine sediment cores can accurately establish the history and sequence of environmental conditions in aquatic systems. Biochemical composition of organic materials derived from plants or animals reflects the atmospheric or ecoclimatic conditions organisms experienced during life. Here we look in more detail at the use of reconstructing environments of micro-particles, whether structural remains of microscopic plant (pollen, diatoms) or animal (foraminifera) species, plant phytoliths, plant or sponge spicules and charcoal particles; biochemical signatures and stable isotope ratios of C3 and C4 plants or other organic materials; and genetic techniques for tracing biological lineages.
Pollen grain outer walls are made up of a very strong and chemically inert substance that is well preserved in soil and sediments; pollen grain shapes, sizes, and surface markings are characteristic of the plant species from which they came. Microscopic examination of lake or marine sediment cores makes it possible to characterize the parent species and the proportional species composition of the pollen “rain” contributing to successive layers of sediment. Taphonomic work looking at the way in which present-day pollen rain is constituted and the likely distances traveled by different species’ pollen makes it possible to reconstruct the surrounding landscape and vegetation likely to have given rise to the pollen rain preserved in sediment. These pollen-based (palynological) techniques can thus be used across time periods spanning from the present far back through past geological eras. Though degrees of resolution dwindle with increasing time depth, they offer a useful line of evidence for historians primarily interested in environmental changes within the last 10,000 years. Classic work on East African lake cores has led to a broad understanding of the spatial and temporal variability of East African Quaternary climates and vegetation types and established initial correspondence between higher-latitude ice ages and cold dry conditions in tropical and subtropical Africa, with warm wet periods corresponding to the higher-latitude interstadials.10 Such studies have proliferated. Pollen analyses from the Niger Inland Delta and from the Dahomey Gap separating the main bodies of West African tropical rain forest are disentangling the influences of climate change and human settlement on the landscape. The Gulf of Guinea pollen record—from both sea bottom and continental pollen cores—details a 40,000-year history of the West African equatorial lowland forests, demonstrating the continued persistence of secondary and rain forests in the Niger catchment during the last glacial when dry conditions prevailed, before postglacial warming and increased monsoons over West Africa led to forest expansion.11
Other species- or process-specific microscopic structures can be similarly diagnostic. Diatoms are single-celled algae which form tough, ornate, characteristically shaped external walls of silica, which are preserved on death. With over 200 genera, some 100,000 extant species and an evolutionary history dating back some 65 million years to the beginning of the Age of Mammals, specific diatom species are associated with different aquatic environments in oceans, seas, lakes, streams, soils, and wetlands, making it possible both to date and to reconstruct specific aquatic conditions, including temperature, chemical, and nutrient environment. Recent work by Verschuren on cores taken from Lake Challa, a volcanic crater on the Kenya/Tanzania border, uses pollen, diatom and other measures, alongside carbon dating and stratigraphy, to reconstruct paleoenvironments, climates, and hydrology from around 160,000 BP to the present, demonstrating correlations with, for example, major El Niño/La Niña events, and showing that the area was wet during the northern hemisphere’s Little Ice Age.12 Alongside pollen and diatoms, other resilient organic microstructures include spicules (from plants, or sponges), cuticles from grass, and microscopic fragments of charcoal, all used to reconstruct past environments. Grasses and monocotyledonous plants often produce characteristically shaped phytoliths of silica and/or calcium, readily preserved, and mineralizing, under the right conditions, to opal. Each of these components, their presence/absence and proportional contribution to the overall core, constitute proxy measures of environmental characteristics. Total volumes of phytoliths, such as have been retrieved from oceanic cores off west equatorial Africa, may indicate changing wind strengths. Sahelian dust plumes across the east Atlantic show correlations in the north/south boundary shifts of the Sahara/Sahel/savanna zones, with dust deposition in oceanic sediments and their associated micro-botanical record. Following their first basic classification, broad structural groupings can indicate tropical rainforest versus savanna, or woody cover versus open grassland.
In general, the pollen signature responds to temperature and moisture. Warmer, wetter periods correspond to a spread of forest, with montane forest extending to lower altitudes. Arid periods are marked by the disappearance of forest taxa from pollen cores, a shift to grassland pollen, and, in montane areas, a cessation of peat growth. Despite broad correlations, it can be hard to disentangle the effects of global as opposed to local or regional climatic changes on ancient vegetation patterns; anthropogenic effects complicate this picture further.
Chemical Signatures of Environmental Context
As well as diatom species identifications, the chemical composition of diatomaceous silica gives us information about environmental conditions. Oxygen has a stable heavy isotope, 18O, as well as the more common “light” isotope, 16O. The oxygen isotope composition of resilient organic materials, such as diatomaceous silica, connects through a complex chain of biophysical processes to conditions prevailing at the time of the organism’s death. Water molecules containing “light” oxygen evaporate relatively more easily, while those containing the heavy isotope condense more readily from vapor back to liquid water. Exact ratios vary with latitude, but when temperatures are cooler (such as during ice ages) heavy oxygen water condenses more easily in low latitudes near the equator; water with light oxygen evaporates preferentially, leaving increasingly high concentrations of heavy isotope water in low-latitude water bodies, and diatoms there incorporate higher proportions of 18O. Conversely water vapor with a correspondingly higher proportion of light oxygen eventually condenses and falls on polar ice sheets and high-latitude mountains. By contrast, during warmer, wetter periods with higher rainfall and higher lake levels in tropical latitudes, diatoms there incorporate lower proportions of 18O. At the same time, on a global scale, ice sheets melt returning light-isotope water into circulation. The biogenic incorporation of 18O is a complex process, and the calculations involved in working back to environmental conditions require arcane corrections, often involving triangulation with other elements such as calcium and strontium, whose respective incorporation by organisms also reflects environmental conditions. For oceanic cores, directly comparable methods use foraminifera—single-celled marine animals with a calcareous exoskeleton, whose shape and size and surface markings are again species-specific and characteristic of particular environmental conditions.
Other processes of differential incorporation into organic material help elucidate environmental changes. For example, photosynthesizing plants incorporate carbon, but different groups of plants use different metabolic pathways to do this, and the different pathways are correlated with environmental conditions where these different plant groups thrive. Most plants are “C3.” They function well in relatively cool and wet environments but shut their stomata in dry conditions to reduce water loss, effectively shutting down photosynthesis. A smaller group of “C4” plants, however, are able to fix carbon more efficiently under conditions of drought, high temperatures, and limited nitrogen or CO2. Crassulacean acid metabolism (CAM) plants display yet another carbon fixation pathway, which evolved in some plant groups as an adaptation to arid conditions. C3, C4, and CAM plant remains leave characteristic carbon compounds (three-carbon compounds for C3, four-carbon for C4, etc). They can also be identified by their differential incorporation of different carbon isotopes. Just as there are stable, non-decaying heavier and lighter oxygen isotopes, there are stable, non-decaying heavier and lighter carbon isotopes (13C and 12C, respectively) as well as the unstable, decaying 14C carbon isotope used in dating organic remains. C4 and CAM plants are more likely to fix heavier CO2 with more of the 13C heavier stable isotope. The different 13C:12C ratios that result can be used to distinguish C3 from C4 plants. Each of these pathways leaves an isotopic signature in the organic carbon compounds left when plants die and decompose, preserved in plant micro-particle remains and/or in the tissues of herbivorous animals that ate those plants. Ultimately, soils where C4 plants grew have a heavier carbon isotopic signature than those dominated by C3 plants. Changing C3:C4 ratios may track shifting boundaries between drier grasslands and more humid conditions with higher-density tree cover.
Shifting ratios between woody cover and open grassland shown by pollen or stable isotope signatures can also be correlates of anthropogenic activity, such as the emergence or arrival of new technologies, as well as or instead of biophysical, ecoclimatic causes. For example, the arrival of Bantu iron working, mining, and smelting activities in the Great Lakes region of East Africa had a major impact on woody vegetation, reflected in both pollen signature and charcoal particles. Here and elsewhere, the rise of microscopic particles of charcoal in lake sediment cores helps distinguish such impacts from climate-driven change. Elsewhere, high concentrations of micro-charcoal and other organic waste have generated “terra preta” fertile dark earths indicative of long human occupation, much researched in South America but increasingly identified in, for example, West African forest environments formerly thought pristine.13
Protein Traces and Genetic and Other Molecular Materials
There has been an explosion of lab analytical methods picking up traces of people’s presence, production, and consumption patterns from proteins and genetic materials in the deposits they leave. For example, detection of traces of mare’s milk on the inner walls of containers helped push back the date of horse domestication in Central Asia. Similarly, different plant proteins left on the surface of utensils can be traced and identified to species or genus level; the biochemical composition of human teeth and skeletal remains may not only indicate what that person ate but also identify the geographical areas where that person spent their time, with transformational implications for our understanding of migration, trade, and communication.
One whole field of genetic research deals with chromosomal and DNA analysis. Widely used in forensics, these approaches complement historical work by, for example, tracing the regional and even specific local ancestral African origins of present-day Afro-Caribbean individuals or the rapid westward spread (introgression) of genetic markers from male zebu cattle (introduced from South Asia to East Africa) through sub-Saharan African cattle populations.14 Forensic analysts tend to work on the basis of whatever DNA material they can find at the scene of investigation, largely nuclear DNA making up an individual’s main genome. Research applications of genetic approaches to historical questions have over the last few years moved beyond “general” DNA to focus on the Y chromosome in males and on mitochondrial DNA (mtDNA), which is not part of the nuclear DNA but is found in the mitochondria that power all energy metabolism from their location in the cytoplasm of every cell. The reason for this interest in Y chromosomes and in mtDNA lies in the fact that both types of genetic material remain, for different reasons, apart from the usual process of recombination (mixing of parental genes) that takes place during sexual reproduction. Because the Y chromosome cannot pair up completely with its partner X chromosome in males, a large part is transmitted unchanged through successive generations of male descendants. Similarly, because mtDNA is non-nuclear and uninvolved in recombination, maternal mtDNA is transmitted largely unchanged from generation to generation by its presence in the cytoplasm of the egg. The sperm is comparatively tiny with relatively speaking few mitochondria, so subsequent multiplication in the course of embryonic cell division, growth, and development means the mtDNA of the offspring is overwhelmingly dominated by maternal mtDNA. Both the Y chromosome and mtDNA are subject to normal processes of change by copy error mutation, but this takes place at a far slower rate than changes introduced to nuclear DNA lineages by sexual recombination. The Y chromosome and mtDNA thus effectively allow tracing back through many generations of unbroken patriline and matriline, respectively. While these are tools of fascinating potential in tracing human lineages, their contribution to the history of African geography, landscapes, and environmental change lies primarily in their potential for tracing movements (individual and population) and biophysical ancestries and the appearance and spread of plant and animal species associated with different production systems.15 More broadly, molecular genetics can be used to trace the historical origins, emergence, and progression of, for example, diseases such as HIV as well as “environmental” (usually vector-borne) diseases such as malaria, dengue, or trypanosomiasis.16
Modeling
Historians of Africa often seek alternative methods because of the relative paucity of written records going back in time. Where data are lacking, disciplines increasingly use modeling to work with the few data available to explore possible pasts (and futures) and to establish relatively more or less probable scenarios of change. Modeling can work in many ways, from the purely qualitative and conceptual “if-then”; to quantitative approaches building on statistical associations observed between variables under different circ*mstances; to going beyond simple statistical association to explore causal associations; to extrapolating these into regions lying beyond available data to allow scenario testing. Modeling, by definition, involves simplifying complex situations but offers thoughtful ways to do this through evidence-based understanding of the relative importance of different influences under different circ*mstances and careful construction of the processes and linkages interrelating the action of those different influences.
One example of both the potential and the limitations of such modeling focuses on the role of fire in African historical geography, landscapes, and environmental change. Controlled use of fire in Africa dates back half a million years. African peoples have over millennia used fire to create conditions for hunting and gathering and, increasingly over the last 10,000 years, for clearing farmland and creating pasture to graze domestic livestock. Global models suggest that fire in Africa has, over the last millennium, shifted from being primarily driven by rainfall, with nonlinear effects in different ecosystems, from montane forest (where cool wet conditions limit fire) to savanna (where heavy rain supports correspondingly heavy growth of subsequently very flammable biomass), to being increasingly anthropogenic-driven, to a present and near future where higher temperatures become ever-more-important in triggering and sustaining fires, irrespective of human activity. For the recent past, remote sensing captures fire patterns across Africa from 1975 onward with high spatial and temporal resolution, and GIS analyses make it possible to relate these patterns statistically to ecoclimatic zone, rainfall, settlement, cultivation, and other environmental variables. For historians exploring changing land use, modeling fire histories can offer a rich vein of supporting information. Models of past biomass burning use proxy measures of charcoal in sediment and soil, tree rings, biomarkers, and historical documents, though charcoal data for the African tropics are scarce and chronological and taphonomic issues limit interpretation.17
Ever-more-sophisticated local, regional, and global climate modeling on a grand scale builds on known historical variation in solar radiation; CO2 levels and their radiative forcing effects; ocean currents; ice caps; wind and weather systems; intertropical and other convergence zones; rainfall regimes; lake levels; drainage systems; and evidence from pollen, diatom, and similar micro-studies to reconstruct past eco-climatic environments across Africa. Such models support our understanding of the Sahara morphing from a well-watered, well-vegetated zone some 8,000 years ago, with extensive wetlands and permanent rivers inhabited by “aquatic” fisher/forager societies and traversed by livestock-herding groups showing continuities with modern West African pastoralist societies, to post-4500 BP arid desert with fossil river valleys like the Azawagh and scattered mountain massifs acting as a network of stepping stones for later waves of Berber-speaking peoples moving across the Sahara, pushed southward from North Africa by Arab invaders. On a more local scale, simpler (but still technically demanding) models can explore the local ecoclimatic and social drivers or implications of land use change, forest conversion to farmland, and shifting erosion rates under different local land use regimes, past, present, and future.
Discussion of the Literature
Each of these methods requires sophisticated technical knowledge to apply, and each brings its own problems of understanding the data it generates, their relevance and their limitations. For example, there is a relative paucity of ground truthing of remotely sensed imagery for many African landscapes (nothing at all exists for most historical images), making interpretation of, for example, changing land cover vegetation types potentially unreliable.18 Operational distinctions between vegetation zones (and thus measurement of changing proportions of such zones) are then very much subject to personal judgment. Remotely sensed images give proxy measures of, for example, green biomass or groundwater, and those proxy measures may map closely but not precisely to the desired variable. Cloud cover may obscure locations and periods of interest; abnormally reflective objects (e.g., tin roofs) can corrupt part of the image. There are technical issues around projections complicating accurate measurement of distance or altitude, with such measurements and their calculations requiring expert analytical skills.
Methods relying on analysis of objects, organic remains, or micro-particles must all allow for the fact that they can be transported by wind or water over long distances and are subject to mixing in soils as well as being open to sampling bias. These techniques can pinpoint environmental change but may be silent as to its cause, and without further dating methods may be ambiguous as to timing and sequence. This commentary looks first at the inherent limitations of these physical science and technology-based approaches, then goes on to consider issues of interpretation: the frameworks used to interpret such data are inevitably socially constructed, offering lenses that insert their own biases into the process.
The physical science- and technology-based approaches set out above, from macro remote sensing to micro-particles and isotope analyses, all offer valuable information on African geography, landscapes, and environmental change. Some span millennia; others focus on narrower bands of time past, within the life span of a tree or a coral reef, or home in on the recent past, as with satellite images available from the mid-1970s. The common feature of all is that they map and quantify proxies of the actual landscape states and changes that historians wish to understand. Those proxies map more or sometimes less closely to the actual object of interest. NDVI and EVI pick up electromagnetic radiation wavelengths that correspond quite closely but imperfectly to the presence of green biomass; the ratio of heavy to light oxygen fixed in organic remains tells us at several removes, and after complex correcting calculations, something about temperature and rainfall of their associated carbon-dated past. These are all indirect, indicative measures. The more that independent lines of evidence offered by different techniques can be triangulated, the more likely it is that we can use this information to make meaningful inferences about past environments.
Also, the proxies being considered are all subject to stochastic, probabilistic processes. The proportional makeup of pollen types in a lake sediment core tell us something about the likely composition of surrounding vegetation at different times. But this is just one of many possible results: there are many processes of wind and water and maybe other influences, which mean these proportions derived from a single core may not be representative. Other points in the same lake, or in adjacent lakes, or others across the wider region might be subject to different combinations of those influences, different microclimates, different preservation conditions—resulting in a very different pollen profile, despite having broadly comparable vegetation in their environs. Proxy measures other than pollen—relative proportions of different diatom species, say, or phytoliths—might suggest a rather different picture. Ultimately, it is important to remember that quantitative results based on a small sample may be very accurate as far as that sample is concerned but nonetheless give such a partial and limited picture as to be misleading with respect to the broader state of landscape and environmental change. The best way to deal with this is to use multiple different proxy measures and to take samples from multiple sites: not surprisingly, multi-proxy and multi-site studies give more robust pictures of past landscapes. Only when a consistent local or regional picture starts to emerge confirmed by these multiple independent lines of evidence can one begin to trust the results as robust. Ultimately, one must be very clear about the way in which sampling has been carried out, considering carefully the environmental and other variation it does or does not encompass and bearing in mind the central limitation of these methods: as with any research, quantitative methods can only tell you about the sample(s) on which they are based.
Political Ecology and the Pitfalls of Socially Constructed Interpretation
Beyond the limitations of proxy measures and of quantitative sampling, there remain major limitations of interpretation inherent in the socially constructed nature of scientific understanding. This is of immense significance in understanding environmental change in African landscapes, where multi-layered, emotive, and conflicting concepts of degradation, whether forest loss or desertification of semi-arid areas, continue to cause confusion about present-day processes. As is the case elsewhere, power inequalities have privileged the socially constructed understandings of elites over the lived experience and place-based knowledge of more marginalized local people. Across Africa and throughout the continent’s history, who you are determines whether your knowledge counts. This socio-political bias may lead observers—historians and non-historians alike—to misread landscapes and landscape histories.19
Historians of Africa are familiar with narratives that assume the environmentally damaging and economically unproductive nature of indigenous local land use practices. They are also aware that “development narratives” are shaped by and serve concerns and agendas beyond their ostensible technical remit.20 This overview considers such narratives in relation first to political ecology, outlining the role they may play in consolidating control and justifying interventions, and second, through a discussion of the long-running debate over perceptions of environmental degradation in African drylands, to the emergence of ecological science in western temperate ecosystems, prior to its export to fundamentally different African ecosystems.
To natural scientists, ecology is the study of the interactions between an organism or species and its environment and is an inherently apolitical pursuit. However, since the environment itself is politicized, judgments of its state are inescapably political. Political ecology reflects this dimension and seeks to unravel the political forces at work in environmental access, management, and transformation. It takes account of the hierarchies of power controlling access to/use of natural resources by different people and analyzes the ways these interest groups operate through narratives and discourse, how their perspectives are structured into the fabric of social institutions and governance, and how the power dynamics underpinning different narratives play out in the acceptance or suppression of different people’s knowledge.21
Political ecology has major implications for understanding different perspectives on landscapes and environmental change. For example, forest loss, as measured (relatively objectively) by remotely sensed data and ground truthing, is increasingly understood as a product of multi-scalar political and economic factors, with many actors and drivers, from local to global. Even in remote places, and going back a long way in time, distant markets drive local extraction. Hierarchies of power, access, and control create winners and losers in the process, and, ultimately, marginalized peoples may be scapegoated as the proximate agents of forest loss and/or displaced and pushed further to exploit marginal environments. But forest loss, as confidently diagnosed by generations of colonial and postcolonial administrators commenting on destructive local land use practices, has in at least one case been turned on its head. Fairhead and Leach’s study of savanna forest islands in Guinée showed, through a combination of both natural and social science sources (including old aerial photographs, pollen analyses, ethnography, and oral and archive-based history), that an entrenched narrative of forest loss was instead a story or “reversed history” of forests created and regenerated in an otherwise grassland environment by the action of villagers concentrating nutrients around their settlement, introducing and fostering the growth of desired tree species.22
Political ecology has multiple roots, drawing on theories of common property resource (CPR) management, political economy, and the relationship between knowledge and power; on cultural ecology; and on peasant and gender studies. CPR theories build on the empirical evidence of functioning systems, against erroneous but still influential “Tragedy of the Commons” assumptions.23 Political economy analyzes the ways in which privatization encroaches upon and dismantles commons, creating inequalities through accumulation by dispossession, driving the degradation of both environment and community and creating dependency. Foucault’s insights into knowledge and power lead us to question how scientific knowledge is constructed: post-colonial studies lead us to ask whose science it is. In interpreting technical evidence from quantitative biophysical methods, it is all too easy to be drawn into accepting what seems to be an objective truth but is, in fact, merely one possible interpretation of findings, framed within the dominant paradigm rather than reflecting sometimes more insightful local, place-based, knowledge. Said’s concept of Orientalism applies equally to the externally constructed and projected preconceptions enshrined in outside observers’ perspectives on African environmental processes.24 State elites, both colonial and post-colonial, have been able to exercise cultural and coercive power to pursue vested interests by deploying their perspectives and interpretations of environmental change to demonize local land use practices. Cultural ecology, however, which works through empirical studies focusing on production systems within wider landscapes, has long recognized the value of local ecological knowledge. It challenges development orthodoxies, based on less critical environmental histories, and presents a view of change that includes the impacts of conquest, settlement, introduced species, population dynamics, and rural-urban linkages.
Debates around Dryland Dynamics: Equilibrium Versus Disequilibrium
“Equilibrium theory” is based on the idea that in any population at low density in a favorable environment, individuals can easily find food and shelter. They grow rapidly, mature early, reproduce successfully, and have low mortality rates from, for example, infectious diseases not easily transmitted at low population densities. However, as the population grows, it becomes increasingly hard for individuals to find food, shelter, and breeding sites, and disease transmission increases. Individuals grow more slowly, mature later, and have fewer offspring, themselves less likely to survive to reproduce. Overall, fertility declines and mortality rises. Population growth rate slows, and in due course population numbers stabilize, fluctuating round an equilibrium level or “carrying capacity,” representing a stable and predictable maximum population size given the resources available in that environment. These processes are termed “density-dependent,” as they predict that rates of growth or decline are tightly linked to population density at any given point.
Equilibrium thinking sees whole ecosystems as progressing through a linear sequence or succession of vegetation stages from the colonization of bare landscapes to the development of a climax vegetation type for local conditions. Generalist species with colonizing strategies give way to slow-maturing, slow-reproducing specialists with limited dispersal abilities, adapted to later successional stages and locked into intricate symbiotic interrelations with other species. According to this thinking, species populations that overshoot carrying capacity disrupt climax communities and push the ecosystem back to a lower successional stage, which may be less productive or less diverse. The population crashes and eventually stabilizes at a reduced carrying capacity. In this model, temporarily high populations exert a destructive impact on resources, effectively resetting carrying capacity at a lower level, perhaps through destruction of particular food species or through erosion and loss of topsoil. The boom-and-bust patterns characteristic of both vegetation production and livestock numbers in African arid and semi-arid lands have been understood in this way, and it remains enshrined in most African states’ official policy documents, as well as in South Africa’s environmental science discourse.25
From the 1980s on, however, major challenges to equilibrium thinking arose from both empirical evidence and theoretical developments.26 Equilibrium models are now seen as derived from frameworks developed in relatively predictable temperate ecosystems and therefore as poorly suited to extremely variable and unpredictable conditions. “Disequilibrium” thinking focuses on the empirically demonstrable fact that the lower the mean annual rainfall, the greater the variability and unpredictability of that rainfall in space and time. Plant-available moisture is the main determinant of forage growth in arid and semi-arid environments and is closely determined by rainfall. Up to around 1,000 mm annual rainfall, there is a linear increase in plant biomass production for every unit increase in rainfall, with soil nutrient availability determining how steeply biomass increases.
Arid and semi-arid tropical systems are characterized by extreme variability in rainfall, plant production, vegetation state, and animal biomass, with extreme fluctuations involving rapid build-up, great mobility, and catastrophic die-offs of both wild and domestic herbivores, driven by chaotically interacting random natural events. The observed vegetation states depend on the past trajectories of and current interplay between rainfall, fire, epidemic disease, grazing and browsing herbivore densities and impacts and the recruitment, regeneration, maturation, and decline of woody species. Both wild and domestic herds move over great distances, with migratory and transhumance systems tracking pasture and other resources under these conditions. Temporary heavy grazing may leave bare landscapes, but these can return rapidly to peak production with the next rains. As well as grass growth, woody vegetation cover may vary dramatically between years due to a combination of rainfall, fire, recruitment, browsing, large mammal damage, senescence, and partial, alternative successional trajectories.
Disequilibrium thinking sees African and other tropical arid and semi-arid ecosystems as largely driven by physical factors such as climate or fire, independent of grazer population density, and, as a result, as undergoing chaotic, nonlinear shifts between multiple, alternative, relatively stable states. The more variable rainfall is, the more variable plant production will be and the greater the potential fluctuations in animal numbers, whether domestic or wild. Where there is great intra- and inter-annual variability of rainfall and a close linkage between rainfall and plant production, “carrying capacity” must fluctuate violently within and between years, making the concept of little use in predicting and managing stocking rates in an arid ecosystem. Savannas may then be poorly represented by equilibrium models that predict a natural progression along a sequence of successional vegetation stages toward a single climax, easily pushed to a state of ecological collapse by the impact of heavy grazing pressure. On the contrary, disequilibrium thinking suggests that, while grazing pressure and anthropogenic fire contribute to the observed state of arid and semi-arid rangeland systems at any given time, they are unlikely to be driving any long-term or irreversible trend and are only two of many influencing factors, the most powerful of which are probably density-independent and non-anthropogenic.
Individual case studies of grazing and/or browsing impacts are complex and differentiated. They neither simply refute nor fully bear out one or another view of rangeland processes. The two opposed models may represent extremes along a continuum rather than mutually exclusive theories, and different situations may correspond to different positions along that continuum. The more arid and unpredictable the ecosystem, the more closely it may correspond to a “disequilibrium” model. However, even in relatively less arid systems such as tropical woodland and forest, disequilibrium dynamics clearly has some explanatory power. At the other extreme, the overwhelming forces of a migrant labor economy and settlement restrictions driving the need to invest in livestock in, for example, South African homelands, may have led to such high stocking rates that the conventional wisdom to an extent holds good. Additionally, both models together may contribute complementary insights where either alone could limit understanding.27
These alternative views of rangeland ecology align with political inequalities. Powerful groups commonly invoke equilibrium theory to support claims that local land use (e.g., pastoralism, swidden farming, floodplain fisheries) drives environmental degradation and to justify intervention. Non-equilibrium ideas undermine the expert, superior positioning of the scientist by emphasizing unknowability in terms of predicting the behavior of complex systems. They create problems for conservationists wishing to clear landscapes of people and livestock and return these spaces to an imagined original state of nature.28 By emphasizing the significance of local and historical specificities, they also affirm devolved land use and management as the most appropriate match between people and environment, thus reducing the legitimacy of state-centric, expert-led, top-down policy and planning. In a final twist, equilibrium theory’s hypothetical sequence of overgrazing and degradation has been linked by some to a postulated vicious cycle whereby overgrazing and increasingly bare ground could drive regional climate downturn through the impact on surface temperatures, evaporation, and rainfall, suggesting that overgrazing might, for example, cause Sahelian droughts. Such effects are indeed observed on a very local and minor scale, but the alternative theory, consistent with the very different “disequilibrium” understanding of dryland dynamics and overwhelmingly substantiated and now widely accepted, is that Sahelian droughts and other extreme climatic events of tropical arid and semi-arid Africa correlate closely with El Niño/La Niña events corresponding to the surface temperature anomalies of the southern Oceans.29 The implications of these two alternative theories could not be more different. Rather than causing their own climatic problems by poor land use, overstocking, and overgrazing, Sahelian pastoralists’ livestock and livelihoods are in fact jeopardized by global climatic and atmospheric processes, driven largely by fossil fuel consumption in industrialized countries. The two alternative models suggest diametrically opposed management approaches and lead to very different policy decisions with far-reaching implications.
The orthodoxy of equilibrium thinking drove development interventions and colonial policies across Africa from at least the 1930s and continues to dominate state policy on natural resource management.30 It has led not only to persistent misreading of landscape but also to misguided efforts at rehabilitation that have disrupted ecologies and communities. The struggle over carrying capacity and destocking outlined above is one case in point, but there is a large literature beyond this concerning parallel shifts in the understanding of the dynamics of wetlands and floodplain fisheries, forests, agro-ecosystems, and fire use and about the implications for African landscapes and environmental change of contests between official natural resource management policies and local land use practices.31
This brings us full circle back to the pitfalls of politically shaped frameworks for the interpretation of environmental data and to the relationship between natural science and history. Environmental historians have been using and re-interpreting colonial technical data for some time.32 They have also examined the politics of expertise and the role of experts—the empire of science.33 However, their main interest has tended to lie with the human rather than the natural impact of colonial interventions, in the social rather than mechanical engineering involved in large-scale development schemes and in the resistance stirred up by small-scale interference.34 If historians wish to draw on technical data generated by natural science to critique the scientific bases of state interventions, they must consider not only the political positionality of observers but also the dominance of different narrative and interpretive frameworks through which observations are filtered and processed. This requires a level of cross-disciplinary engagement between social and natural sciences based on a mutual understanding of basic methodology and research techniques this article hopes to promote.35
