Population and Environment

, Volume 35, Issue 1, pp 68–97

High-resolution mapping of rural poverty and famine vulnerability in the Sahel: a possible approach for the Republic of Niger

Authors

    • Department of GeographySaginaw Valley State University
Original Paper

DOI: 10.1007/s11111-012-0180-6

Cite this article as:
Grolle, J. Popul Environ (2013) 35: 68. doi:10.1007/s11111-012-0180-6
  • 526 Views

Abstract

Conventional approaches to mapping poverty and vulnerability rely on statistical techniques for analyzing national census results in conjunction with much smaller auxiliary data sets. The resulting low-resolution maps offer only limited explanations for the causes of poverty/vulnerability, especially those related to the physical environment. Scientists have mapped land cover performance (or “land degradation” or “desertification”) on national scales using satellite remote sensing, but their efforts have rarely been incorporated into poverty/vulnerability mapping endeavors. This paper describes an alternative, comparatively low-cost approach that could make possible poverty/vulnerability mapping at high resolution across the inhabited territories of entire Sahelian countries. The new hybrid approach calls for close collaboration between a team of remote sensing scientists and a team of field workers engaged in social science, earth science, and biomedical research. Maps produced through this approach should help in targeting programs to alleviate poverty and could improve the efficacy of famine early warning and mitigation.

Keywords

Poverty mappingFamine vulnerabilityRemote sensingFieldworkSahel

Introduction

For more than a decade, international development organizations have been engaged in mapping poverty and vulnerability at national and sub-national scales. Mapping specialists hope that geographically targeted aid will reduce overall poverty/vulnerability to a greater degree than country-wide efforts directed at certain disadvantaged socioeconomic groups irrespective of their locations (see Baker and Grosh 1994). Other goals include achieving a better understanding of the causes of poverty and offering a check on political pressures that may divert assistance from the neediest places. Improvements in food security monitoring and in targeting food aid or other emergency response programs are additional potential major benefits (Henninger and Snel 2002; Davis 2003; Devereux et al. 2004).

Most poverty mapping specialists use statistical and econometric techniques to analyze national census results in tandem with accessory, spatially limited data sets such as household income-expenditure surveys. A frequently expressed concern with these methods is that, owing to internal correlations among data sets, the resulting maps have limited capacities to explain the causes of poverty and vulnerability. Additional shortcomings stem from different data limitations. Spatial resolutions are usually no finer than third-tier administrative units (comprising, for example, 4,000–20,000 households), and attempts to incorporate data on physical environments into the cartographic enterprise have been difficult and rather tentative.

This paper outlines an approach that could complement and strengthen both food security monitoring and poverty mapping programs. Ideally, this approach would make possible relatively low-cost poverty/vulnerability mapping at high resolution across the inhabited territories of entire Sahelian countries. It would enhance the spatial resolution of early warning systems (EWS) and the capacity to target relief and mitigation activities. It also presents to the poverty mapping community new possibilities for incorporating data on the physical environment. Leading roles are proposed for remote sensing science and for fieldwork that would require interaction with the rural poor and vulnerable themselves.

The Republic of Niger is the primary focus here because it is listed as next to last (186th) in the global UN Human Development Index rankings and a widespread, highly publicized famine occurred there in 2005. The northern three-fourths of this landlocked country is part of the Sahara, while the southern quarter is mainly the Sahel, a swath of semiarid steppe. Probably 80 percent of Niger’s 16.5 million people live in rural areas and are directly engaged in food production. “Pure” pastoralists (“nomads”), those not involved at all in farming, likely account for less than 10 percent of the rural population. The vast majority are grain farmers who integrate livestock husbandry into their agricultural systems. They live in nucleated villages whose hinterlands differ remarkably in natural resource endowments. Niger suffered a major famine about once every decade during the twentieth century; the famines of 1972–1974 and 1983–1985 captured the attention of international news media. The need for poverty/vulnerability mapping has been stressed by officials of Niger’s Système d’Alerte Précoce (SAP; early warning system), and CARE International and the UN World Food Programme had been planning to initiate mapping projects in the country.

This presentation is based on a literature review and examinations of published and unpublished documents supplemented by selected fieldwork by the author in rural localities of south-central Niger and in a Local Government Area of northern Nigeria. Interviews and correspondence with Government of Niger and other officials also contributed. The next section briefly considers poverty, vulnerability, and possible interactions between them before discussing the significance of the physical environment for mapping. Section three reviews an attempt to map poverty in Uganda primarily with satellite data, and an attempt to use data registered in an incipient Geographic Information System (GIS) for high-resolution poverty/vulnerability mapping in Niger. The results are encouraging, but the greatest unmet need is for fieldwork. The fourth section describes a novel, hybrid mapping approach that would meld satellite remote sensing and field research. Satellite-based maps of land cover performance (LCP) would inform a field research agenda comprising interviewing, environmental surveys of villages’ hinterlands, and assessments of child nutritional status. An example from fieldwork in two northern Nigerian localities demonstrates how interview series and natural resource assessments can detect differences in vulnerability to famine. Section five discusses how this new approach could guide poverty reduction programs, and how it could improve the efficacy of EWS by evaluating a geographic famine model and by identifying sentinel villages.

Poverty, vulnerability, and variations at local scales

In the myriad works on globalization, a common assertion is that absolute location in geographic space is becoming increasingly insignificant. That place continues to exert powerful influences on people’s opportunities for subsistence and advancement is viewed as an obsolete notion in some quarters (see de Blij 2009). The concept of “action space” was put forward in an early attempt to explicate the declining importance of place in the West African Sahel (Painter et al. 1994; see also Manvell 2005, 2006). Intended to controvert national government programs that compel villagers to devote more labor to environmental conservation and agricultural intensification, the concept is defined as “… the geographical and temporal distribution of the ensemble of opportunities and constraints, both local and distant, that individuals exploit and address as they endeavor to survive and improve their lives” (Painter et al. 1994:452). The action spaces of villages, families, and individuals may be situated along a continuum, ranging from small and tightly focused on a few settlements to large and dispersed. Within many localities, commuting for agricultural wage labor, traditional building construction, and trade at rural periodic markets is common. Dry season labor migration to rural and urban places in the West African savanna and coastal zones is the most prevalent type of extra-local movement. Some action spaces encompass mutual support networks between home village families and kin who have relocated to other rural places and to often faraway cities.

Rain’s (1999) thorough study of labor migration in south-central Niger concluded that rural villages’ aggregate vulnerability to famine is not significantly reduced by mobility and kinship ties across space, although individuals’ food security may be enhanced with more widespread contacts and greater options. The need for large action spaces to contend with vulnerability is conditioned to a great extent by villages’ natural resource endowments, most of which could be evaluated and mapped with satellite remote sensing techniques and field research. Before discussing the qualities of village hinterlands and their influence on poverty and vulnerability, it is important to try to elucidate these two phenomena and how they may be related.

A broad international consensus about the nature of poverty and how it should be combated is represented by the eight Millennium Development Goals and 21 associated targets. Oft-cited works on vulnerability emphasize both the risk of exposure to a shock such as drought-induced famine, and households’ resources for coping with, resisting, and recovering from its impacts (Chambers 1989; Watts and Bohle 1993; Wisner et al. 2004). How poverty and vulnerability are related is a subject of considerable disagreement. In Sen’s (1981) analysis, vulnerability is synonymous with poverty mainly because coping abilities derive from households’ “entitlements,” or ownership of tangible resources (labor, land, animals) that can be exchanged for food. Other scholars contend that vulnerability and poverty are not the same. Swift (1989, republished 2006) argued that a broader appreciation for households’ non-material assets can help explain why some poorer people may not be as vulnerable to food insecurity as those with more abundant observable assets. “Investments” in education and environmental stewardship, and “claims” on other families, patrons, and governments (social capital or the “moral economy”), are important assets that should be considered in developing a clearer perspective on the relationship between poverty and vulnerability.

However complex the relationship might be, it seems likely that vulnerability to famine or other episodic shocks is predisposed by poverty resulting from adverse long-term trends. Case study findings, though, are mixed. Watts (1983), for instance, showed that in one northern Nigerian village, the most detrimental and risky types of famine coping—farmland sales and permanent household distress migration—were resorted to by the poorest of three socioeconomic classes. In Darfur during the mid-1980s famine, de Waal (1989) determined that mortality rates were nearly identical across socioeconomic strata, but found statistically significant differences among villages. These were attributable partly to village-wise differences in access to safe drinking water.

The Darfur study documented that wells and other water sources are among the strongest indicators of poverty and vulnerability to famine. Another is the presence in or absence from villages’ hinterlands of (usually) small parcels of clayey, moisture-retentive soils suitable for specialized farming during the long dry season. Land cover in the form of uncultivated bushlands, fallows, and human-fostered trees, hedges, and pastures is a third significant place-based indicator of vulnerability and poverty.

Domestic water problems

Though the exact figure is unknown, Sahelian villages bereft of a perennial source of water for domestic purposes certainly number in the hundreds. Their residents must retrieve water from wells in neighboring settlements. The calories burned in this arduous task are considerable, and the hours required for it otherwise could be devoted to economic activities, including food production (see Whittington et al. 1990, for a Kenya case study, and Mehretu and Mutambirwa 1992a, b, for case studies in Zimbabwe). Where patrilocal marriage is customary, young men in water-deficient settlements may have difficulty finding a wife (Reij et al. 2005); thus, kinship ties with families in other villages, an important dimension of action space, can be restricted. Some households have donkeys employed mainly in retrieving water, but they need to be fed grain, and could be used more frequently for transporting compound manure to farms and for other, remunerative porterage work. Case studies have documented that in communities with chronic water shortages, personal and household hygiene is compromised, diarrheal diseases are more frequent and severe, and malnutrition rates are higher (Tompkins et al. 1978; Nyong and Kanaroglou 1999).

Clayey lands

Soils also vary importantly at local scales. Millet and sorghum, the crops from which people get the bulk of their calories, are grown during the 3- to 4-month rainy season on expansive sandy (and in places rocky) upland soils. But the hinterlands of some villages contain relatively small tracts of moist, clayey soils known in the Hausa language as fadama. They typically occur in and near the beds of dry or ephemeral streams and ponds, as well as in depressions. Some fadama are sources of clay for making bricks and water vessels for sale. During the rainy season, farmers may grow sorghum and corn on their fadama parcels, or cash crops of cotton and rice. At least as important are the irrigated dry season crops of wheat, onions, and other vegetables. If their villages are located within a reasonable distance to a market, farmers with fadama may have the chance to earn cash incomes. Local people without access to these parcels sometimes benefit from opportunities for dry season agricultural wage labor. In some villages with suitable lands, intensifying dry season farming is an important strategy for coping with food shortage. Since the 1970s, work on some fadama farms has been impaired by flooding and alluvial deposits, while others have dried up and sand has filled parts of them (Luxereau and Roussel 1997).

Land cover

In addition to the low-lying clayey parcels, other local land cover variables influence poverty and vulnerability. These include the presence or absence of uncultivated bushlands and fallows, pastures, barren rocky ground, and active sand dunes. In some locales, but apparently not in others, dunes have caused problems for agriculture (Warren et al. 2003; Mortimore and Adams 1999). Hard, barren expanses in a village’s hinterland may, as Milich and Weiss (2000) found, be alarming signs of anthropogenic land degradation. Or they may occur naturally, as exemplified in Kuyewa (Fig. 1), where a lunar landscape stretches from one sector of the village’s outskirts to the horizon. Whatever their origin, such lands are of no imaginable utility.
https://static-content.springer.com/image/art%3A10.1007%2Fs11111-012-0180-6/MediaObjects/11111_2012_180_Fig1_HTML.gif
Fig. 1

South-central Niger and Northern Nigeria showing places referred to in the text and Table 1

Fallows, bushlands, and pastures are the bases of important coping strategies and non-farm incomes. They contain wild food reservoirs, whose composition and abundance vary from locality to locality. Wild food plants account for significant proportions of normal diets in some villages, and assume even greater importance as grain stores dwindle and prices rise, both annually and with the threat of crop failure (Humphrey et al. 1993). After harvest failures, people in some villages have subsisted almost entirely on the wild legume Boscia senegalensis (Delehanty 1988; Grolle 1995). Several wild plant species contain protein, vitamins, and minerals that protect against deficiency and infectious diseases (Freiberger et al. 1998; Kim et al. 1997), while sale of traditional medicines made from specific leaves, bark, roots, and seeds is a source of income for some individuals (see Etkin and Ross 1982). Natural vegetation also is used to make mats, rope, granary covers, kitchen utensils, and other crafts. Firewood and livestock fodder are key resources derived from bushlands and fallows.

Differences in the importance of vegetation-based and other famine coping strategies are evident among villages at similar latitudes (Table 1; Fig. 1). For the people of six sample villages, reliance on wild plant foods is the number one strategy; in six others, it is ranked only as fourth or fifth, and is absent from the coping repertories of six villages. Sale of firewood and fodder is the second strategy in four villages, and in four others is a fourth or fifth strategy. Reportedly, it is not an option in 15 villages. Six villages in Table 1 do not have a common rainy season pasture in their hinterlands. Livestock sales is a top-ranked strategy in only one of these. The eight other villages that rely heavily on livestock sales all have local pasture. In six of the seven Zinder Département villages, though, informants cited the decline in pasture size and the disappearance of or decline in certain fodder species as major environmental problems.
Table 1

Basic environmental data and famine coping strategies, 24 villages in Maradi and Zinder Départements, Niger

Village

Latitude N

Well(s)

Clayey lowlands (dry season cultivation)

Local pasture (rainy season)

Informants’ rankings of strategies’ importance

#1

#2

#3

#4

#5

Bader Kaka

14°43′

Y

N/SC

Y

WF

UF

M

LS

B

Azogor

14°41′

Y

N

Y

M

PT

AWL

PWL

MA

Kuyewaa

14°39′

Y

N/SC

N

MA

WF

M

FA

LS

Sabon Kafia

14°38′

Y

Y

Y

WF

M

LS

FA

Zangon Allegas

14°36′

N

N

Y

WF

M

LS

MA

Zangon Mallama

14°34′

N

N

Y

LS

AWL

WF

L

M

Dan Marke Gaya

14°30′

Y

N/SC

Y

LS

AWL

L

UF

RO

Gidan Illo

14°26′

Y

Y

Y

WF

UF

M

LS

B

Roura

14°03′

Y

N

Y

AWL

L

PT

WF

W/FS

Araouraye

13°55′

Y

N

Y

L

AWL

M

LS

W/FS

Gidan Boka

13°53′

Y

ND

N

WF

UF

UF

M

UF

Dargue

13°52′

Y

N/SC

N

AWL

W/FS

LS

MA

WF

Djoutchi

13°45′

Y

N/SC

N

LS

AWL

L

PT

MA

Kolori Koloa

13°43′

Y

Y

Y

LS

DSF

FA

M

Guidguira

13°40′

Y

Y

Y

LS

W/FS

UF

WF

FS

Foura Guirke I

13°37′

Y

N/SC

ND

AWL

W/FS

M

LS

WF

Tabouka

13°37′

Y

Y

Y

MA

LS

WF

W/FS

AWL

Dan Mallam

13°37′

Y

N

Y

LS

UF

UF

WF

M

Drouma

13°34′

Y

Y

Y

DSF

LS

WF

M

Kaima Peuhl

13°30′

Y

N

Y

LS

W/FS

WF

MS

AWL

Sakata

13°29′

Y

N

N

MA

AWL

C

LS

W/FS

Foura Guirke II

13°27′

Y

Y

N

WF

DSF

RO

W/FS

M

Badetta

13°20′

Y

N/SC

Y

MA

M

AWL

LS

WF

Takassaba Saboua

13°16′

Y

Y

Y

R

ND

UF

DSF

LS

Sources CARE International (1997, 1998)

Y/N yes/no, N/S no, but cultivation of secondary (rainy season) crops including calabash, squash, cassava, corn, okra, melon, AWL agricultural wage labor, B begging, C crafts production intensified, DSF dry season (lowland) farming intensified, FA food aid, FS farmland sale, L loans, LS livestock sales, M migration: expanded participation in/duration of labor migration and/or migration of whole families, MA mutual aid, MS milk sales, ND no data, PT petty trading, PWL pastoral wage labor, R rationing/reducing food consumption, RO religious occupations, W/FS wood and/or fodder sales, WF wild foods, UF uncustomary foods (sauce without millet/sorghum; millet bran; millet bran and wood shavings; millet bran, clay, and milk; grain from dug up termite mounds; livestock carcases)

aAuthor’s fieldwork, Zinder Département villages (2002)

The diminution of bushlands, fallows, and pastures is widely viewed as resulting from farmers’ attempts to feed growing populations by bringing more land under cultivation. In adapting to the loss of uncultivated lands, people have pursued strategies collectively referred to as intensification. One development entails closer integration of farming and animal husbandry as livestock are fed crop residues and weeds inside family compounds. Another practice is the preservation and planting of trees, hedges, and windbreaks on farms to provide firewood, fodder, and building materials. Small enclaves of fallow on individual farms offer similar benefits (Mortimore and Adams 1999; Mortimore et al. 2001; CARE 1997, 1998). Increases in tree cover have indeed been documented in parts of rural Niger (Raynaut 1997; Mahamane 2001), and Olsson et al. (2005) speculated that intensification may help to explain the recent satellite-detected “greening” of the Sahel. Yet there may be problems involved with intensification. Compound feeding with crop residues keeps certain key nutrients from being returned to the soil (Hiernaux and Turner 2002), and the use of millet stalks and dried livestock dung for cooking fuel has been noted with some concern (Mahamane 2001). In central Maradi Département, people apparently were struggling with the transition to more intensive systems (Mortimore et al. 2001). Intensification may have the effect of reducing families’ flexibility in farm management and in their pursuit of non-farm incomes (Adams and Mortimore 1997). In Rain’s (1999) study, informants saw a direct link between loss of bushlands and fallows and diminished capacities for coping with drought in their home villages. In one c. 15-square-km south-central Niger locality, he determined that the more recently established villages were situated farther from wells than the older villages, and had inferior farmlands, fewer trees, smaller fallows and bushlands, and active sand dunes. Rates of participation in seasonal migration were much higher in the newer settlements, especially those founded during or in the aftermath of the 1972–1974 drought-induced famine. Newer settlements also have been the main origins of famine-impelled distress migrations.

Attempts to map poverty/vulnerability with remote sensing and geographic information systems

A review of recent attempts to map poverty/vulnerability using GIS or remote sensing indicates both the potentials and shortcomings of these approaches. The most rigorous effort to map poverty with satellite data was undertaken for Uganda (Rogers et al. 2006; Robinson et al. 2007). Socioeconomic data, derived principally from government surveys, included household expenditures, human population densities, access to markets, tsetse fly probabilities, and domestic livestock densities. Statistical relationships between independent, satellite-detected environmental variables and poverty measures were at least as strong as relationships exposed through the analysis of internally correlated socioeconomic data sets. The maps, produced at resolutions ranging from 1.1 to 110 square kilometers, are predictive poverty “risk” maps. They identify poverty density “hot spots”—relatively small areas having large numbers of very poor people—within six of the country’s districts. Gaps in the socioeconomic data limited the “training set” for establishing relationships between poverty and satellite data, and the maps’ accuracy decreases as their resolutions become higher. No fieldwork was involved in the mapping effort. Less systematic attempts to use satellite data in poverty mapping were discussed by Kristjanson et al. (2005) for a district in southern Kenya and by Legg et al. (2005) for two zones of Nigeria.

During the past decade, a team from Stone Environmental, an international development consulting firm, tried using GIS-registered data to map village-wise poverty and vulnerability for Niger. The goal was to help meet the conditions of the World Bank’s debt forgiveness program, which included reducing poverty by 15 percent by 2015. Documents and the resulting maps are available at www.stone-env.com. The document titled “Using GIS to Help Understand Poverty in Africa” contains the frankest discussion of the project’s shortcomings and needs. One major limitation stemmed from errors in the national census: approximately 1,000 villages were incorrectly located. Wells were included as indicators of poverty/vulnerability, but no data on their status (perennial, seasonal, dry) were available. Villages within five km of a well were considered to have access to drinking water. Another indicator was access to irrigable farmland as defined by various distances from villages to watercourses. Information on the status of the watercourses, and whether the adjacent lands can actually support dry season cultivation, was missing. Distances to health clinics, schools, markets, and roads were additional indicators of villages’ vulnerability. No attempt was made to assign weights to the indicators. On the maps, villages are designated as having 0–2, 3–4, or 5–6 unspecified indicators of poverty/vulnerability.

A better approach for mapping could begin by adopting remote sensing scientists’ methods for discerning and mapping “desertification” or “land degradation.” Whether semiarid lands are in fact being turned into desert by human activities is highly controversial. Using the term “land cover performance” (LCP) sidesteps this often emotionally charged debate. Detecting the spatial variation in response of vegetation to rainfall is the essence of LCP mapping. Such efforts require a team of expert remote sensing scientists, several data sets, and sufficient computing capacity. A fundamental problem is that, owing to satellites’ orbital parameters and to cloud cover, high-resolution data are available rather infrequently, while data available on a weekly or daily basis have much lower resolutions. Several methods have been developed to contend with this problem and others, such as how to estimate rainfall for areas too far from surface meteorological stations. An interactive program of field research and remote sensing work would determine which method(s) is most relevant to poverty/vulnerability mapping.

One promising approach for LCP mapping on national scales has been devised by Hountondji et al. (2005, 2006) for Niger and for Burkina Faso. The only satellite data used were the coarse spatial (7.8 km × 7.8 km) but high temporal resolution data from the US NOAA’s polar orbiters. To assess and map LCP, Normalized Difference Vegetation Index (NDVI), a satellite-derived measure of vegetative vigor, had to be averaged (integrated) over 17 years and analyzed in conjunction with monthly rainfall records from weather stations (128 for Burkina Faso, 109 for Niger). The resulting maps show LCP only in the vicinity (10 square km) of the stations, leaving unmapped the large majority of the countries’ inhabited territories. As suggested by the authors, one means of achieving comprehensive LCP mapping would be to use rainfall estimates based on cold cloud measurements from the European Meteorological Satellite (METEOSAT). Another would be to use kriging, a sophisticated mathematical technique, to interpolate rainfall totals for all of the areas between surface stations. Social science field data could determine which method for comprehensive LCP mapping would be most relevant in mapping poverty/vulnerability on a (low resolution) locality-by-locality basis.

Another approach required data from three different satellites, rainfall data, two research phases, and a team of six remote sensing scientists to map LCP for the entire territory of Senegal (Li et al. 2004; Budde et al. 2004). In the first phase, Li et al. kriged rainfall data from about 40 surface stations to produce a rainfall map of the whole country. This provided the basis for analyzing 16 years of integrated coarse resolution NDVI from the NOAA satellites. The resulting LCP map has a resolution of 8 km × 8 km. To achieve higher resolution in the second research phase, Budde et al. analyzed 1 km × 1 km resolution NOAA NDVI data, which are available only for selected years, as well as comparable data for later years from the French satellite SPOT. Data with 30 m × 30 m resolution from NASA’s LANDSAT were used for detailed evaluation of the preliminary 1 km × 1 km map. The final map shows every square kilometer of Senegal that displayed anomalous LCP (negative or positive) in more than 4 years of the seven-year period of analysis. Further investigations, including site-specific (presumably field) investigations, are called for to assess the 1 km classifications of LCP (Budde et al. 2004).

The maps of Senegal, Burkina Faso, and Niger all show striking differences in LCP within relatively small areas, indicating that vegetation-based coping strategies, and therefore vulnerability, vary on local scales. In poverty/vulnerability mapping efforts, interaction between teams of remote sensing scientists and field researchers should take place as different maps are produced at various resolutions with different techniques. A possible problem is that in recent years, the meteorological services of many African countries, including Niger’s, have been charging prohibitive sums for rainfall data (see Nicholson 2005). Perhaps for a good cause, data could be made available at substantially reduced costs.

An innovative mapping approach

The new hybrid approach to poverty/vulnerability mapping calls for collaboration between teams of field researchers and remote sensing scientists. Fieldwork involves interviewing at increasingly fine spatial scales, and in selected settlements, assessments of natural resources and child nutritional status. The steps in fieldwork, from canton (third-tier administrative unit) center, to locality center, to village, are those required to secure permission for carrying out rural-based research of any sort.

The first step for the remote sensing team is to produce a land cover performance (LCP) map for a section of south-central Niger using coarse resolution NOAA satellite data. Two or three cantons that exhibit large variations in LCP within them would be selected for the first round of field research, which involves series of interviews with groups of local experts in canton centers (large older villages where traditional authority presides). The interview data should be analyzed with reference to the coarse resolution LCP map, and localities then selected for the second phase of fieldwork. Localities are constellations of villages and hamlets under the authority of paramount “chiefs,” who usually reside in the largest of these settlements. Series of group interviews would be conducted in them. After this phase, remote sensing scientists would produce a LCP map based on 1 km NOAA data. The result of this work would inform the selection of individual settlements for intensive fieldwork. They would represent a full range of poverty/vulnerability levels as indicated by the analysis of the LCP maps and data from the canton and locality center interviews.

In the selected villages, series of group interviews are conducted with prominent household heads. An important activity for the groups becomes stratifying village households, including those headed by women, using indigenous terms for socioeconomic status (wealthy, “middle class,” peasant, extremely poor; see Hill 1972; Watts 1983; Lennihan 1987). Household heads randomly selected from the three or four socioeconomic groups are then interviewed privately. Topics include access to fadama parcels (if applicable), how livestock are fed, and sources of cooking fuel. More sensitive subjects concern the extent to which they are self-sufficient in staple foods; relatives in nearby and distant settlements; sources of cash income; coping strategies, including wild foods and help from relatives and friends; temporary migrations, and earnings or remittances from them; sale (or purchase) of farmland or other major assets; and distress migrations. Natural resource surveys in villages’ hinterlands, and assessments of child nutritional status (contextualized by interviews with mothers), would be accomplished while the interviewing with household heads is under way. The third remote sensing effort would integrate 1 km NOAA data, 30 m NASA LANDSAT data, and data from hinterland surveys to develop a training set for linking remotely sensed variables with natural resource and socioeconomic variables. Proposed improvements in the fieldwork and detail on linking satellite and field data are presented in later sections. The next section is a brief account of fieldwork that demonstrates the effectiveness of two activities—interviewing and natural resource assessments—in detecting villages’ vulnerability to famine.

Local knowledge and the physical environment in detecting poverty/vulnerability

Series of group interviews with native experts were organized in two localities east and northeast of Illela (Fig. 1). Selected in consultation with traditional authorities, participants included prominent farmers, local grain and livestock traders, butchers, oral historians, and men who hold various traditional offices. Owing to their occupations and social status, most of these informants travel frequently to settlements within and beyond their localities. A topographic map helped to focus discussions. Community mores in the Muslim far north of Nigeria did not afford women the chance to participate. (In some Niger communities, women would be allowed to participate in group interviews. Women researchers could assemble groups composed of midwives, oral historians, grain traders, praise singers, and prominent farmers. They may be permitted to participate in interviews together with the male informants.)

In the locality centers, settlement history and twentieth century famines were initial foci, followed by the status of wells and clayey fadama lands. The groups then developed village-by-village assessments of land cover variables and the coping strategies based on them, including resort to wild plant foods, fuelwood and fodder sales, livestock sales, and crafts manufacture. They also arrived at village-wise estimates of involvement in dry season labor migration. Two of the most difficult topics were addressed toward the end of the interviews. After sometimes contentious deliberations, the groups came up with what proved to be accurate village-by-village estimates of the prevalence of farm sales and household distress migrations during recent famines. Based on these interview data, 11 additional villages within the localities were selected for group interview series.

Analysis of the interview data found that one of the least vulnerable villages was Amarawa, where intensive field research on other topics had been initiated. The most vulnerable villages were Lakoda and Kadadin Buda.1 Interviewing in Amarawa determined that seven families had migrated in distress during the mid-1980s famine, of which three had returned within 3–5 years. In fact, two families from villages in Niger migrated to Amarawa, as did another family from a village located about 10 km to the east. During interview series in Lakoda, informants revealed that a total of 30 families migrated, and only twelve had returned after the famine. Lakoda experienced no immigration.2

Both Amarawa and Lakoda were founded in pre-colonial times and have populations of similar size (450–500 taxpayers, 3,000–3,500 people). Amarawa is situated on a tarred road three km south of Illela. Lakoda is 15 km to the northeast and seven km from a tarred road. With guidance from a topographic map, basic reconnoitering and transects on foot resulted in sketch maps of the natural resource constituents of the two villages’ hinterlands.

The extent of the villages’ sandy millet- and sorghum-producing uplands is comparable, but Amarawa has fewer farms with rocky soils (fako or debagi in Hausa) and at least twice as many trees. Boundaries between Lakoda’s farms are designated mainly by rock lines, Amarawa’s by shrubs and grasses. Very little fallow or bushland exists in either hinterland, though each has a livestock grazing reserve nearby where farming is prohibited. The reserves are shared with neighboring villages. Amarawa has numerous good wells that yield water throughout the year. Lake Kalmalo, with a surface area of about seven square km, is about two km to the west. Fadama lands extending from the lake’s eastern and southern shores cover approximately 15 square km, and most Amarawa families have access here to one or more small irrigable parcels. The lake’s fishery is a source of incomes and protein. Although the lake reportedly dried up during the 1984–1985 drought and famine, some farmers were able to produce special crops of sweet potatoes and melons on the lake bed and surrounding areas. As is the case elsewhere in the Sahel (Luxereau and Roussel 1997), this fadama has better wild food reservoirs than the uplands.

Lakoda had lands that were suitable for dry season farming, but they dried up following the 1970s drought years. Wells also became perennially dry, and people were forced to retrieve water for domestic purposes from wells in three neighboring settlements (Fig. 2). Table 2 shows Lakoda’s water provisioning profile for 1 day during the rainy season. Whether on foot, by bicycle, or with donkey or camel, a trip rarely takes less than 1 h. Wait times at the wells vary. The total daily time commitment for the village must exceed 1,000 h. Return trips from Gaeti and Lafani are uphill (35–40 m of local relief), although the return from Zangon Lakoda is slightly downhill (15 m of relief) until the acclivity from Lakoda’s outskirts up to its habitations (15 m of relief). The sandy soil is difficult to walk on, and daily high temperatures during the rains average between 36 and 38 °C with high relative humidity. Women and elderly men usually headload a bucket with a 16 or 18 l capacity, and small children a plastic container holding 8 l. Most donkeys are burdened with two 40-l plastic containers and camels with three or four 40- or 50-l containers.
https://static-content.springer.com/image/art%3A10.1007%2Fs11111-012-0180-6/MediaObjects/11111_2012_180_Fig2_HTML.gif
Fig. 2

Lakoda village and neighboring settlements

Table 2

Domestic water provisioning, Lakoda Village one day (late June 1990)

 

Number of trips by

Women/elderly men/small children on foot

Bicycles

Donkeys (managed by boys)

Camels (managed by older men)

Well Location

 Zangon Lakoda

126

1

109

1

 Lafani

112

12

134

17

 Gaeti

44

7

70

14

 Totals

282

20

313

32

Source Field data

The major commitment of time and energy to procuring such a basic human need certainly restricts people’s capacity to exercise famine coping strategies. The lack of fadama and the sparsity of trees and other natural vegetation precluded some options. Opportunities for selling fodder, manufacturing crafts, and consuming wild plant foods were drastically fewer than in Amarawa. The differences in land cover between the two village hinterlands can be discerned on a June 25, 1984 cloud-free 30-m-resolution LANDSAT image.3

Proposed improvements

The main improvement in the interviewing component of fieldwork would be to include women to the greatest extent possible. Improvements in hinterland resource assessments would entail using airphotos and Global Positioning System (GPS) receivers, conducting systematic vegetation surveys, and involving villagers in mapping. Groups of informants have produced useful maps of their villages’ hinterlands that show the approximate extent and locations of various constituents including wells, depressions, pastures, livestock rights-of-way, fallows, active sand dunes, and barren rocky ground (Répubilque du Niger 1997a, b, c). Nutritional assessments of children would be introduced as an additional important means of quantifying vulnerability and poverty.

Hinterland environmental surveys

Hinterland maps produced by villagers and available airphotos (from 1975, and/or 1979, 1984, 1987, 1996) serve as a basis for GPS measurements of larger hinterland constituents. Trees on farmlands can be mapped (Mahamane 2001). Comparisons of the tree surveys and GPS measurements with airphotos will be important in gauging environmental change in hinterlands. Greater numbers of trees, diminished bushlands, and declines in pasture size might be signs of intensification, which could mean less flexibility in coping with drought impacts. Surveys of natural vegetation can be conducted in pastures, patches of bushlands, and any fallow enclaves on farms. In Senegal, Gonzalez (2001) developed a method that enabled him to survey tree species at the rate of one village hinterland per day. At a similar rate, a team of two trained fieldworkers can survey in the uncultivated lands trees, shrubs, and grasses with particular attention to wild food plants and other valuable species.

Nutritional assessments

One-time anthropometric surveys have documented spatial variations in child nutritional status among villages at similar latitudes and within relatively small areas (Table 3). Possible reasons for variations are rarely discussed in the literature. The nutritional status of children in households whose heads were selected for interviewing would be assessed with anthropometry and a Bioelectrical Impedance Analyzer (BIA).4 Interviews with mothers would provide information necessary for contextual analysis of anthropometric and BIA data: season of birth, child spacing, age (determined with reference to a local chronology), recent illnesses, measles vaccination records, health clinic attendance, and use of oral rehydration salts, anti-malarial drugs, antibiotics, and aspirin. The extent to which mothers are involved in farming also may be important.
Table 3

Spatial variation in child nutritional status as indicated by one-time surveys

Source

Region

Year/season

Approximate spatial extent of study

Summary

Tompkins et al. (1978)

Northern Nigeria

1977, April–May (early rainy)

Villages within 100 km2

Significantly greater prevalence of acute malnutrition (<60 % weight/age, <80 % weight/height) in villages with unprotected or dry water sources

Patel (1994)

Western Sudan

1991, March–May (hot dry, early rainy)

Three localities at similar latitudes 100–200 km apart

Locality-wise prevalence of acute malnutrition (<80 % weight/height) ranged from 18 to 38 %

Dettwyler (1991)

Southwest Mali

1989, season not specified

Six villages within smallest administrative unit (arrondissement)

For girls, significant village-wise differences in malnutrition (weight/age, height/age); for boys, no significant village-wise differences

Quelin et al. (1991)

South-central Niger

1987, April (hot dry)

Six localities at similar latitudes on average 50 km apart; maximum distance apart 200 km

Locality-wise prevalence of acute malnutrition (<80 % weight/height) ranged from 6.8 to 16.1 %

Brett-Smith (1984)

Central Mali

1984, September (rainy, but drought year)

10 villages in three localities; localities average 80 km apart

Village-wise prevalence of borderline malnutrition (arm circumference) ranged from 40 to 71 %

Ebomoyi et al. (1991)

West-central Nigeria

Year, season not specified

Villages in wooded savanna locality and villages in adjacent guinea savanna locality; the two groups of villages about 150 km apart

For boys, prevalence of moderate malnutrition (as indicated by five anthropometric measures) 14.9 % in wooded savanna and 27.7 % in guinea savanna; for girls, prevalence of moderate malnutrition nearly the same (13.4, 14.2 %)

Nkamany et al. (1980)

Western Congo (former Zaire)

1978, September (end dry; drought 1977 up to 11/1978)

Groups of villages in six districts; maximum distance between groups 250 km, average 50 km apart

Average district-wise prevalence of acute malnutrition (<80 % weight/height) ranged from 2.1 to 12.0 %

Jacobsen (1978)

Southern Tanzania

1977, May–June (end rainy)

23 villages in six districts; villages in same districts on average 20 km apart

Village-wise prevalences of malnutrition (<80 % weight/age) within districts ranged from 28–31, 33–44, 27–53, 34–55, 54–59, 32–48 %

Seaman et al. (1978)

Southeast Ethiopia

1974, May–July (rainy, following drought year)

44 villages and other sampling points in three districts at similar latitudes; total area 90,000 km2

Average district-wise prevalence of malnutrition (<80 % weight/height) 23.4, 10.9, 12.6 %

Lindtjorn (1990)

Southern Ethiopia

1985–1986; number of months of consecutive measurement ranged from one to 22

19 food distribution sites in Arero province (40,000 km2); five food distribution sites in Borana province (30,000 km2)

In Arero, site-wise prevalence of wasting ranged from 1.81 % to 15.73 %; in Borana, range was 5.35–20.40 %

Measurements of the same children are needed during discrete seasons (December–January, the cool post-harvest period; the hot dry season of March–April; and the peak of the rainy season in July–August). The goal of the series is to determine the severity of the annual “hungry season,” which among farmers corresponds to the rainy season when grain stores are low or exhausted and demands for agricultural labor are at their peak. The relatively few longitudinal studies confirm spatial variations in hungry season impacts. Among 13 rural sites in semiarid West Africa, average weight loss by adults during the hungry season varied from 5 to 2 kg (Ferro-Luzzi and Branca 1993). Two other field studies found comparable differences among proximate villages (Hunter 1967, for adults; Wandel and Holmboe-Ottensen 1992, for children).

The extreme and far-reaching effects of seasonality have been revealed by fieldwork spanning 48 years in rural Gambia. People born during the rains are nearly ten times more likely to die as young adults of cancers and infections than people born at other times (Moore et al. 1997). If season of birth predicts mortality, then it seems logical to propose that the most severe annual nutritional stress occurs in communities that have the highest chronic vulnerability to the impacts of famine and other shocks. Studies supporting this idea have shown that several of the causes of seasonal stress also predispose famine and that famine evolves from hungry season adversity (Ogbu 1973; Glantz 1989).

Developing and evaluating the training set

Various statistical tests could be performed to determine the strength of relationships between or among remotely sensed variables and the village-level field data. Key determinations would include
  1. 1.

    the strength of the relationships between data on domestic water sources and clayey lands and data on coping strategies, temporary and distress migrations, and child nutritional status;

     
  2. 2.

    the relationships between satellite-derived land cover variables, on the one hand, and GPS measurements of dunes, barren expanses, bushlands, pastures, and vegetation survey data on the other;

     
  3. 3.

    how strongly measurements of village hinterland constituents are related to socioeconomic variables, including famine coping strategies and child nutritional status;

     
  4. 4.

    how strongly changes in village hinterlands (as shown by comparisons of GPS measurements and airphotos) are related to coping strategies and other socioeconomic variables and to child nutritional status;

     
  5. 5.

    whether differences in child nutritional status are greater among villages or between socioeconomic classes;

     
  6. 6.

    the strength of relationships between the independent, external satellite-derived land cover variables, socioeconomic variables, and child nutritional status.

     
Satellite-derived land cover performance, whether symptomatic of land degradation or intensification, likely will be related to vulnerability/poverty as indicated by data on coping, other socioeconomic data, and child nutritional status. The statistical analyses should identify different LCP “signatures” of different poverty/vulnerability levels, which would lead to production of a provisional, high-resolution map for the entire Sahel zone of Niger. Localities or villages designated on the map as very, moderately, or less vulnerable or poor could be selected for field research. Several rounds of this evaluative fieldwork with analysis of the resulting data should confirm that certain remotely sensed variables accurately represent different degrees of poverty and vulnerability (Fig. 3).
https://static-content.springer.com/image/art%3A10.1007%2Fs11111-012-0180-6/MediaObjects/11111_2012_180_Fig3_HTML.gif
Fig. 3

Sketch of alternative approach for poverty/vulnerability mapping

Benefits of an intensive mapping approach

After multiple evaluations and refinements, the predictive poverty and vulnerability map of Niger would offer significant benefits. Produced through analyses of satellite and field data, the map could serve as a tool for specifying the causes of and remedies for vulnerability and poverty, rather than simply identifying neglected localities and villages. It also could assist early warning systems (EWS) by improving their monitoring capabilities, and could help with better coordination and targeting of emergency relief.

The map might document that dry wells and desiccated clayey lands are indeed common, strong predictors of poverty and vulnerability. In the Central Plateau region of Burkina Faso, major investments in soil and water conservation, which include contour stone bunds and rock dams for gully rehabilitation, have resulted in rising water tables and rejuvenated wells. Crop yields have increased dramatically, and the additional stalks and natural fodder have helped better the integration of agriculture and animal husbandry. Farmers reported that vulnerability to food insecurity has decreased as a consequence. Covering several hundred villages and 100,000 ha, the costs since the mid-1980s have amounted to about US $ 200 million (Reij et al. 2005). Perhaps costs could be constrained and the benefits more effectively distributed if the map enabled targeting of investments to villages where they would have the greatest impacts.

The map should identify localities and villages where land cover variables significantly contribute to predicting poverty and vulnerability. Diminished bushlands and fallows limit the scope for vegetation-based coping strategies and income-earning pursuits. In some places where this is the case, assistance in diversifying economic activities may be the only realistic approach for reducing vulnerability and poverty. The possibility that agricultural intensification may be restricting opportunities for diversification and limiting action spaces would require more investigation (Adams and Mortimore 1997; Mortimore and Turner 2005). Another possibility for redressing poverty or vulnerability linked to land cover would involve place-specific modifications of the “Code Rurale.” To minimize disputes over land, the code stipulates that uncultivated parcels will belong to those who “put them in value.” Some people now feel compelled to clear bushlands and limit fallows, but relaxing the code could help restore vegetation-based coping strategies.

USAID’s Famine Early Warning System Network (FEWS NET) is the most comprehensive and sophisticated EWS operating in Africa. Criticisms of FEWS NET center on the low spatial resolution of its assessments of different current vulnerability levels, and limited knowledge of the relationships between indirect indicators (grain and livestock prices, meteorological satellite rainfall estimates, coarse resolution satellite measures of vegetation) and actual conditions at the village level. Resolutions usually are limited to second-tier administrative units (arrondissements in Niger) and/or to “livelihood zones” (a total of 10 have been specified for Niger). In 2005, the system was criticized for not having detected famine in the country’s most populous livelihood zone, the southern grain-farming region (Crombé 2009).

The Niger poverty/vulnerability map could help to improve early warning and mitigation by identifying “famine epicenters,” the primary elements of a geographic famine model. This model holds that famine begins in some communities before it evolves in others, and from these probably numerous but identifiable epicenters of high chronic vulnerability, famine diffuses to become a regional crisis as families migrate to nearby settlements to find food and water. In some cases, refugees tax kinship support networks and village hinterland resources to the point where host communities no longer can sustain some of their own members. Then, both refugees and some host community families must seek food from farther afield. Evidence for this first, locality-scale phase of diffusion is presented in Table 4. An alternative to the geographic model is that centers of famine involution exist where people lacking in situ coping strategies or social capital in other settlements have no choice but to sell assets and purchase expensive grain.
Table 4

Data indicating the spatial diffusion of famine through inter-village distress migration

Source

Place

Time period

Approximate distance covered

Summary

Cutler (1984)

Northeast Ethiopia

1982–1983

50–75 km

Migration from famine “epicenters” to more prosperous areas on periphery of “famine zone”; low grain prices at these destinations began to rise, reportedly due to aggregate demand, however constrained, of refugees

Brooke (1967)

Northwest Tanzania

1943–1950

50–100 km

Family migration to unaffected rural areas; refugees worked on hosts’ farms in exchange for share of food crops, and/or were allowed to clear bushland for own farms

Vaughan (1987)

Southern Malawi

1949

10–80 km

As community-based support deteriorated, groups of men and sometimes women migrated to kins’ villages where they were given food and opportunities to work for food; “many” died during ordeal; migration implicated in smallpox epidemic

Colson (1979)

Southern Zambia

1940–1973

N/A

Migration to more productive areas to sell crafts and livestock at higher prices; family migration to kins’ villages for assistance

de Waal (1989)

Western Sudan

1984–1985

100 km

Mass migrations from northerly famine-affected villages to villages farther south; refugees encamped in host villages’ margins engaged in alms seeking, low status trades, agricultural wage labor; health crisis observed; most refugees returned home at start of 1985 rains

  

1983–1986

100 km

Mass migrations from villages to “famine camps” near small town; later arriving refugees displaced those who arrived earlier from low status economic niches; health crisis observed; nearly all 1985 arrivals returned home

  

1984–1985

Local inter-village

Families left several villages due to lack of food and dry wells; refugees received food from host village, which itself was receiving food assistance from relatives in villages c. 250 km to the south

  

1984–1985

Local inter-village

Refugees encamped on host village outskirts received external food relief (mediated by hosts); destitution averted, but health crisis observed; most refugees returned home in 1985

Chastanet (1983)

East-central Senegal

1914–1945

Local inter-village; beyond locality

Inter-village assistance adequate for coping with relatively brief subsistence crises of 1917–1918, 1919–1920, 1920–1921; longer crisis required expansion of “zone of survival,” that is, the need to migrate farther from home village to find food

Turton and Turton (1984)

Southwest Ethiopia

1983

Local inter-village

Study village “descended upon” by kin from nearby villages that had experienced lean harvests; hosts suffered as a consequence of having given away substantial portions of own average harvests

Autier et al. (1989)

Northwest and central Mali

Mid-1980s

Local inter-village

Families migrated to neighboring villages to seek food; strategy reportedly not effective if crisis more severe, widespread

Spittler (1977)

South-central Niger

Early 1930s

Local inter-village

Migrations after poor harvests to seek support in villages with more food

Faulkingham and Thorbahn (1975)

South-central Niger

1969–1973

20–50 km

Families from four villages migrated to kins’ villages in northern Nigeria

Mortimore (1989)

North-central Nigeria

1973–1974

Local inter-village

20 of 71 villages surveyed reported receiving migrants from nearby villages during famine

  

1984

80–200 km (?)

After harvest failure, 22 families from Niger villages arrived at kins’ small village (pop. 750) where harvests had been average; refugees survived initially through charity, water carrying, fuelwood and fodder collecting; plots for compounds allocated immediately to refugees; fallow lands were expected to be allocated by start of 1985 rains

Delehanty (1988)

South-central Niger

1984–1985

25 km

Retreat of Aseveral@ families southward to reside with kin; refugees’ village of origin founded c. 30 years before by migrants from host village

Grolle (1995)

Northwest Nigeria; south-central Niger

1913–1988

Local inter-village to 300 km

Migrations to seek relief and/or employment from kin, friends, or patrons; interviews revealed (1) loss of life in encampments on villages’ outskirts; (2) emigration of some host village families attributed to refugee influx

N/A Not available

Evidence for a subsequent, broader diffusion phase comes from research in central Ethiopia, where the early 1980s famine spread sequentially through five regions. By mid-1983, an historically famine-free “corridor” had been overwhelmed by a massive influx of refugees from more vulnerable places (Pankhurst 1992; Mesfin 1984; Kumar 1990). In Niger during the mid-1980s famine, an estimated 400,000 people migrated in search of water and food (UN in Timberlake 1985). Informants in villages in the more favored agricultural environments of southern Niger attributed the severity of that famine to the presence in their communities of refugees from northern farming areas (USAID 1992; field data from the villages of Droum and Guidguir; Fig. 1). Archival sources and field research implicate refugee movements in the diffusion of famine in northern Nigeria (Van Apeldoorn 1981; Grolle 1995).

Distress migration by nutritionally compromised people and their exposure to infectious diseases in refugee congregations are key factors in the health crisis model of famine mortality (de Waal 1989; Shears and Lusty 1987; Toole and Waldman 1990). Continuum models of the temporal progression of famine had specified family distress migration as a last-ditch strategy resorted to after productive assets had been liquidated in a buyer’s market in order to purchase increasingly expensive grain (Watts 1983; Corbett 1988). Early warning systems had used such models to justify their emphasis on monitoring grain and livestock prices as recorded at regional markets (Hutchinson 1998). Later work demonstrated that coping strategies are not discrete and neatly sequenced, but exist in clusters, with several strategies being taken up at the same time (Devereux 2001). Moreover, family distress migration has occurred early on during a crisis in a deliberate attempt to preserve the productive assets that constitute the basis of people’s livelihoods in their home villages (de Waal 1989; Grolle 1995). Though migration may be effective in helping families safeguard their future livelihoods, it also accounts for the spread of disease and mortality.

Some researchers have suggested that near real-time data from sentinel communities would significantly improve EWS (Jaspars and Young 1995; Davies 1996; Adams et al. 1998; Mock et al. 1993). Devereux et al. (2004) have advocated nutritional surveillance in selected communities as part of a comprehensive vulnerability mapping program. Earlier and more geographically precise warnings of famine gestation could be developed if village-level data were analyzed together with grain and livestock prices and other coarse resolution indicators. Yet little has been written thus far about how sentinel communities should be chosen and with what justification. Research for the Niger poverty/vulnerability map would have identified villages that were epicenters during recent famines and perhaps additional villages that were centers of famine involution. The map itself would enable rational, nationwide predictions of epicenter communities and other highly vulnerable ones. In Rain’s (1999) study locality, epicenters might be found among the newer, resource-poor settlements, and an older, less vulnerable village could be monitored for comparison. In other areas, epicenter status and lesser degrees of chronic vulnerability may be attributable to different factors.

If the geographic model satisfactorily explains the gestation and spread of famine across space, then a potent relief response targeted at epicenter communities might arrest famine diffusion and disease epidemics. An Agence Française de Developpment report on the 2005 Niger famine documented that social targeting (i.e., of “vulnerable groups”) was too complicated and not widely attempted, while geographic targeting was poorly coordinated and impeded by local politics (Olivier de Sardan et al. 2007). The worst example was of one village that received food aid from different NGOs on five occasions, though many villages in the vicinity received nothing.

Summary and conclusion

Several approaches for mapping degradation or land cover performance have been employed by remote sensing scientists. While the relationship between land cover and poverty or vulnerability is not yet widely established, the remote sensing maps, if evaluated and refined through analysis of field data, could serve as the basis for extensive poverty/vulnerability mapping. The Uganda mapping effort demonstrated that external, independent remotely sensed land cover variables are at least as effective in mapping poverty as the conventional statistical techniques. Village-level fieldwork in northern Nigeria showed that two environmental variables—wells and clayey lands—may have robust predictive powers.

The suggested mapping approach should confirm that location is strongly related to poverty and vulnerability. Systematically demonstrated relationships between land cover variables and poverty/vulnerability would illuminate the role of natural environments in rural people’s subsistence and survival strategies and in their general material well-being. Location likely explains, to a great extent, the necessity for people to expand action space as well as their capacity to do so. Analysis of interview, land cover, and nutritional survey data would determine whether location better accounts for poverty/vulnerability than membership in a particular socioeconomic class. Links between high levels of poverty/vulnerability and marked changes in land cover since the 1970s might indicate the need for altering the Code Rurale’s provisions regarding uncultivated arable lands. Sahelian governments’ emphasis on village-level conservation efforts might be justified where significant declines in land cover are associated with higher poverty and vulnerability.

The suggested mapping effort would yield important benefits to famine early warning and mitigation. It would make possible objective and strategic selection of sentinel communities, and spatial resolution refined to the village level could enable earlier warnings of famine gestation. The map would also provide an objective basis for targeting food relief and other mitigation activities. Devereux et al. (2004) argued that inclusion of nutritional monitoring in a food security mapping system would result in convincing and timely warning of incipient famine if seasonal, site-specific baseline data were available. They estimated that developing a comprehensive mapping program for the “average” poor Sub-Saharan African country would cost US $ 1–2 million per year over 5–10 years. The project outlined above could be accomplished for much less. Remotely sensed data are available at little or no charge, and salaries for fieldworkers would be modest. Compensation for remote sensing scientists also would be a relatively minor expense. The greatest expense would be for vehicles, as many study villages would be located away from improved roads.

Footnotes
1

Kadadin Buda is located just east of the Local Government Area within which permission to conduct fieldwork had been secured. In a related research effort focused on former famine refugees, eight household heads from Kadadin Buda were interviewed in a village they had helped to establish in the Nigerian “Middle Belt,” a well-vegetated, relatively sparsely populated ecological zone south of the Sahel. They recounted that environmental conditions in the Kadadin Buda hinterland were similar to those in Lakoda.

 
2

The same methods of interviewing and environmental assessment were employed in localities in Zinder Département, Niger. The results point out some complexities and underscore the need for fieldwork. For example, the collection of hamlets known as Zongon Allegas has no permanent water source or fadama. Camels and donkeys are used to retrieve water from Sabon Kafi, a round trip of 15 km. The hinterland is punctuated with rocky outcrops, and vegetation is remarkably sparse. Wild food plants were quickly depleted as the mid-1980s crisis began. More than half of this settlement’s population reportedly migrated in distress. Zangon Mallam also has no fadama, and wells are located 2–3 km away. The land is considerably less rocky, though, and vegetation, including some wild food and fodder species, is more abundant. The residents’ main famine coping strategy was to move to the larger, nearby village of Gangara, where water and opportunities for wage labor on fadama farms were available. Distress migration reportedly was limited to three families.

 
3

Web-based access to the entire NASA LANDSAT archive began in August 2009. LANDSAT scenes that used to cost as much as US $ 1,200 are now free to the public. Interested readers can log on to landsat.usgs.gov. Using the GLOVIS function, they can call up the c. 185 km × 170 km scene that includes the northern Nigerian study localities. The scene ID is LT51910501984177XXX03. The coordinates for the center of Lakoda village are 13.80 N, 5.38 E; Amarawa’s are 13.71 N, 5.30 E. Analysis other than basic inspection would require familiarity with ArcGIS or comparable software.

 
4

The Bioelectrical Impedance Analyzer (BIA) is a durable, hand-held device for non-intrusive assessment of nutritional status. It measures, to three significant figures, basal metabolic rate, intra- and extra-cellular water content, and lean v. adipose tissue mass. Anthropometric data fortified with BIA data detect deteriorating health and nutrition sooner than anthropometrics alone. (Glew et al. 2001; VanderJagt et al. 2000; VanderJagt et al. 2001).

 

Acknowledgments

Many thanks to David Helgren and Evelyn Ravuri for their very helpful comments on earlier versions of the manuscript. The criticisms of Population and Environment’s editor and anonymous reviewers proved to be very valuable as well. Thanks also to Scott Youngstedt for sharing a document from Niamey, and to Ellen White for preparing the figures. Among the several hundred Nigerians and Nigeriens who helped to make this work possible, Mohammed Iliya (Department of Geography, Usman Danfodio University, Sokoto), Bagudu Natilli Kalgo (field assistant), and Mahamane Goni Boulama (consultant) deserve special thanks. Fieldwork in northern Nigeria was supported by a Fulbright-Hays doctoral dissertation award. The National Geographic Society’s Research and Exploration Committee funded research in Niger.

Copyright information

© Springer Science+Business Media, LLC 2012