Reading the Rains: Local Knowledge and Rainfall Forecasting in Burkina Faso

Carla Roncoli (University of Georgia), Keith Ingram (University of Georgia), Paul Kirshen (Tufts University).

Abstract: This paper describes how farmers of Burkina Faso predict seasonal rainfall and examines how their forecasts relate to those produced by meteorological science. Farmers’ forecasting knowledge encompasses shared and selective repertoires. Most farmers formulate expectations from observation of natural phenomena. Cultural and ritual spiritualists also predict rainfall from divination, visions, and dreams. Rather than positing local and scientific knowledge as self-exclusive, our research shows that farmers operate in multiple cognitive frameworks. Moreover, they are interested in receiving scientific information because they perceive local forecasts as becoming less reliable as a result of increasing climate variability. Some aspects of local forecasting knowledge, such as those stressing the relationship between temperatures, wind, and rainfall, can help explain meteorology-based forecasts. But significant discordance remains between scientific and local forecasts. The former predict total rainfall quantity at a regional scale, whereas the latter stress rainfall duration and distribution, and are more attuned to crop-weather interactions. Local systems of thought stress the relationship between knowledge and social responsibility. This emphasizes the need for scientists to integrate information dissemination projects with efforts to improve farmers’ capacity to respond to forecasts and to cope with suboptimal climate impacts.

Keywords: farmer knowledge, indigenous meteorology, seasonal rainfall forecasts, climate variability, environmental indicators, Burkina Faso.

Introduction

Communication of scientific information to lay users must hinge on a thorough understanding of pre-existing cognitive systems (Kempton et al. 1995). In order to guide ongoing efforts to disseminate meteorology-based climate forecasts, this paper presents ethnographic data on rainfall forecasting knowledge held by farmers of Burkina Faso. Attention to differences and similarities between farmers’ and scientists’ forecasts can help introduce the latter in ways that are culturally appropriate and locally relevant. Comparison between forecasts can also generate important theoretical insights into the relationships between knowledge systems.

In exploring the interface of different knowledge systems we do not embrace a rigid dichotomy of self-exclusive categories, such as technoscience vs. ethnoscience, modern vs. traditional, or western vs. indigenous (Agrawal 1995, Nader 1996b, Strathern 1996, Sillitoe 1998b, Gray 2000). Although aspects of farmers’ knowledge may challenge deep-seated assumptions or well-founded theories that underpin modern scientific rationality, local knowledge need not be posited as an incompatible alternative to science (Nader 1996a). Rather, evidence shows that local knowledge can and must be integrated with research-generated information and technology in efforts to improve rural livelihoods (Brokensha et al. 1980, Richards 1985, Gladwin 1989, McCorkle 1989, Thrupp 1989, Flora 1992, Moock and Rhoades 1992, Warren et al. 1995, Chambers 1996, Sillitoe 1998a).

Terms such as ‘scientific’ and ‘indigenous’ remain problematic, but for brevity’s sake we will clarify our terminology without delving into the surrounding debate (Antweiler 1998). We adhere to the common definition of science as knowledge generated by experts using recognized and rigorous approaches to observation and experimentation. We avoid acronyms, such as ITK (indigenous technical knowledge) or TEK (traditional environmental knowledge), that tend to reify diverse and fluid cognitive dimensions into an inflexible package of disembodied know-how. We avoid ‘indigenous knowledge,’ which connotes colonizing discourse and policies in much of Francophone Africa. Rather, we prefer the terms ‘farmers’ knowledge’ or ‘local knowledge,’ which evoke the performance element of knowledge and the contextual aspect of its practice (Richards 1993).

The term ‘local’ need not isolate African communities from the fast-evolving and far-reaching west (Hobart 1993, Appadurai 1995, Fardon, 1995). Rather, ‘local’ stresses the site-specificity of farmers’ environmental knowledge and the place-based dimension of cultural and social meanings associated with such knowledge. Farmers’ knowledge is intimately interwoven with local and regional processes of ecological and historical change that impinge on local landscapes and lifeways. These processes shape fields of power where different world views intersect and claims to authority are made by social actors as they negotiate rights and responsibilities in a context of growing uncertainty (Long and Long 1992, Thompson and Scoones 1994).

Frameworks of global and local knowledge

In recent years, meteorological science has made enormous progress in predicting climate. The realization that sea surface temperatures (SSTs) influence global atmospheric circulation enables scientists to formulate forecasts of seasonal rainfall. These are presented as the probability of the seasonal rainfall being in the above normal, below normal, or normal compared with historical trends. In West Africa seasonal rainfall relates to the three months of July, August, and September, during which 90% of total annual rainfall occurs.

The reliability and potential utility of this information varies according to geographical region, that is, the area’s position in relation to oceanic-atmospheric circulation, and according to the political, economic, and social context which shapes its dissemination and application (Orlove and Tosteson 1999, Stern and Easterling 1999). In Africa, analysis of the relationship of seasonal rainfall forecasts to crop yields and food security has shown mixed results (Phillips et al. 1999, Broad and Agrawala 2000). Even in Latin America where the predictive capacity of SST-based forecasts is greater than for other regions, effective integration of forecasts into local production and livelihood systems remains a challenge (Broad 2000, Nelson and Finan 2000, Otterstrom 2000).

In West Africa, efforts to disseminate and apply forecasts are at an experimental stage. At the forefront of these efforts, the Climate Forecasting and Agricultural Resources (CFAR) project has been conducting multidisciplinary research to identify points of entry and feasible modalities for integrating seasonal rainfall forecasts into agricultural production systems of the Sahel-Sudan region (Ingram et al. 2002). This paper reports a part of the CFAR project research conducted among farmers of the Central Plateau of Burkina Faso between 1998 and 1999.

Bonam village is in the heart of the Central Plateau, about 100 km northeast of Ouagadougou and 18 km northwest of Boulsa, the administrative center of Namentenga Province (Figure 1: INSERT MAP). The UNDP rated Namentenga Province among the poorest in the country, it suffers from economic marginality, poor transport infrastructure, land degradation, and extreme demographic pressure (PNUD 1997). Most of Bonam’s 2,800 inhabitants are Mossi, the largest ethnic group in the country, and most make their living from rainfed agriculture.

Rains fall during a single season, lasting from May to October, characterized by extreme inter-annual variability and non-uniform distribution within seasons. Mean annual rainfall (1964-1998) is 674 mm, but year-to-year variation is great. For instance, Boulsa received 1050 mm rain in 1994, but only 491 mm in 1997, the second driest year in three decades. In 1998 seasonal rainfall was 617 mm, near the historic average.

Farmer production strategies primarily manage risks and reduce losses by diversifying field locations and cropping systems and by minimizing investments. At the onset of the rainy season, farmers make crucial decisions about what, when, and where to plant. Based on their expectations for seasonal rainfall, they strive to match the water retention capacity of different soils and landscape positions with the water requirements of crops and crop varieties in ways that suit a variety of rainfall scenarios. Farming systems center around cultivation of grain crops (sorghum, millet, maize, rice) and leguminous crops (peanut, bambara nut, cowpea, sesame), grown on non-contiguous fields with household labor and low levels of external inputs (Roncoli et al. 1999).

Although subsistence agriculture remains the primary occupation, Bonam’s social and cultural landscape is diverse, including a wide spectrum of knowledge frameworks. These knowledge frameworks include different ethnic traditions, along with such modern hallmarks as development ideology, agricultural extension, medicine, formal education, and monotheistic religions. Bonam is situated a road in poor condition that was formerly an important link between northern livestock production areas and the market town of Puytenga, which attracts traders from neighboring countries. Bonam has a health post, maternity ward, and primary and middle schools that receive funds and supplies from French and Dutch church and school groups. Several NGOs have been active in development and relief interventions, including school feeding programs and delivery of low-cost grain during famine years, such as 1998.

The religious landscape is characterized by a relatively peaceful, somewhat overlapping coexistence of customary beliefs and monotheistic religions. For the majority of the Mossi population, spiritual life revolves around lineage ancestors and earth spirits that mediate the relationship between human and divine. Spirits may inhabit specific places or fetishes, such as hills, rocks, and groves, and are believed to intervene in human affairs, sometimes arbitrarily, sometimes as judgment for moral transgressions. Senior men (buud kasma) are responsible for performing sacrifices honoring their family’s ancestors. Overall spiritual authority rests with an earth-priest, or Tengsoba (teng = land, sob = lord, owner). He is the eldest member of the clan (Nyononse) that originally settled in the area and has ritual authority over the land. The Tengsoba of Bonam is very influential because his ancestors established the town of Boulsa, so he retains spiritual jurisdiction over the Namantenga province. But some residents, who attribute problems affecting the village, including drought, to his failure to adequately perform sacrifices to the ancestors, are contesting the legitimacy of the present Tengsoba.

Village sections surrounding the market are inhabited mostly by Muslims and host small mosques and practicing marabouts (Islamic clerics). In the nearby Fulani village of Tangyoco, a regionally famous Islamic sheik ministers from a mosque constructed with Saudi funds, runs a Franco-Arabic school, and dispenses blessings to pilgrims, which include national politicians. The Catholic Church in Bonam is the largest building in the village, where a local deacon, or occasionally French priests from Boulsa, conduct services. Bonam also hosts a chapel and pastor for the Assemblies of God, generally known as the "Protestants." They draw converts from the ranks of educated and ambitious young village residents, who are attracted by the leadership opportunities offered by prayer groups and by its criticism of traditional practices, such as arranged marriages.

Regardless of role or rank, everyone in Bonam engages in agriculture, though not all perform the actual work themselves. Given the high cost of staple grains and falling purchasing power of local currency, even the few salaried workers in Bonam must farm to make ends meet. Hence, everyone has a stake in the nature of rainfall and eventually acquires or develops ways of linking perceptions and expectations regarding the season to come.

Like scientific inquiry, local rainfall forecasts rely on observation and interpretation of specific phenomena. Such phenomena may occur in the surrounding landscape, including trees, animals, and sky, or they may be spiritually manifested in the form of divination, visions, or dreams. Environmental indicators are based on experience and can be learned by anyone who listens to elders, although work spaces and vantage points of daily life shape one’s ability to observe and interpret these signs. Generally, elderly male farmers are considered to know more than younger men or women farmers, but knowledge varies greatly among elders. In addition women may note different phenomena than men and herders may seek different information than farmers. Spiritualists, who have inherited powers or acquired skills by virtue of initiation or election by the spirits, hold knowledge that is not available to all. Shared knowledge and spiritual knowledge are neither clearly defined nor mutually exclusive categories. Some elders have reputations for expertise in reading stars based on keen scrutiny and spiritual insight.

Shared forecasting knowledge

Environmental indicators that farmers use to predict the coming rainy season become available for observation at different times of year, beginning immediately after harvest and continuing until into the new rainy season. Indicators on which farmers rely most are fruit production of certain trees at the onset of the rainy season and temperatures during the dry season. They also observe the intensity and direction of winds, and behavior of birds and insects throughout the year. In a survey of farmers (N=23) conducted in July 1998, half referred to flower and fruit production of local trees and one-third referred to temperatures during the dry season as predictors for the upcoming rainy season. A few mentioned insect behavior and predictions by ritual specialists. Based on these indicators, most farmers correctly forecast a ‘good’ rainy season for 1998.

Dry season temperatures

Farmers structure rainfall expectations and evaluations around an archetypal classification of the calendar year with two principal periods: a seven-month dry season and a five-month rainy season. The dry season includes a cold-dry period (waodo) and a hot-dry period (toologo). The rainy season (seogo) includes an early part (signoya), the main part of the rainy season (seogo), and a late part (bombiongo) during which staple crops mature. During each period, farmers expect natural phenomena such as temperatures, winds, clouds, and rain to conform to a certain pattern that they define as the norm. As Osunade (1994) found in Nigeria, Bonam farmers consider deviations from the idealized norm to be inauspicious, resulting in abnormal rainfall and poor crop performance.

Among the earliest indicators are the starting date, intensity, and duration of the cold-dry period. If the cold-dry period begins early or ends late farmers expect the rains to do likewise. Elderly farmers predict the onset of rains by counting 182 days from the beginning of the cold-dry period. That is, farmers count 91 days for the cold-dry period plus 91 days for the hot-dry period before the next season should start. Several informants correctly predicted the early onset of the 1998 rains by this method. Night temperatures may fall to 15 C or below during cold-dry season (late November to early February). According to farmers, intense cold during this time corresponds to abundant rainfall in the following season. Interruptions to the cold, such as several days of warm temperatures, predict drought spells during the rainy season.

Following the cold-dry period is a hot-dry period from late February to early May. Intensely hot temperatures are also believed to predict abundant rainfall. Morning fog or rain in April cools temperatures when they should be hot and, therefore, are considered to be a bad sign. Some farmers qualify these interpretations with greater specificity. For instance, they stressed that the shift from the cold to the hot part of the dry season should be marked by a series of cold night and hot days. Cold temperatures that persist beyond early morning indicate rain failures. Violent winds during the dry season also predict aborted rain events. On the other hand, strong but not destructive eastward winds and hot but not scorching temperatures at the beginning of the rainy season signal a regular and favorable seasonal rainfall. During the rainy season, farmers expect rain after a sultry day and if winds blow toward the east. Farmers also expect hail early in the rainy season and believe that occurrence of hail predicts a good performance of leguminous crops.

Flower and fruit production of local trees

Most local trees that farmers used to forecast rainfall begin flowering between February and March and produce fruits between April and June (Table 1). According to farmers, good yields from trees such as the taanga, sibga, or even the exotic mango (Magnifera indica L.) predict abundant rainfall and a favorable season. On the other hand, good yields from nobga and sabtuluga indicate that drought resistant crops, such as millet, will yield more than sorghum, the preferred staple. In 1997 there was such a surplus of sabtuluga fruit that they rotted on trees.

Farmers recognized that other factors might affect the formation of buds and flowers and subsequent fruit tree production. For instance, taanga needs a succession of cold nights and hot days to flower properly. Violent heat or wind can cause flowers to dry and fall, as happened in 1998 when a harmattan (seasonal wind that blows sand) blew fiercely in February and when April temperatures reached 47ºC. Farmers reported that these phenomena, as well as the 1997 drought, caused low taanga (sheanuts) yields in 1998. Therefore, they did not interpret poor taanga production to predict poor rains in 1998, especially considering that other indicators, such as early rain onset, predicted good seasonal rainfall.

Variations in fruit production patterns are also influence farmers’ expectations. A preponderance of fruits on one side of the tree is believe to indicate areas where sorghum will yield best, others believe that it points to areas where rains will begin first. In 1998 many sibga trees produced two flushes of fruit, which farmers interpreted to indicate that the season would have two distinct planting periods, both of which would produce satisfactorily.

Not all trees in a species flower or produce fruit at the same time. Nor do all trees in a species have equal productivity. When using trees to forecast seasonal rainfall farmers do not on base their expectations on generic observations. Rather most farmers base expectations on trees that are located near their farms or homes and which they have been able to observe over time.

Certain trees are also used in guiding agricultural work. When sibga begin fruiting and when sabtuluga lose their leaves, farmers know it is time to get ready for planting. Kankanga grows where the water table is near the soil surface, signaling herders where to dig wells to water their cattle and farmers where they can plant water demanding crops, such as cotton.

Other rainfall forecasting indicators

Several additional indicators become available early in the rainy season, which tend to be noted by those people who are socially and spatially positioned to observe them. For example, herders and farmers said that if water levels in streams and ponds remain flooded after the first rains, then the season will be favorable because good rainfall at the onset is believed to lead to a regular unfolding and ending of the rains. Herders who pasture animals in uncultivated areas watch the nesting of a koobaagi (Fulani name for a small quail-like bird; koobre in Moré) in the early rainy season. They believe that when nests hang high on trees then rains will be heavy, when nests hang low rains will be scarce.

In both farming and herding groups, the gender division of labor ascribes to women mostly activities within and around household compounds, for example, threshing, cooking, and cleaning. Collecting water and firewood expands the boundaries of their observations to the uncultivated bush. Women note fluctuations in water levels in ponds and wells and in the production of wild fruit that they collect to cook or sell in the market. They also observe insect behavior at water sources and in rubbish heaps. For instance, bugvaré are black insects of the Orthoptera sp. that dig concave nests in rubbish heaps outside compounds. After the first rains, larvae emerge, filling the nests with dirt. A good season is expected if bugvaré fill their nests to the brim with dirt, which symbolizes a full granary. The appearance of large numbers of mayflies during maize planting (June-July) is also considered as a good sign. Because mayflies proliferate after abundant rains (sag nyanga), their appearance could simply indicate a correlation between abundant rains and production of staple grains.

Date and nature of rainy season onset

Most farmers formulate their expectations for rainy season at the onset of the rains. Early onset, especially with a regular succession of rains that enable good crop establishment, is by far the most widely reported predictor of a good season. More than half of the farmers who predicted a good rainfall season in 1998 did so on the basis of its early onset. Number of times farmers must plant is also key in evaluating the nature of a season. When discussing the 1997 drought farmers stressed that they had to replant up to seven times as the crop repeatedly failed to establish. On the basis of this indicators they had known a food crisis loomed ahead about half a year before official early warning systems (Roncoli et al. 2001a).

In a good season, rains begin from the South and move Northward. If rains progress according to this pattern farmers expect satisfactory crop production, even if total amount of rainfall is below normal. In the past, seasonal migrants returning from Ghana brought news about the onset and nature of rainfall to the South, which helped farmers prepare for planting (Elliott Skinner, personal communication).

The planting period ends in early August and is marked by several indicators. For instance, when magpies (sãsanga) crackle loudly as they splash in dew that accumulates on tall grass, farmers say the magpies are mocking slow farmers who have not finished planting. Gray hornbills (silãkwe) fly south at the end of planting and returning before harvest. The fact that farmers identify early August as the last dates for planting millet and sorghum is not surprising; this time coincides with the critical day-length that triggers flowering of local varieties. Plants sown later will flower after a shortened vegetative period, leading to reduced grain yield potential.

Star and moon movements

The relationship between astronomy and rainfall is a more specialized domain, in which some elders are especially knowledgeable. These elders can predict the optimal time for planting from constellation movements and lunar phases. For instance, the visible phases of the moon, especially the full moon, is expected to be drier than dark phases of the moon because moonlight is believed to exert a force that prevents rain from falling. Constellations that appear between April and July signify suitable planting periods for different fields and crops. The appearance of Souci (Pleiades) in early May indicates that it is time to prepare valley bottom fields for planting. The position of the Budb Kutoega (planting hoe, also called Gobo; Ursa Major) identifies the period for planting sorghum. When positioned as a hoe striking the ground in June, Budb Kutoega indicates that it is time to begin planting. When Budb Kutoega rotates so that it looks like a hoe resting on the compound wall in early August, it indicates the end of the planting period.

Changes in star and moon appearance provide a framework for sequences of expected rain events that mark key points in crop growth. Ideally, the rainy season starts during dark phase of the sixth moon (May) and is marked by a protracted nocturnal rain that leaves the soil evenly moist, followed by lighter rains two to three days apart, then two or three heavier rains about five days apart. The total failure of this sequence in 1997 meant that early in the rainy season farmers already knew that the rains would be poor and yields would be low. Some of them took precautionary measures to stock grain or limit farm expenses, such as not engaging labor groups for weeding.

Farmers speak of drought (waré) in reference to a dry year, a prolonged dry period, or the failure of any rain event they expect. During the visible phase of the eighth moon, the appearance of Tatba, a four-star constellation in the shape of a roofing mat, marks the end of a 10-12 day dry period that farmers expect in late June and early July. According to farmers, Tatba signifies the beginning heavy rains that will continue through July and August. Farmers usually wait to plant maize until this dry period is broken by a type of rain known as sag nyanga, characterized by steady and abundant nocturnal precipitation that deposits 20 to 40 mm of rain and leaves the soil moist for several days (Table 2). If two or three more sag nyanga rains fall after planting maize, farmers expect good yields, not only for maize, but for all grain crops. Farmers reported that sag nyanga rains have become rare in recent years.

Selective forecasting knowledge

Interpretation of star movements and lunar phases is available to anyone who is willing to observe them systematically over time. In contrast, divination and other spiritual practices are the prerogative of select groups or individuals. Spiritualists command different degrees of expertise that is unavailable to common people. Some of them are also believed to have power over rainfall. Depending on their religious inclinations, farmers give greatest credence to customary or Islamic spiritualists, but during crises they consult a variety of sources.

The most authoritative among customary leaders is the Tengsoba, who performs sacrifices to mediate between the living and the ancestral and earth spirits that inhabit fetish sites, some of whom influence the rains. The Tengsoba can formulate predictions from the behavior of sacrificed animals, how long it flutters for before falling, which direction it falls, and where the blood spills. Rain falling after a ritual is considered a sign that the fetish has accepted the offering and granted the accompanying petition. The Tengsoba and other members of the Nyononse clan might also receive dream messages or visions from ancestors or divinities concerning rainfall and crop performance.

Bagbugda (sing. baga) form another class of spiritualists who foresee but do not influence rainfall. Bagbugda are men who patrilineally inherit their powers and who oversee inherited sacred sites. Bagbugda gather annually for a bagaré ceremony, during which they receive prophecies, including rainfall predictions. Various bagbugda receive messages that differ in scope and subject. As is the case of other spiritualists, some bagbugda are reputed to be are more powerful than others, but none of them can claim to have the complete picture. In the case of negative premonitions, bagbugda consult one another and with the Tengsoba and buud kasma to identify sacrifices that need to be performed to avert or to alleviate the anticipated misfortune.

While Tengsoba and bagbugda practice for the welfare of the community within their jurisdiction, farmers consult other spiritualists, such as getba (sing. geta) for matters of personal concern. Getba do not inherit or derive their skills, which consists of casting cowries or manipulating ritual items, from a relationship with sacred sites and spirits. Rather, they acquire them through apprenticeship from either a relative or another practitioner.

Marabouts are Islamic spiritualists who are also believed to have the capacity to foresee but not control the future. These spiritualists range from conventional Islamic clerics (Limam) who are versed in the Koran and preside over services at local mosques, to local healers and diviners who mix Islam with indigenous beliefs and practices. Some marabouts practices resemble those of getba, with Islamic verses replacing cowries. Most marabouts receive gifts or money for their services, with the most respected and renowned marabouts drawing substantial income from their practice. However, these practices are generally disapproved by Islamic orthodoxy.

The Zambende ceremony, which marks the beginning of the Muslim liturgical year, is the key venue for marabouts to develop and issue their prognostications. Zambende is a ten-day Islamic feast that occurs three months after the end of Ramadan, the month of ritual fasting. Marabouts base their predictions on the day of the week that marks the first day of Zambende. Each day is associated with a different prophet. The year that follows is characterized by symbolic events in the life or time of that prophet. Prophecies are written in Arabic texts that Koranic teachers pass on to their students, along with instructions for ritual offerings and other measures to stave off inauspicious events. Each year marabouts gather to produce and discuss the prophecies before they communicate them to the faithful during Friday prayers.

Unlike the above mentioned customary and Islamic spiritualists who can only seek to alleviate predicaments through intercessory prayer, there are other practitioners, known as sa tatta, who are believed to have direct command over the rains. Many belong to the Gurmantché ethnic group that inhabits eastern Burkina Faso, but some Mossi chiefs are also believed to have such power. Farmers explain some extreme rainfall conditions as being associated with the installment or death of powerful chiefs.

. Rainmakers aim to enhance their own power rather than the welfare of the community. Farmers are reluctant to call upon rainmakers’ power because rainmaking is considered to be dangerous to both practitioners and the community. Invoked rains are believed to be mostly of saraogo type, short but violent downpours accompanied by heavy wind, sharp thunder, and lightening. They often localized or abnormally distributed with destructive consequences for farms, dwellings, animals, and people. Farmers consider all saraogo rains as bad omens because they damage rather than benefit crops. Therefore, spiritual leaders and elders were anxious about a cloud seeding project that the government implemented in parts of Burkina Faso at the onset of the 1998 rainy season. They saw as an attempt to force the hand of nature and blamed it as the cause for a long dry spell that damaged newly planted maize following the onset of the rains.

Some spiritualists foresaw the 1997 drought and their responses have implications for their potential role in diffusing scientific forecasts as well. The Tengsoba admitted having had dire premonitions during the dry season, but he did not reveal them. Farmers confirmed that, while the Tengsoba normally shares the content of messages he receives during sacrifices, in 1997 he was silent. Other elders and leaders echoed his conviction that "not all truths are for the telling." In particular, they do not publicize dire premonitions because to do so "would be like launching a curse" against those under their authority. That is, verbalizing negative forecasts would reify them into an inevitable outcome, voiding any possibilities for supplication and negotiation with divinities to avert or alleviate the predicament ahead.

Likewise, the sheik of Tangyoco resists any attempt by his followers and visitors to elicit predictions from him. In line with Islamic orthodoxy, he adamantly strives to elude popular attribution of sainthood, insisting that he can only dispense blessings and offer intercessions. According to him, predicting the future is an expression of vanity and pride, a sinful attempt to interpose oneself between God and the faithful, because "knowledge makes you like God." He warned that too much knowledge could divert people from dutiful worship and weaken their faith in God’s mercy. According to the sheik, whatever the prediction, "God can always change things." If the future could be predicted, he said, "People would act on what they know and neglect to pray on Fridays."

The theme of destiny as the outcome of negotiation between human intercessions and the judgment of spiritual entities was particularly salient in the course of discussions of forecasts with farmers. For instance, some informants who predicted good rainfall in 1998 explained their expectations by stating that God would not send more adversity than people could bear, that is, a second year of drought. Failure to alter inauspicious predictions by interceding with ancestors or divinities on behalf of the living may also undermine the legitimacy of those spiritualists responsible for mediating that interface. This can be especially problematic for customary leaders, such as the Tengsoba, whose credibility is already being eroded by a multiplicity of pressures at work in local society. A bleak forecast may also discourage people from farming and induce them to migrate, further undermining the social order on which the elders’ authority rests.

Because they provide alternative frameworks of authority that people can invoke to contest customary power arrangements, elders denounce formal education and monotheistic religions as the reason for declining compliance with ritual obligations and, consequently, for increasing drought and other climatic predicaments. They also deplored the fact that animals needed for sacrifices must now be purchased rather than being contributed by the community. Because ancestors disdain the gift purchased animals, sacrifices may fail to bring about the desired outcome of petitions. In the past the Tengsoba could seize any animal, which its owners could not refuse to surrender because it was to be sacrificed to the ancestors for the welfare of the entire community. Now owners can oppose the confiscation by invoking Islamic or Christian identity and they can report the illegal appropriation to the police and demand monetary compensation. Because animals are assets that families can sell to buy food during times of shortage, voluntary contributions to ceremonial offerings are often insufficient.

Bridging knowledge systems: local and scientific forecasts

Economic and ecological processes interact with societal transformations to erode customary orders of authority and knowledge. Production from household fields, traditionally controlled by family heads, is increasingly insufficient to meet subsistence needs. Contributions from field of wives and young men are becoming more significant to family survival, especially during drought and other crises. Consequently, the authority that family heads formerly held as guarantors of family subsistence and continuity has weakened. Family heads can no longer autocratically decide farming practices; they must now negotiate decisions with their subordinates. In selecting varieties to plant, household heads now consult their dependents, including wives, in order to achieve a diversified mix on household and personal fields and to spread risk and labor demands.

When asked whether they perceive any change in climate that has occurred during their lifetime, most farmers interviewed responded that climate variability has increased. In their view, it rains less than before, rains begin late or end prematurely, and dry spells during the season are more protracted and more frequent. In other words, rains have become like the national lottery, whereby one can randomly gain or lose but in which one loses most often. Climate variability has weakened farmers’ confidence in local forecasts of rainfall patterns. Some elders recalled that in the past they were able to predict the rain onset so accurately that they could mobilize family labor plant on dry soil, knowing that the rains would soon follow, but now their sons refuse to go to the field until it actually rains.

Because they have lost confidence in their ability to predict rainfall, farmers are open to and keenly interested in alternative sources of rainfall information. They do not resist the introduction of scientific information or regard it as threatening the integrity of local cultural traditions. In fact, the cognitive landscape of Bonam farmers is quite diverse and dynamic, as are most local knowledge systems (Barth 1995, Sillitoe 1998b). Farmers pragmatically mix traditional farming knowledge with extension advice, local technology with development innovations. Even in the case local forecasting knowledge, farmers are used to combining a variety of environmental observations and spiritual traditions.

A plurality of knowledge frameworks is not necessarily destabilizing. But some types of knowledge have a lot more ideological weight than others because of the power they signify and the resources that embody. Farmers’ time-honored wisdom and experience has a lot to offer but also a lot to lose in the encounter with knowledge repertoires, such as formal education, agricultural technology, and scientific information, that are backed by foreign financing and government endorsement. Scientific forecasts must be presented in ways that do not devalue but build on local expertise and conform to cultural notions about the nature of knowledge, its production and validation, and its relationship to society. Culturally sensitive dissemination of scientific knowledge can have political and programmatic benefits. For instance, the fact that local systems of thought stress the incompleteness and provisional nature of knowledge and they conceptualize destiny as the outcome of negotiation between human and divine (inscrutable) will can help explain the probabilistic dimension of scientific forecasts.

There are several aspects of method and content where local and scientific forecasting knowledge either converge or follow analogous patterns. In formulating predictions, spiritualists sometimes use approaches reminiscent of scientific practice. Zambende prophecies derive from the exegesis of textual material and from consultations among spiritualists, who then officially communicate them to the lay public. Just as scientists develop climate forecasts from global circulation models and adapt information from models to national conditions through analysis of and extrapolation from historic data bases, marabouts predict seasons from a body of global knowledge, the Islamic calendar and history, and adapt predictions to the local context through sacrifice and Mossi ritual.

In the shared domain of environmental knowledge, farmer forecasts resemble scientific methods in their reliance on the systematic observation of natural phenomena. Observation-based knowledge production is consistent with cultural learning styles where children learn from watching adults rather than through verbal instruction or asking questions. In Bonam, seniority is a mark of authority because those who have lived many years have heard and seen a great deal. Elders interviewed enumerated the many historical events they had ‘seen’, such as famines, chiefly successions, colonial intervention taxation, forced labor, and war conscription, as a way of shoring up the authority of their words. The trope of ‘knowing as having seen’ operates in both environmental and spiritual knowledge domains. One sees signs in the landscape in as well as in dreams, in the outcome of sacrifices, or in the casting of cowries.

But, unlike scientific forecasts that may be transferred unchanged from outlying centers of knowledge to sites of application, the production of local forecasts is deeply localized. They derive from sustained scrutiny and intimate interaction with a meaningful microenvironment whose rhythms are intertwined with the cycles of family and community life. It is not the generic sibga or taanga that farmers considered in predicting rainfall, but specific trees located on one’s farm or near one’s home that have been objects of engaged ‘beholding’ in making making decisions of livelihood significance. This ability to comprehend environmental clues and apply them to farming choices is performative knowledge that is intuitive and ingrained in practice, rather than articulated in a set of abstract principles or calculations (Richards 1993, Sillitoe 1998a).

Meteorologists could build on local understanding of the relationship between dry season temperatures and seasonal rainfall to explain technical aspects of scientific forecasts based on sea surface temperatures. Farmers’ interpretations of wind patterns also recognize the ocean as the origin for rain. During the dry season, farmers expect winds to blow westward, that is, to go to the ocean to pick up water, and then return blowing eastward at the onset of the rainy season. Farmers predicted and explained drought from the absence of such winds.

Farmers’ forecasts diverge from scientific ones in important ways, particularly the parameters and scale they address. Local forecasts focus on rainfall characteristics that are most relevant to farmers, such as time of onset, duration, and distribution (Roncoli et al. 200lb). To evaluate seasonal rainfall, farmers consider the number, type, and timing of rains rather than total rainfall quantity, the key variable in scientific forecasting. In farmers’ understandings and decisions, a forecast of ‘abundant’ seasonal rainfall often translates into an expectation of a longer season (Roncoli et al. 2001b). Currently, science is unable to predict reliably either duration or distribution of seasonal rainfall, but the integration of scientific forecasts with local knowledge might allow some inferences in this regard. For instance, the abnormally heavy rains that fell in July and August 1999 could have been predicted by combining farmers’ predictions of delayed onset with the scientific forecast for above normal seasonal rainfall.

Farmer’s forecasts also differ from scientific ones in that they address a local rather than regional scale and crop-climate interactions rather than precipitation per se. Farmers recognize that rainfall may have different implications for each crop species according to when and how it occurs. For instance, the same amount of rainfall can lead to radically different production outcomes if it occurs by means of a sa nyanga or a saraogo rain event (Table 2). Water deficit periods that occur during establishment or heading will cause more damage than those that occur during other crop growth stages.

Contextualizing forecasts: program and policy implications

A comparison between local and scientific forecasting knowledge leads to two key questions: Are they accurate and are they reliable? Assessment of forecast accuracy encounters both philosophical and methodological challenges. Agrawal (1995) criticized efforts to evaluating local knowledge by scientific methods as way to reaffirm the hegemony of ‘Western’ science. In many instances, including this case study, evaluation of local knowledge is hindered by the lack of quantitative data, such as fruit production by local trees. Data on dry season temperatures may be obtained for some areas, but farmers’ forecasts are based on subjective evaluations rather than objective measurements. In 1998 most farmers interviewed correctly predicted good rainfall, although expectations might have been biased by hope for relief from the food shortage caused by the previous year’s drought. Emerging empirical evidence appears to confirm and validate farmers’ understandings of environmental relationships. Findings from Zimbabwe suggest that some local indicators have predictive ability (FEWS 1998) and a recent analysis by Orlove et al (2000) points to a correlation between El Nino-related reduced rainfall and diminished Pleiades visibility, an indicator used by Peruvian farmers in making planting decisions.

Farmers and scientists have different requirements for and definitions of accuracy. In consonance with local knowledge frameworks that recognize the inevitable incompleteness of human understanding, Bonam farmers do not consider any one indicator or piece of information as entirely reliable. Local forecasting systems encompass a range of indicators and methods, which become available to different people and at different times. Farmers do not retroactively revise hypotheses derived from observation of earlier phenomena on the basis of later ones, nor do they try to reconcile contradictions in order to develop a cogent set of forecasts. Discrepancies among forecasts or between forecasts and outcomes are not failures. Farmers explained such discrepancies in terms of microenvironment diversity and differences among cropping systems. In making resource management decisions, farmers balance predictions for the upcoming rainy season balanced against experiences of the previous season, observations of when and how the rains begin, and the expectation of great variability. Farmers’ ability to adapt predictions to the local context may help mitigate the risks entailed in the dissemination of probabilistic scientific forecasts in an uncertain environment (Roncoli et al. 2001b).

This pragmatic flexibility in shifting among knowledge systems is not unique to how Bonam farmers deal with rainfall predictions. Research on applications of local and scientific knowledge to management of agricultural and natural resources indicates that farmers consider elements from various repertoires in making decisions, sometimes drawing from one knowledge framework to diagnose a problem and from another to find solutions (Brodt 1999, Millar and Curtis 1999). Huber and Pederson (1998) poignantly depict Tibetan yak herders gathering around short-wave radios to listen to a weather report, while revering local gods, which they believe control the weather. In their view, scientific and local knowledge are neither comparable nor in competition because they pertain to different domains, namely ecology and morality.

Bonam spiritualists and local leaders also stressed the moral dimension of knowledge in reference to their potential roles in dissemination of scientific forecasts. In a cultural context where specialized knowledge is intimately linked to social responsibility, delivering ominous knowledge obliges assistance or participation in coping with the consequences of that omen. The cultural perception of obligation associated special knowledge emphasizes the importance that scientists clearly communicate the risks and limits of scientific forecasts, and that forecast dissemination efforts should go beyond a simple provision of information to include assistance in negotiating a mitigated outcome of predicted scenarios. Forecast delivery must be integrated with efforts to foster strategic programs and a policy environment that support users’ capacity to respond optimally to a forecast (Stern and Easterling 1999, Broad 2000, O’Brien et al. 2000, Nelson and Finan 2000). Especially needed are resources that enhance the flexibility of farmers’ responses, such as agricultural credit, seed of varieties adapted to local conditions, and technologies, such as plows, that accelerate crop establishment (Ingram et al. 2002).

Conclusions

Our findings show that neither the experiences of bridging knowledge systems nor the concepts of rainfall forecasting are alien to farmers of Burkina Faso. Local forecasts converge with scientific ones in some aspects of content and method, but also diverge in terms of practical significance and moral meanings. These contrasts challenge science on two fronts. The specificity of local forecast parameters urges science to be more relevant and responsive to farmers’ information needs. Cultural emphasis on the relationship between knowledge and responsibility also calls for scientists to be more aware of and accountable for the intended and unintended impacts of scientific products. There is ample opportunity and growing need for introducing scientific forecasts to support production capacity and food security. To be effective and sustainable, forecast dissemination must be integrated into a concerted strategy that ensures communication between scientists and farmers and enables farmers to respond to forecasts in ways that are socially acceptable and environmentally sound.

References

Agrawal, A. 1995. Dismantling the divide between indigenous and scientific knowledge. Development and Change 26:413-439.

Antweiler, C. 1998. Local knowledge and local knowing: an anthropological analysis of contested cultural products in the context of development. Anthropos 93:469-494.

Appadurai, A. 1995. The Production of locality." Counterworks: Managing the diversity of knowledge, ed. R. Fardon, 204-225. New York: Routledge.

Barth, F. 1995. Other knowledge and other ways of knowing. Journal of Anthropological Research 51: 65-68.

Broad, K. 2000. El Niño and the anthropological opportunity. Practicing Anthropology, 22:20-23.

Broad K. and S. Agrawala. 2000. The Ethiopia food crisis - uses and limits of climate forecasts. Science 289:1693-1694.

Brokensha, D., D. Warren, and O. Werner eds. 1980. Indigenous knowledge systems and development. Waltham, MD: United Press of America.

Brodt, S. 1999. Interaction of formal and informal knowledge systems in village based tree management in Central India. Agriculture and Human Values 16:355-363.

Chambers, R. 1996. Whose reality counts? Putting the first last. London: Intermediate Technology Publications.

Fardon, R. 1995. Introduction: Counterworks. In Counterworks: Managing the diversity of knowledge, ed. R. Fardon, 1-22. New York: Routledge.

FEWS 1998. October country updates. Zimbabwe. Famine Early Warning System Project, Arlington, VA.

Flora, C. 1992. Reconstructing agriculture: the case for local knowledge." Rural Sociology 57:92-97.

Gladwin, C. 1989. Indigenous knowledge systems: the cognitive revolution and agricultural decision making. Agriculture and Human Values 3:25-32.

Gray, L. 2000. Reconciling local knowledge with Western scientific notions of soil fertility decline in southwestern Burkina Faso. Paper presented at the African Studies Association Annual Meeting, Nashville:15-19 November 2000.

Hobart, M. 1993. Introduction: The growth of ignorance? In An anthropological critique of development: The growth of ignorance, ed. M. Hobart, 1-30. New York: Routledge.

Huber, T., and P. Pedersen. 1998. Meteorological knowledge and environmental ideas in traditional and modern societies: The case of Tibet. Journal of the Royal Anthropological Institute 3:577-598.

Ingram, K. C. Roncoli, and P. Kirshen,. 2002. Opportunities and constraints for farmers of West Africa to use seasonal precipitation forecasts with Burkina Faso as a case study. Agricultural Systems (in press).

Kempton,W., J. Boster, and J. Hartley. 1995. Environmental values in American culture. Cambridge: MIT Press.

Long, N., and A. Long eds. 1992. Battlefields of knowledge: the interlocking of theory and practice in social research and development. London: Routledge.

McCorkle, C. 1989. Towards a knowledge of local knowledge and its importance for agricultural RD&E.. Agriculture and Human Values 4:4-13.

Millar, J. and A. Curtis. 1999. Challenging the boundaries of local and scientific knowledge in Australia: Opportunities for social learning in managing temperate upland pastures. Agriculture and Human Values 16:389-399.

Moock, J. and R. Rhoades ed. 1992. Diversity, farmer knowledge, and sustainability. Ithaca NY: Cornell University Press.

Mutiso, S.K. 1997. Indigenous knowledge in drought and famine forecasting in Machakos District, Kenya. In Indigenous knowledge and change in African agriculture, eds. W.M. Adams and L.J. Slikkerveer, 67-86. Ames IA: Center for Indigenous Knowledge for Agriculture and Rural Development.

Nader, L. 1996a. Anthropological inquiry into boundaries. In Naked science: Anthropological inquiry into boundaries, power, and knowledge, ed. L. Nader, 1-28. New York: Routledge.

Nader, L. 1996b. The three-cornered constellation: Magic, science, and religion revisited. In Naked science: Anthropological inquiry into boundaries, power, and knowledge, ed. L. Nader, 259-276. New York: Routledge.

Nelson, D. and T. Finan. 2000. The emergence of a climate anthropology in northeast Brazil. Practicing Anthropology, 22:6-10.

Ngugi, R. 2000. Use of contemporary and indigenous climate forecasting information in farm level decision-making in Mwingi District, Kenya. Paper presented at the symposium on Forecasting, Food, and Farming in Africa. New York: Institute of African Studies, Columbia University, 24 March 2000.

O’Brien, K., L. Sygna, L. O. Naess, R. Kingamkono, B. 2000. Is information enough? User responses to seasonal climate forecasts in southern Africa. Report to the World Bank, AFTE1-ENVGC. Adaptation to climate change and variability in sub-Saharan Africa, Phase II. Oslo: Center for International Climate and Environmental Research, University of Oslo.

Orlove, B, J. Chiang, and M. Cane. 2000. Forecasting Andean rainfall and crop yield from the influence of El Nino on Pleiades visibility. Nature 403:68-71.

Orlove, B. and J. Tosteson. 1999. The application of seasonal to interannual climate forecasts based on El Niño-Southern Oscillation (ENSO) events: lessons from Australia, Brazil, Ethiopia, Peru, and Zimbabwe. WP 99-3, Institute of International Studies, University of California, Berkeley.

Osunade, A. 1994. Indigenous climate knowledge and agricultural practice in southwestern Nigeria. Malaysian Journal of Tropical Geography 25:21-28.

Otterstrom, S. 2000. Variation in coping with El Niño droughts in northern Costa Rica. Practicing Anthropology, 22:15-19.

Phillips J., E. Makaudze, L. Unganai, J. Makardho, and M. Cane. 1999. Current and Potential Use of Climate forecasts for Farm Management in Zimbabwe. Paper presented at the Society for Applied Anthropology Meetings, Tuscon, AZ, April 22^nd-25^th, 1999.

PNUD. 1997. Rapport sur le dévéloppement humain durable: Burkina Faso. Programme des Nations Unies pour le Developpement. Ouagadougou, Burkina Faso.

Richards, P, 1985. Indigenous agricultural revolution: Agriculture and food production in West Africa. London: Hutchinson.

Richards, P. 1993. Cultivation: Knowledge or performance? In An anthropological critique of development: The growth of ignorance, ed. M. Hobart, 61-78, New York: Routledge.

Roncoli, C., K. Ingram, and P. Kirshen. 2001a. The Costs and Risks of Coping: Household Impacts and Farmer Responses to Drought in Burkina Faso. Climate Research (in press).

Roncoli, C, K. Ingram., P. Kirshen, and C. Jost. 2001b. Modernity meets marginality: Impacts of seasonal rainfall forecasts on rural livelihoods in Burkina Faso. Paper presented at the Society for Applied Anthropology Meetings, Merída, Mexico, March 22^nd-25^th, 2001.

Roncoli, C. S. Bahadio, and S. Boena. 1999. The role of rainfall information in farmers’ decisions: ethnographic research in the Central Plateau (Burkina Faso). Technical Report n.1, Climate Forecasting and Agricultural Resources Project. Athens GA: University of Georgia.

Skinner, Elliot, Department of Anthropology, Columbia University, New York, personal communication, November 17, 2000.

Sillitoe, P. 1998a. The development of indigenous knowledge: A new applied anthropology. Current Anthropology 39:223-235.

Sillitoe, P. 1998b. What, know natives? Local knowledge in development. Social Anthropology 6:203-220.

Stern, P. and W. Easterling eds. 1999. Making climate forecasts matter. Washington DC: National Research Council, National Academy Press.

Strathern, M. ed. 1995. Shifting contexts: Transformations in anthropological knowledge. New York: Routledge.

Thompson, J. and I. Scoones. 1994. Challenging the populist perspective: Rural people’s knowledge, agricultural research and extension practice. Agriculture and Human Values 11:58-76.

Thrupp, L. A. 1989. Legalizing local knowledge: from displacement to empowerment for Third World people. Agriculture and Human Values 6:13-24.

Warren, M., L.J. Slikkerveer, and D. Brokensha eds. 1995. The cultural dimension of development: Indigenous knowledge systems. London: Intermediate Technology Publications.

Table 1

Local Trees Bonam Farmers Use as Indicators for Forecasting Rainfall

Moré	French	Scientific*	Fruits ripen	Indicates
kankanga	figuier	Ficus gnanphalocarpa (Miq.) Steud. ex A. Rich.	May/June	Water table near ground surface
sibga	raisinier	Anogeissus leiocarpus (DC) Guill & Perr	May/June	Abundant fruit yield = abundant rainfall
lenga	citronnier de mer	Xymenia americana L	May.	"
taanga	karité	Butyrospermum parkii (G. Don) Kotschy	June	"
roanga	carroubier africain	Parkia biglobosa ( Jacq.) R.Br. ex G. Don f.	March/April	"
pusga	tamarinier	Tamarindus indica L.	June	"
sabtuloga	bouleau d’Afrique	Lannea acida A. Rich.	April	Abundant fruit yield = scarce rainfall
nobga	prunier	Sclerocarya birrea(A. Rich.) Hochst.	May	"

*Source: Information provided by Dr. Pierre Ouédraogo, Direction Provinciale de l’Environnement, Eaux et Forêts, Bouls

Table 2

Typology of Rainfall Events

sag kenga

Big rain. Refers to amount, duration, or distribution.

sag serdem

Fine rain, lasting several hours. Causes good infiltration

sigri saaga (sig saaga)

Planting rain, rain that marks onset of rainy season.

Lasts the whole night: 6-8 hrs.

Steady precipitation, not violent, without wind or thunder.

Leaves soil moist for several (5+) days.

sag sika

Very localized, with one or few clouds.

Without wind or cool temperatures.

Can fall any time of day, but mostly in August.

sag nyaanga

Abundant precipitation, mostly mornings or evenings.

Lasts several hours.

Accompanied by gentle thundering.

Leaves soil soaked, good infiltration, moisture lasts long time.

Characterizes favorable rainy season and predicts good harvest.

Mostly in July-Aug, but increasingly rare in recent years.

saraogo

Short but violent downpour, usually early morning or at sunset.

Localized or oddly distributed.

Accompanied by heavy wind, sharp thunder, and lightening.

Damages animals, people, dwellings, crops.

Believed to be invoked or attracted.

Characterizes unfavorable rainy seasons and predicts poor harvest.

Mostly in July-Aug, increasing frequency in recent years.

bind saaga

Literally "excrement rain" because it washes dried excrement off plants in the bush.

Also called "mango rains" because it washes mango fruits.

Early rains in April.

Abundant in the past, used to fill streambeds.

Enables people to use mud to repair walls.

Stimulates growth of young fibers of roofing grass or bagana’ (Philostigma sp.).