Chapter III
AGRICULTURE
I. LIVESTOCK
The total number of livestock was estimated by the Centre for Ecological Monitoring (CEM) from Systematic Aerial Reconnaissance (SAR) flights made in 1989 and 1990 across Senegal (see "rapport de campagne : décompte du bétail", CSE,1990).
Sampling is carried out according to rules which guarantee its representativity with some degree of accuracy : the distance between the set flight lines, the utilization of a Global Positioning System (GPS) that helps localize the observations made, the utilization of photographies and the consideration of altitude and speed. The data obtained from SAR are computerized and treated statistically, particularly through a test of accuracy. This technique shows that the data have a margin of confidence less than 30%, which represents a high reliability. The same method was used for the determination of woody cover (Chapter IV, section II).
A complete air coverage of Senegal was achieved in 1989 and 1990, resulting in the following assessment of the total number of livestock as follows :
1.1 Cattle
A total number of 1,765,546 head was identified at the national level.
The highest numbers are found in the Provinces of Linguère (169,816), Tambacounda (163,077), Matam (152,415), Bakel (141,700), Kaffrine (126,750) and Podor (103,048). These six Provinces account for almost 50 % of the inventoried cattle. The Province of Linguère where cattle breeding is the dominant activity contains 10 % of the cattle.
When comparing regional data, the Region of Saint-Louis (in the North) is by far ahead, with more than 20 % of the total; followed by the Regions of Tambacounda (in the East) and Louga (in the North) with 18 % and 16 %, respectively. These three Regions account for 50 % of the cattle.
Field surveys on the composition of the herds took three months. These showed that dairy cows represent 37.5 % of the total.
1.2 Small Ruminants
There are more small ruminants than cattle, with a total of 2,161,059 head.
The leading Province is Linguère (in the sylvopastoral zone ) with 321,860 head, 15 % of the national animal population. The number of small ruminants in this Province is higher than in the other Provinces, except for Louga and Saint-Louis. The highest concentrations are found in the sandy Ferlo. The importance of small ruminants in such an adverse environment "highlights the remarkable resistance of small ruminants, their capacity to quickly re-establish their numbers, their role as food and for market oriented production and, especially, their paramount importance for those who live in such a constraining and uncertain environment".
The Provinces of Podor and Matam have 288 534 and 284 840 heads respectively, more than 13 % of the total, followed by Kaffrine, further South, with 218,418 small ruminants : 10 % of the total. number.
However, numbers of small ruminants have decreased in the North, since 1989, when Mauritanians stopped their transhumances in that part of the country.
The Central-Western rural Provinces have the highest densities with 33 head/km² in Bambey and Diourbel and 29 in Mbacké.
As for cattle, the regional analysis of data shows the dominance of the Region of Saint-Louis which has 678,287 : 32 % of the national population of small ruminants. It is followed by Louga with 540,701 head : 25 % of the national population. Of course, the smallest numbers are found in Lower Casamance and in the Province of Vélingara in Higher Casamance, which are more oriented to agriculture and forestry than animal husbandry.
1.3 Horses and Asses
The figures relating to these two species have not been considered in this study due to their poor representativity by comparison with the former, and also to the unreliability of the results of their counting via the nation-wide Systematic Aerial Reconnaissance (some of these animals are kept in enclosures inside compounds). The figures emanating from the Department of Animal Husbandry have been used for these two types of animals, for which field census is relatively easy (the Plan of Action for Livestock) : 365,285 horses and 302,455 asses. The same applies for poultry, for which estimates are very approximate.
1.4 Poultry
Poultry statistics (which are impossible to estimate from an aircraft) are rather complex because of the great inter-annual variation in their numbers (mortality caused by diseases for traditional poultry, rapid reconstitution of their populations). According to official estimates (the Plan of Action for Livestock), Senegal had 10,500,000 in 1991, the reference year for this study.
II. RICE CULTIVATION
Rice production is an important component of Senegalese agriculture. The production areas are located in the North in Senegal River basin, the South-East in the Anambé bassin (irrigated regimes), and in the South where there is enough rainfall to flood shallow areas for rainy season crops. The cropping systems are rather numerous :
- Submerged and upland rice cultivation
- Plain upland rice cultivation with short cycle varieties
- Upland rice cultivation supported by an easyly accessible water table
- Shallow submerged rice cultivation with long or medium term varieties
- Aquatic rice cultivation in saline areas
- Irrigated rice cultivation with two production cycles per year.
The latter system is the most productive and involves continuous flooding, unlike the other two which involves intermittent flooding (little or no supervised traditional rice cultivation for subsistence).
The surface area and production of each of the two main cropping systems (flooded and intermittently flooded) were determined from the official statistics collected by the supervisory services and the regional agricultural services. Field missions made it possible to verify data accuracy and cropping systems. However, it should be noted that in recent years there has been a significant increase in the area cultivated in the north of Senegal (continuous flooding system). This phenomenon is linked to the damming of the Senegal River and the enrolment of farmers in specialised organizations.
III. SAVANNAH BURNING
Remote sensing was used to estimate the surface areas affected by savannah burning in 1991. The approach is based on an interactive supervised classification, validated in the field with a Global Positioning System (GPS).
The spatial analysis of the results shows a close relationship between fire occurrences and the exploitation patterns of the natural resources of the areas concerned.
3.1. Method of Estimation of Burned Areas Through Remote Sensing
The savannah burning map developed during this study is the result of the processing of NOAA-AVHRR images.
The distortions of the raw images are corrected using the conventional methods of geometric correction. This results in a corrected image, recorded in the five channels of the AVHRR sensor. The best prints are then selected, taking into account the cloud coverage criterion which is the main impediment to monitoring savannah fires with daytime images. In Senegal the sky is cloudy for much of the year, due to the circulation of the tropical trade wind which builds up between December and April, corresponding to a large part of the fire season; this is concludes with the first rains in June.
3.2. Classification of Burned Areas
The approach consists in a semi-automatic identification of burned areas. The starting point is a coloured image obtained after placing the channels of thermal (3), visible (1), and near infrared (2) in the red, blue and green, respectively. The resulting image permits the identification of distinct and dark spots representing burned areas, different from the permanent elements of the physical environment in that they suddenly appear and vanish progressively under the action of wind, trampling and several other factors. This method of identification of burned savannah has been applied to wider areas in the sub-region, by resorting to NOAA-GAC images whose spatial resolution is far more reliable than AVHRR.
Classification takes place interactively on the screen, using the CHIPS parallelepiped classification which uses thresholds taken from the specific frequency histogrammes. Such an approach is practical for two reasons :
- visual identification is made at a very early stage thanks to the channel combination, which allows for the adjusment of the thresholds used in the classification;
- the identified areas are those which have burned recently, giving remarkably low reflections, and which have not had enough time to evolve to a point that one can be dubious about their nature or area.
Carte des feux de savanne
3.3. Identification and Mapping of Burned Areas
The result of image classification is superimposed on the map delimiting the sub-provinces. This operation is possible with the cross-tabulation module available in the Idrisi geographic information system software. The final map permits the localization of the areas burned as a result of the passage of fires in the different sub-provinces of Senegal. In total the burned areas amount to 483,800 hectares, about 2.4 % of the national territory. It can be assessed either globally or by taking into account specific criteria such as the dates of occurrences (seasonality), local influences (variations...).
3.4. Analysis of Savannah Burning Distribution
a) Chronological analysis
The distribution map of savannah burning in 1991 (figure 1) shows a chronological evolution. The first big fires occured in the North in December, creating significant losses in grass cover. In February-March, the fires progressively shifted Southward in a more scattered way. Minor fires were observed a little further North in the second fortnight of May.
This evolution in time reflects the dryness of the grass cover, in keeping with a North-South gradient, due to the progessive arrival of the dry season following the withdrawal of the ITF (Inter Tropical Front). The influence of local conditions in the geography of burned areas is clear in two cases :
- in the North, in the Province of Linguère, the area and continuity of burned areas can be explained by the dominance of a continuous grass cover.
- in the South, the scattered structure of burned areas is due to the greater presence of trees which bring more discontinuity to a grass cover in which late drying perenials predominate, as well as a more dense human settlement.
b) Spatial Analysis
The study of the map shows that the burned areas are concentrated in the Central-Eastern part of the country, east of a line linking the Sub-provinces of Ross-Béthio in the North and Paoskoto in the South. This imaginary line isolates the major towns and a large part of the groundnut basin to the West; burned areas are located in the East (see figure 1), in the poorly populated regions (shrub savannah, forest/tree savannah zones), which sheds more light on the rural specificity of the savannah burning phenomenon. However, there are no burned aeras in the far North which is essentially rural.
Burned areas are located in two major zones : the Southern part of the Province of Linguère; and the large corridor starting East of the Region of Tambacounda through The Gambia to the Provinces of Kolda and Vélingara. The demarcation line between these two entities goes along the northern part of the Provinces of Koumpentoum, Koungheul and Koussanar.
- Central-Western fires
These represent 35.5 % of the total area burned in 1991 and are distributed between the shrub savannah (196,800 hectares) the transition zone (141,500 hectares) and the low density agricultural zone (9,500 hectares). The pastoral zone in the North has a small share (300 hectares), which can be explained by the low level of plant production following a poor rainy season in 1991 in the Northern part of the country (very late rainfalls, inadequate distribution in time and space).
Fires are widely tributed in this Centre-West eco-geographic zone (previous table) which is a vast traditional cattle breeding area. The frequent population movements for grazing purposes constantly endanger plant resources, otherwise highly needed. They occur relatively early, since at the end of November the vegetation is so dry that fires can cover massive areas, if nothing is done to stop their progression as is often the case, due to the lack of maintenance of firebreaks.
Quantitative distribution of burned areas (1000 Hectares)
Source : Ecological Monitoring Centre.
- Forest zone fires
This second zone presents different characteristics to the previous one. In 1991 burned areas accounted for 64.5 % in this zone. The Provinces of Koumpentoum (70,800 ha) Makacoulibantang (21,800 ha), Koussanar (19,000 ha) and Goudiry (14,000 ha) are among the most affected. South of The Gambia (the Regions of Kolda and Ziguinchor) where 30 % of the land is devoted to agriculture, 84,200 ha burned. The use of fire for clearing purposes and the abundant combustible materials have largely contributed to the periodicity of fires in this part of the country.
In the classification, the Province of Kédougou is not taken into account, for reasons related to the high density of the plant cover which saturates the signal received by the AVHRR sensor. This should not significantly affect the results of this study, since fires have been relatively rare and strictly confined to a few areas in this zone where natural conditions (topography and moisture mainly) are little conducive to their spreading.
Finally, the statistics relating to fires in The Gambia have been taken into account in the graph (figure 3) so as to highlight the importance of the spatial continuity in fire dynamics.
Province-specific share of total burned areas
Source : Ecological Monitoring Centre.
Overall, the time-space analysis of the fire map provides valuable data on the distribution of burned areas and the local variability of the phenomenon, which seems to be highly related to ecological and geographical conditions. It also suggests the existence of a close relationship between the way natural resources are managed by populations and repeated fires.
3.5. Quantity of Biomass Effectively Burned by Fires
The biomass effectively burned essentially relates to the grass vegetation; the contribution of the ligneous biomass (above ground) accessible to fires (dead leaves on the ground, shrubs, dry low leaves) is not easy to determine in the Sahelian context.
Superimposition of the 1991 biomass map on that of savannah burning made it possible to determine the total biomass burned. The CEM undertakes an annual survey of the biomass produced at the end of the rainy season. The methodology is based on the determination of decadal and monthly vegetation indices (NDVI) during the growth period. A map of integrated NDVI is produced at the end of the season, and then calibrated on the ground from field measurements of biomass at 35 control sites, chosen for their representativeness, distributed across the national territory. Cropping areas are not taken into account, since the biomass exposed (crops) is, to a large extent, spared by the fire, fields being cleared before the dry season and residues burned taken into account under "Land Use". The spatial distribution of the biomass effectively burned according to the type of savannah is indicated in the table below. Total biomass burned in savannah areas is 557 766 tonnes of dry matter, that is 1.4 % of the overall production of savannah areas, and 2.7 % of the grass biomass (in the calculation of gas emissions, about 80 % of this biomass is considered to be burned). In contrast to the North, where the grass cover has not grown very much, the shrub savannah has been more seriously affected.
Zonal Distribution of Burned Biomass
Affectation des Biomasse totale Biomasse exposée Biomasse touchée sols par les feux Zone pastorale 2 867 523 t/ms 1 714 200 t/ms 266 t/ms Savane arbustive 3 454 600 t/ms 1 843 500 t/ms 188 000 t/ms Savane 21 440 500 t/ms 11 016 800 t/ms 105 100 t/ms arborée/forêt Zone de transition (savane arbustive associée à une agriculture très 2 502 600 t/ms 1 316 000 t/ms 136 300 t/ms localisée) Zone 9 175 900 t/ms 4 731 500 t/ms 128 100 t/ms agro-forestière Total 39 441 123 t/ms 20 622 000 t/ms 557 766 t/ms
Source : Ecological Monitoring Centre.3.6. Origin of Savannah Fires Field surveys made it possible to identify some causes of fires; they can be classified into two categories : a). Anthropogenic Causes Almost all of the fires identified are anthropogenic, either through carelessness or lack of concern.In Northern Senegal (highly pastoral) fires very often result from campfires that are not completely put out by herdsmen constantly on the move. Field investigations unveiled also the existence of fires deliberatly started in order to : - speed up gum exudation from Sterculia setigera or Acacia Senegal; - clear crop fields (quite generalized utilization of fire for clearing and preparing the fields for next crop year); - enrich and renew grazing lands; - clear roads and tracks; - destroy agricultural and animal (e.g.,ticks) pests.Cigarette ends thrown away by careless smokers are also said to have triggered several fires along main roads.In the Southern part of the country (a forest zone mainly, with a dense cover) natural resource husbandry is the prime cause of savannah fires. Indeed with the depletion of wood resources in the North, following drought and massive felling, the exploitation zone has progressively been focussed in the South which is endowed with a more dense wood cover and a greater variety of species. On top of the considerable utilization of fire in many resource uses - including charcoal production, poaching in the game reserve of Niokolo Koba and hunting nearby, and honey harvesting - there are several socio-economic considerations that suggest a popular will to adopt a passive attitude as regards the regulations limiting their access to these very resources.There are also other considerations : adjacency of The Gambia which creates a cross border dynamic (smuggling); the recurrence of fires along main roads, and especially the railroad; as well as the starting up of roadworks within the framework of the opening-up policy of the Region of Tambacounda (the opening of the road Dialakoto-Kédougou which goes through the game reserve of Niokolo-Koba). b). Natural Causes They are related to several factors, particularly lightening and rockslides. These factors contribute little to the area of savannah fires.IV. BURNING OF AGRICULTURAL RESIDUES IN THE FIELDSAgricultural residues include groundnut leaves, rice straw, millet and maize leaves and stalks, cotton leaves and stalks. In general, rural populations in the Sahel have several uses for these residues (cattle feed in the dry season, household and enclosure fences). In addition, the clearing work which gives rise to field burnings happens at the end of the dry season. The residues remain on the field for nine months, which explains why the fraction which has not been used in the ways mentionned loses its nutritive values partly through trampling from cattle, or rot on the spot. Thus, the burned fraction is very small for the country's main crops (groundnuts, millet-sorghum and maize). - Groundnut leaves : they provide an important source of animal feed and at least 90 % (taking into account transport losses) are collected for this purpose by farmers and stored for the animals. The fraction burned on the field is negligible, or non-existent. - Millet leaves and stalks : stalks are partly used to make fences, the rest is left on the spot. With consumption by livestock, deterioration through trampling (they remain on the field for the whole dry season) and the fraction that decomposes, the quantity burned when clearing the fields is estimated at 10 - 20 % . The leaves are used to feed the animals. - Maize leaves and stalks : either the leaves are used as cattle feed or they progressively lose their nutritive values. But the stalks are not used for making fences as is the case with millet. Apart from the proportions consumed by livestock (about 30 to 40 %), they are mainly burned on the spot. - Rice straw : In the rice growing schemes of the Senegal River Valley, where 70 % of local production come from, due to the supervision provided to the local population, 80 % of the straw is burned. The reason is to avoid a prolonged presence of wandering cattle in rice fields, as the latter are seen as the main vectors for the dissemination of wild rice (seeds contained in dung). The remainder is either gathered in the form of enriched (with molasse) husks, is used to feed the animals in the fold, or buried in the ploughed fields as organic fertilizer. In Lower and Middle Casamance, some rice residues are left in the fields for grazing, the remainder being also buried for soil improvement. The fraction burned is negligible. - Cotton seeds : nearly one third are used to feed livestock, the remainder for vegetal oil production. - Cotton stalks : they are not consumed by livestock and do not quickly lose their nutritive value because of their solidity. Apart from the small fraction used by local people for energy purposes (availability of several other combustibles in the production areas), more than 80 % of cotton stalks are burned on the field.Residue/production ratios used in the calculation of gas emissions have been computed from field research by agricultural researchers.V. FERTILIZERS In order to obtain locally produced, as well as imported quantities, the previous approach was used with the producer and customs. 5.1. ProductionFertilizers are produced by ICS (Chemical Industries of Senegal).
NPK Fertilizers
Year Total production (tonnes) 1989 116 780 1990 138 658 1991 171 045 1992 159 606 1993 130 643
Source : ICS.5.2. ConsumptionFertilizers are marketed in Senegal by SENCHIM/ICS Fertilizer Sales in Senegal
Type % des éléments constitutifs Total
Azote (N) Phosphore Potassium (tonnes)
Diamonium de
phosphore 18 46 0 2 482
(DAP)
Diamonium de
phosphore 10 10 20 1 662
(DAP)
Diamonium de
phosphore 6 20 10 6 137
(DAP)
Diamonium de
phosphore 8 18 27 1 439
(DAP)
DAP soufré et 8 18 27 8 700
boré
Trisuperphos-ph 0 40 0 1 897
ate (TSP)
Double
superphosphate 0 30 0 2 093
(DSP)
Double
superphosphate 14 22 12 5
(DSP)
Urée 46 0 0 7 548
Source : SENCHIM / ICS.5.3. Imports
Type d'engrais Importations (tonnes)
Nitrate de sodium 8
Nitrate d'ammonium 23 207
Urée 22 298
Autres engrais chimiques ou minéraux
azotés 151
SENCHIM / ICS.5.4. Total Consumption of Fertilizers in Senegal
Importations 45 664 Production 31 963 Total (tonnes) 77 627
Source : SENCHIM / ICS.As can be seen on the pie chart below, more than half of the fertilizers consumed in Senegal are imported.Source : SENCHIM / ICS.VI. APPLICATION OF IPCC/OECD METHODOLOGY SERIE DE TABLEAUXVII. COMMENTSTABLE 4.1-A :The number of dairy cows was determined from a survey by the Department of Animal Husbandry on the cattle structure of herds of cattle. They account for 37.5 % of the cattle.Using census by aircraft, it is difficult to discriminate between sheep and goats. As a result the two have been counted together and refered to as "small ruminants". However, the difference between the emission factors is very small and has no fundamental bearing on the calculations.TABLE 4.3-A :The biomass density of each type of savannah and the biomass exposed to burning, have been directly determined from NOAA-AVHRR satellite images validated by field measurement. This means that biomass exposed is not a result of the multiplication of average biomass per hectare by burned area (in ha) as proposed by the IPCC/OECD. It has been directly extracted from the biomass map, created by the Centre for Ecological Monitoring, using a module of the Idrisi GIS software. The method consists of superimposing the fire map on that of plant production, permitting the extraction of the fraction effectively burned. The figure obtained was multiplied by 0.8 as indicated in the methodology.The share of woody biomass out of the reach of fires has not been taken into account.TABLE 4.4-A :Groundnut leaves have not been taken into account in the computations since they are not burned. They are almost entirely consumed by livestock. The quantities of burned crop residues are very small in the Sahelian context due to the multiple uses for these products (making fences, feeding cattle) and to the great losses in nutritive value during the eight to nine months of the dry season (cattle trampling, rotting, wind action..). Burning the small quantities remaining takes place when ploughing the fields for the next crop, that is nearly eight months after harvest.TABLE 4.4-B:In the Senegalese statistics there is no difference between millet and sorghum which have always been accounted together since 1960. However, we have estimated that millet represents about 80 % of the total, leading to the calculation of the Nitrogen-Carbon Ratio on the basis of a combination : (80 % of 0.016 + 20 % of 0.02).The carbon fraction contained in cotton is not available, not even the default value; consequently, the potato coefficient was used. The same applies for the Nitrogen-Carbon Ratio, where that of soy-bean was used because the characteristics of these two products are quite similar.As far as cotton is concerned, no study has been carried out in Senegal on the Residue/Production Ratio, or on the Dry Matter Content of stalks. The figures used resulted from discussions with professionals in the cotton sector.