The profitability of traditional and innovative mulching techniques using millet crop residues in the West African Sahel

Agriculture Ecosystems & Environment ELSEVIER

Agriculture, Ecosystems and Environment 67 (1998) 23-35

The profitability of traditional and innovative mulching techniques using millet crop residues in the West African Sahel John Lamers a, * Michael Bruentrup a, Andreas Buerkert b Institute of Agricultural Economics and Social Sciences in the Tropics (490), Unil,ersiO, of Hohenheim, 70593 Stuttgart, Germany b ICRISAT, B.P. 12404, Niamey, Niger

Accepted 16 May 1997

Abstract

To try to maintain soil fertility in the West African Sahel (WAS), research and development organisations promote the use of mineral fertilizers, especially phosphate (P), and an improved management of crop residues (CR). While researchers have achieved considerable increases in yield using broadcast mulch rates of 2000 kg CR ha l, farmers traditionally apply CR only on selected microsites with low productivity. The economic analyses in this paper were based on the results of an on-station field experiment presented previously. The yields were analysed according to a set of hypothetical application methods, that is assuming that: (1) all fertilisers were broadcasted evenly, (2) all fertilisers were applied to poorly producing microsites only, and (3) all fertilisers were applied to good producing microsites only. Linear programming models were used to estimate the comparative advantages of traditional and innovative mulching techniques for typical farms in western Niger. The following scenarios were tested: (1) application of CR only to poorly producing microsites when the latter comprised 10%, 30% and 50% of the total farm land; (2) applying CR mulch to poorly and good producing microsites or broadcasting it with or without the addition of P fertilizers; and (3) the amount of CR available for mulching was compared between sites where there was or was not a market for buying and selling CR. The data came from an experiment conducted between 1991 and 1992 at the ICRISAT Sahelian Centre in Niger and from several household surveys used to characterise a typical farm in western Niger. The optimum solutions to the linear equations suggest that farmers' concentrated microsite techniques are more profitable than broadcast applications due to the higher productivity of the traditional techniques per unit of labour input. At a proportion of 10% poor microsites, it is not profitable to treat the poorly producing parts. However, as the proportion of the low fertility land increases the profitability of treating these areas increases. High relative costs of P fertilizer and low millet grain prices render P application beneficial only if applied to good producing microsites. In the short run, the farmers' mulching techniques are currently rational from an agronomic and economic viewpoint. Yet, as the proportion of low fertility land increases, traditional techniques need more CR which can only be assured by P inputs. A more favourable agricultural price policy may improve adoption of P fertilizers by Sahelian farmers. © 1998 Elsevier Science B.V. Keywords: Sahel; Farm economics; Fertility management; Millet; Erosion control; Crop residues; indigenous knowledge

* Corresponding author. Brabantstr. 28, D-52070 Aachen, Germany. Tel.: +49-241-502566: fax: +49-241-502566. 0167-8809/98/$19.00 © 1998 Elsevier Science B.V. All rights reserved. PII S0167-8809(97)00092-3

24

J. Lamers et al. / A griculture, Ecosystems and Encironment 67 (1998) 23-35

I. Introduction The West African Sahel (WAS) is an area characterised by continuously declining soil fertility and increasing soil degradation through wind erosion. Assuring food production for the rapidly growing population is a constant challenge to agricultural research. Repeatedly it has been shown (Bationo et al., 1994) that mineral fertilizers and the application of millet crop residues (CR) improve the production of pearl millet (Pennisetum glaucum L. R. Br.) on the acid sandy soils characterised by low levels of organic matter, low cation exchange capacity and low levels of minerals such as phosphorus and nitrogen (West et al., 1984). However, a review showed that in Sahelian regions receiving less than 800 mm rainfall, there is an extremely variable response of millet to mineral fertilizers (Deuson and Sanders, 1988). Also they are relatively costly and not always available. Mulch, broadcasted at 2000 kg CR ha M, increases grain and stover yields of millet due to the mechanical (Michels et al., 1993) and the nutritional effects associated with the decomposition of crop residues (Bationo and Mokwunye, 1991a; Rebafka, 1993). Lamers and Feil (1995) observed that farmers applied crop residues only to selected areas in fields characterised by low productivity rather than by broadcasting. The phenomenon of spatial variability, which results in good and poor millet growth in close proximity, has received much attention. Most studies describe the spatial soil and crop growth variability in terms of physical and chemical soil properties (e.g., Brouwer et al., 1993; Geiger and Manu, 1993; Herrmann et al., 1994). Brouwer et al. (1993) defined this as: " . . . variability over relatively short distances (approximately 2-50 m), with differences in soil properties which do affect hydrological behaviour and plant growth " Spatial variability in research experiments leads to high coefficients of variation during statistical analyses which makes the interpretation of results difficult (Buerkert, 1995). Here, for simplicity, we use the term 'good' microsites for areas with high inherent productivity and "poor' microsites for those with low productivity. The agronomic efficiency of the farmers' microsite application was recently demonstrated in an on-station experiment (Buerkert, 1995). Abso-

lute millet grain and stover yields on good microsites were significantly higher than on poor microsites. In contrast, the relative yield increase due to CR applications was highest on poor microsites. Scarcity of labour, particularly during weeding, is a major constraint in the traditional millet cropping systems (Baidu-Forson, 1988). Weed control during the first 30-40 days after planting avoids losses in millet grain yields of 10-30 kg ha-I for every day of weeding delay (McIntire et al., 1989). Yet, the application of 2000 kg CR ha-~ stimulates weed growth and increases the labour needed for weeding by 25-30% (Lamers and Bruentrup, 1996). This is decisive for the profitability of a broadcast soil mulch. Farmers' microsite applications of CR not only augment the yields per unit of applied CR, but also reduce the increase in time required for weeding because the mulched surface to be weeded is smaller. Little is known concerning the most economic use of CR in the WAS. McIntire et al. (1992) postulated that feeding CR to livestock is the most beneficial use in the long run. Lamers and Bruentrup (1996) showed that the current practice of Nigerien farmers of using stover for many purposes, such as feed, mulch, heating, construction, instead of for one a single objective, is the most profitable alternative farming in the short and long run (Bruentrup et al., 1997). Yet, these studies compared only broadcast mulching with alternative uses and neglected farmers' strategies. This paper examines the comparative advantages of farmers' microsite applications with CR and broadcast mulching for a typical farm in western Niger.

2. Methodology The economic analyses in this paper were based in part on the results of an on-station field experiment with a 3 × 2 factorial design reported by Buerkert et al. (1995). Details on the experimental design, the definition of microsites with a 'good' and 'poor' productivity, the specific soil characteristics of both as well as biomass production on those microsites are not presented here. This ex post economic analysis is a companion to the publication from Buerkert et al. (1995). Buerkert et al. (1995) reported results of several treatments. The first set of treatments (CR uses)

J. Lamers et al./ Agricuhure, Ecosystems and Ent,ironment 67 (19981 23-35

were: (1) apply millet CR at 500 kg ha ~ (500), (2) apply millet CR at 2000 kg ha -~ (2000), and (3) burn 2000 kg millet CR ha -~ and apply the ash (ash). The second set of treatments (phosphates) were: (1) apply 13 kg P ha ~ (P) as 167 kg of single super phosphate (SSP) and (2) apply no phosphate. A control was also run which received none of these treatments. The parameters measured for each combination of treatments were the yields (kg ha -~) of grain and crop residues. These yields were analysed according to a third set of hypothetical application methods, that is assuming that (1) all fertilisers were broadcasted evenly (broad), (2) all fertilisers were applied to poor microsites only (poor), and (3) all fertilisers were applied to good microsites only (good). Tables 1 and 2 show the average values obtained for the observation years 1991 and 1992 and the derived values for an economic analysis. The profitability of the treatments was analysed by linear programming (LP) under the resources and demands of western Niger households (Table 3). The LP model maximizes the aggregated gross margins

25

of the various mulching techniques, with or without P, which were estimated as the gross outputs minus the variable costs. Gross margins are expressed as return per unit land, total labour and labour for weeding. The labour requirements for the various treatments were recorded in a different experiment but one that was conducted on the same site and during the same years (Lamers and Bruentrup, 1996). The observed yield levels in this study were generally lower than those in the experiment of Buerkert 11995) so the latter were adjusted (Tables 1 and 2). On this basis it was estimated that one adult harvested 40 kg grain h J and 180 kg stover h -~ (Lamers and Bruentrup, 1996). Baidu-Forson et al. (1994) found that weeding time increased with increased application of SSP. Therefore. the labour demand of 21 h ha J relating to SSP use (the application and additional weeding demands) was derived from the experiment that had used the same quantity of SSP (Baidu-Forson et al., 1994). In the LP models it was assumed that a maximum of 10% of the cash income generated through millet

Table 1 Gross margin estimates as return per unit land, total labour and labour for weeding, for different mulch techniques with crop residues without an application of phosphorus fertilizer Input

Unit

Total labour Weeding CR P Purchase CR Total

h ha -~ h ha- i kg ha- ~ kg ha ~ kg ha ] CFA ha- ~

Control

500broad

500poor

500good

179 52 0 0 0 0

171 55 500 0 0 0

152 55 500 0 0 0

190 55 500 0 0 0

2000broad 194 64 2000 0 670 7906

2000poor 193 64 2000 0 670 7906

2000good 194 64 2000 0 670 7906

ashbroad 189 52 2000 0 705 8319

ashpoor ashgood 193 52 2000 0 880 10 340

185 52 2000 0 530 6254

469 1165

750 1475

550 1100

950 1850

1025 1900

1000 1900

1050 1900

850 1850

750 16011

950 2100

CFA ha- ] CFA ha- i CFA ha- I

39 680 9623 49 303

60 000 6284 66 284

44 000 3186 47 186

76 000 9381 85 381

82 000 0 82 000

80 000 0 80 000

84 000 0 84 000

68 000 0 68 000

60 00(1 0 60 000

76 000 0 76 (X)0

CFA ha-~ CFA ha -] CFA ha-]

49303 275 948

66284 387 1205

47 186 311 858

85381 448 1552

74094 382 1158

72094 373 1126

76094 391 t 189

59681 315 1148

49616 256 954

69746 377 1341

Output Grain CR

kg ha-i kg ha i

Value c~fOutput Grain CR Total

Gross margin Land Total labour Weeding

Average grain and stover yields as effected by crop residues (CR) and phosphorus (P) according to poor and good producing microsites are adapted fi'om Buerkert (I 995). Crop residue uses: 500 kg CR ha I 2000 kg ha- i and the ashes of 2000 kg CR ha- i burned applied broadcasted (broad~ and on good and poor producing microsites.

26

J. Lamers et al. / Agriculture, Ecosystems and Em,ironment 67 (1998~ 23-35

Table 2 Gross margin estimates as return per unit land, total labour and labour for weeding, for different mulch techniques with crop residues and an application of phosphorus fertilizer Input

Unit

P500broad P500poor P500good P2000broad P2000poor P2000good Pashbroad Pashpoor Pashgood

Total labour Weeding CR P Purchase CR Total

h ha-I h ha- i kg hakg hakg ha-I CFA ha- i

231 73 500 165 0 10725

212 73 500 t65 0 10725

250 73 500 165 0 10725

206 82 2000 165 285 14088

206 82 2000 165 355 14914

205 82 2000 165 215 13 262

204 70 2000 165 163 12 643

208 70 2000 165 390 15 327

201 70 2000 165 0 10725

1075 2275

900 1900

1250 2650

1250 2450

1150 2350

1350 2550

1350 2625

1200 2300

1500 2950

CFA ha CFA ha CFA ha-

86000 12 892 98 892

72000 9794 81794

100000 15 989 115 989

100000 0 100000

92000 0 92000

108000 0 108000

108000 0 108000

96000 0 96000

120000 767 120767

CFA ha CFA haCFA ha-

88167 382 1201

71069 335 968

105 264 422 1434

85912 418 1043

77086 374 936

94738 46t 1150

95358 466 1355

80673 388 1146

110042 547 1563

Output

Grain CR

kg hakg ha-

i t

Value of output

Grain CR Total Gross Margin

Land Totallabour Weeding

Average grain and stover yields as effected by crop residues (CR) and phosphorus (P) according to poor and good producing microsites are adapted from Buerkert (1995). Crop residue uses: 500 kg CR ha- 1 2000 kg ha- t and the ashes of 2000 kg CR ha- i burned applied broadcasted (broad) and on good and poor producing microsites.

production could be spent to hire labour at a daily w a g e rate o f 1000 F C F A . M o n e y was m a d e available in increments o f 2000 F C F A up to a m a x i m u m of 5 0 0 0 0 F C F A . Previous studies in the same region s h o w e d that e x p e n s e s for hiring labour and purchas-

Table 3 Main features of households in western Niger Feature

Amounts Unit

Total farm size In fallow

10 30

ha %

Family size Working members

[I 36

person c/c

Subsistence requirements 200 Self-sufficiency level 90

kg millet per person per year %

Total working capacity Weeding time capacity

days per person days per person

Livestock possession

200 45 3

Tropical Livestock Unit (TLU)

Source: Baidu-Forson, 1988: McIntire et al., 1989.

ing inputs vary extremely, but are always a major part o f the annual expenses (Baidu-Forson, 1988; McIntire et al., 1989). In line with their results, in the f o l l o w i n g expenses b e t w e e n 2 0 0 0 - 1 6 000 F C F A per year are considered low, expenses b e t w e e n 16 000 and 3 4 0 0 0 F C F A m e d i u m , and cash expenses of m o r e than 34 000 F C F A per year are defined as high. The prices for grain, stover and S S P were 80, 11.8 and 65 F C F A kg ~ respectively and labour cost 700 F C F A for a 6-h day. Prices and r e v e n u e s are presented in F C F A at the rate prior to the devaluation of this currency in January 1994, w h e n US$1 was e q u i v a l e n t to 270 F C F A . M o r e details on the LP m o d e l are presented e l s e w h e r e (Lamers and Bruentrup, 1996). In the base case of the LP model, the o p t i m u m farm organisation was analysed assuming 30% o f the available surface was poorly producing (thus 70% good) with or without the possibility for the farmer to sell or buy CR. The effects o f the three C R uses 500, 2000, ash and a bare control were c o m p a r e d concurrently. The m o d e l investigated the C R uses in

27

J. Lamers et al. /Agriculture, Ecosystems and Enuironment 67 (1998) 23-35

combination with the three hypothetical application methods, i.e., 500poor, 2000poor and ashpoor on poorly producing microsites; 500good, 2000good and ashgood on good producing microsites; and 500broad, 2000broad and ashbroad for broadcast application. Finally the CR uses and methods of application were further examined in combination with the phosphate (P) treatment to give P500broad, P500poor, P500good, P2000broad, P2000poor, P2000good, Pashbroad, Pashpoor, and Pashgood. To test the sensitivity of the optimum solutions for different proportions of poor and good producing land, two scenarios of 10% and 50% poor producing microsites were analysed. At the onset of the rainy season 2000 kg CR ha i were not available on those treatments requiring these amounts (Tables 1 and 2). In these cases it was assumed that the shortfall in CR supplement was purchased.

3. Results 3.1. The economics of" the mulching techniques with crop residues

Details of the gross margin estimates used in the base case and the sensitivity analyses are given in

Table 1. Without the use of P, the highest return per unit land was found with 500good and 2000good, whereas the lowest return was associated with 500poor, Highest labour productivity was obtained for 500good and the lowest for the bare control. Highest return per unit weeding was recorded for 500good and ashgood and lowest for 500poor. Based on these estimates, the CR uses, which depended on the cash available to hire labour, showed that only 500good, 2000good, ashgood and 2000poor could be considered as optimum solutions under the assumptions made in the base case (30% poor producing microsites and without trading CR) (Fig. 1). At 30% poor microsites, a maximum of about 3.4 ha could be cultivated when using family labour only. The optimum solution consisted of ashgood and 500good, because applied in combination they gave the best results when scarcity of" labour for weeding was a constraint. At low (between 0 and 16000 FCFA) and medium (between 16000 and 34 000 FCFA) cash expenses for hiring labour, it was beneficial to cultivate more land with a low labour demanding mulching technique (500good) as it had high returns to labour per unit of land (Table 4). In contrast, at high cash amounts (more than 34000 FCFA) profitability could be increased by selecting

Crop residue uses as influenced by cash available for hiring labour without buying crop residues. 30 % poor spots

6 5

g4

1 I ~

= i o

0

10000

F ~ 500 good

-

20000 30000 cash available(FCFA) 2000 poor

.=, 2000 good

40000

~ ash good

--

50000

total a r e a

I

Fig. 1. Crop residue uses as influenced by cash availablefor hiring labour without buying crop residues. 30% poor microsites.

28

J. Lamers

et

aL / Agriculture. Ecosystems and Em,ironment 67 (1998) 23-35

Table 4 Summary of crop residues uses (CR) a in the optimum solution as influenced by cash expenses of the households, the purchase of CR, different proportions of poor producing microsites (10%, 30%, 50%) and an applicationof phosphorus (0 and 13 kg P ha i )

P application

Proportion

Cash expenses without buying crop residues

Cash expenses with buying crop residues

poor

Lowb

High

Low

Medium

High

10%

ashgood; 500good

500good; 2000poor; 2000good

500good

500good; 2000good; 2000poor

500good: 2000poor; 2000good

500good: 2000poor; 2000good 2000poor; 500good; 2000good

500good; 2000poor

500good: 2000good; 2000poor 500good: 2000good: 2000poor

500good; 2000good 2000good 500good; 2000poor

500good; Pashgood

Pashgood; 500good; 2000good:

500good: Pashgood

Pashgood: 500good; 2000good; 2000poor

Pashgood; 2000good: 500good: 2000poor Pashgood: 2000poor; 500good; P500good; 2000good

500good; 2000poor: Pashgood; Pashgood; 2000poor P500good; 500broad

microsites

0 kg P ha-I

30%

50%

13 kg P ha -I

Medium

500good; 2000poor; 2000good;ashgood ashgood;500good; 500good: 2000poor 2000poor; 2000good ashgood; 500good; 500good: 2000poor: 2000poor 2000good:ashgood

10%

Pashgood;ashgood: 500good

Pashgood: 2000good: 500good

30%

Pashgood;ashgood:500good

Pashgood; 2000good; 500good: 2000poor

Pashgood; 2000good: 500good: 2000poor Pashgood; 2000poor; 500good; P500good

50%

Pashgood;ashgood;50Ogood: 2000poor

Pashgood; 500good: P500good: 500good

2000poor; Pashgood: 500broad; P500good:

500good; 2000poor

~'Crop residue uses: 500 kg CR ha i 2000 kg ha- I and the ashes of 2000 kg CR ha- ~ burned applied broadcasted (broad) and on good and poor producingmicrosites. bLow expenses: less than 15000 FCFA. Medium: between 16000 and 34 000 FCFA. High: more than 34 000 FCFA.

those mulching techniques that gave high yields (2000poor and 2000good), but which had high labour demands as well. In all cases, 500good was part of the optimum solution since it was the best CR supplier for the CR consuming techniques 2000good and 2000poor. The sensitivity analysis showed that the combinations of the optimum CR uses remained constant, even though proportions of poor microsites other than 30% subjected to different CR uses were considered (Table 4). At higher proportions (50%) of poor producing microsites, 2000poor entered the optimum solution at low expenditures whereas at 10% poor producing microsites 2000poor became one of the optimum solutions only at medium cash expenses. At 50% poor microsites it was profitable to

apply CR to the poor producing areas, however at 10%, applying it to the good producing parts was preferable. The optimum mulching technique with CR changed with the possibility of buying or selling CR (Fig. 2). At 30% poor microsites, only 500good entered the solution and CRs were sold since the shadow price of an internal use is lower than the recorded market price of 11.8 FCFA kg-1 (Fig. 3). The revenues resulting from CR sales are used to hire labour. At low cash expenses, 2000poor in combination with 500good was the most profitable option and therefore CR was partly transferred from 500good to 2000poor and partly sold (Fig. 3). At medium cash expenses, 500good remained the main CR supplier for 2000poor and 2000good and CR

J. Lamers et al. /Agriculture, Ecosystems and Encironment 67 (1998) 23-35

29

Crop residue uses as influenced by cash available for hiring labour with buying crop residues. 30 % poor spots 8

"O

'~

2

'::

'> :2:-..:

~'

•

°

°

°

•

•

•

•

:,

o

0

•

10000

20000

50000

4OO00

30000

cash available (FCFA)

: =L 500 good

•

2000 poor

'= 2000 good

total area

n I

Fig. 2. Crop residue uses as influenced by cash available for hiring labour with buying crop residues. 30% poor microsites.

income and trade of commodities as influenced by cash available for hiring labour with buying crop residues. 30 % bad spots 5000

TI 700000 f3

•

o ]

4000

~

L~

;600000

c:

: -500000

::~ 3000

':

~

<

-400000 -300000

~=2000

~200000

~

1000 9100000

-7

I

0,~ 10000

20000

30000

40000

50000

cash available (FCFA)

total income ,= sale of grain .... sale of crop residues (CR) + purchase of grain i purchase of crop residues (CR)

I

Fig. 3. Income and trade of commodities as influenced by cash available for hiring labour with buying crop residues. 30% poor microsites.

J. Lamers et aL /Agriculture, Ecosystems and Environment 67 (1998) 23-35

30

sales were reduced to less than 500 kg. At high cash expenditures, the optimum solution consisted of 500good and 2000poor only and CR was transferred from the good to the poor microsites. Burning CR was not at any point a profitable option. At 10% and 50% poor producing parts, mulching strategies remained constant irrespective of the proportion of poor producing microsites, though this depended on the cash availability (Table 4). At 10% poor producing microsites and low cash expenses, only 500good entered the solution. At 50%, 2000poor entered the solution at low expenses (from 2000 FCFA onwards), whereas 2000good was only lucrative at medium expenses. At 50% poor microsites and low expenses (between 2000 and 14 000 FCFA) CR was purchased. A CR purchase never occurred in any other case (Fig. 3). At a cash availability of 46 000 FCFA, CR was transferred and sold.

3.2. Crop residues in combination with phosphate fertilizer The gross margin estimates for the different CR uses in combination with P are presented in Table 2. At 30% poor microsites and without the option to

trade CR, all optimum solutions at low, medium and high cash expenses are dominated by Pashgood (Fig. 4). Also included in the solution from the onset is ashgood, however, it is completely replaced by 2000poor and 2000good at medium expenses. At 38000 FCFA, 2000good is replaced by P500good. Pashgood and P500good have high returns to land and labour during weeding (Table 2). At higher or lower proportions of poorly producing microsites, the same uses of CR entered the solution as at 30% (Table 4). At 50% poorly producing microsites, these CR uses appear profitable even at low cash expenses but at 10% poor microsites they are profitable only at medium and higher cash availability. Compared to the same situation without P (Fig. 1), land was not limiting when using P at low, medium or high cash availability. Thus even when household labour was complemented by hired labour, the total cultivated area did not increase to the maximum possible (7 ha). A broadcast use of CR appeared as a solution for the first time at 46000 FCFA and 50% poorly producing microsites (Table 4), but only in combination with P application (Pashbroad). This option has a high return to total labour (Table 2).

Crop residue and fertiliser uses as influenced by cash available for hiring labour without buying crop residues. 30% poor spots 7

6 5 .C

v 4 "10 r-

z3

I

~..~-

1

~

~

to

~

S . . . . . . . . . . .

0

~:='Z

~ ........... ~,, . : - . . . , : : , ~ " , ' ~

20000

10000

~

500 good

-

P500 good - ~

2000 poor

o

o

~

~

30000

cash a v a i l a b l e ,

e

~

',

Pash good m

........

40000

:

50000

(FCFA)

2000 good

~

ash good

total area

Fig. 4. Crop residue and fertilizeruses as influencedby cash availablefor hiring labourwithoutbuyingcrop residues. 30% poor microsites.

J. Lamers et al, / Agriculture, Ecosystems and Environment 67 (1998) 23-35

31

Crop residue and fertiliser uses as influenced by cash available for hiring labour with buying crop residues. 30 % poor spots 7

5

.

6 i

0

.

.

.

.

~

10000

20000

30000

40000

50000

cash available(FCFA)

.... 500 good

° 2000 poor .~ 2000 good l Pash good .,- total area

, ~- P500 good ~

I

Fig. 5. Crop residue and fertilizer uses as influenced by cash available for hiring labour with buying crop residues. 30% poor microsites.

At 30% poor producing microsites and with the possibility to trade CR, the cropping pattern consisted of 500good and Pashgood at low cash expenses (Fig. 5). At 38000 FCFA, 2000good is replaced by P500good. At 10% poorly producing microsites, the optimum solutions are more or less similar to 30%, except that 2000good remained in the solution and P500good was not included (Table 4). At 50% poorly producing microsites, 2000poor was included at low expenses (6000 FCFA) and P500good at medium expenses whereas Pashbroad became part of the optimum solution only at cash expenses of 46000 FCFA and more. Compared to the situation without P, land was not limiting and the possible 7 ha were never fully used for cultivation.

4. Discussion 4.1. The profimbili~, of the mulching techniques with crop residues The data which formed the basis for this study are supported by on-farm outcomes which showed that millet grain yields on more productive sites in farm-

ers fields, corresponding to micro-highs, are 120% higher than on micro-lows (Geiger and Manu, 1993). These authors thus recommended that the management of sites with different productivity potentials should be adapted to their specific conditions as they may react differently to inputs such as fertilizers. A synopsis of the mulching techniques that emerged in the optimum solutions shows that farmers' microsite-like applications are, in the short-term, and under the current input-output price relationships more profitable than broadcast CR applications (Table 4). Broadcasting CR did not appear in the optimum solutions even though certain combinations of CR uses such as 2000poor and 2000good (e.g., Fig. 1) indicated that under special conditions a broadcast application (2000broad) may become economical. This situation occurred at medium to high cash expenses, but the proportion of 2000poor and 2000good in the optimum solution were unequal. The broadcast application was at a disadvantage since the observed labour demands are still slightly higher than with 2000poor or 2000good. Lack of cash limited the expansion of the high input mulching techniques, such as 2000broad or even 2000good. The second highest return to land is

32

J. Lamers et al. /Agriculture, Ecosystems and Encironment 67 (1998) 23 35

obtained with 2000good but due to its high labour demands it was not profitable to use 2000 kg CR ha-~ with 50% poor producing microsites, at least not when CR could be traded. At a low cash availability, the uses with the highest land to labour efficiency, ashgood and 500good, became part of the optimum solution as long as scarcity of labour for weeding was a constraint. The importance of weeding in the labour budgets of smallholder farmers in western Niger is illustrated by Baidu-Forson (1988). He showed that about 60% of household expenditures for cropping activities, varying between 4000 to 13000 FCFA per household~ were used to hire manpower for weeding. Similar results were found for traditional maize cultivation in Togo and B~nin (Koch and Lutzeyer, 1992). It may thus be generalised that improved technologies with higher labour demands, particularly during peak periods such as weeding, or technologies which need high capital inputs will have reduced chances of being adopted by smallholders in West Africa despite the gains in output and despite farmers' desire to counteract soil erosion and declining soil fertility. The farmers' traditional strategy to apply high concentrations of organic material on poor producing microsites (Lamers and Feil, 1995), simulated by 2000poor, is selected in most solutions. Also in accordance with farmers' strategies, the LP model selected only 2000poor and never 500poor or ashpoor. As soon as good producing microsites become limiting, 2000poor is an economically attractive option because it has a high marginal net return to labour and land. At higher proportions of poorly producing microsites, the necessity to cultivate these microsites increases, whereas at 10% poorly producing microsites, it is more beneficial to exploit the better producing parts. Yet, given the population growth rate of above 3% per year, the proportion of low fertility land under cultivation will increase and consequently traditional farming techniques will need more CR than is currently produced. To sustain these techniques, more CR must be produced. Recent studies demonstrated that Nigerien households particularly in the proximity of urban centres may increase their income through the sale of CR (Baidu-Forson, 1994: Speirs and Olsen, 1992), leaving many fields bare and thus more prone to erosion. When trading CR was incorporated in the LP model,

ashgood, which dominated under the assumption that CR could not be traded, did not enter the solution at any point. This illustrates that farmers' strategies which involve burning CR make sense (to reduce labour demands during the weeding peak periods), as long as selling CR is not possible. However, when there is a high proportion of poorly producing microsites, the shadow price of an internal use of CR rises. The economic use of CR then becomes the transfer of surplus CR from 500good to 2000poor rather than burning or selling it. Yet, this option is only valid as long as farmers own sufficient capital to cover the associated increases in labour demand. The fact that farmers normally burn CR after satisfying livestock and domestic needs is thus an economically logical strategy, at least in the short run and as long as labour for weeding is a constraint. Moreover, burning CR prevents infestation by the stem borer Coniesta ( Acigona) ignefusalis which may cause major crop losses (Ma'iga and Issa, 1988). Unfortunately, the complete removal of millet crop residues for 6 - 8 yrs consecutively reduced millet grain yields to 39 kg ha J owing to the continuous export of P and potassium, which in traditional millet cropping are not replaced (Rebafka, 1993). Although the export of potassium is reduced when CRs are burned rather than physically removed, much more P is removed by burning CRs than by mulching them (Buerkert, 1995).

4.2. The profitabilio, of crop residues combined with a phosphate fertilizer In the short run, farmers' traditional mulching strategies appear rational, yet it should be remembered that farmers do not have many alternatives to a concentrated use of CR on selected microsites. Without the use of mineral fertilizers, CR yields are between 800-1500 kg ha-~ (McIntire et al., 1992). In addition, due to the many uses of CR the material left on the field is often insufficient for a broadcast mulch and certainly at a rate of 2000 kg ha (Lamers and Feil, 1995). It may thus be argued that high levels of stover production would ease the competition between the alternative uses. Therefore, P application becomes a key input to make more CR available and to satisfy the increasing CR demands. However, resource poor farmers will not implement

J. Lamers et al. / Agriculture, Ecosystems and Em,ironment 67 (1998) 23-35

innovations if these are not economically viable, particularly since 13 kg P ha-~ alone did not always guarantee a CR production of 2000 kg ha- 1 (Bationo and Mokwunye, 1991 a). Only the high yields on good microsites could cover the costs of applying P fertilizers. The relatively high fertilizer prices, the low returns due to low millet grain prices and the unpredictable agroclimatic conditions, all act as disincentives for farmers to invest in millet production and militate against the use of the external inputs, such as P fertilizers, necessary for improving millet production. It should be noted however that the yield increases with the P application rate used in this study are lower than those found in previous studies (Bationo and Mokwunye, 1991b). This may be due to the fact that in contrast to the experiment presented here, nitrogen was also applied. On the acid sandy soils in the southern Sahel, nitrogen next to phosphate is the most limiting nutrient for millet production (Buerkert, 1995). Under traditional millet cropping practices, it is evident that the proportion of land with low productivity (poor microsites) will increase. Moreover, due to the high opportunity costs of CR, farmers may use them to increase their present income and thereby sacrifice the long-term production capacity of the soil. The use of P fertilizer alleviated the need for more land which consequently did not become limiting. The use of P may thus reduce overcultivation and soil degradation, two prevailing problems in the WAS (Mclntire et al., 1992). However, the cultivable area ceased to be limiting mainly at medium and high cash expenses, a situation which occurs only in the smaller, richer part of the Nigerien farming community (Baidu-Forson, 1988). The absence of profitable returns cannot of itself justify a decision against the use of mineral fertilizers, especially in a region which is characterised by soils of low inherent fertility that are prone to erosion. There are compelling reasons for investing in fertilizers. Postponing the use of yield increasing fertilizers is questionable because the opportunity costs of farmers' foregone income are high. Moreover, if nothing is done to sustain the soil fertility now, the Sahelian population will require support

33

from external sources in the future. This future support would in all likelihood exceed the present cost of making fertilizers accessible. Financial incentives are therefore justified to stop further depletion of land in the WAS. Price policy measures should be taken such as subsidising mineral fertilizers and supporting floor grain prices in good production years as suggested earlier by Deuson and Sanders (1988). Under a more favourable agricultural price policy farmers are likely to respond to the new technologies of resource conservation such as P application and broadcast CR mulching.

5. Concluding remarks It cannot be expected that farmers will implement innovations proposed by research if these are not economically viable, even if they counteract important problems such as wind erosion or the further decline in the already low fertility of these sandy soils. The most economical mulching techniques with millet crop residues (CR) depend on the proportion of poor and good producing microsites in the farmers' fields and the possibility of trading CR. As long as the proportion of poor producing land is low, the farmers" strategy of mulching the selected poor microsites is more profitable than broadcasting CR. An increase in the proportion of these poor areas renders broadcast applications more profitable and the shadow price for the internal use of residues makes it more profitable to use it as a mulch than to sell it or burn it. The use of mineral phosphate fertilizer was not economical in any combination of CR uses when applied to poor producing microsites. In contrast, its use on good producing microsites became viable in combination with 500 kg CR ha-I (P500good) or when 2000 kg CR ha -j were burned (Pashgood). When considering the use of P, available land area was no longer a limiting factor and consequently the maximum of 7 ha was not cultivated. A use of P may thus allow for a reduction in the cultivable area and of over-use and therefore of soil degradation. Given the need to use external inputs such as mineral fertilizers, policy makers must create conditions that

34

J. Lamers et al. / Agriculture, Ecosystems and Environment 67 (1998) 23-35

favour the implementation of innovations for decreasing soil degradation, and increasing soil fertility and productivity. The economic evaluation of the different uses of CR in this study fixed on short-term aspects. At present little information exists on long-term processes, yet such a database may help in future modelling approaches to better predict farmers' decision-making processes and to assist policy makers in deciding upon the best soil conservation techniques in the WAS.

Acknowledgements This study was supported by the German Research Foundation (DFG), via the Special Research Programme 308 of the University of Hohenheim. The authors thank Dr. P. Lawrence for editing their English, Prof. Dr. H. Marschner for commenting on an earlier draft, and to anonymous reviewers for their skilful and supporting comments.

References Baidu-Forson, J., 1988. Characteristics of farm households and economy in western Niger villages: Some evidence from baseline surveys. Resource Management Program, Economics Group, Progress Report 87. International Crops Research Institute for the Semiarid Tropics, Niamey, Niger, 37 pp. Baidu-Forson, J., 1994. Uses and economic value of crop residues and herbaceous plants to rural households in western Niger. Resource Management Program Progress Report 5. International Crops Research Institute for the Semiarid Tropics, Niamey, Niger, 20 pp. Baidu-Forson, J., Renard, C., Deuro, N., Reddy, C., Kadi, M., 1994. Evaluation agro-rconomique de nouveaux syst~mes de production en milieu paysan au Niger. Resource Management Program Progress Report no. 4. International Crops Research Institute for the Semiarid Tropics, Niamey, Nigeria, 22 pp. Bationo, A., Mokwunye, A.U., 1991a. Role of manures and crop residue in alleviating soil fertility constraints to crop production: With special reference to the Sahelian and Sudanian zones of West Africa. Fertiliz. Res. 29, 117-125. Bationo, A., Mokwunye, A.U., 1991b. Alleviating soil fertility constraints to increased crop production in West Africa: The experience in the Sahel. Fertiliz. Res. 29, 95-115. Bationo, A., Buerkert, A., Sedogo, M.P., Christianson, B.C., Mokwunye, A.U., 1994. A critical review of crop-residue use as soil amendment in the West African semiarid tropics. In: Powell, J.M., Fernfindez-Rivera, S., Williams, T.O., Renard,

C. (Eds.), Livestock and Sustainable Nutrient Cycling in Mixed Farming Systems of sub-Saharan Africa, Vol. II: Technical papers. ILCA, Addis. Brouwer, J., Fussell, L.K., Herrmann, L., 1993. Soil and crop growth microvariability in the West African semiarid tropics: A possible risk-reducing factor for subsistence farmers. Agric. Ecosyst. Environ. 45, 229-238. Bruentrup, M., Lamers, J.P.A., Herrmann, L., 1997. Modelling long-term effects of crop residue management for sustainable farming systems. In: Teng, P.S., Kropff, M.J., ten Berge, H.F.M, Dent, J.B., Lansigan, F.P., van Laar, H.H. (Eds.), Applications of Systems Approaches at the Farm and Regional Levels, Vol. I. Kluwer Academic Publishers, Great Britain, pp. 53-64. Buerkert, A., 1995. Effects of crop residues, phosphorus, and spatial soil variability on yield and nutrient uptake of pearl millet (Pennisetum glaucum L.) in southwest Niger. Ulrich Grauer Verlag, Stuttgart, 273 pp. Buerkert, A., Stern, R.D., Marschner, H., 1995. Poststratification clarifies treatment effects on pearl millet growth in the Sahel. Agron. J. 87, 752-761. Deuson, R., Sanders, J.H., 1988. Technology development and agricultural policy in the Sahel-Burkina Faso and Niger. In: Unger, P.W., Sneed, T.V., Jordan W.R., Jensen, R. (Eds.), Challenges in Dryland Agriculture--A Global Perspective. Texas Agricultural Experiment Station, Amarillo/Bushland, TX, pp. 610-612. Geiger, S.C., Manu, A., 1993. Soil surface characteristics and variability in the growth of millet in the plateau and valley region of western Niger. Agric. Ecosyst. Environ. 45,203-211. Herrmann, L., Hebel, A., Stahr, K., 1994. Influence of microvariability in sandy Sahelian soils on millet growth. Z. Pflanzenernaehr. Bodenk. 157, 111-115. Koch, W., Lutzeyer, H.-J., 1992. Untersuchungen zu Ackerunkraut- und Brachevegetation in Togo und B~nin (Studies on weeds and fallow vegetation in Togo and B~nin). Geobot. Kolloq. 8, 19-34. Lamers, J.P.A., Bruentrup, M., 1996. Comparative advantage of single and multipurpose uses of millet stover in Niger. Agric. Syst. 50 (3), 273-285. Lamers, J.P.A., Feil, P.R., 1995. Farmers" knowledge and management of spatial soil and crop growth variability in Niger, West Africa. Neth. J. Agric. Sci. 43 (4), 375-389. Ma'iga, S., Issa, H., 1988. Les principaux insectes nuisibles aux cultures pluviales. In: Reddy, K.C., Berrada, A., Bonkoula, A. (Eds.), Manuel de l'exp~rimentation en plein champ, h Fusage des cadres de d~veloppement agricole, l~re ~dition. Institute National de Recherches Agronomiques du Niger (INRAN), Niamey, pp. 47-68. McIntire, J., Hopkins, J., Lamers, J.P.A., Fussell, L.K., 1989. The millet system of western Niger: I. Crop production. ILCA, Addis Ababa, Ethiopia (unpubl.), 135 pp. McIntire, J., Bourzat. D., Pingali, P., 1992. Crop-livestock interaction in sub-Saharan Africa. World Bank Regional and Sectoral Studies. The World Bank, Washington, DC, 246 pp. Michels, K., Sivakumar, M.V.K., Allison, B.E., 1993. Wind ero-

J. Lamers et al. / Agriculture, Ecosystems and Eneironment 67 (1998) 23-35

sion in the southern Sahelian Zone and induced constraints to pearl millet production. Agric. Forest. Meteorol. 67, 65-77. Rebafka, F.P., 1993. Deficiency of phosphorus and molybdenum as major growth limiting factors of pearl millet and groundnut on an acid sandy soil in Niger, West Africa. Hohenheimer Bodenkundliche Hefte no. 9. Universit~it Hohenheim, 155 pp. Speirs, M., Olsen, O., 1992. Indigenous integrated farming sys-

35

tems in the Sahel. World Bank Technical Paper 179, Africa Technical Department, Washington, DC, 66 pp. West, L., Wilding, L.P., Landeck, J.K., Calhoun, F.G., 1984. Soil Survey of the ICRISAT Sahelian Center, Niger, West Africa. Soil and Crop Sciences Department/TropSoils, Texas A&M University, 66 pp.