Plant and Soil 123, 51-58 (1990), © Kluwer Academic Publishers. Printed in the Netherlands. PLSO 8292

The effect of soil tillage and fertilizer use on pearl millet yields in Niger

C.B. CHRISTIANSON, A. BATIONO and W.E. BAETHGEN

Agro-Economic and IFDC-West Africa Divisions, International Fertilizer Development Center (IFDC), P.O. Box 2040, Muscle Shoals, AL 35662, USA

Received 16 June 1989. Revised January 1990

Key words: millet, Niger, partially acidulated phosphate rock, phosphorus placement, phosphate rock, tillage

Abstract

Farmers in Niger generally do not plow their fields and are therefore unable to incorporate phosphate. Experiments were conducted in Niger to assess the effect of soil tillage, P source, and fertilizer placement on yields of pearl millet (Pennisetum glaucum [L.] R. Br.). Treatments included single superphosphate (SSP) or ground Tahoua phosphate rock (PRT) incorporated into the soil during tillage or SSP surface applied after tillage. In plots which were not tilled, P sources (SSP, PRT, and PAPR-partially acidulated rock) were broadcast on the soil surface with no incorporation. In order to improve P efficiency under zero tillage, P was point placed in the soil near the plant with either broadcast or point-placed urea. Treatments in which tillage was used showed a slight though nonsignificant yield increase over untilled plots. The yield increase did not appear to be due to phosphate incorporation but rather to direct tillage effects on early plant growth. In a comparison of SSP with PRT or PAPR broadcast on soils not receiving tillage, PRT performed poorly relative to the other P sources. SSP outyielded PAPR and PRT in 1986, but in subsequent years, no signficant difference was found between PAPR and SSP. Point placement of P or N near the plant did not significantly increase yields over broadcast treatments even though the millet was planted with wide 1 x 1 m spacing.

Introduction

Significant responses to fertilizer application have been shown in the production of pearl millet (Pennisetum glaucum [L.] R. Br.) in the sandy soils of Niger. In this region, response to P is pronounced (Traor6, 1974) though, even in the presence of adequate phosphate, nitrogen response may be limited by low planting density and poor rainfall distribution (Bationo et al., 1990; Christianson et al., 1990).

Niger has two phosphate deposits, the Parc W rock of low reactivity (citrate soluble P 9.4% of total P) and the Tahoua rock (PRT) of moderate reactivity (citrate solubility 11.8% [citrate solu-

51

ble P 9.4% of total P]). Through PRT can be applied directly, the Parc W rock must be acidu- lated before use as a P source for crops. Through the process of partial acidulation, Parc W phos- phate rock can be treated with half the sulfuric acid necessary to produce single superphosphate (SSP). This partially acidulated phosphate rock (PAPR) costs less to produce than SSP and contains half of its P in a water-soluble form (58% citrate soluble). Typically, experiments conducted in Niger have shown PAPR made from Parc W phosphate rock to be 80% as efficient as SSP (IFDC, 1988).

Most fertilizer experiments in Niger have been conducted on experimental stations where the

52 Christianson et al.

land has been plowed and/or disced, generally using tractor-drawn implements. This land prep- aration allows the establishment of a good seed- bed and the thorough incorporation of preplant phosphate fertilizer. Incorporation of P has been shown to be an important factor in fertilizer efficiency in other parts of the world, especially for ground PR.

However, the small farmer in Niger generally does little to prepare the land before planting other than to remove small shrubs from the site. Mechanized land preparation is extremely rare, and in the region in which this study was con- ducted, plowing with animal traction is also quite unusual. Farmers plant with wide spacing (1 pocket of 3 plants per 2m 2 [Mclntire, 1986]) and, because it is impractical to incorporate fer- tilizer by hand over such a thinly spaced crop, they usually apply the fertilizer directly to the soil surface. Since fertilizer is relatively expen- sive, the farmers often do not apply it before planting but rather apply both N and P to the crop 2-3 weeks after germination and then only to the plants that appear most likely to survive.

The soils of Niger are generally very sandy; they have a low P fixation capacity (Mokwunye et al. 1986) and very high water infiltration rates. This could allow some movement of P and N into the subsoil with fertilizer. Because of their sandy nature, these soils are also highly suscep- tible to wind erosion, a problem that could potentially be exacerbated with plowing. These problems can be especially severe just before planting when the soils are dry after a 7-month period of no rain and high air temperatures (reaching 47°C in April [Sivakumar, 1986]). Any yield advantage achieved by extensive land prep- aration before planting must be balanced by the risk of soil loss by erosion and the cost of purchasing and maintaining the animals and equipment required.

The objectives of the experiments reported in this paper were (1) to determine the yield benefit and response to P applied as SSP and PRT that can be achieved with preplant tillage and phos- phate incorporation, (2) where tillage is not practiced, to determine which fertilizer manage- ment of N and P will optimize yields, and (3) to compare a variety of P sources available in Niger under conditions of minimal land preparation.

Materials and methods

The experiment was conducted in the village of Gobery, 120 km southeast of Niamey, Niger, on a sandy psammentic Paleustalf (95% sand, 1.5% clay, pH [KCI] 4.2, 0.3% organic matter, ECEC 0.64 cmol(+) kg -1, 73 mg P kg -1, and 2 .4mgPkg -~ Bray 1). The annual rainfall re- ceived in 1986, 1987, and 1988 was 551,509, and 732mm, respectively, compared with a long- term average of 605 mm (Sivakumar, 1986). The study was carried out in a farmer's field that had no previous history of fertilizer use. Pearl millet (Pennisetum glaucum (L.) R. Br. cv CIVT) was planted in 5m × 10m plots at the nationally recommended density of 10,000 pockets ha -1 (3 plants per pocket). Where it was required, tillage to a depth of 10 cm was achieved with a rotary tiller (1986, 1987) or by donkey-drawn imple- ments (1988).

The experiment consisted of nine sets of treat- ments (Table 1) with various combinations of tillage, P fertilizer sources, and P and N fertilizer placement. Phosphate was applied in each treat- ment at rates of 15, 30, and 45 kg P205 ha -1 with appropriate control treatments. Urea applied at 30 kg N ha -1 was broadcast in two equal splits at 2 weeks after planting (WAP) and 6 WAP. The trial was conducted using a randomized block design with six replications. In the tilled treat- ments, SSP or PRT was incorporated before planting. In order to separate the effect of tillage per se from that of phosphate incorporation, a third treatment consisted of SSP applied to the soil surface after tillage. In the untilled plots, a comparison was made amongst SSP, PAPR, and PRT, all surface applied without incorporation. In an attempt to simulate the farmer's practice, 3 treatments involved point placement of the N and P fertilizer in holes 6-8 cm deep, 10 cm from the pocket, at 2 WAR Nitrogen in these treat- ments was either point placed or broadcast.

Analysis of variance was performed with the grain and total dry matter yields for each of the 3 years, and linear contrasts were used to de- termine significant differences amongst the treat- ment sets. Regression analysis was then used to describe the millet response to P fertilizer. Orthogonal contrasts were used to study the trend (linear or quadratic) of the crop response

Millet response to tillage and fertilizers in Niger 53

Table 1. Treatments in experiments

Method of P Method of N Treatment P source d application b appication ~

1. Land tilled before planting A SSP BI BR B SSP BR BR C PRT BI BR

2. Land not tilled before planting D SSP BR BR E PAPR BR BR F PRT BR BR G SSP PP BR H SSP PP PP I SSP BR PP

a SSP = single superphosphate, PRT = Tahoua phosphate rock, PAPR = partially acidulated Parc W phosphate rock (50% acidulation). b BI = broadcast and incorporated, BR = broadcast without incorporation.

PP = point placed near pocket. Note: P was applied to 0, 15, 30, and 45 kg P205 h a ~. N was applied at 30 kg N ha x as urea in 2 splits.

to fertilizer application for each of the nine treatment sets.

Results and discussion

Precipitation in 1986 and 1987 approached long- term averages, and yields were generally good though a dry period in August 1987 diminished yields slightly (Tables 2 and 3) (Figures 1, 2, and 3). In 1988, heavy precipitation, especially in August and September (total 497 mm), resulted in reduced yields due to disease and insect ( Raghuva albipunctella, Joannis) damage.

Effect of tillage on yields and on P response

In all 3 years, grain and dry matter yields tended to be higher in tilled plots than in untilled plots, but only in 1987 was this difference statistically significant for grain yield (Fig. 1). This positive response to tillage in 1987 may have been due to drought conditions that developed in August. Tillage promotes early root development in these sandy soils (ICRISAT, 1986) which, in turn, may have resulted in a crop more resistant to drought stress. The effect of tillage was not related to the effect of P incorporation. In the comparison of treatments in which SSP was either incorporated

during tillage or applied to the soil surface imme- diately after tillage, no significant difference in yield response was found between treatments. In Niger tillage has been shown to decrease soil bulk density and thus enhance early plant growth (ICRISAT, 1986). In the first 4 weeks of the trial, plants in the tilled treatments appeared to be larger and more vigorous though these differ- ences were not evident at harvest.

At equivalent rates of P application, PRT did not perform as well as did SSP; yield increases for PRT were 50%-60% lower than those for SSP (at 30kg P205 ha-l) . As in the SSP treat- ments, tillage improved PRT efficiency slightly though these effects were not significant for any of the 3 years.

The climate of Niger is harsh and, as discussed previously, the average farmer in the study vil- lage grows his crops at a subsistence level. In view of these constraints, the yield advantage found with tillage does not appear to have offset its additional costs. Bullocks are expensive to purchase and require feed to help them survive the 7-month dry season in condition to work in May and June. In this study this cost could not be recouped in increased yield gains. In fact, it may be more advantageous from the standpoint of erosion control not to till these soils. The soils in this study had a very low silt content and

54 Christianson et al.

Table 2. Regression model parameters for grain and stover yield in 1986-88

Grain Treatment

Year set /30 /31 /32 R 2 SE

Stover

r0 r1 r2 R 2 SE

1986 A 443 46.8 -0 .49 0.98 127 B 475 49.3 -0 .60 0.99 17 C 411 10.8 - 0.88 96 D 314 50.6 0.62 0.99 63 E 349 17.3 - 0.99 38 F 349 8.6 - 0.89 71 G 311 38.9 -0 .46 0.99 76 H 314 38.9 -0 .45 0.99 64 I 339 24.4 - 0.99 15

1987 A 523 49.0 -0 .52 0.99 24 B 487 66.0 -0 .86 0.98 138 C 499 14.8 - 0.99 39 D 387 53.2 -0 .70 0.98 112 E 384 44.3 -0 .49 0.98 123 F 452 12.6 - 0.96 60 G 392 42.5 -0 .42 0.99 91 H 412 12.4 0.19 1.00 1 I 405 41.4 0.30 0.99 32

1988 A 313 43.4 -0.61 0.96 129 B 264 38.4 -0 .49 0.98 89 C 275 17.9 -0 .17 0.99 39 D 287 34.9 -0 .48 0.99 47 E 304 25.3 -0 .24 0.99 32 F 316 11.3 -0 .08 0.91 83 G 320 29.2 -0 .36 0.96 103 H 295 32.0 -0 .39 0.99 7 I 308 38.8 -0 .49 0.99 48

1,365 57.9 -0 .88 0.93 208 1,408 57.0 -0.65 0.85 403 1,349 27.6 -0 .07 0.97 139 1,496 60.0 -0 .77 0.91 284 1,572 58.9 -1 .00 0.99 55 1,584 27.8 -0 .23 0.97 108 1,610 68.0 -1 .08 0.93 228 1,514 41.0 -0 .40 0.94 205 1,610 44.5 -0 .30 0.96 225

1,338 85.9 -1 .22 0.96 246 1,294 89.6 -1 .13 0.99 49 1,369 26.9 - 0.89 223 1,407 83.8 -1.11 0.98 194 1,539 56.0 -0 .57 0.87 399 1,422 9.8 0.28 0.97 123 1,364 73.6 -0 .85 0.91 387 1,413 30.3 - 0.99 78 1,447 44.2 - 0.96 211

1,365 57.9 -0 .88 0.93 208 1,555 27.7 - 0.78 352 1,365 24.3 - 0.97 101 1,670 25.4 - 0.78 317 1,572 58.9 -1 .00 0.99 55 1,636 17.4 - 0.93 106 1,611 68.0 -1 .08 0.93 228 1,605 22.8 - 0.89 194 1,677 31.2 - 0.94 185

Note: Treatment sets are described in Table 1.

Table 3. Linear contrasts of the various treatment sets

1986 1987 1988

Grain Stover Grain Stover Grain Stover yield yield yield yield yield yield

Ef fec t o f tillage

SSP till vs no till NS NS ** NS NS NS PRT till vs no till NS NS NS NS NS NS SSP BI vs SSP BR both till NS NS NS NS NS NS SSP vs PRT both till ** ** ** ** ** **

Ef fec t o f P source - no till

SSP vs PAPR * * NS NS NS NS NS SSP vs PRT ** ** ** ** ** NS PAPR vs PRT ** ** ** * ** NS

Ef f ec t o f fer t i l i zer m a n a g e m e n t - n o till

SSP (broadcast vs PP) - N broadcast * NS NS NS NS NS SSP (broadcast vs PP) - N point placed NS ** ** ** * NS Mean yield, kg ha -1 917 2,405 1,120 2,441 762 2,261 SE, kg ha-I 263 583 278 604 178 530

Note: SSP = single superphosphate, P R T = T a h o u a phosphate rock, P A P R = partially acidulated Pare W phosphate rock: BI = broadcast and incorporated, BR = broadcast - no incorporation, PP = point placed. NS = not significant, * = significant at P = 0.05, ** = significant at P = 0.01.

Millet response to tillage and fertilizers in Niger 55

1800

1600

1400

1200 ug

1000

6oo

600

400

200

1986 ~" o ~ - - - - o

i / . i / / / 7

/

/ / / / * - " x

//. - " _~ - : : TILI, a,.~ ( I )

/ ~ I I ~___+ ~ ( s ) _ , TmL PaT (I)

~- --o No TUl SSP (s)

r~- --x NO TILL PRT (s)

; 1'o 1'~ ~'o 2's 3'0 ~'s io 26 P F e r t i l i z e r Ra te (kg P 2 0 5 / h a )

50

1800

1600

1400

1200

lOOO

800

600

400

200

o o 5

F E i i , 1" , i ~ i

1 9 8 7 / " ~

. , / . I : . r ~ ~ (x)

+_ _ _+ TILL SSP (S)

_, T~ aRT (I)

ff---o NO TILL ~P (S)

x-- -~ NO rmn P~r (s)

1'o 1'~ ~'o ~'s ~'o 3; io ,'5 so P F e r t i l i z e r Ra te (ka P 2 0 5 / h a )

1800 . . . . . . . . .

1600

1400

1200

1000

800

600

400

200

0 0 5

1 9 8 8

. _ , 11LLP~rr (I)

~- --o No TU,L SSP (s)

x-- --x NOTIIIPRT(S)

1'o 1'6 2'0 2's go 3's io is P F e r t i l i z e r Rate (kg PSOS/ha)

50

Fig. I. Effec t of t i l lage and P source on mi l l e t y ie ld , fer t i l izers i nco rpo ra t ed ( I ) or surface app l i ed (S). N broadcas t . T r e a t m e n t s

ful ly desc r ibed in Tab le 1.

therefore did not crust after rainfall. Soils in the more humid areas in southern Niger and Mali are, however, prone to crusting, resulting in reduced crop emergence and increased water runoff. Such soils would be expected to show significant effects of tillage.

Effect o f P source in untilled treatments

In this comparison of P sources, all fertilizers were surface applied to untilled soil in order to

simulate the conditions found in farmers' fields. In 1986, SSP outperformed PAPR and PRT sig- nificantly and at 30 kg P205 ha -1 showed a yield advantage of 400 kg over PAPR and 700 kg over PRT (Fig. 2). In 1987 and 1988, PAPR perfor- mance improved to become equal to that of SSP though PRT continued to give significantly lower yields. In that the trial location was not changed each year and phosphate fertilizers give residual responses in the season following initial applica- tion, the improved performance of PAPR may have been due to this residual P effect. PRT, at

56 Christianson et al.

1800

1600

1400

1200

1000

~' coo

600

400

200

0 0

1 9 8 6

_ ~ PRT

1'0 15 20 216 310 315 40 4~5 50 P Fertilizer Rate (kg P205/ha)

1 8 0 0

1 6 0 0

1400

1200

1000

soo

k 600

400

200

0 0 5 0

1987

4- - - - .+ P l l l~

~. _ $ P l~

P Fertil izer Rate (kg P2OS/ha)

1 8 0 0 . . . . . . . , ,

1 6 0 0

1400

1200

~ 1 0 0 0

600

600

400

2O0

0 0

1 9 8 8

~ s ~ s ~ ~ $

4 - - - + PkPR

_ s P R Y

; 1'0 1; 3'0 ;6 20 26 60 P Ferti l izer Rate (kg. P2Of/ha)

Fig. 2. Effect of P source on millet y i e ld - no tillage. Treatments fully described in Table 1.

the low application rates used in this study, did not exhibit a pronounced residual response.

Fertilizer use was shown to result in large yield increases in both tilled and untilled treatments. PAPR performed well when surface applied, and, because it can be made from Nigerien phosphate from the Parc W deposit, its use may be economically attractive in the future when compared with imported SSP. However, previ- ous studies have shown that the high iron and aluminum content of Tahoua phosphate rock makes it unsuitable for partial acidulation (Ham- mond et al., 1989). Because P from PRT is less

available than that of the more water-soluble sources, it will be necessary to use more PRT than PAPR or SSP to achieve equivalent yields. However, in that PRT is not acidulated and can be applied to the soil directly as a ground pow- der, the farmer will realize the attendant saving in cost per kilogram of P205, which will help offset the higher rates of application.

Effect of point placement of fertilizer

The wide (1 m x I m) spacing used in millet pro- duction results in 50% of the fertilizer being

25 cm or farther from the nearest plant. Point placement brought all the fertilizer near to the plant and resulted in its placement in the moist subsoil. Previous studies using calcium am- monium nitrate (CAN), a nitrogen fertilizer not susceptible to ammonia volatilization, have shown that point placement increased plant N uptake by 75% over a broadcast treatment at a planting density of 10,000 pockets ha -~. This effect appeared to have been due to improved spatial availability of the fertilizer (IFDC, 1987). It was therefore postulated that point placement

Millet response to tillage and fertilizers in Niger 57

of SSP might improve P uptake in a similar manner. However, no difference in yield was found when P was either point placed or surface broadcast in combination with N that had been point placed or broadcast (Fig. 3). The possible yield advantage of applying the fertilizer near the plant may have been diminished by the 2-week delay in P application associated with this treat- ment. Early crop growth is less rapid in the absence of sufficient P; thus, there is reduced tolerance to short-term drought and sand dam- age prior to rainfall events, factors which may

1800 , , , , , , , , ,

1600

1400

~ 1200

1000

800

r~ 600

400

200

0 0

1 9 8 6 7

• ///

/

: - s s r (s~) N (~) _ _ + ~ P (PP) N ( B R )

_ $ ~ (PP) N (PP)

O - - - O ~ (rot) N (PP)

' i'0 ~ 2tO ' ' ' ' ' 5 1 25 30 35 40 45 50

P F e r t i l i z e r Ra te (kg P 2 0 5 / h a )

1800

1600

1400

1200

1000

800

800

400

200

i ~ i i i J l i i /

/o / /

1987

/ ~ / / / . -

: - ss~ (sR) N (sR) + . _ _+ SSP (PP) N (BR)

_ ~ ~ P (PP) N (PP)

e - - -o sse ( ~ ) ~ (PP)

' 'o 1'5 zo ' ' 3 ' 5 ' ' 5 1 25 30 40 45

P F e r t i l i z e r Ra te (kg P 2 0 5 / h a )

1800 . . . . , , , ,

50

1600

1400

1200

1000

800

600

400

200

1988

~------JO i~ .I" 0

........." • - ~ - -- ---.-~--11,

o/

S (~z) N (sa) 4_ _ _+ ss~ (pp) s (m~)

_ . ssP (Pe) ~ (PP) e - - -o ss~ (m~) N (PP)

' 1'0 ; 2 0 ' ' ' ' ' 5 1 25 30 35 40 45

P Fertilizer Rate (kg P205/ha)

50

Fig. 3. Effect of P and N p l a c e m e n t on mi l l e t y i e l d - n o t i l lage , fer t i l izers po in t p l aced (PP) or surface b roadcas t (BR) .

T r e a t m e n t s ful ly desc r ibed in Tab le 1.

58 Millet response to tillage and fertilizers in Niger

have diminished the effect of point placement. In addition, the pearl millet root system is exten- sive, and, in a sandy soil of such low fixation capacity, P may have been moved into the soil by infiltrating water and intercepted by the root.

Though farmers tend to plant at a wide range of densities, the nationally recommended density is 10,000 pockets ha -1, and this was used in the experiment. Though placement of all the P fertil- izer near the plant did not improve P response at this density, point placement may still be a useful technique in fields of those farmers who plant at densities as low as 3,000-5,000 pockets ha -1. In such cases, pearl millet at this wide spacing may exhibit a response to point-placed fertilizers.

Conclusion

In this study, tillage did not show any significant effect on efficiency of fertilizer use. Thus, on the basis of the data presented in this study and the present lack of PAPR in the Nigerien market, SSP broadcast on the untilled soil surface with application of nitrogen will result in significant economic millet yield increases for the farmers in the region of Gobery, Niger. In the event that PRT is substituted for SSP, significantly higher rates of P application will be required to achieve yields equivalent to those found with SSP use.

References

Bationo A, Christianson C B and Baethgen W E 1989 Effect of planting density and N fertilizer rates on millet yield in the sandy soils of Niger. Agron. J. In press.

Christianson C B, Bationo A, Henao J and Vlek P L G 1990 Fate and efficiency of N fertilizer applied to millet in Niger. Plant and Soil In press.

Hammond L L, Chien S H, Roy A H and Mokwunye A U 1989 Solubility and agronomic effectiveness of partially acidulated phosphate rocks as influenced by their iron and aluminum content. Fert. Res. 19, 93-98.

International Crops Research Institute for the Semi-Arid Tropics (ICRISAT) 1986 ICRISAT Annual Report, 1985.

International Fertilizer Development Center (IFDC) 1988 IFDC Annual Report, 1987, Muscle Shoals, AL 35662, USA

International Fertilizer Development Center (IFDC) 1987 IFDC Annual Report, 1986, Muscle Shoals, AL 35662, USA.

McIntire T 1986 Constraints to fertilizer use in sub-Saharan Africa. In Management of Nitrogen and Phosphorus Fer- tilizers in Sub-Saharan Africa. Eds. A U Mokwunye and P L G Vlek. pp 33-58. Kluwer Academic Publishers, Dor- drecht, The Netherlands.

Mokwunye A, Chien S H and Rhodes E R 1986 Reactions of phosphate with tropical Africa soils. In Management of Nitrogen and Phosphorus Fertilizers in Sub-Saharan Af- rica. Eds. A U Mokwunye and P L G Viek. pp 253-282. Kluwer Academic Publishers, Dordrecht, The Nether- lands.

Sivakumar MVK 1986 Climate of Niamey- Agroclimatology Progress Report 1, ICRISAT Sahelian Center, Niamey, Niger.

Traor6 M F 1974 Etude de la fumure mineral azotic intensive des crrrales et due rrle sprcific de ia matiere organique darts la fertilit6 des sols au Mali. Agron. Trop. 29, 567- 586.