HOME HELP FEEDBACK SUBSCRIPTIONS ARCHIVE SEARCH TABLE OF CONTENTS		QUICK SEARCH:   [advanced] Author: Keyword(s): Year:  Vol:  Page:

This Article Abstract Figures Only Full Text (PDF) Free Alert me when this article is cited Alert me if a correction is posted Services Similar articles in this journal Similar articles in ISI Web of Science Alert me to new issues of the journal Download to citation manager Citing Articles Citing Articles via ISI Web of Science (18) Citing Articles via Google Scholar Google Scholar Articles by Bielders, C. L. Articles by Rajot, J.-L. Search for Related Content PubMed Articles by Bielders, C. L. Articles by Rajot, J.-L. GeoRef GeoRef Citation Agricola Articles by Bielders, C. L. Articles by Rajot, J.-L.

DIVISION S-6-SOIL & WATER MANAGEMENT & CONSERVATION

On-Farm Evaluation of Ridging and Residue Management Practices to Reduce Wind Erosion in Niger

^a Faculté des Sciences Agronomiques, Université Catholique de Louvain, Place Croix du Sud 2/2, B-1348, Louvain-la-Neuve, Belgium
^b Institute for Plant Production and Agro-Ecology in the Tropics and Subtropics, Univ. of Hohenheim, Stuttgart, Germany
^c IRD, LISA-Université Paris 12, 61 Avenue du Général de Gaulle, F-94010 Créteil Cedex, France

bielders{at}geru.ucl.ac.be

	ABSTRACT

TOP ABSTRACT INTRODUCTION Materials and methods Results and discussion Conclusion REFERENCES

 
Wind erosion is regarded as a major contributor to the desertification process in the Sahel, yet little quantitative information is available for that region on soil losses by wind erosion under different land management practices. A 3-yr, on-farm experiment was, therefore, set up to assess the effect of ridging and either banded or broadcast millet stover mulches (2000 kg ha^-1) on soil loss in a millet-cowpea intercrop. For wind directions approximately perpendicular to the orientation of ridges and residue bands, sediment mass balances were calculated from the change in horizontal sediment mass fluxes measured across the experimental plots with Big Spring Number Eight sand traps. Mass balance calculations for 16 events over 3 yr indicated an average soil loss of 17.5, 15.4, and 18.0 Mg ha^-1 on control plots, and deposition of 15.5, 15.3, and 7.4 Mg ha^-1 on banded residue plots in 1995, 1996, and 1997, respectively. Broadcast and banded residue mulches were not significantly different (P = 0.05) in terms of their sediment trapping efficiency. During the same time period, ridges reduced soil losses by an average of 57% compared with the control plots, but their efficiency was reduced to less than 15% after 100 mm of cumulative rainfall as ridges collapsed. Linear regression analysis using the incoming sand fluxes as the independent variable was used to estimate potential soil losses for all events with sediment fluxes <25 kg m^-1 irrespective of wind direction. The calculations indicated potential soil losses of up to 79 Mg ha^-1 on control plots and deposition of 41 Mg ha^-1 on broadcast residue plots in a single year. For wind erosion control, broadcast millet stover mulching constituted the most effective control technique because it effectively protected the soil against erosion and its trapping efficiency is expected to be independent of wind direction.

Abbreviations: BSNE, big spring number eight

	INTRODUCTION

TOP ABSTRACT INTRODUCTION Materials and methods Results and discussion Conclusion REFERENCES

 
IN THE SAHELIAN ZONE of Western Niger, the dominantly sandy soils, with low soil organic matter contents and low soil cover at the time of occurrence of the most erosive winds, create an environment with a high potential wind erosion hazard. In the traditional millet cropping systems of this region, the most predominant forms of wind erosion-induced damage are seedling burial under sand deposits and the loss of topsoil (Michels et al., 1993, 1995a; Sterk and Stein, 1997). Although significant damage due to seedling burial is not observed every year, loss of topsoil is a recurrent problem.

Soil mass flux measurements from on-station experiments in Niger have been reported for the sandy soils traditionally used for millet cropping for low-input management conditions as well as after implementation of wind erosion control measures such as ridging and mulching. Michels et al. (1995b, 1998) and Buerkert et al. (1996) reported a decrease by 36 to 67% in horizontal sediment fluxes at 0.1m above ground following the application of a 2000 kg ha^-1 broadcast millet stover mulch. Banzhaf et al. (1992) reported no significant effect of ridges on sediment fluxes measured between 0.05 and 0.5 m above ground. The results of these studies cannot, however, be interpreted in terms of soil losses since mass flux measurements were in all cases limited to a single measurement point in the center of the experimental plots.

On the basis of measurements of surface elevation, Michels et al. (1995b) reported a relative difference in surface elevation of 33 mm after 1 yr between bare millet plots and plots mulched with 2000 kg ha^-1 millet stover as a result of wind erosion and sediment deposition. Buerkert et al. (1996) reported a loss of 12 mm of topsoil over a 10-mo period in unmulched plots, equivalent to a loss of soil of approximately 190 Mg ha^-1 assuming a bulk density of 1500 kg m^-3. In this latter case, however, water erosion may have contributed to the total soil loss because of specific topographic conditions and the presence of extensive surface crusts. Buerkert et al. (1996) also reported a net deposition of 270 Mg ha^-1 of wind blown sediment over the same time period on plots covered with a 2000 kg ha^-1 millet stover mulch. Following intensive monitoring of sand fluxes in a 40- by 60-m experimental plot, Sterk and Stein (1997) estimated total soil losses from only four sand storms in 1993 at 45.9 Mg ha^-1. These measurements were made on a field planted with millet with a 800 kg ha^-1 broadcast mulch of millet stover, a mulching rate which reflects on-farm conditions under favorable circumstances at the start of the growing season when no effort is made to protect residue against free grazing livestock (Baidu-Forson, 1995; McIntire and Fussell, 1989).

Whether millet stover application rates <2000 kg ha^-1 can be used effectively for wind erosion control under Sahelian conditions remains uncertain. Michels et al. (1995b) report that application rates of 500 kg ha^-1 are ineffective for wind erosion control. Sterk and Spaan (1997) have observed that the wind erosion reduction efficiency of millet stover mulches applied at rates of 1000 and 1500 kg ha^-1 decreases with increasing wind velocity. For application rates of 1000 kg m^-1, the authors observed an increase in erosion for wind speeds in excess of 11 m s^-1. At the rate of 1500 kg ha^-1, extrapolation of the available data suggests that the reduction efficiency becomes 0 at wind speeds of 16 m s^-1 and is less than 50% at wind speeds >10.8 m s^-1 (Sterk and Spaan, 1997).

The available data points to the high sensitivity to wind erosion of the soils from Western Niger, and the potential of surface mulches applied at rates of 2000 kg ha^-1 for reducing this hazard. However, the high soil loss rates reported by Michels et al. (1995b) and Buerkert et al. (1996) remain somewhat controversial because they are derived either from limited number of measurement points of changes in surface elevation or because they may have been affected by both wind and water erosion.

All of the above-mentioned studies were limited to on-station conditions. No data are currently available on soil losses for on-farm conditions in the Sahel both for bare soils and following the implementation of common wind erosion control measures such as mulching or ridging. In this paper, we present the results of 3 yr of investigation of the effect of two types of residue management and ridging on soil losses by wind estimated from direct measurements of airborne sediment fluxes.

	Materials and methods

TOP ABSTRACT INTRODUCTION Materials and methods Results and discussion Conclusion REFERENCES

 
Study Site
The experimental field (13°31'8''N, 2°39'5''E) was located in western Niger near the village of Banizoumbou, approximately 60 km east of the capital city Niamey. Average annual rainfall is about 500 mm, and rainfall was 523, 495, and 415 mm in 1995, 1996, and 1997, respectively. The climate, as in most of the Sahel, is characterized by a prolonged dry season from October to May. During this period northeasterly Harmattan winds occur, but these are less erosive than the easterly convective storms typically observed from May to September (Michels et al., 1995b). This latter period corresponds with the cropping season. The soils at the experimental site are classified as sandy siliceous isohyperthermic psammentic Paleustalfs with 95% sand and 3% clay in the top 0.05 m. Soil organic carbon content in the top 0.15 m is 2 g kg^-1.

Experimental Layout
The 3-yr experiment was initiated in 1995 and consisted of five treatments in a randomized block design with four replications. The 20 plots, each 15 by 20 m in size, were aligned in a single, 300 m long, north–south oriented strip on the western end of a farmer's field. The farmer's field extended for more than 380 m to the east of the experimental field. Prior to the experiment, the entire field had been planted annually with pearl millet [Pennisetum glaucum (L.) R. Br.] since 1992 and managed by the farmer. Farmer management practices were limited to sowing the untilled, unfertilized land with a hand-hoe, and to two manual weeding operations per cropping season.

During the experiment, the farmer's field to the east of the experimental field was managed by the farmer as in previous years. All treatments in the experimental field were cropped with a millet–cowpea [Vigna unguiculata (L.) Walp.] intercrop planted in alternating, north-south oriented rows spaced 0.75 m apart. Both millet and cowpea were sown in hills spaced 1 m apart within rows and were thinned to three plants per hill approximately 3 wk after sowing. Millet was sown on 21 June 1995, 4 June 1996, and 28 May 1997, after the first rain exceeding 15 mm. Cowpea was sown 3 to 6 wk after the millet, on 8 July 1995, 26 June 1996, and 8 July 1997. The crops were harvested in the second half of October each year.

The experimental treatments consisted of (i) an unmulched and unridged control, (ii) a surface mulch of 2000 kg ha^-1 of banded millet stover (stems and leaves; 0.2–0.3 m wide bands) applied in the cowpea rows at the start of the rainy season prior to millet sowing, (iii) 0.2-m-high ridges built in the cowpea rows immediately before cowpea sowing with a traditional, handheld hoe, and (iv) a combined mulch–ridge treatment with 2000 kg ha^-1 banded millet stover applied in the cowpea rows and buried in ridges. For this latter treatment, residue was applied at the start of the rainy season in 1995 and at the end of the previous growing season in 1996 and 1997, and always immediately followed by ridging. The fifth treatment differed between 1995 and the following years. In 1995, ridges were built in both the millet and cowpea row immediately before millet and cowpea sowing, respectively, and no mulch was applied. In 1996 and 1997, this treatment was replaced with broadcast mulch using 2000 kg ha^-1 millet stover without ridges.

In Treatments 2 to 4, residue application in the banded and broadcast residue plots was done at the start of the growing season rather than at the end of the previous season because the latter case would require protective measures against free grazing animals throughout the dry season. When residue is buried in ridges as in Treatment 4, grazing is not a problem. All aboveground cowpea (pods and fodder) and millet (heads, leaves, and stems) biomass was removed from the plots each year, except for the millet stover application on the mulched treatments.

Data Collection and Analysis
We measured soil mass fluxes on the eastern and western side of each plot using big spring number eight (BSNE) sand traps (Fryrear, 1986) with 0.001-m² vertical openings placed at 0.1 and 0.35 m above ground and 0.5 m outside plot boundaries. Because of limitations in the number of traps available, only three of four blocks were equipped in this way. Sediment was collected from the traps after each storm and dried at 105°C. When two sandstorms occurred in the same day, traps could generally not be emptied in time and sand catches for the two storms were pooled. Sand mass fluxes (q) at the eastern and western plot boundaries were calculated for each storm by fitting an exponential equation of the type q = a exp (bz) to the 0.10 and 0.35 m height (z) data (Fryrear and Saleh, 1993), and then integrating the equation between 0 and 0.35 m height. Sandstorms were monitored prior to and during the main part of the growing season, starting 1 June 1995, 6 June 1996, and 23 May 1997 until 31 August each year. Because the trapping efficiency of BSNE sand traps varies with the soil texture (Fryrear, 1986) and no site specific trapping efficiency tests could be performed, the trapping efficiency of the BSNE traps was assumed to be unity in the present study. This is a conservative estimate since the use of a unit trapping efficiency results in underestimation of horizontal sediment fluxes and therefore of erosion/deposition rates in mass balance calculations.

Wind velocity profiles and wind directions were recorded automatically for each storm at a weather station located approximately 150 m east of the experimental field and averaged over 5-min intervals. Average wind velocity and direction at 2 m above ground were calculated only for the period preceding rainfall during which saltation occurred, which was determined from an automatic rainfall gauge and from saltation measurements with a Saltiphone (Spaan and van den Abeele, 1991). To minimize interference from sediment fluxes from adjacent plots, only storms with an average wind direction comprised between 69.5 and 110.5° (north = 0, measured clockwise) were taken into account for sediment mass balance calculations. For those storms, soil loss or deposition was estimated from the difference between the east–west components of the incoming (eastern side) and outgoing (western side) soil mass fluxes. In the event that two sandstorms occurred within the same day, the sandflux data was discarded, unless one of the two sandstorms clearly dominated the other in duration and average wind velocity. In such cases, the average wind direction of the most intense storm was taken into account for calculation of the east–west components of the sandfluxes.

Analysis of variance for treatment effect was carried out in Genstat (Lawes Agricultural Trust, 1996) for each sand storm separately. For calculation of the seasonal balance, occasional missing data were estimated by means of Genstat using the procedure described by Healy and Westmacott (1956). Linear regression analysis was carried out for the relationship between incoming and outgoing soil mass flux by the linear fitting "FIT" directive of Genstat for simple linear regression analysis. Fitting of the ldl rational function (y = a+ b 1/(1+cx) to the erosion reduction efficiency of ridging was carried out by the "FITNONLINEAR" directive of Genstat for nonlinear regression analysis.

	Results and discussion

TOP ABSTRACT INTRODUCTION Materials and methods Results and discussion Conclusion REFERENCES

 
Wind Characteristics
Average wind velocity, direction and duration for each recorded storm are shown in Table 1 . Forty-nine percent of the observed storms lasted less than 15 min. Six of 67 storms exceeded 1 h. The measured wind velocity averaged per event did not exceed 14.0 m s^-1 at a 2-m height. For similar soils and under dry bare soil conditions, the threshold wind velocity for the initiation of saltation has been estimated at approximately 8.5 m s^-1 (Sterk, 1997, p. 23). Of the 67 storms with recorded wind directions in the three observation years, 85% came from directions between northeast and south (45–180°).

View larger version (18K): [in this window] [in a new window]   Fig. 1 Dust storms in 1995, 1996, and 1997 and corresponding horizontal soil mass fluxes. Values correspond to the average sediment mass flux density of 15 sand traps located 0.1 m above ground on the eastern side of the experimental field. Events marked with an arrow correspond to dust storms with average wind direction comprised between 69.5 and 110.5° (North = 0, measured clockwise), used for subsequent sediment mass balance calculations

The greatest soil fluxes at the eastern side of the experimental field were recorded before 15 July each year, after which only minor events were recorded (Fig. 1). This overall trend of decreasing fluxes over time reflects the changes in both soil erodibility and wind erosivity during the course of the rainy season. Under Sahelian conditions and for sandy soils, the vegetative development of the millet crop and weeds in the experimental field is likely to be the main determinant of the change in land erodibility. As a result of the low planting densities in farmers fields (about 5000 hills ha^-1), crop development becomes significant in terms of ground cover and silhouette area only 4 to 6 wk after sowing, which corresponds to the tillering–booting development stage. Since millet was planted in the farmers field in late May through early June each year (20 June 1995, 4 June 1996, 28 May 1997), the low saltation fluxes observed after mid-July may, therefore, largely result from a reduction in ground level wind velocity due to the vegetative development of the millet crop on the eastern side of the experimental field. In addition, the second weeding operation in the farmer's field took place each year towards the middle of August. Substantial weed development may, therefore, have occurred prior to the weeding operation and may have further contributed to a reduction in saltation fluxes.

Of the recorded events, 6, 6, and 4 events had average wind directions between 69.5 and 110.5° which were suitable for mass balance calculations in 1995, 1996, and 1997, respectively. These selected events represent 20, 30, and 23% of the total seasonal sediment flux densities measured on the eastern side of the experimental field in 1995, 1996, and 1997, respectively.

Soil Mass Balance of Selected Erosion Events
Average soil loss or deposition for the 16 dust storms with a wind direction comprised between 69.5 and 110.5° are presented in Table 2 . Soil loss was observed on control and on ridged plots consistently over time, except for one event in 1997 where a net deposition of 0.6 Mg ha^-1 was measured on ridged plots. Deposition of wind blown sediments occurred on banded residue and broadcast residue plots for all events except for small amounts of erosion on 7 July 1997 for banded residue plots and on 29 July 1997 on banded residue and broadcast residue plots.

For dust storms that had suitable wind directions for mass balance calculations, soil losses on the control plots amounted to 17.5, 15.4, and 18.0 Mg ha^-1 in 1995, 1996, and 1997, respectively (Table 2). The highest soil loss of 8.3 Mg ha^-1 from a single event was measured on 30 June 1997. This corresponds to the most intense storm of all those used for mass balance calculations, with an average wind velocity of 14.0 m s^-1 over a 10 min. period (Table 1). On plots with banded residue, deposition amounted to a total of 38.2 Mg ha^-1 over 3 yr (Table 2). The broadcast residue plots trapped 65% less sediment than the banded residue plots in 1996 but 45% more in 1997. Up to 6.2 Mg ha^-1 of sediment deposition was measured for a single event for both banded residue and broadcast residue plots. The annual mass balance for control plots was significantly different (P = 0.05) from plots with a millet stover mulch. However, no significant difference (P = 0.05) was observed between both types of residue management for wind directions approximately perpendicular to the residue bands.

Because convective sandstorms in the Sahel are frequently immediately followed by rainfall, the amounts of airborne sediment caught in sand traps close to the soil surface may be overestimated because of the contribution from rainsplash. However, because sand traps were located 0.5 m outside the plot boundaries on bare soil, it is assumed that this contribution of splash to the total mass of sediment trapped per storm is independent of the treatment and fairly constant across the field. It is, therefore, unlikely that splash resulted in a large systematic bias in the mass balance calculations per treatment or significantly affected the mass balance calculations.

The magnitude of soil losses reported here for bare soils compares favorably with those reported by Sterk and Stein (1997), who measured soil losses of 45.9 Mg ha^-1 in just four sand storms in 1994 on a 40- by 60-m plot covered with 0.8 Mg ha^-1 of millet stover mulch. On the basis of repeated measurements of surface topography, Buerkert et al. (1996) also observed net soil losses on bare millet plots, but they reported much greater values of soil loss or deposition than presented here. This may be attributed to the fact that their measurements integrate soil losses over a 10-mo period. In addition, their soil loss measurements may have been affected by water erosion as a result of surface and topographic conditions. On the contrary, the results presented here are only for a limited number of selected events and strictly a result of wind erosion. As mentioned above, the selected events used for the mass balance calculations presented here represent only between 20 and 30% of the incoming sand fluxes measured between late May to early June and 31 August of each year. Whereas the figures presented by Buerkert et al. (1996) are, therefore, likely to overestimate soil losses by wind erosion on bare plots, those presented here certainly underestimate reality.

For an application rate of 2000 kg ha^-1, surface coverage of broadcast residue has been estimated at 7% by Michels et al. (1995b). For broadcast residue at such low surface coverage, sediment trapping can be expected to occur only in the immediate vicinity of the millet stems. In the case of banded residues, the stems are packed on top of each other, creating a dead volume between the stems were wind velocity is effectively reduced and sediment deposition can occur. This is in agreement with field observations that banded residue favors the formation of natural ridges by trapping wind blown sand. However, despite the very different placement of broadcast and banded residue, it appears that, for wind directions perpendicular to the orientation of residue bands, the trapping efficiency of banded residue and broadcast residue is similar.

For the period following ridge making, ridges without mulch reduced erosion rates by 57% on average compared with the control plots. However, this average reduction factor, based on a limited number of events, does not accurately represent ridge effectiveness. Indeed, the efficiency of ridges at reducing wind erosion will decrease over time as the ridges collapse under the influence of rainfall, soil translocation by wind and weeding operations (Fryrear, 1984). This is particularly true for the structurally unstable sandy soils poor in organic matter that characterize the study site.

The average efficiency of ridges at reducing wind erosion was found to be strongly related to cumulative rainfall since ridge making (Fig. 2) . After 100 mm of rainfall, the erosion rate on ridged plots was only 15% lesser on average than on the control plots. In Fig. 2, the measurement point corresponding to 99 mm of cumulative rainfall was excluded from the regression analysis. This corresponds to a fairly minor event on 29 July 1997. For small events, small errors in the mass balance calculations lead to large errors in the efficiency factor. The negative efficiency after 170 mm of rainfall indicates that ridged plots were on average more erodible than control plots after the ridges collapsed. For erosion events preceding ridging, the average reduction efficiency of ridged plots vs. control plots was -22%, indicating that ridged plots before ridging were indeed, on average, more erodible than control plots a result of inter-plot variability. The need to take into account variations in intrinsic erodibility even between closely spaced plots was also pointed out by Sterk and Spaan (1997).

View larger version (15K): [in this window] [in a new window]   Fig. 2 Average erosion reduction efficiency of ridged plots compared with control plots, as a function of cumulative rainfall

The presence of millet stover in the ridges at the end of the growing season increased the effectiveness of the ridges for wind erosion control. Ridges with incorporated residue reduced soil losses by wind erosion by 87% on average compared with the control plots, as opposed to 57% for ridged plots without residue. On the basis of the available data, there was no evidence that the efficiency of this treatment varied over time or as a function of cumulative rainfall. The greater effectiveness of ridges with incorporated residue may be attributed to a slower collapse of the ridges over time. Furthermore, even after some of the ridge has been blown or washed away, the exposed residue maintains a greater level of surface roughness than on the control or ridged plots, thus reducing wind velocity. There was no indication (P = 0.05) that burial of residue in ridges at the end of the previous growing season (1996–1997) was less effective than burial early in the growing season (1995), despite the fact that the latter technique should have resulted in a greater surface roughness early in the season. This may be due to the slow biological activity during the dry season in these sandy soils, which ensures that residue buried at the start of the dry season will still be largely undecomposed by the start of the following rainy season.

The above mentioned erosion reduction efficiencies for ridged and ridge + banded residue plots are valid only for storms with an average wind direction approximately perpendicular to the ridges. As the average wind direction deviates from the east-west direction, the efficiency can be expected to decrease rapidly and erosion rates may approach that of control plots for wind directions parallel to the ridge direction.

In discussing the relative efficiencies of the various management techniques, it is difficult to separate the direct effects of the technique on surface roughness and wind velocity from the indirect effects on crop development and its resulting impact on wind velocity. Differences in crop growth may partly explain why the ridged treatments with buried residue remain effective over time, whereas sole ridges do not. Indeed on control and ridged plots, plant growth and development was very limited because of the rapid degradation of the soil (Fig. 3) . However, on plots with residue—whether incorporated or not—plant growth was normal and, therefore, must have contributed to the overall efficiency of the treatment. Whereas total above ground biomass (stems and leaves of millet and cowpea) at the end of the growing season was similar for control (534 kg ha^-1) and ridged (705 kg ha^-1) plots each year, it was 2.1 times greater on ridge + banded residue plots than on ridged plots (s.e.d. = 134 kg ha^-1). However, the importance of above ground biomass production on the sensitivity of various treatments to wind erosion should not be overemphasized because the most erosive storms always occurred before crop development reached a significant level, i.e., within the first 4 to 6 wk after sowing (Fig. 1 and 3).

View larger version (22K): [in this window] [in a new window]   Fig. 3 Millet growth during the 1997 growing season for the five experimental treatments

In the present experiment, millet stover application rates of 2000 kg ha^-1 with a proven efficiency for wind erosion reduction were used (Michels et al., 1995b, 1998; Buerkert et al., 1996). Under the low input, Sahelian conditions of Western Niger (no fertilizer input) such levels of crop residue are difficult to achieve on-farm. On the basis of the results of this experiment, it would be of interest to test the effectiveness of lower application rates for the most effective treatment, i.e., for broadcast millet stover mulch.

Potential Seasonal Soil Mass Balance
The above mass balance calculations do not provide a means to estimate the total soil loss during the entire monitoring period since the calculations are restricted to storms with an average wind direction of 69.5 < < 110.5°. However, for those storms, a linear relationship exists for the combined 3 yr between incoming (eastern side) and outgoing (western side) soil mass fluxes for the three control plots (Fig. 4) . For each of the three plots, only the slopes of the linear regressions are significantly different from each other (P < 0.01). The intercepts are not significantly different from 0 (P = 0.05). On the basis of the available data, the relationship appears to be independent of the year and time of year. The slope of the relationship between incoming and outgoing mass fluxes varied from 1.10 (Plot 19) to more than 2.28 (Plot 8). The results indicate that the soil erodibility of small areas (300 m²) can vary widely within distances of 50 to 100 m, something also observed by Sterk and Spaan (1997) under on-station conditions. This may be due to differences in surface crusting, millet growth, and topography.

View larger version (18K): [in this window] [in a new window]   Fig. 4 Incoming (eastern side) vs. outgoing (western side) soil mass fluxes (1995–1997) for the three control plots, for storms with an average wind direction comprised between 69.5 and 110.5° (North = 0, measured clockwise)

View larger version (18K): [in this window] [in a new window]   Fig. 5 Incoming (eastern side) soil mass flux vs. soil mass balance on plots mulched with 2 Mg ha-1 of millet stover applied (a) as bands (1995–1997) or (b) broadcast (1996–1997), for storms with an average wind direction comprised between 69.5 and 110.5°

The mass balances reported here are based on sediment mass fluxes between heights of 0 and 0.35 m. This is likely to include most of the sediment transported by saltation (Chepil, 1945) but probably excludes a substantial fraction of the suspended sediment fluxes. Although this suspended fraction may to some extent influence the nutrient balance of the plots as a result of the relative enrichment of the finer particles in nutrients (Zobeck and Fryrear, 1986; Sterk et al., 1996), it is unlikely to greatly affect the total mass balance.

The present results are based on the assumption that reliable estimates of sand fluxes between 0 and 0.35 m above ground can be obtained from measurements at two heights and integration of the fluxes between 0 and 0.35 m by an exponential function. On the basis of measurements of sediment fluxes (Sterk, 1997, unpublished) at 7 heights between 0.05 and 1 m from 20 modified Wilson and Cook sand catchers (Sterk et al., 1996), we have estimated that the method used here underestimates sediment fluxes between 0 and 0.35 m height by 32 ± 7% (mean ± SD) on average. This error results primarily from the underestimation of sediment fluxes below 0.1 m where no measurements can be made using conventional BSNE sand traps. Addition of 1 or 2 sand traps at heights >0.35 m does not improve the estimation of sediment fluxes below 0.1 m and therefore would still result in an overall underestimation of fluxes. Because the sediment fluxes reported here may substantially underestimate sediment fluxes between 0 and 0.35 m, the calculated mass balances are likely to underestimate actual soil losses or deposition.

	Conclusion

TOP ABSTRACT INTRODUCTION Materials and methods Results and discussion Conclusion REFERENCES

 
On the sandy soils from western Niger, wind erosion constitutes a serious hazard to cropland in the absence of adequate surface mulches. Under such conditions, soil losses of up to 18 Mg ha^-1 yr^-1 were measured on bare plots planted with a millet–cowpea intercrop, but actual losses may have exceeded 79 Mg ha^-1 yr^-1.

The application of 2 Mg ha^-1 of millet stover as mulch was highly effective at controlling wind erosion. Broadcast or banded residue were equally effective for wind directions approximately perpendicular to the residue bands. Two to 3 mm of aeolian sediment may have been deposited each year on plots with a broadcast millet stover mulch. Because the broadcast residue is equally effective as banded residue but its efficiency is expected to be independent of wind direction, broadcast residue mulches should be recommended over the use of banded residue for wind erosion control purposes.

Because ridges are unstable on these sandy soils, ridges spaced 1.5 m apart were effective at reducing wind erosion only for the first few storms following ridge making. After 100 mm of cumulative rainfall, the erosion rate on ridged plots was only 15% lesser than on bare control plots. Incorporating residue in the ridges constitutes an effective means to prolong the effectiveness of ridges on these soils. Ridges with incorporated residue reduced the erosion rate by 87% on average compared with bare plots. This latter treatment offers the additional advantage that it can be carried out at the start of the dry season (post-harvest) rather than at the start of the rainy season (pre-sowing), thereby making the labor requirements for the implementation of this management practice less critical. However, the choice of proper ridge orientation in the Sahelian environment where both wind and water erosion can be major land degradation processes needs to be carefully considered.McIntyre Fussel 1989

	ACKNOWLEDGMENTS

 
A contribution from ICRISAT. This research was funded in part by the Netherlands Directorate General for International Cooperation (DGIS). The authors are thankful to the Soil and Water Management team of ICRISAT-Niamey, Mr. S. Maharou, Anton Vrieling and several Nigerien trainees who have contributed to the success of this study. We are also thankful to Dr. G. Sterk for providing some of his sediment mass flux density data from Niger.

Received for publication April 25, 1999.

	REFERENCES

TOP ABSTRACT INTRODUCTION Materials and methods Results and discussion Conclusion REFERENCES

 

• Baidu-Forson J. Determinants of the availability of adequate millet stover for mulching in the Sahel. J. Sustain. Agric. 1995;5:101-116.
• Banzhaf J., Leihner D.E., Buerkert A., Serafini P.G. Soil tillage and windbreak effects on millet and cowpea. I. Wind speed, evaporation, and wind erosion. Agron. J. 1992;84:1056-1060.[Abstract/Free Full Text]
• Brouwer, J., and J. Bouma. 1997. Soil and crop growth variability in the Sahel: Highlights of research (1990–95) at ICRISAT Sahelian Center.Information Bulletin No 49. ICRISAT, Patancheru, India.
• Buerkert A., Lamers J.P.A., Marschner H., Bationo A. Inputs of mineral nutrients and crop residue mulch reduce wind erosion effects on millet in the Sahel. In: Buerkert B., et al. , ed. Wind erosion in West Africa. The problem and its control. Weikersheim, Germany: Margraf Verlag, 1996:145-160 Proc. Intern. Symp., Univ. Hohenheim, Stuttgart, Germany. 5–7 Dec. 1994..
• Chepil W.S. Dynamics of wind erosion: 1. Nature of movement of soil by wind. Soil Sci. 1945;60:305-320.
• Fryrear D.W. Soil ridges, clods and wind erosion. Trans. ASAE 1984;27:445-448.
• Fryrear D.W. A field dust sampler. J. Soil Water Conserv. 1986;41:117-120.
• Fryrear D.W., Saleh A. Field wind erosion: vertical distribution. Soil Sci. 1993;155:294-300.
• Hagen L.J., Armbrust D.V. Aerodynamic roughness and saltation trapping efficiency of tillage ridges. Trans. ASAE 1992;35:1179-1184.
• Healy M.J.R., Westmacott M.H. Missing values in experiments analysed on automatic computers. Appl. Statistics 1956;5:203-306.
• Lawes Agricultural Trust. Genstat 5, release 3.2. Clarendon Press, Oxford, UK: Oxford Science Publications, 1996.
• McIntyre J., Fussel L.K. On-farm experiments with millet in Niger: Crop establishment, yield loss factors and economic analysis. Exp. Agric. 1989;25:217-233.
• Michels K., Armbrust D.V., Allison B.E., Sivakumar M.V.K. Wind and wind-blown sand damage to pearl millet. Agron. J. 1995;87:620-626 a.[Abstract/Free Full Text]
• Michels K., Lamers J.P.A., Buerkert A. Effects of windbreak species and mulching on wind erosion and millet yield in the Sahel. Exp. Agric. 1998;34:449-464.
• Michels K., Sivakumar M.V.K., Allison B.E. Wind erosion in the Southern Sahelian Zone and induced constraints to pearl millet production. Agric. Forest Meteor. 1993;67:65-77.
• Michels K., Sivakumar M.V.K., Allison B.E. Wind erosion control using crop residue: I. Effects on soil flux and soil properties. Field Crops Res. 1995;40:101-110 b.
• Spaan W.P., van den Abeele G.D. Wind borne particle measurements with acoustic sensors. Soil Technology 1991;4:51-63.
• Sterk G. Wind erosion in the Sahelian zone of Niger: Processes, models, and control techniques. Wageningen, the Netherlands: Wageningen Agricultural University, 1997 Tropical Resource Management Papers #15..
• Sterk G., Stein A. Mapping wind blown mass transport by modeling variability in space and time. Soil Sci. Soc. Am. J. 1997;61:232-239.[Abstract/Free Full Text]
• Sterk G., Spaan W.P. Wind erosion control with crop residue. Soil Sci. Soc. Am. J. 1997;61:911-917.[Abstract/Free Full Text]
• Sterk G., Hermann L., Bationo A. Wind-blown nutrient transport and soil productivity changes in southwest Niger. Land Degrad. Dev. 1996;7:325-335.
• Zobeck T.M., Fryrear D.W. Chemical and physical characteristics of windblown sediment. II. Chemical characteristics and total soil and nutrient discharge. Trans. ASAE 1986;29:1037-1041.

This Article Abstract Figures Only Full Text (PDF) Free Alert me when this article is cited Alert me if a correction is posted Services Similar articles in this journal Similar articles in ISI Web of Science Alert me to new issues of the journal Download to citation manager Citing Articles Citing Articles via ISI Web of Science (18) Citing Articles via Google Scholar Google Scholar Articles by Bielders, C. L. Articles by Rajot, J.-L. Search for Related Content PubMed Articles by Bielders, C. L. Articles by Rajot, J.-L. GeoRef GeoRef Citation Agricola Articles by Bielders, C. L. Articles by Rajot, J.-L.