SPATIAL VARIABILITY OF SOIL  UNDER VETIVER GRASS STRIPS
 
BY
 
OKON, PAUL BASSEY
B.Agric. Hons. (Calabar), M.Sc (Ibadan)
MATRIC No. 108559
 
A Project Report in the Department of Agronomy submitted to the Faculty of Agriculture and Forestry in Partial fulfilment of the Requirement for the Degree of

MASTER OF SCIENCE
UNIVERSITY OF IBADAN
MARCH, 2002
 

ABSTRACT

The variability of surface soils on the same field is influenced largely by soil erosion and runoff deposition. The spatial variability of soils on runoff plots under the influence of  vetiver grass strips (VGS) was therefore studied using both the conventional ìFisherî statistics and geostatistics.  The VGS was established at 20 m spacing two years earlier hence some soil had accumulated on the upper slope side at a mean rate of 71.0 mm per year.  The 720 m2 area under study was on a 5% slope.  There were greater faunal activities around the VGS.

Grid (5 m x 3 m) systematic sampling was used and the 120 samples collected at depths of 0 - 5 cm and 5 ñ 10 cm were processed and analyzed physically and chemically.

The use of classical statistics characterized by mean, standard deviation, coefficient of variation (CV), and range allowed grouping of variabilities into low (CV = 15%) e.g. total sand (712 - 866 g kg-1), soil pH (4.0 - 6.9), coarse sand, and bulk density (B.D.); medium (CV = 15 - 35%) e.g. clay, silt, gravel content, organic C, and total N; and high (CV = 36 - 70%) e.g. available P, fine sand, and silt at 5 - 10 cm depth.  Available P(0.48 - 13.28 mg kg-1) ranked highest in variability (CV = 46.28 - 66.43%).  Static soil properties such as sand and B.D. had low CV, and dynamic soil properties such as nutrient and silt had high CV.

For the geostatistics, considered along a south-north transect, the frequencies of all soil properties under study followed normal distribution; less clay was found in-between two VGS than in lags not bound by VGS at 0-5 cm depth.  Random variable functions (rvf) indicated that 10 m apart was ideal distance lag to take samples in the field. B.D. (1.24 - 1.51 g cm-3) and gravel content (50.4 - 175.1g kg-1) were not significantly different amongst plot in the field.  The semivariances approached the overall variance, Ko, within distance lags of 5 - 20 m for clay at 5-10 cm depth, total sand and soil pH at 0-5 cm depth. Values beyond the variance distances, Ko were considered independent of each other. B.D., Available P, and soil pH revealed no spatial dependence on the VGS effect.  The observed periodic fluctuations were related to soil conservation strategy.  The autocorrelograms of most parameters indicated random pattern without specific trends at the scale estimated.

The effects of VGS on properties of soils bounded by it on both sides were compared with the properties of soils without VGS on both ends. The results showed that clay and bulk density were lesser with VGS than without; 83.83 g kg-1 clay with VGS  and 90.5 g kg-1 without VGS; bulk density of 1.40 g cm-3 with VGS and 1.43 g cm-3 without VGS both at 0 - 5 cm depth. The reverse was observed for sand content, though, their effects were linked to slope, erosion and pedogenic factors rather than VGS. Among the chemical properties, organic C and total N were higher in parcels of land bounded by the VGS than those without. At 5 - 10 cm depth, org. C was 14.25 g kg-1 with VGS and 11.35 g kg-1 without. Similar rise was observed for soil pH; 5.61 with VGS and 5.25 in the control without VGS at 0 - 5 cm depth. In contrast available P was lower in lags bounded by VGS than those without VGS.

The use of the VGS technology in combination with mechanical measures such as along water ways gullies, embankments to facilitate crop production and the integration of soil variability with problem solving were recommended and it was concluded that the application of geostatistics was not necessary because of the small volume of the data involved per plot, and since the observations were normally distributed and no good deal of autocorrelation existed, conventional statistics was enough to fully quantify the variability.