- Email: [email protected]
Evaporation from soils below sparse crops in contour hedgerow agroforestry in semi-arid Kenya
Agricultural and Forest Meteorology 130 (2005) 149–162 www.elsevier.com/locate/agrformet
Evaporation from soils below sparse crops in contour hedgerow agroforestry in semi-arid Kenya J.M. Kinama a, C.J. Stigter b,*, C.K. Ong c, J.K. Ng’ang’a d, F.N. Gichuki e b
a TTMI-Project, National Dryland Farming Research Centre, KARI, Katumani, P.O. Box 340, Machakos, Kenya TTMI-Project, Department of Environmental Sciences, Wageningen University, Duivendaal 2, 6701 AP Wageningen, The Netherlands c International Centre for Research in Agroforestry, P.O. Box 30677, Nairobi, Kenya d TTMI-Project, Department of Meteorology, University of Nairobi, P.O. Box 30197, Nairobi, Kenya e Department of Agricultural Engineering, University of Nairobi, P.O. Box 30197, Nairobi, Kenya
Received 27 April 2004; received in revised form 9 February 2005
Abstract In many agricultural systems in the semi-arid tropics, crops use only a small fraction of the total rainfall. Agroforestry can greatly reduce some losses, especially on hill slopes, where soil evaporation, runoff and soil losses are important. This paper reports on soil evaporation from a rotation of intercropped maize and cowpea between contour hedgerows of pruned Senna siamea trees as well as trimmed Panicum maximum grass strips on a 14% hill slope at a semi-arid site in Machakos, Kenya. There were five treatments in order to separate effects of Senna mulch, hedges, and grass strips. Micro-lysimeters were placed between crop rows for three seasons. It followed from their results that, for the three seasons concerned, tree prunings as mulch reduced soil evaporation as percentage of rainfall in the measuring period by absolute values of 9%, 4% and 6% compared to the control sole maize and cowpea with bare soil. The influence of the hedge added to this only insignificantly, even at 1 m distance. The non-mulched plots had soil evaporation reduced by only between on average 1% and 4% in absolute values compared to the control over all the seasons, with a maximum of 5% close to the hedge in the first season. Mulch apparently is the main evaporation reducing factor. Soil evaporation reached the highest percentage of rainfall in the long rains of 1994, becoming 65% in sole maize. It was 50% for sole cowpea in the 1994/1995 short rains and for sole maize in the next long rains. The highest value, although an upper limit could largely be understood from highest early season evaporative demands, rainfall distributions and low crop cover. The other values were in line with earlier reports for dry areas. Some advantages and disadvantages of these agroforestry systems are reviewed. # 2005 Elsevier B.V. All rights reserved. Keywords: Soil evaporation; Micro-lysimeters; Agroforestry systems; Hedgerow intercropping; Mulch; Senna siamea; Grass strips; Maize; Cowpea
1. Introduction * Corresponding author. Tel.: +31 317 483 981; fax: +31 317 482 811. E-mail address: [email protected] (C.J. Stigter).
Decisions on land use, such as ‘‘being forced to use sloping land for production’’, belong to the strategic needs in tropical Africa, and choices of cropping
0168-1923/$ – see front matter # 2005 Elsevier B.V. All rights reserved. doi:10.1016/j.agrformet.2005.03.007
150
J.M. Kinama et al. / Agricultural and Forest Meteorology 130 (2005) 149–162
systems, such as ‘‘one that reduces soil loss and water runoff on slopes’’ to the tactical needs (Olufayo et al., 1998). An important goal of combining trees and crops in agroforestry systems is to make better use of the environmental resources required by plants to grow (e.g. Mungai et al., 2001; Kinama et al., in press). One of the means of achieving this is to utilize resources that would otherwise be lost from the system (e.g. Ong and Black, 1992; Cannell et al., 1996). Wallace (1991) suggested that introduction of trees into a crop may lead to an overall increase in the proportion of rainfall that is used as transpiration. Van Noordwjik and Ong (1999) estimated by modeling that the potential increase in transpiration due to scattered trees in the Sahel is only about 8% of total rainfall. Although there is little information on the role of agroforestry in reducing soil evaporation, it is expected that maximum gain will be made in semi-arid and arid environments where soil evaporation forms a substantial part of the soil water balance (e.g. Stigter, 1994). Under west African conditions, Wallace et al. (1988) and Bley et al. (1991) showed that soil evaporation is a major part of the soil water balance that depends on soil wetness. The influence of scattered tree shade on soil evaporation was studied in the Sahel savanna of Africa (Johnson, 1995; Belsky et al., 1993; Kinyamario et al., 1995). Higher topsoil moisture under woody canopies than in treeless sites appears to be common during the rainy season and sometime afterwards, as exemplified in Boffa (1999). Wallace et al. (1999) modeled the reduction of soil evaporation under a tree canopy in Kenya to be 35% on an annual basis. In general, shading has beneficial effects but total shade area may only account for about 10% of total land area. Measurements of soil evaporation under 6–8 m tall Grevillea robusta trees with a leaf area index of 2 at Machakos, using a micro-lysimeter technique, showed that soil evaporation is reduced by 23%, largely attributed to tree shade (Jackson and Wallace, 1999). Maintaining a large leaf area, however, is not a viable option for water limited environments since competition for soil water by dominant trees results in frequent crop failures (Lott, 1998). Pruning of branches and roots may of course reduce competition. Thus, there is a trade-off between the advantages of shade on lowering soil evaporation and improving crop aerial climate versus the negative effect of water competition from tree roots. The same
applies to advantages of soil loss and run off reductions of hedgerows (Kinama et al., in press). In our research, water lost by evaporation from soil deserved separate attention from water lost by run off in the erosion causing processes. It is a considerable loss and the influence of mulch (and to a much lesser extent the contour hedge rows) is due to completely different processes compared to their influence on run off losses. Soil cover is considered to be the dominant factor of all parameters that affect erosion. Research on contour hedgerows on sloping land in semi-arid Kenya has quantified the impact of Senna siamea mulch and tree barriers on runoff and soil loss (Kiepe, 1995; Kinama et al., in press). S. siamea mulch is ideal for mulching because of its slow decomposition rate (Mugendi et al., 1994). Measurements by Kiepe (1995) in a young S. Siamea based agroforestry system over 3 years showed that mulching alone reduced runoff by 50%, even during heavy storms, and soil loss by more than 80%. Kinama et al. (in press) showed that in four seasons of this same rotation of maize and cowpea with S. siamea, in the mulched hedgerow plots cumulative soil loss reduced from 100 to only 2 t ha1, and runoff from 100 to 20 mm. However, due to competition between crops and hedgerows, yields have better to be supported by fertilizers, otherwise losses are almost prohibitive. Ageing systems show increased competition (Kinama et al., in press). Critical studies exist of alley cropping in the semiarid tropics on flat land (Rao and Westley, 1989; Mungai, 1995; Sanchez, 1995; Mungai et al., 2001). But there is much more scope for increasing the use of limited rainfall in contour hedgerows on sloping lands. Hedge width and tree and hedge spacing and pruning as well as the use made of mulch are issues together with other economic ones (Kiepe, 1995; Ong et al., 1996; Kinama et al., in press), that may also affect soil evaporation. Understanding soil evaporation is necessary in the evaluation of the water balance of these agroforestry systems, just like that has been noted for rainfall interception losses (Jackson, 2000). 2. Materials and methods 2.1. Instrument This paper describes the use of mini- or microlysimeters to quantify the influence of S. siamea
J.M. Kinama et al. / Agricultural and Forest Meteorology 130 (2005) 149–162
mulch, contour hedgerows and grass strips on soil evaporation on sloping land at the semi-arid site in eastern Kenya detailed below. Boast and Robertson (1982) and Daamen et al. (1993) have extensively described the technique used in the present study. 2.2. Experimental site The on-station trials were conducted at ICRAF’s Research Station at Machakos, which is about 70 km South East of Nairobi and 7 km from Machakos town. The station lies between latitudes 18300 and 18350 S and longitudes 378 and 378150 E. It has an altitude of 1560 m above sea level with slopes ranging from 0% to 22%. The experimental plots were established on sloping land of about 14%. The site receives 410 mm on average for the short rains (SRs), which are from mid-October till January, and 360 mm on average for the long rains (LRs), which are from mid-March to July. Seasonal rainfall and rainfall distribution over the season are highly variable. The site received from 1989 to 1995 between 290 and 810 mm in the SRs and between 110 and 620 mm in the LRs (Kinama et al., in press). During SR94/95 the rainfall was above average, while those in the two LRs were below average. The soils are sandy clay loams over sandy clay developed in situ on rocks of the Precambrian basement complex. The soils are about 150 cm deep and have originally been classified as chromic luvisols (Kibe et al., 1981), but revised as Haplic Lixisols (FAO/UNESCO, 1988) or Kanhaplic Rhodustaff (Soil Survey Staff, 1990). They are dark reddish brown, sandy clay loam becoming sandy clay at the lower horizons (Mbuvi and Van de Weg, 1975; Barber et al., 1979; Kilewe and Ulsaker, 1984). They are shallow due to the presence of a pentroplinthite (Murram) horizon (Marimi, 1979). Due to low structural stability, the soils are prone to slaking, highly erodible and prone to surface capping by intense rainfall. This risk is enhanced by low sub soil permeability (Kiepe, 1995). 2.3. Experimental design The experimental plots (Fig. 1) were on land which had been under alley cropping with hand hoe cultivation, with long term runoff/soil erosion monitoring since the establishment of the hedgerows
151
in 1988, as described by Kiepe (1995) and Kinama (1997). Grass strips (Panicum maximum) were established earlier, in 1984. The plant rows, the grass strips and the S. siamea hedgerows were planted along the contours, lying roughly E–W. The experiments covered in this study were for three out of six cropping seasons studied. During the SRs cowpea (Vigna unguiculata, cv. K80 or SK-27) was planted while maize (Zea mays, cv. Katumani composite B) was sown during the LRs. S. siamea, a non-modulating leguminous tree, was chosen because it was among the few multi-purpose trees/shrubs considered suitable for the area as contour hedgerow barriers. The tree species is drought tolerant and suited to the local semi-arid conditions, as reported by Rao and Westley (1989) and Mugendi et al. (1994). Its mulch is suitable for erosion control purposes because of the high amounts of tannin in the mulch (Kiepe, 1995), making disintegration rates low (Mugendi et al., 1994). Only S. siamea loppings from the hedgerows were spread uniformly on the soil surface. The quantities of mulch applied therefore depended on the rainfall conditions prior to their application. The hedgerows which had reached about 1.5 m in height, were cut down to 25 cm 2 weeks before the expected onset of the rains, delivering 1.7 0.3 t ha1 of mulch (Kinama et al., in press). The grass strips were cut down to 25 cm at harvest (from about 1 m height) and 2 weeks before planting (from about 50 to 60 cm height). The total biomass (2.5 0.4 t ha1) was carried off the plots for other purposes (Kinama et al., in press). No fertilizer was used during the six seasons of crop measurements. The study consisted of five treatments with no replicates. The plots measured 10 m width 40 m downslope and it was the sampling procedure which was replicated. This means that there were various sampling points in each plot. The following treatments were used (Fig. 1a): Treatment 1: Maize or cowpea control (C). Treatment 2: Maize or cowpea + S. siamea mulch (+M). Treatment 3: Maize or cowpea + S. siamea hedgerows + mulch (H + M). Treatment 4: Maize or cowpea + S. siamea hedgerows without mulch (H M). Treatment 5: Maize or cowpea + grass strips without mulch (G M).
152
J.M. Kinama et al. / Agricultural and Forest Meteorology 130 (2005) 149–162
Because local farmers never use Panicum grass cuttings as mulch, but use them for other purposes, such as roof thatching and animal fodder, there was no G + M plot (Kinama et al., in press). The +M plot had its mulch obtained from the H M plot, while the H + M plot had mulch from its own hedgerows. There were four rows of maize in each of the alleys formed by the S. siamea hedgerows (Fig. 1b). These hedgerows were 4 m apart and within row tree distance was 25 cm. The closest maize row to these hedgerows was 50 cm (Fig. 1b). The spacing of the maize that was completely added to the hedges, that is with no adaptation of spacing to the intercrop situation (Baldy and Stigter, 1997), was 100 cm 27 cm. This gave a density of just over 37,000 plants ha1. The G M treatment, with five rows of maize between each pair of seven grass strips, had a population of 33,670 maize plants ha1 because the seven grass
strips occupied an area of about 70 m2. The width of the strips was more than 1 m and their centres were between 6 and 7 m apart. The distances from grass strips to first maize rows were about 50 cm. The six rows of cowpea between each pair of the eleven hedgerows were planted at a spacing of 60 cm 20 cm but with a distance of 1 m between the two rows of cowpea at each side of the hedges (Fig. 1b), so not completely added (Baldy and Stigter, 1997). This gave a plant density of 83,333 plants ha1 in the C and +M plots and of 75,000 plants ha1 in the H + M and H M plots. With nine rows of cowpea between the seven grass strips, the G M plot had a lower plant density per unit area of crop, which would become bigger if calculated with 70 m2 as taken up by the grass strips, so calculated per unit area of land. This illustrates the semantic difficulties in calculations of yields per hectare in such agroforestry experiments
Fig. 1. (a) Field layout at Machakos field station showing the sampling points for measuring soil evaporation in the various treatments. Vertical lines are plot edges along the slope direction. Horizontal lines are the hedgerows respectively the grass strips. Collecting tanks were used for soil and water runoff determinations; (b) planting layout for the hedges with their respective intercrops, (A) maize; (B) cowpea.
J.M. Kinama et al. / Agricultural and Forest Meteorology 130 (2005) 149–162
153
Fig. 1. (Continued ).
and we should use row countings and yields per row for that purpose (Kinama et al., in press). 2.4. Soil evaporation quantification Soil evaporation was measured by using microlysimeters (Boast and Robertson, 1982; Daamen et al., 1993; Kinama et al., 1999) made of a PVC cylinder measuring 10.5 cm outer diameter 15 cm depth. The inner diameter was 10.25 cm. The PVC cylinders were commercially constructed in Nairobi with the assistance of ICRAF technicians, from a joint design by the TTMI-Project and ICRAF. This cylinder held a soil core. The cylinder was encased by a PVC outer cylinder of slightly bigger diameter but of similar depth to the inner measuring cylinder (Fig. 2). In order to obtain a soil core, the inner cylinder was driven into the soil using a wooden hammer. This was done causing minimum soil disturbance so that the soil core obtained remained as far as possible in the same condition as the surrounding soil. Given the crustforming nature of the Machakos soil (Allen, 1990),
this was always done when the soil was fairly wet, when the seal rarely cracks. The soil core was then carefully removed using a local panga (machete) and trimmed at the bottom with a sharp knife. The soil core was tightly closed at the bottom end by a cylindrical glassy encasing material of slightly smaller diameter than that of the inner cylinder, reinforced by plastic Sellotape (Daamen et al., 1993). The soil core (micro-lysimeter) was placed back into the soil with the external cylindrical encasing such that the micro-lysimeter was slightly above the soil level. This was to ensure that no runoff and splash water or soil particles entered the micro-lysimeter. The microlysimeter was weighed by a portable balance and replaced back into the soil (Fig. 2). The weight of the internal cylinder, etc. was determined and recorded before the preparation of the soil core. Micro-lysimeters were placed at each of the six sampling points in each of the five treatment plots (Fig. 1a). In the C and +M plots, two micro-lysimeters were installed at 10 m downslope from the top of the plots, 25 m downslope and 35 m downslope respectively, with those in the +M plot having mulches
154
J.M. Kinama et al. / Agricultural and Forest Meteorology 130 (2005) 149–162
Fig. 2. Micro-lysimeter designed for estimating soil evaporation, with (a) external cylinder; (b) internal cylinder; (c) wire handle; (d) raised part of the soil core; (e) bulk of the soil core and (f) bottom sealing encasing.
representatively placed on them. At these positions, the two micro-lysimeters were placed 1 m apart parallel to the crop rows, across the slope and halfway between these plant rows. In the H + M, H M and G M plots, the microlysimeters were installed (Fig. 1a) at 1 m (H2 and G2) and 2 m (H3 and G3) below the fourth, above the seventh and below the ninth hedgerow and below the third, above the fifth and below the sixth grass strip, respectively. In these positions, they were placed midway between the rows of maize. For the cowpea, they were placed between the first and second row and between the third and fourth row below and above the same hedgerows and grass strips. More details on these placements with respect to the cowpea rows can be found in Kinama (1997). Those in H + M plots had mulch placed on them to represent those treatments. To measure soil evaporation, the micro-lysimeters were weighed using a portable balance (0.1 g) both in the morning 0900 h and in the afternoon 1600 h local time, and the differences in weight represented the water loss. Because the surface area of the microlysimeter was 82.7 cm2, 0.1 g is equivalent to a depth of water of 0.012 mm. Such data were obtained for up to a maximum of seven rainfree days after every rainfall event, following the well-researched protocol proposed by Daamen et al. (1993). Daamen et al. (1993) determined that not changing the lysimeters at the end of each of the first 2 days after rainfall gave an
overestimation in the accumulated evaporation for the period concerned. We made that assessment also for our soils and conditions and found experimentally that this was never more than 10% of the measured values. The accumulated differences between treatments are even appreciably less affected, because the errors are always in the same direction. Daamen et al. (1993) also found that for microlysimeters more than 100 mm deep such as ours, after about 7 days the conditions in the micro-lysimeters are no longer representative of the surrounding soil. This was in our case also the limit after which evaporation could be considered almost negligible because of root extractions and self-mulching (formation of a dry soil layer that acts as an insulating mulch) of the drying soil. Such days were normally measured with unrenewed micro-lysimeters, as suggested by the part of the protocol followed. They had generally values of or below 0.4 0.1 mm. Results during rainy days have generally been found unreliable (Allen, 1990; Daamen et al., 1993), so the soil cores were replaced when it rained within the measuring period. The total soil evaporation loss was obtained by taking the sum of all the water losses for each treatment during the measurements, together with the estimates of soil evaporation during rainy days. Several microlysimeter measurements were carried out during rainy days with clear portions of the wet days allowing for measurements. This was done to form a basis for assumptions for the soil evaporation estimates during wet days. On average the evaporation losses during such dry periods with wet surfaces were about 4 mm per day. This was therefore used throughout for measuring days interrupted by rainfall. Within treatment measurement variation between places such as averaged in the columns of Table 1 were always within 0.2 mm for the larger values (between 2 and 3.9 mm) and each average had therefore an appreciably lower relative error. For the lower values there were rarely differences outside the balance absolute accuracy. Cumulative relative errors were therefore appreciably smaller than the differences between treatments and always less than 1% in the totals of Table 2. Because of the discontinuity and limitations of soil evaporation measurements from wet soil, trends in evaporative demand over the season were represented by daily weather station measurements of a class-A evaporation pan.
J.M. Kinama et al. / Agricultural and Forest Meteorology 130 (2005) 149–162
155
Table 1 Example of daily soil evaporation (mm), with subtotal for the period ‘‘2–12 January’’ for the short rains of 1994/1995, when cowpea was grown, for the five treatments and the positions concerned between the hedge rows Treatment
C
+M
H2 + M
H2 M
H3 + M
H3 M
G2 M
G3 M
Day type (a) (b) (b) (b) (b) (b) (b) (b) (c) (c) (c)
4.0 4.0 3.5 3.0 2.8 2.5 1.3 0.6 0.3 0.2 0.4
4.0 3.4 3.0 2.6 2.4 2.0 0.9 0.3 0.1 0.0 0.2
4.0 3.2 2.9 2.4 2.3 1.9 0.8 0.2 0.2 0.0 0.2
4.0 3.3 3.0 2.5 2.4 2.1 0.9 0.2 0.3 0.0 0.2
4.0 3.6 3.3 2.7 2.5 2.3 1.0 0.5 0.2 0.1 0.4
4.0 3.7 3.4 2.8 2.6 2.4 1.2 0.6 0.2 0.2 0.3
4.0 3.6 3.4 2.9 2.6 2.4 1.3 0.6 0.3 0.2 0.3
4.0 3.8 3.4 2.9 2.7 2.5 1.3 0.6 0.3 0.2 0.4
Sub total
22.6
18.9
18.1
18.9
20.6
21.4
21.6
22.1
(a) Estimates made during rainy days; (b) measurements within 7 days after rainfall and (c) measurements after the representative measuring periods of up to 7 days. With C = control (no hedge or grass strip); +M = mulch (no hedge or grass strip); H2 + M = 1 m from the hedge; H3 + M = 2 m from the hedge; H2 M = 1 m from the hedge; H3 M = 2 m from the hedge; G2 M = 1 m from the grass strip; G3 M = 2 m from the grass strip.
example for a period of 11 days during which the most important three types of measuring days all occurred. They include the estimates made during rainy days (a), measurements within 7 days after rainfall (b) and measurements taken immediately after the representative measuring periods of up to 7 days (c). The second day had its considerable rainfall very early and was therefore used to test the 4 mm assumption for dry periods of rainy days. The
3. Results 3.1. Soil evaporation results for the long rains of 1994 Rainfall and pan evaporation distributions for the LR94, SR94/95 and LR95 seasons are given in Figs. 3–5. Table 1 presents soil evaporation losses from the five plots for a representative detailed
Table 2 Totals of soil evaporation losses (mm) and rainfall (mm) for the measuring periods given in the long rains of 1994 (LR94), the short rains of 1994/ 1995 (SR94/95) and the long rains of 1995 (LR95) Treatment
C
+M
H2 + M
H3 + M
H2 M
H3 M
G2 M
G3 M
Rainfall
Period
Total LR94 % of rainfall
160.7 66.3
139.4 57.5
134.1 55.3
140.1 57.8
149.1 61.5
154.4 63.7
152.0 62.7
159.0 65.6
242
18 March 1994 31 May 1994
549
13 October 1994 20 January 1995
285
18 March 1995 5 June 1995
[H2(G2) + H3(G3)]/2 % of rainfall Total SR94/95 % of rainfall
137.1 56.6 275.1 50.1
253.4 46.2
247.3 45.0
[H2(G2) + H3(G3)]/2 % of rainfall Total LR95 % of rainfall [H2(G2) + H3(G3)]/2 % of rainfall
252.2 45.9
151.8 62.6 264.3 48.1
249.8 45.5 138.8 48.7
122.1 42.8
117.0 41.1
122.0 42.8 119.5 42.0
269.8 49.1
155.5 64.2 268.8 48.9
267.1 48.6 128.6 45.1
133.4 46.8 131.0 46.0
273.2 49.8 271.0 49.4
129.6 45.5
134.3 47.1 132.0 46.3
With C = control (no hedge or grass strip); +M = mulch (no hedge or grass strip); H2 + M = 1 m from the hedge; H3 + M = 2 m from the hedge; H2 M = 1 m from the hedge; H3 M = 2 m from the hedge; G2 M = 1 m from the grass strip; G3 M = 2 m from the grass strip.
156
J.M. Kinama et al. / Agricultural and Forest Meteorology 130 (2005) 149–162
Fig. 3. Seasonal rainfall together with evaporation pan data for the long rains of 1994 (LR94).
assumption for these drying conditions gives slightly too high values and somewhat reduces the differences between treatments. Table 1 shows results for the period 2–12 January 1995, which was part of the SR94/95 (Fig. 4). A breakdown into several periods confirmed considerably higher daily soil evaporation losses in all the plots in the wetter parts after mid-March and in early April (Fig. 3), compared to the drier remainder of April and May (Kinama, 1997). The latter were also
lower because the crop had now fully developed, which further reduced the area of soil exposed to full solar radiation and wind. The end of this period was characterised by very small changes in the weight of the micro-lysimeters. This pattern is illustrated in the last days of Table 1. The overall results for LR94 in Table 2 show that there were generally somewhat higher soil evaporation losses recorded in the non-mulched C, H M and G M plots compared to the mulched plots +M and
Fig. 4. Seasonal rainfall together with evaporation pan data for the short rains of 1994/1995 (SR94/95).
J.M. Kinama et al. / Agricultural and Forest Meteorology 130 (2005) 149–162
157
Fig. 5. Seasonal rainfall together with evaporation pan data for the long rains of 1995 (LR95).
H + M, respectively. The individual periods such as exemplified in Table 1 also generally confirm this. It was of course as a result of less solar radiation reaching the mulched soil and less water vapor leaving it than for the non-mulched plots. The differences found are, however, rather small due to mulch amounts limited to about 2 t ha1 (Kinama et al., in press). At the same time, the areas at H2 + M, H2 M and G2 M also showed on average somewhat lowered soil evaporation losses compared to their neighbouring areas at H3 + M, H3 M and G3 M, respectively. This must be due to the extended shading from the hedgerows and grass strips. Cumulative relative errors in soil evaporation were derived above as being less than 1% although in absolute sense soil evaporation was overestimated as explained. Even the above smaller differences over the periods involved in Tables 1 and 2 were therefore larger than these non-biased parts of the skewed error margins, although appreciably smaller than those caused by the mulching. The soil evaporation loss from the mulched plot was clearly lower compared to that from the C plot throughout the season, while the soil evaporation at the H2 + M was always slightly lower than in the +M plot, again due to the additional hedgerow shade at the H2 + M. The overall results of Table 2 show also soil evaporation expressed as a percentage of total rainfall for the measuring period (242 mm).
Table 2 shows that, after averaging over the plots, the mulched plots had lower soil evaporation by just below 10% (in absolute values) than the control plot, while the non-mulched plots were only between 2% and 4% lower than the control. The soil evaporation on the whole ranged from 57% in H + M to 66% in the C plot control. Even if these values may be somewhat too high, they show that soil evaporation losses account for a very high proportion of water loss in the water balance equation for the semi-arid areas of Kenya. 3.2. Soil evaporation results for the short rains of 1994/1995 The results for SR94/95 are presented in Table 2, with rainfall and pan evaporation presented in Fig. 4. For this season below optimal mulch was available and this reduced soil evaporation as a percentage of rainfall in the measuring period by an absolute value of only 4% in the +M plot and in the H + M plot by only 5% (Table 2). There was on average negligible reduction in such soil evaporation by in the order of 1%, not much higher than the accuracy limits suggested, in the H M and G M compared to the C plot. Table 2 shows that in this season from 45% to 50% of the total rainfall in the measuring period (549 mm) appeared to be used up by soil evaporation as an upper limit (Section 2.4).
158
J.M. Kinama et al. / Agricultural and Forest Meteorology 130 (2005) 149–162
3.3. Soil evaporation for the long rains of 1995 With the rainfall picture in Fig. 5, soil evaporation losses are presented in Table 2. They show similar trends to those for LR94. One period apart, there was relatively little soil evaporation in May/June because the soil, covered with a well-established maize canopy, was quite dry. The total soil evaporation losses, as a percentage of the total rainfall in the measuring period, are presented in Table 2. The mulched plots +M and average H + M had reduced soil evaporation by 6% and 7% (in absolute values) compared to the C plot. The unmulched plots had soil evaporation only lowered by between 2% (G M) and 3% (H M) over the whole season compared to the control. There were still minor but consistently similar differences in soil evaporation between the points of measurements at H2 + M and H3 + M, H2 M and H3 M and at G2 M and G3 M, respectively. On average the seasonal soil evaporation losses ranged here from 45–47% of rainfall (285 mm) in the non-mulched plots to 41–43% in the mulched plots, compared to the 49% in the C-plot.
4. Discussion 4.1. Additional considerations In Table 1, one can see the three stages of drying distinguished in older soil science work, where the second stage represented evaporation from a not completely wet nor completely dry soil surface, where patches had already dried up (e.g. Stigter, 1994). The complete ‘‘self mulching’’ of stage 3 normally reduces the evaporation from soil considerably. The first stage, with completely wet soil, gave maximum evaporation. When the soil was drying at and very near the surface, the evaporation reduced till about half the maximum value in the second stage (Table 1). After which it suddenly dropped for all treatments in the third stage, when water vapor had to come from within the soil and an appreciable additional diffusion resistance was quickly added. One may argue that differences in the length and surface conditions of the second drying stage between micro-lysimeters and the soil they are supposed to represent could easily spoil this representation, but nobody has looked into that issue so far. What may
comfort us in this respect is the occurrence of the third drying stage relatively simultaneously for the different treatments, as again the data of Table 1 demonstrate. From our Table 2 data, the influence of the hedges and strips may be observed when comparing H2 and H3 respectively G2 and G3 measurements. These differences appear almost independent of the presence of mulch, that itself exerts the largest influence on soil evaporation. The fact that +M and H3 + M (with lysimeters in the middle between plant rows) hardly differ shows that H3 is unaffected by the hedge. The small to very small differences between G3 M and C confirm this for the strips. Considering also the height and composition of the hedgerows and strips, the differences between H3 M and G3 M are therefore unlikely to be due to differences in shading patterns. They must be due to differences in wetting and drying patterns caused by differences in airflow and overland water flow, due to differences in shrub/ tree, grass and soil interactions with water and wind. This is an additional reason for not considering our measurements as pure replicates but as samples of which the average should be representative for the average of the field as a whole. One alternative way to look at the soil evaporation data is to determine with an ANOVA separation of means the statistical significance of differences in daily average soil evaporation obtained from the succession of measurement periods. This is done in Table 3. These data show a statistical significance of the differences between the Control plot and all mulched plots at the P = 0.05 level. They show the same for the differences between all mulched plots and all non-mulched plots, with the exception of H2 M and G2 M in LR95. Some more of the differences that were physically significant are not always so statistically at the level considered. For example the differences between H2 M and H3 M, respectively H2 + M and H3 + M and those between G2 M and G3 M are always in exactly the same direction, but sometimes statistically significant and sometimes not. This shows the wellknown superiority of a good physical error analysis. 4.2. Discussion and comparison of the results Determination of soil evaporation leads to the separation of the transpiration of plants from the total
J.M. Kinama et al. / Agricultural and Forest Meteorology 130 (2005) 149–162 Table 3 Results of ANOVA statistical analysis—separation of daily means for soil evaporation for different treatments for the long rains of 1994 (LR94), the short rains of 1994/1995 (SR94/95) and the long rains of 1995 (LR 95) LR94
SR94/95
LR95
Treatment
Mean
Treatment
Mean
Treatment
Mean
H2 + M +M H3 + M H2 M G2 M H3 M G3 M C
1.79 1.86 1.87 1.99 2.03 2.06 2.12 2.14
H2 + M H3 + M +M H2 M G2 M H3 M G3 M C
2.50 2.55 2.56 2.67 2.72 2.73 2.76 2.78
H2 + M +M H3 + M H2 M G2 M H3 M G3 M C
1.46 1.53 1.53 1.61 1.62 1.67 1.68 1.73
d d d c bc bc ab a
f e e d c bc ab a
c bc bc b ab a a a
Differences between averages followed by a different letter or letters were statistically significant at P = 0.05. For LR94, F = 23.0; for SR94/95, F = 63.7, for LR95, F = 8.9. With C = control (no hedge or grass strip); +M = mulch (no hedge or grass strip); H2 + M = 1 m from the hedge; H3 + M = 2 m from the hedge; H2 M = 1 m from the hedge; H3 M = 2 m from the hedge; G2 M = 1 m from the grass strip; G3 M = 2 m from the grass strip.
evapotranspiration of a crop. The soil evaporation values assisted in the determination of the water use efficiency of crops (Kinama, 1997). In Kenya, Muchena (1986) noted that soil evaporation was quite high in the dry areas of Eastern Kenya. Daamen et al. (1995) indicated reduced seasonal soil evaporation of 14 2% from sparse millet crops on a sandy soil compared to bare soil. Wallace (1991) estimated soil evaporation losses from millet in semi-arid West Africa as 30–50% of the seasonal rainfall. Cooper et al. (1983) showed that evaporation losses in dryland barley in Syria were as high as 50–60% of total rainfall. Leuning et al. (1994) calculated that soil evaporation under a wheat canopy in Australia was almost 50% of total rainfall. In a temperate humid climate, on bare loamy sand in spring in Denmark, Plauborg (1995) noted soil evaporation as high as 65% and 50% of accumulated potential evapotranspiration in drying periods of 13 and 23 after wetting respectively. In semi-arid Kenya, modeling of evapotranspiration in intercropping of Senna spectabilis with maize or cowpea on flatlands indicated that soil evaporation ranged from 42% to 58% and was not significantly different from that of monocropped maize or cowpea (Mclntyre et al., 1996). The lack of difference was explained by the high proportion of rain occurring during the first
159
weeks of the rainy season, when vegetation cover was small. Moreover, tree prunings were removed from the site and not used for mulching to reduce soil evaporation. Compared to the above our upper limit soil evaporation losses from unmulched maize/senna or grass strip cropping for LR94, between 60% and 65%, were indeed high. The instrumental contribution to overestimation was never higher than 10% of the measured value. Moreover, this effect, due to water extraction outside the lysimeters by the plant/tree roots (Daamen et al., 1993), is smallest early in the season where evaporative demands were highest (Fig. 3). One other reason for somewhat high evaporation losses may have been the assumption of a maximum soil evaporation during rains. However, when we look indeed at the pan evaporation as a measure of evaporative demand, the main reason for the high values in this season is real. They are, as in the case reported above (Mclntyre et al., 1996), due to very high evaporative demands early in the season, in a period with an appreciable contribution to the lowest seasonal rainfall (Fig. 3) and still little soil cover from young plants and recently pruned hedges. This is confirmed by unmulched soil evaporation losses of 45– 50% for SR94/95 and LR95, which were well within ranges reported above for dry areas of the world. Treatment differences in soil evaporation of H2 respectively G2 plots, mainly caused by grass strips, hedges and mulch, were particularly due to shading of the soil and insulation effects where air movement is reduced by the hedges and within and below the mulch. Additionally, wind movement may have a prevailing direction towards hedgerows, which can also lead to asymmetrical shading of the rows near the hedge or grass (Mungai et al., 2000). To assess the influence of crop shade - for the mulched fields in addition to mulch shade, and closest to the hedges in addition to hedge shade—as well as the influence of changing crop shade over the season, would have demanded a much higher sampling density. Now the lysimeters were all located (close to) midway between crop rows (Section 2.4), to make them comparable. It would have needed a sampling pattern across crop rows and within crop rows for each present measuring point to representatively follow influences of crop shade.
160
J.M. Kinama et al. / Agricultural and Forest Meteorology 130 (2005) 149–162
5. Concluding remarks The low mulch rates reduced soil evaporation as percent of seasonal rainfall by absolute amounts of nearly 10% or less in LR94, a maximum of 5% for SR94/95 and in-between those values in LR95. The maximum saving in soil evaporation provided roughly an additional 20–25 mm for crop transpiration, the lowest value for the driest season. In addition to the above, mulch contributed appreciably to the reduction of runoff and soil losses, while it reduced in the H + M plots high cumulative 1993–1995 maize grain yield losses – which are due to hedge/grass strip competition – by 35% compared to the H M and G M plots (Kinama et al., in press). Another analysis, of ‘‘water cost’’ as tree transpiration of contour hedgerows and the ‘‘water cost’’ of producing mulch in order to ‘‘save’’ unproductive water losses as runoff or soil evaporation on sloping lands, indicated that the former ‘‘water cost’’ exceeds the amount saved by three to four times (Ong et al., 1996). Therefore, fodder or fruit trees would be economically more preferable for farmers than trees which just provide a long lasting mulch such as S. siamea. The fact that Kenyan farmers prefer the more competitive grass strips above tree hedgerows is due to advantages of using its biomass, and its quality as a durable stabiliser for erosion control embankment (Kinama et al., in press). However, it must be concluded from the results of our research reported here that for farmers under these semi-arid conditions with low mulch production, the choice remains between low but sustainable yields in hedgerow agroforestry or quickly deteriorating yields on unprotected sloping lands, due to soil erosion. The use of younger agroforestry systems or root pruned ageing systems combined with fertilizers, where possible with mulch brought in from outside the system, will increase the sustainable yields obtainable in this type of hedgerow agroforestry.
Acknowledgements The authors are grateful to the Traditional Techniques of Microclimate Improvement (TTMI) Project, funded by the Directorate General for International Cooperation (DGIS), Ministry of For-
eign Affairs, The Netherlands, at Wageningen University and the University of Nairobi, for providing equipment and co-supervision for J.M. Kinama’s PhD research. We thank the Director, Kenya Agricultural Research Institute (KARI) for his paid study leave and use of research facilities, and the Swedish International Development Agency (SIDA) for supporting field work. Additional support was provided by the International Centre for Research in Agro-Forestry (ICRAF). We thank the TTMI-Project and Mr. A. Khan (ICRAF) for support in development of instrumentation and Mr. John Mailu (ICRAF) for field assistance.
References Allen, S.J., 1990. Measurement and estimation of evaporation from soil under sparse barley crops in Northern Syria. Agric. For. Meteorol. 49, 291–309. Baldy, C., Stigter, C.J., 1997. Agrometeorology of Multiple Cropping in Warm Climates. INRA/Science Publishers/ Oxford & IBH Publishers, Paris/Enfield (USA)/New Delhi, 237 pp. Barber, R.G., Thomas, D.B., Moore, T.R., 1979. The erodibility of two soils from Kenya. J. Soil Sci. 30, 579–591. Belsky, A.J., Mwonga, S.M., Amundson, R.G., Duxbury, J.M., Ali, A., 1993. Comparative effects of isolated trees on their understorey environment in high and low rainfall savannas. J. Appl. Ecol. 30, 143–155. Bley, J., Van der Ploeg, R.R., Sivakumar, M.V.K., Allison, B.E., 1991. A risk-probability map for millet production in south–west Niger. In: Sivakumar, M.V.K., Wallace, J.S., Renard, C., Giroux, C. (Eds.), Soil Water Balance in the Sudano-Sahelian Zone. IAHS Publ. No. 199, IAHS Press, Wallingford, pp. 571–581. Boast, C.W., Robertson, T.M., 1982. A ‘‘microlysimeter’’ method for determining evaporation from bare soil: description and laboratory evaluation. Soil Sci. Soc. Am. J. 46, 469–696. Boffa, J.M., 1999. Agroforestry Parklands in Sub-Saharan Africa. FAO Conservation Guide 34. FAO, Rome. Cannell, M.G.R., Van Noordwijk, M., Ong, C.K., 1996. The central agroforestry hypothesis: the trees must acquire resources that the crop would not otherwise acquire. Agrofor. Syst. 34, 27–31. Cooper, P.J., Keatinge, J.D.H., Hughes, G., 1983. Crop evapotranspiration—a technique for calculating its components by field measurements. Field Crops Res. 7, 299–312. Daamen, C.C., Simmonds, L.P., Wallace, J.S., Laryes, K.B., Sivakumar, M.V.K., 1993. Use of micro-lysimeters to measure evaporation from sandy soils. Agric. For. Meteorol. 65, 159– 173. Daamen, C.C., Simmonds, L.P., Sivakumar, M.V.K., 1995. The impact of sparse millet crops on evaporation from soil in semi-arid Niger. Agric. Water Manage. 27, 225–242.
J.M. Kinama et al. / Agricultural and Forest Meteorology 130 (2005) 149–162 FAO/UNESCO, 1988. Soil Map of the World 1: 5,000,000. Revised Legends. FAO, Rome. Jackson, N.A., 2000. Measured and modelled rainfall interception loss from an agroforestry system in Kenya. Agric. For. Meteorol. 100, 323–336. Jackson, N.A., Wallace, J.S., 1999. Soil evaporation measurements in an agroforestry system in Kenya. Agric. For. Meteorol. 94, 203–216. Johnsson, K., 1995. Agroforestry in dry savanna areas in Africa: interactions between trees, soils and crops. PhD Thesis. Swedish University of Agricultural Sciences, Umea, Sweden. Kibe, J.M., Ochung, H., Macharia, P.N., 1981. Soils and vegetation of the ICRAF experimental farm (Machakos District). Ministry of Agriculture, National Agricultural Laboratories, Kenya Soil Survey, Detailed Soil Survey Report No. D 23. Kiepe, P., 1995. No Runoff, No Soil Loss: Soil and Water Conservation in Hedgerow Barrier Systems, Tropical Resource Management Papers 10. Wageningen Agricultural University, The Netherlands, 156 pp. Kilewe, A.M., Ulsaker, L.G., 1984. Soil physical characteristics and their application to agriculture. E. Afr. Agric. For. J. 44, 247– 255. Kinama, J.M., 1997. The effects of hedgerows (alley cropping) on microclimate, soil and water conservation and competition for sustainable land use on the sloping areas of Machakos, Kenya. PhD Thesis. University of Nairobi, Kenya. Kinama, J.M., Ng’ang’a, J.K., Ong, C.K., Stigter, C.J., Gichuki, F.N., 1999. Quantification of soil evaporation in cowpea/maize or grass strip systems in semi-arid eastern Kenya. Paper Presented at the National Soil and Water Management Conference, KARI, Nairobi, Kenya. Kinama, J.M., Stigter, C.J., Ong, C.K., Ng’ang’a, J.K., Gichuki, F.N., in press. Contour hedgerows and grass strips in erosion & runoff control in semi-arid Kenya, and their competitive effects. Resubmitted after revision to Agric. Ecosyst. Environ. Kinyamario, J.I., Trlica, M.J., Njoka, T.J., 1995. Influence of tree shade on plant water status, gas exchange, and water use efficiency on Panicum maximum Jacq. and Themeda triandra Forsk in a Kenyan Savanna. Afric. J. Ecol. 33, 114–123. Leuning, R., Condon, A.G., Dunin, F.X., Zegelin, S., Denmead, O.T., 1994. Rainfall interception and evaporation from soil below a wheat canopy. Agric. For. Meteorol. 67, 221–238. Lott, J.E., 1998. Resource capture and use in semi-arid overstorey agroforestry systems. PhD Thesis. University of Nottingham, UK. Marimi, P.M., 1979. The effects of some tillage methods and cropping systems in conserving rainfall in a semi-arid area of Eastern Kenya. In: Presented at a Soil and Water Conservation Workshop. Kabete. University of Nairobi, Nairobi, Kenya. Mbuvi, P.J., Van de Weg, R.F., 1975. Some preliminary notes on the soils of Katumani, Kampi-ya-mawe, Embu and Murinduko Agricultural Research Stations. Report No. P25, Kenya Soil Survey, Nairobi, Kenya. Mclntyre, B.D., Riha, S.J., Ong, C.K., 1996. Light interception and evapotranspiration in hedgerow agroforestry systems. Agric. For. Meteorol. 81, 31–40.
161
Muchena, F.N., 1986. Soil fertility constraints in improving cereal yields in soils of the arid and semi-arid areas of Kenya. In: Paper Presented at Organisation of African Unity (OAU) Drought Symposium, Kenyatta International Conference Centre, OAU, Nairobi, Kenya. Mugendi, D.N., Mochoge, B.O., Coulson, C.L., Stigter, C.J., Arap Sang, F.K., 1994. Decomposition of Cassia siamea loppings in semi-arid Machakos, Kenya. Arid Soil Res. Rehabil. 8, 363–372. Mungai, D.N., 1995. A micrometeorological approach to understanding maize yield performance in alley cropping in the semi-arid areas of Machakos District, Kenya. In: Stigter, C.J., Wangati, F.J., Ng’ang’a, J.K., Mungai, D.N. (Eds.), The TTMI-Project and the ‘‘Picnic’’ Model. An Internal Evaluation of Approaches and Results and of Prospects for TTMI-Units. Wageningen Agricultural University, The Netherlands, pp. 111–123. Mungai, D.N., Stigter, C.J., Coulson, C.L., Ng’ang’a, J.K., 2000. Simply obtained global radiation, soil temperature and soil moisture in alley cropping systems in semi-arid Kenya. Theor. Appl. Climatol. 65, 63–78. Mungai, D.N., Stigter, C.J., Coulson, C.L., Ng’ang’a, J.K., Netondo, G.W.S., Umaya, G.O., 2001. Understanding yields in alley cropping maize (Zea mays L) and Cassia siamea (Lam) under semi-arid conditions in Machakos, Eastern Kenya. J. Environ. Sci. 13, 291–298. Olufayo, A.A., Stigter, C.J., Baldy, C., 1998. On needs and deeds in agrometeorology in tropical Africa. Agric. For. Meteorol. 92, 227–240. Ong, C.K., Black, C.R., 1992. Complementarity in resource use in intercropping and agroforestry systems. In: Paper Presented at the 52nd International Easter School on Resource Capture by Crops, School of Agriculture, Sutton Bonington, UK. Ong, C.K., Black, C.R., Marshall, F.M., Corlett, J.E., 1996. Principles of resource capture and utilisation of light and water. In: Ong, C.K., Huxley, P. (Eds.), Tree–Crop Interactions: A Physiological Approach. CAB International, Wallingford, UK, pp. 73–185. Plauborg, P., 1995. Evaporation from bare soil in a temperate humid climate. Measurements using micro-lysimeters and time domain reflectometry. Agric. For. Meteorol. 76, 1–17. Rao, M.R., Westley, S.B., 1989. Agroforestry for the African semiarid zone. Agrofor. Today 1, 5–11. Sanchez, P.A., 1995. Science in agroforestry. Agrofor. Syst. 30, 5– 55. Soil Survey Staff, 1990. Keys to Soil Taxonomy, 4th ed. S.M.S.S. Technical Monograph 19. Virginia Polytechnic and State University, Blacksburg, VA, 422 pp. Stigter, C.J., 1994. Micrometeorology. In: Griffiths, J.F. (Ed.), Handbook of Agricultural Meteorology. Oxford University Press, (Chapter 5), pp. 59–72. Van Noordwjik, M., Ong, C.K., 1999. Can the ecosystem mimic hypothesis be applied to farms in African savannahs? Agrofor. Syst. 45, 131–158. Wallace, J.S., 1991. The measurement and modelling of evaporation from semi-arid land. In: Sivakumar, M.V.K., Wallace,
162
J.M. Kinama et al. / Agricultural and Forest Meteorology 130 (2005) 149–162
J.S., Renard, C., Giroux, C. (Eds.), Soil Water Balance in the Sudano-Sahelian Zone. IAHS Publ. No. 199, IAHS Press, Wallingford, pp. 131–145. Wallace, J.S., Gash, J.H.C., McNeil, D.D., Sivakumar, M.V.K., 1988. Evaporation from sparse dryland millet crops in Niger West Africa. In: Unger, P.W., Sneed, T.V., Jordan, W.R., Jensen,
R. (Eds.), Proceedings of the International Conference on Dryland Farming, Texas Agricultural Experiment Station, Armavillo, Bushland, TX, pp. 325–327. Wallace, J.S., Jackson, N.A., Ong, C.K., 1999. Modelling soil evaporation in an agroforestry system in Kenya. Agric. For. Meteorol. 94, 189–203.
Evaporation from soils below sparse crops in contour hedgerow agroforestry in semi-arid Kenya J.M. Kinama a, C.J. Stigter b,*, C.K. Ong c, J.K. Ng’ang’a d, F.N. Gichuki e b
a TTMI-Project, National Dryland Farming Research Centre, KARI, Katumani, P.O. Box 340, Machakos, Kenya TTMI-Project, Department of Environmental Sciences, Wageningen University, Duivendaal 2, 6701 AP Wageningen, The Netherlands c International Centre for Research in Agroforestry, P.O. Box 30677, Nairobi, Kenya d TTMI-Project, Department of Meteorology, University of Nairobi, P.O. Box 30197, Nairobi, Kenya e Department of Agricultural Engineering, University of Nairobi, P.O. Box 30197, Nairobi, Kenya
Received 27 April 2004; received in revised form 9 February 2005
Abstract In many agricultural systems in the semi-arid tropics, crops use only a small fraction of the total rainfall. Agroforestry can greatly reduce some losses, especially on hill slopes, where soil evaporation, runoff and soil losses are important. This paper reports on soil evaporation from a rotation of intercropped maize and cowpea between contour hedgerows of pruned Senna siamea trees as well as trimmed Panicum maximum grass strips on a 14% hill slope at a semi-arid site in Machakos, Kenya. There were five treatments in order to separate effects of Senna mulch, hedges, and grass strips. Micro-lysimeters were placed between crop rows for three seasons. It followed from their results that, for the three seasons concerned, tree prunings as mulch reduced soil evaporation as percentage of rainfall in the measuring period by absolute values of 9%, 4% and 6% compared to the control sole maize and cowpea with bare soil. The influence of the hedge added to this only insignificantly, even at 1 m distance. The non-mulched plots had soil evaporation reduced by only between on average 1% and 4% in absolute values compared to the control over all the seasons, with a maximum of 5% close to the hedge in the first season. Mulch apparently is the main evaporation reducing factor. Soil evaporation reached the highest percentage of rainfall in the long rains of 1994, becoming 65% in sole maize. It was 50% for sole cowpea in the 1994/1995 short rains and for sole maize in the next long rains. The highest value, although an upper limit could largely be understood from highest early season evaporative demands, rainfall distributions and low crop cover. The other values were in line with earlier reports for dry areas. Some advantages and disadvantages of these agroforestry systems are reviewed. # 2005 Elsevier B.V. All rights reserved. Keywords: Soil evaporation; Micro-lysimeters; Agroforestry systems; Hedgerow intercropping; Mulch; Senna siamea; Grass strips; Maize; Cowpea
1. Introduction * Corresponding author. Tel.: +31 317 483 981; fax: +31 317 482 811. E-mail address: [email protected] (C.J. Stigter).
Decisions on land use, such as ‘‘being forced to use sloping land for production’’, belong to the strategic needs in tropical Africa, and choices of cropping
0168-1923/$ – see front matter # 2005 Elsevier B.V. All rights reserved. doi:10.1016/j.agrformet.2005.03.007
150
J.M. Kinama et al. / Agricultural and Forest Meteorology 130 (2005) 149–162
systems, such as ‘‘one that reduces soil loss and water runoff on slopes’’ to the tactical needs (Olufayo et al., 1998). An important goal of combining trees and crops in agroforestry systems is to make better use of the environmental resources required by plants to grow (e.g. Mungai et al., 2001; Kinama et al., in press). One of the means of achieving this is to utilize resources that would otherwise be lost from the system (e.g. Ong and Black, 1992; Cannell et al., 1996). Wallace (1991) suggested that introduction of trees into a crop may lead to an overall increase in the proportion of rainfall that is used as transpiration. Van Noordwjik and Ong (1999) estimated by modeling that the potential increase in transpiration due to scattered trees in the Sahel is only about 8% of total rainfall. Although there is little information on the role of agroforestry in reducing soil evaporation, it is expected that maximum gain will be made in semi-arid and arid environments where soil evaporation forms a substantial part of the soil water balance (e.g. Stigter, 1994). Under west African conditions, Wallace et al. (1988) and Bley et al. (1991) showed that soil evaporation is a major part of the soil water balance that depends on soil wetness. The influence of scattered tree shade on soil evaporation was studied in the Sahel savanna of Africa (Johnson, 1995; Belsky et al., 1993; Kinyamario et al., 1995). Higher topsoil moisture under woody canopies than in treeless sites appears to be common during the rainy season and sometime afterwards, as exemplified in Boffa (1999). Wallace et al. (1999) modeled the reduction of soil evaporation under a tree canopy in Kenya to be 35% on an annual basis. In general, shading has beneficial effects but total shade area may only account for about 10% of total land area. Measurements of soil evaporation under 6–8 m tall Grevillea robusta trees with a leaf area index of 2 at Machakos, using a micro-lysimeter technique, showed that soil evaporation is reduced by 23%, largely attributed to tree shade (Jackson and Wallace, 1999). Maintaining a large leaf area, however, is not a viable option for water limited environments since competition for soil water by dominant trees results in frequent crop failures (Lott, 1998). Pruning of branches and roots may of course reduce competition. Thus, there is a trade-off between the advantages of shade on lowering soil evaporation and improving crop aerial climate versus the negative effect of water competition from tree roots. The same
applies to advantages of soil loss and run off reductions of hedgerows (Kinama et al., in press). In our research, water lost by evaporation from soil deserved separate attention from water lost by run off in the erosion causing processes. It is a considerable loss and the influence of mulch (and to a much lesser extent the contour hedge rows) is due to completely different processes compared to their influence on run off losses. Soil cover is considered to be the dominant factor of all parameters that affect erosion. Research on contour hedgerows on sloping land in semi-arid Kenya has quantified the impact of Senna siamea mulch and tree barriers on runoff and soil loss (Kiepe, 1995; Kinama et al., in press). S. siamea mulch is ideal for mulching because of its slow decomposition rate (Mugendi et al., 1994). Measurements by Kiepe (1995) in a young S. Siamea based agroforestry system over 3 years showed that mulching alone reduced runoff by 50%, even during heavy storms, and soil loss by more than 80%. Kinama et al. (in press) showed that in four seasons of this same rotation of maize and cowpea with S. siamea, in the mulched hedgerow plots cumulative soil loss reduced from 100 to only 2 t ha1, and runoff from 100 to 20 mm. However, due to competition between crops and hedgerows, yields have better to be supported by fertilizers, otherwise losses are almost prohibitive. Ageing systems show increased competition (Kinama et al., in press). Critical studies exist of alley cropping in the semiarid tropics on flat land (Rao and Westley, 1989; Mungai, 1995; Sanchez, 1995; Mungai et al., 2001). But there is much more scope for increasing the use of limited rainfall in contour hedgerows on sloping lands. Hedge width and tree and hedge spacing and pruning as well as the use made of mulch are issues together with other economic ones (Kiepe, 1995; Ong et al., 1996; Kinama et al., in press), that may also affect soil evaporation. Understanding soil evaporation is necessary in the evaluation of the water balance of these agroforestry systems, just like that has been noted for rainfall interception losses (Jackson, 2000). 2. Materials and methods 2.1. Instrument This paper describes the use of mini- or microlysimeters to quantify the influence of S. siamea
J.M. Kinama et al. / Agricultural and Forest Meteorology 130 (2005) 149–162
mulch, contour hedgerows and grass strips on soil evaporation on sloping land at the semi-arid site in eastern Kenya detailed below. Boast and Robertson (1982) and Daamen et al. (1993) have extensively described the technique used in the present study. 2.2. Experimental site The on-station trials were conducted at ICRAF’s Research Station at Machakos, which is about 70 km South East of Nairobi and 7 km from Machakos town. The station lies between latitudes 18300 and 18350 S and longitudes 378 and 378150 E. It has an altitude of 1560 m above sea level with slopes ranging from 0% to 22%. The experimental plots were established on sloping land of about 14%. The site receives 410 mm on average for the short rains (SRs), which are from mid-October till January, and 360 mm on average for the long rains (LRs), which are from mid-March to July. Seasonal rainfall and rainfall distribution over the season are highly variable. The site received from 1989 to 1995 between 290 and 810 mm in the SRs and between 110 and 620 mm in the LRs (Kinama et al., in press). During SR94/95 the rainfall was above average, while those in the two LRs were below average. The soils are sandy clay loams over sandy clay developed in situ on rocks of the Precambrian basement complex. The soils are about 150 cm deep and have originally been classified as chromic luvisols (Kibe et al., 1981), but revised as Haplic Lixisols (FAO/UNESCO, 1988) or Kanhaplic Rhodustaff (Soil Survey Staff, 1990). They are dark reddish brown, sandy clay loam becoming sandy clay at the lower horizons (Mbuvi and Van de Weg, 1975; Barber et al., 1979; Kilewe and Ulsaker, 1984). They are shallow due to the presence of a pentroplinthite (Murram) horizon (Marimi, 1979). Due to low structural stability, the soils are prone to slaking, highly erodible and prone to surface capping by intense rainfall. This risk is enhanced by low sub soil permeability (Kiepe, 1995). 2.3. Experimental design The experimental plots (Fig. 1) were on land which had been under alley cropping with hand hoe cultivation, with long term runoff/soil erosion monitoring since the establishment of the hedgerows
151
in 1988, as described by Kiepe (1995) and Kinama (1997). Grass strips (Panicum maximum) were established earlier, in 1984. The plant rows, the grass strips and the S. siamea hedgerows were planted along the contours, lying roughly E–W. The experiments covered in this study were for three out of six cropping seasons studied. During the SRs cowpea (Vigna unguiculata, cv. K80 or SK-27) was planted while maize (Zea mays, cv. Katumani composite B) was sown during the LRs. S. siamea, a non-modulating leguminous tree, was chosen because it was among the few multi-purpose trees/shrubs considered suitable for the area as contour hedgerow barriers. The tree species is drought tolerant and suited to the local semi-arid conditions, as reported by Rao and Westley (1989) and Mugendi et al. (1994). Its mulch is suitable for erosion control purposes because of the high amounts of tannin in the mulch (Kiepe, 1995), making disintegration rates low (Mugendi et al., 1994). Only S. siamea loppings from the hedgerows were spread uniformly on the soil surface. The quantities of mulch applied therefore depended on the rainfall conditions prior to their application. The hedgerows which had reached about 1.5 m in height, were cut down to 25 cm 2 weeks before the expected onset of the rains, delivering 1.7 0.3 t ha1 of mulch (Kinama et al., in press). The grass strips were cut down to 25 cm at harvest (from about 1 m height) and 2 weeks before planting (from about 50 to 60 cm height). The total biomass (2.5 0.4 t ha1) was carried off the plots for other purposes (Kinama et al., in press). No fertilizer was used during the six seasons of crop measurements. The study consisted of five treatments with no replicates. The plots measured 10 m width 40 m downslope and it was the sampling procedure which was replicated. This means that there were various sampling points in each plot. The following treatments were used (Fig. 1a): Treatment 1: Maize or cowpea control (C). Treatment 2: Maize or cowpea + S. siamea mulch (+M). Treatment 3: Maize or cowpea + S. siamea hedgerows + mulch (H + M). Treatment 4: Maize or cowpea + S. siamea hedgerows without mulch (H M). Treatment 5: Maize or cowpea + grass strips without mulch (G M).
152
J.M. Kinama et al. / Agricultural and Forest Meteorology 130 (2005) 149–162
Because local farmers never use Panicum grass cuttings as mulch, but use them for other purposes, such as roof thatching and animal fodder, there was no G + M plot (Kinama et al., in press). The +M plot had its mulch obtained from the H M plot, while the H + M plot had mulch from its own hedgerows. There were four rows of maize in each of the alleys formed by the S. siamea hedgerows (Fig. 1b). These hedgerows were 4 m apart and within row tree distance was 25 cm. The closest maize row to these hedgerows was 50 cm (Fig. 1b). The spacing of the maize that was completely added to the hedges, that is with no adaptation of spacing to the intercrop situation (Baldy and Stigter, 1997), was 100 cm 27 cm. This gave a density of just over 37,000 plants ha1. The G M treatment, with five rows of maize between each pair of seven grass strips, had a population of 33,670 maize plants ha1 because the seven grass
strips occupied an area of about 70 m2. The width of the strips was more than 1 m and their centres were between 6 and 7 m apart. The distances from grass strips to first maize rows were about 50 cm. The six rows of cowpea between each pair of the eleven hedgerows were planted at a spacing of 60 cm 20 cm but with a distance of 1 m between the two rows of cowpea at each side of the hedges (Fig. 1b), so not completely added (Baldy and Stigter, 1997). This gave a plant density of 83,333 plants ha1 in the C and +M plots and of 75,000 plants ha1 in the H + M and H M plots. With nine rows of cowpea between the seven grass strips, the G M plot had a lower plant density per unit area of crop, which would become bigger if calculated with 70 m2 as taken up by the grass strips, so calculated per unit area of land. This illustrates the semantic difficulties in calculations of yields per hectare in such agroforestry experiments
Fig. 1. (a) Field layout at Machakos field station showing the sampling points for measuring soil evaporation in the various treatments. Vertical lines are plot edges along the slope direction. Horizontal lines are the hedgerows respectively the grass strips. Collecting tanks were used for soil and water runoff determinations; (b) planting layout for the hedges with their respective intercrops, (A) maize; (B) cowpea.
J.M. Kinama et al. / Agricultural and Forest Meteorology 130 (2005) 149–162
153
Fig. 1. (Continued ).
and we should use row countings and yields per row for that purpose (Kinama et al., in press). 2.4. Soil evaporation quantification Soil evaporation was measured by using microlysimeters (Boast and Robertson, 1982; Daamen et al., 1993; Kinama et al., 1999) made of a PVC cylinder measuring 10.5 cm outer diameter 15 cm depth. The inner diameter was 10.25 cm. The PVC cylinders were commercially constructed in Nairobi with the assistance of ICRAF technicians, from a joint design by the TTMI-Project and ICRAF. This cylinder held a soil core. The cylinder was encased by a PVC outer cylinder of slightly bigger diameter but of similar depth to the inner measuring cylinder (Fig. 2). In order to obtain a soil core, the inner cylinder was driven into the soil using a wooden hammer. This was done causing minimum soil disturbance so that the soil core obtained remained as far as possible in the same condition as the surrounding soil. Given the crustforming nature of the Machakos soil (Allen, 1990),
this was always done when the soil was fairly wet, when the seal rarely cracks. The soil core was then carefully removed using a local panga (machete) and trimmed at the bottom with a sharp knife. The soil core was tightly closed at the bottom end by a cylindrical glassy encasing material of slightly smaller diameter than that of the inner cylinder, reinforced by plastic Sellotape (Daamen et al., 1993). The soil core (micro-lysimeter) was placed back into the soil with the external cylindrical encasing such that the micro-lysimeter was slightly above the soil level. This was to ensure that no runoff and splash water or soil particles entered the micro-lysimeter. The microlysimeter was weighed by a portable balance and replaced back into the soil (Fig. 2). The weight of the internal cylinder, etc. was determined and recorded before the preparation of the soil core. Micro-lysimeters were placed at each of the six sampling points in each of the five treatment plots (Fig. 1a). In the C and +M plots, two micro-lysimeters were installed at 10 m downslope from the top of the plots, 25 m downslope and 35 m downslope respectively, with those in the +M plot having mulches
154
J.M. Kinama et al. / Agricultural and Forest Meteorology 130 (2005) 149–162
Fig. 2. Micro-lysimeter designed for estimating soil evaporation, with (a) external cylinder; (b) internal cylinder; (c) wire handle; (d) raised part of the soil core; (e) bulk of the soil core and (f) bottom sealing encasing.
representatively placed on them. At these positions, the two micro-lysimeters were placed 1 m apart parallel to the crop rows, across the slope and halfway between these plant rows. In the H + M, H M and G M plots, the microlysimeters were installed (Fig. 1a) at 1 m (H2 and G2) and 2 m (H3 and G3) below the fourth, above the seventh and below the ninth hedgerow and below the third, above the fifth and below the sixth grass strip, respectively. In these positions, they were placed midway between the rows of maize. For the cowpea, they were placed between the first and second row and between the third and fourth row below and above the same hedgerows and grass strips. More details on these placements with respect to the cowpea rows can be found in Kinama (1997). Those in H + M plots had mulch placed on them to represent those treatments. To measure soil evaporation, the micro-lysimeters were weighed using a portable balance (0.1 g) both in the morning 0900 h and in the afternoon 1600 h local time, and the differences in weight represented the water loss. Because the surface area of the microlysimeter was 82.7 cm2, 0.1 g is equivalent to a depth of water of 0.012 mm. Such data were obtained for up to a maximum of seven rainfree days after every rainfall event, following the well-researched protocol proposed by Daamen et al. (1993). Daamen et al. (1993) determined that not changing the lysimeters at the end of each of the first 2 days after rainfall gave an
overestimation in the accumulated evaporation for the period concerned. We made that assessment also for our soils and conditions and found experimentally that this was never more than 10% of the measured values. The accumulated differences between treatments are even appreciably less affected, because the errors are always in the same direction. Daamen et al. (1993) also found that for microlysimeters more than 100 mm deep such as ours, after about 7 days the conditions in the micro-lysimeters are no longer representative of the surrounding soil. This was in our case also the limit after which evaporation could be considered almost negligible because of root extractions and self-mulching (formation of a dry soil layer that acts as an insulating mulch) of the drying soil. Such days were normally measured with unrenewed micro-lysimeters, as suggested by the part of the protocol followed. They had generally values of or below 0.4 0.1 mm. Results during rainy days have generally been found unreliable (Allen, 1990; Daamen et al., 1993), so the soil cores were replaced when it rained within the measuring period. The total soil evaporation loss was obtained by taking the sum of all the water losses for each treatment during the measurements, together with the estimates of soil evaporation during rainy days. Several microlysimeter measurements were carried out during rainy days with clear portions of the wet days allowing for measurements. This was done to form a basis for assumptions for the soil evaporation estimates during wet days. On average the evaporation losses during such dry periods with wet surfaces were about 4 mm per day. This was therefore used throughout for measuring days interrupted by rainfall. Within treatment measurement variation between places such as averaged in the columns of Table 1 were always within 0.2 mm for the larger values (between 2 and 3.9 mm) and each average had therefore an appreciably lower relative error. For the lower values there were rarely differences outside the balance absolute accuracy. Cumulative relative errors were therefore appreciably smaller than the differences between treatments and always less than 1% in the totals of Table 2. Because of the discontinuity and limitations of soil evaporation measurements from wet soil, trends in evaporative demand over the season were represented by daily weather station measurements of a class-A evaporation pan.
J.M. Kinama et al. / Agricultural and Forest Meteorology 130 (2005) 149–162
155
Table 1 Example of daily soil evaporation (mm), with subtotal for the period ‘‘2–12 January’’ for the short rains of 1994/1995, when cowpea was grown, for the five treatments and the positions concerned between the hedge rows Treatment
C
+M
H2 + M
H2 M
H3 + M
H3 M
G2 M
G3 M
Day type (a) (b) (b) (b) (b) (b) (b) (b) (c) (c) (c)
4.0 4.0 3.5 3.0 2.8 2.5 1.3 0.6 0.3 0.2 0.4
4.0 3.4 3.0 2.6 2.4 2.0 0.9 0.3 0.1 0.0 0.2
4.0 3.2 2.9 2.4 2.3 1.9 0.8 0.2 0.2 0.0 0.2
4.0 3.3 3.0 2.5 2.4 2.1 0.9 0.2 0.3 0.0 0.2
4.0 3.6 3.3 2.7 2.5 2.3 1.0 0.5 0.2 0.1 0.4
4.0 3.7 3.4 2.8 2.6 2.4 1.2 0.6 0.2 0.2 0.3
4.0 3.6 3.4 2.9 2.6 2.4 1.3 0.6 0.3 0.2 0.3
4.0 3.8 3.4 2.9 2.7 2.5 1.3 0.6 0.3 0.2 0.4
Sub total
22.6
18.9
18.1
18.9
20.6
21.4
21.6
22.1
(a) Estimates made during rainy days; (b) measurements within 7 days after rainfall and (c) measurements after the representative measuring periods of up to 7 days. With C = control (no hedge or grass strip); +M = mulch (no hedge or grass strip); H2 + M = 1 m from the hedge; H3 + M = 2 m from the hedge; H2 M = 1 m from the hedge; H3 M = 2 m from the hedge; G2 M = 1 m from the grass strip; G3 M = 2 m from the grass strip.
example for a period of 11 days during which the most important three types of measuring days all occurred. They include the estimates made during rainy days (a), measurements within 7 days after rainfall (b) and measurements taken immediately after the representative measuring periods of up to 7 days (c). The second day had its considerable rainfall very early and was therefore used to test the 4 mm assumption for dry periods of rainy days. The
3. Results 3.1. Soil evaporation results for the long rains of 1994 Rainfall and pan evaporation distributions for the LR94, SR94/95 and LR95 seasons are given in Figs. 3–5. Table 1 presents soil evaporation losses from the five plots for a representative detailed
Table 2 Totals of soil evaporation losses (mm) and rainfall (mm) for the measuring periods given in the long rains of 1994 (LR94), the short rains of 1994/ 1995 (SR94/95) and the long rains of 1995 (LR95) Treatment
C
+M
H2 + M
H3 + M
H2 M
H3 M
G2 M
G3 M
Rainfall
Period
Total LR94 % of rainfall
160.7 66.3
139.4 57.5
134.1 55.3
140.1 57.8
149.1 61.5
154.4 63.7
152.0 62.7
159.0 65.6
242
18 March 1994 31 May 1994
549
13 October 1994 20 January 1995
285
18 March 1995 5 June 1995
[H2(G2) + H3(G3)]/2 % of rainfall Total SR94/95 % of rainfall
137.1 56.6 275.1 50.1
253.4 46.2
247.3 45.0
[H2(G2) + H3(G3)]/2 % of rainfall Total LR95 % of rainfall [H2(G2) + H3(G3)]/2 % of rainfall
252.2 45.9
151.8 62.6 264.3 48.1
249.8 45.5 138.8 48.7
122.1 42.8
117.0 41.1
122.0 42.8 119.5 42.0
269.8 49.1
155.5 64.2 268.8 48.9
267.1 48.6 128.6 45.1
133.4 46.8 131.0 46.0
273.2 49.8 271.0 49.4
129.6 45.5
134.3 47.1 132.0 46.3
With C = control (no hedge or grass strip); +M = mulch (no hedge or grass strip); H2 + M = 1 m from the hedge; H3 + M = 2 m from the hedge; H2 M = 1 m from the hedge; H3 M = 2 m from the hedge; G2 M = 1 m from the grass strip; G3 M = 2 m from the grass strip.
156
J.M. Kinama et al. / Agricultural and Forest Meteorology 130 (2005) 149–162
Fig. 3. Seasonal rainfall together with evaporation pan data for the long rains of 1994 (LR94).
assumption for these drying conditions gives slightly too high values and somewhat reduces the differences between treatments. Table 1 shows results for the period 2–12 January 1995, which was part of the SR94/95 (Fig. 4). A breakdown into several periods confirmed considerably higher daily soil evaporation losses in all the plots in the wetter parts after mid-March and in early April (Fig. 3), compared to the drier remainder of April and May (Kinama, 1997). The latter were also
lower because the crop had now fully developed, which further reduced the area of soil exposed to full solar radiation and wind. The end of this period was characterised by very small changes in the weight of the micro-lysimeters. This pattern is illustrated in the last days of Table 1. The overall results for LR94 in Table 2 show that there were generally somewhat higher soil evaporation losses recorded in the non-mulched C, H M and G M plots compared to the mulched plots +M and
Fig. 4. Seasonal rainfall together with evaporation pan data for the short rains of 1994/1995 (SR94/95).
J.M. Kinama et al. / Agricultural and Forest Meteorology 130 (2005) 149–162
157
Fig. 5. Seasonal rainfall together with evaporation pan data for the long rains of 1995 (LR95).
H + M, respectively. The individual periods such as exemplified in Table 1 also generally confirm this. It was of course as a result of less solar radiation reaching the mulched soil and less water vapor leaving it than for the non-mulched plots. The differences found are, however, rather small due to mulch amounts limited to about 2 t ha1 (Kinama et al., in press). At the same time, the areas at H2 + M, H2 M and G2 M also showed on average somewhat lowered soil evaporation losses compared to their neighbouring areas at H3 + M, H3 M and G3 M, respectively. This must be due to the extended shading from the hedgerows and grass strips. Cumulative relative errors in soil evaporation were derived above as being less than 1% although in absolute sense soil evaporation was overestimated as explained. Even the above smaller differences over the periods involved in Tables 1 and 2 were therefore larger than these non-biased parts of the skewed error margins, although appreciably smaller than those caused by the mulching. The soil evaporation loss from the mulched plot was clearly lower compared to that from the C plot throughout the season, while the soil evaporation at the H2 + M was always slightly lower than in the +M plot, again due to the additional hedgerow shade at the H2 + M. The overall results of Table 2 show also soil evaporation expressed as a percentage of total rainfall for the measuring period (242 mm).
Table 2 shows that, after averaging over the plots, the mulched plots had lower soil evaporation by just below 10% (in absolute values) than the control plot, while the non-mulched plots were only between 2% and 4% lower than the control. The soil evaporation on the whole ranged from 57% in H + M to 66% in the C plot control. Even if these values may be somewhat too high, they show that soil evaporation losses account for a very high proportion of water loss in the water balance equation for the semi-arid areas of Kenya. 3.2. Soil evaporation results for the short rains of 1994/1995 The results for SR94/95 are presented in Table 2, with rainfall and pan evaporation presented in Fig. 4. For this season below optimal mulch was available and this reduced soil evaporation as a percentage of rainfall in the measuring period by an absolute value of only 4% in the +M plot and in the H + M plot by only 5% (Table 2). There was on average negligible reduction in such soil evaporation by in the order of 1%, not much higher than the accuracy limits suggested, in the H M and G M compared to the C plot. Table 2 shows that in this season from 45% to 50% of the total rainfall in the measuring period (549 mm) appeared to be used up by soil evaporation as an upper limit (Section 2.4).
158
J.M. Kinama et al. / Agricultural and Forest Meteorology 130 (2005) 149–162
3.3. Soil evaporation for the long rains of 1995 With the rainfall picture in Fig. 5, soil evaporation losses are presented in Table 2. They show similar trends to those for LR94. One period apart, there was relatively little soil evaporation in May/June because the soil, covered with a well-established maize canopy, was quite dry. The total soil evaporation losses, as a percentage of the total rainfall in the measuring period, are presented in Table 2. The mulched plots +M and average H + M had reduced soil evaporation by 6% and 7% (in absolute values) compared to the C plot. The unmulched plots had soil evaporation only lowered by between 2% (G M) and 3% (H M) over the whole season compared to the control. There were still minor but consistently similar differences in soil evaporation between the points of measurements at H2 + M and H3 + M, H2 M and H3 M and at G2 M and G3 M, respectively. On average the seasonal soil evaporation losses ranged here from 45–47% of rainfall (285 mm) in the non-mulched plots to 41–43% in the mulched plots, compared to the 49% in the C-plot.
4. Discussion 4.1. Additional considerations In Table 1, one can see the three stages of drying distinguished in older soil science work, where the second stage represented evaporation from a not completely wet nor completely dry soil surface, where patches had already dried up (e.g. Stigter, 1994). The complete ‘‘self mulching’’ of stage 3 normally reduces the evaporation from soil considerably. The first stage, with completely wet soil, gave maximum evaporation. When the soil was drying at and very near the surface, the evaporation reduced till about half the maximum value in the second stage (Table 1). After which it suddenly dropped for all treatments in the third stage, when water vapor had to come from within the soil and an appreciable additional diffusion resistance was quickly added. One may argue that differences in the length and surface conditions of the second drying stage between micro-lysimeters and the soil they are supposed to represent could easily spoil this representation, but nobody has looked into that issue so far. What may
comfort us in this respect is the occurrence of the third drying stage relatively simultaneously for the different treatments, as again the data of Table 1 demonstrate. From our Table 2 data, the influence of the hedges and strips may be observed when comparing H2 and H3 respectively G2 and G3 measurements. These differences appear almost independent of the presence of mulch, that itself exerts the largest influence on soil evaporation. The fact that +M and H3 + M (with lysimeters in the middle between plant rows) hardly differ shows that H3 is unaffected by the hedge. The small to very small differences between G3 M and C confirm this for the strips. Considering also the height and composition of the hedgerows and strips, the differences between H3 M and G3 M are therefore unlikely to be due to differences in shading patterns. They must be due to differences in wetting and drying patterns caused by differences in airflow and overland water flow, due to differences in shrub/ tree, grass and soil interactions with water and wind. This is an additional reason for not considering our measurements as pure replicates but as samples of which the average should be representative for the average of the field as a whole. One alternative way to look at the soil evaporation data is to determine with an ANOVA separation of means the statistical significance of differences in daily average soil evaporation obtained from the succession of measurement periods. This is done in Table 3. These data show a statistical significance of the differences between the Control plot and all mulched plots at the P = 0.05 level. They show the same for the differences between all mulched plots and all non-mulched plots, with the exception of H2 M and G2 M in LR95. Some more of the differences that were physically significant are not always so statistically at the level considered. For example the differences between H2 M and H3 M, respectively H2 + M and H3 + M and those between G2 M and G3 M are always in exactly the same direction, but sometimes statistically significant and sometimes not. This shows the wellknown superiority of a good physical error analysis. 4.2. Discussion and comparison of the results Determination of soil evaporation leads to the separation of the transpiration of plants from the total
J.M. Kinama et al. / Agricultural and Forest Meteorology 130 (2005) 149–162 Table 3 Results of ANOVA statistical analysis—separation of daily means for soil evaporation for different treatments for the long rains of 1994 (LR94), the short rains of 1994/1995 (SR94/95) and the long rains of 1995 (LR 95) LR94
SR94/95
LR95
Treatment
Mean
Treatment
Mean
Treatment
Mean
H2 + M +M H3 + M H2 M G2 M H3 M G3 M C
1.79 1.86 1.87 1.99 2.03 2.06 2.12 2.14
H2 + M H3 + M +M H2 M G2 M H3 M G3 M C
2.50 2.55 2.56 2.67 2.72 2.73 2.76 2.78
H2 + M +M H3 + M H2 M G2 M H3 M G3 M C
1.46 1.53 1.53 1.61 1.62 1.67 1.68 1.73
d d d c bc bc ab a
f e e d c bc ab a
c bc bc b ab a a a
Differences between averages followed by a different letter or letters were statistically significant at P = 0.05. For LR94, F = 23.0; for SR94/95, F = 63.7, for LR95, F = 8.9. With C = control (no hedge or grass strip); +M = mulch (no hedge or grass strip); H2 + M = 1 m from the hedge; H3 + M = 2 m from the hedge; H2 M = 1 m from the hedge; H3 M = 2 m from the hedge; G2 M = 1 m from the grass strip; G3 M = 2 m from the grass strip.
evapotranspiration of a crop. The soil evaporation values assisted in the determination of the water use efficiency of crops (Kinama, 1997). In Kenya, Muchena (1986) noted that soil evaporation was quite high in the dry areas of Eastern Kenya. Daamen et al. (1995) indicated reduced seasonal soil evaporation of 14 2% from sparse millet crops on a sandy soil compared to bare soil. Wallace (1991) estimated soil evaporation losses from millet in semi-arid West Africa as 30–50% of the seasonal rainfall. Cooper et al. (1983) showed that evaporation losses in dryland barley in Syria were as high as 50–60% of total rainfall. Leuning et al. (1994) calculated that soil evaporation under a wheat canopy in Australia was almost 50% of total rainfall. In a temperate humid climate, on bare loamy sand in spring in Denmark, Plauborg (1995) noted soil evaporation as high as 65% and 50% of accumulated potential evapotranspiration in drying periods of 13 and 23 after wetting respectively. In semi-arid Kenya, modeling of evapotranspiration in intercropping of Senna spectabilis with maize or cowpea on flatlands indicated that soil evaporation ranged from 42% to 58% and was not significantly different from that of monocropped maize or cowpea (Mclntyre et al., 1996). The lack of difference was explained by the high proportion of rain occurring during the first
159
weeks of the rainy season, when vegetation cover was small. Moreover, tree prunings were removed from the site and not used for mulching to reduce soil evaporation. Compared to the above our upper limit soil evaporation losses from unmulched maize/senna or grass strip cropping for LR94, between 60% and 65%, were indeed high. The instrumental contribution to overestimation was never higher than 10% of the measured value. Moreover, this effect, due to water extraction outside the lysimeters by the plant/tree roots (Daamen et al., 1993), is smallest early in the season where evaporative demands were highest (Fig. 3). One other reason for somewhat high evaporation losses may have been the assumption of a maximum soil evaporation during rains. However, when we look indeed at the pan evaporation as a measure of evaporative demand, the main reason for the high values in this season is real. They are, as in the case reported above (Mclntyre et al., 1996), due to very high evaporative demands early in the season, in a period with an appreciable contribution to the lowest seasonal rainfall (Fig. 3) and still little soil cover from young plants and recently pruned hedges. This is confirmed by unmulched soil evaporation losses of 45– 50% for SR94/95 and LR95, which were well within ranges reported above for dry areas of the world. Treatment differences in soil evaporation of H2 respectively G2 plots, mainly caused by grass strips, hedges and mulch, were particularly due to shading of the soil and insulation effects where air movement is reduced by the hedges and within and below the mulch. Additionally, wind movement may have a prevailing direction towards hedgerows, which can also lead to asymmetrical shading of the rows near the hedge or grass (Mungai et al., 2000). To assess the influence of crop shade - for the mulched fields in addition to mulch shade, and closest to the hedges in addition to hedge shade—as well as the influence of changing crop shade over the season, would have demanded a much higher sampling density. Now the lysimeters were all located (close to) midway between crop rows (Section 2.4), to make them comparable. It would have needed a sampling pattern across crop rows and within crop rows for each present measuring point to representatively follow influences of crop shade.
160
J.M. Kinama et al. / Agricultural and Forest Meteorology 130 (2005) 149–162
5. Concluding remarks The low mulch rates reduced soil evaporation as percent of seasonal rainfall by absolute amounts of nearly 10% or less in LR94, a maximum of 5% for SR94/95 and in-between those values in LR95. The maximum saving in soil evaporation provided roughly an additional 20–25 mm for crop transpiration, the lowest value for the driest season. In addition to the above, mulch contributed appreciably to the reduction of runoff and soil losses, while it reduced in the H + M plots high cumulative 1993–1995 maize grain yield losses – which are due to hedge/grass strip competition – by 35% compared to the H M and G M plots (Kinama et al., in press). Another analysis, of ‘‘water cost’’ as tree transpiration of contour hedgerows and the ‘‘water cost’’ of producing mulch in order to ‘‘save’’ unproductive water losses as runoff or soil evaporation on sloping lands, indicated that the former ‘‘water cost’’ exceeds the amount saved by three to four times (Ong et al., 1996). Therefore, fodder or fruit trees would be economically more preferable for farmers than trees which just provide a long lasting mulch such as S. siamea. The fact that Kenyan farmers prefer the more competitive grass strips above tree hedgerows is due to advantages of using its biomass, and its quality as a durable stabiliser for erosion control embankment (Kinama et al., in press). However, it must be concluded from the results of our research reported here that for farmers under these semi-arid conditions with low mulch production, the choice remains between low but sustainable yields in hedgerow agroforestry or quickly deteriorating yields on unprotected sloping lands, due to soil erosion. The use of younger agroforestry systems or root pruned ageing systems combined with fertilizers, where possible with mulch brought in from outside the system, will increase the sustainable yields obtainable in this type of hedgerow agroforestry.
Acknowledgements The authors are grateful to the Traditional Techniques of Microclimate Improvement (TTMI) Project, funded by the Directorate General for International Cooperation (DGIS), Ministry of For-
eign Affairs, The Netherlands, at Wageningen University and the University of Nairobi, for providing equipment and co-supervision for J.M. Kinama’s PhD research. We thank the Director, Kenya Agricultural Research Institute (KARI) for his paid study leave and use of research facilities, and the Swedish International Development Agency (SIDA) for supporting field work. Additional support was provided by the International Centre for Research in Agro-Forestry (ICRAF). We thank the TTMI-Project and Mr. A. Khan (ICRAF) for support in development of instrumentation and Mr. John Mailu (ICRAF) for field assistance.
References Allen, S.J., 1990. Measurement and estimation of evaporation from soil under sparse barley crops in Northern Syria. Agric. For. Meteorol. 49, 291–309. Baldy, C., Stigter, C.J., 1997. Agrometeorology of Multiple Cropping in Warm Climates. INRA/Science Publishers/ Oxford & IBH Publishers, Paris/Enfield (USA)/New Delhi, 237 pp. Barber, R.G., Thomas, D.B., Moore, T.R., 1979. The erodibility of two soils from Kenya. J. Soil Sci. 30, 579–591. Belsky, A.J., Mwonga, S.M., Amundson, R.G., Duxbury, J.M., Ali, A., 1993. Comparative effects of isolated trees on their understorey environment in high and low rainfall savannas. J. Appl. Ecol. 30, 143–155. Bley, J., Van der Ploeg, R.R., Sivakumar, M.V.K., Allison, B.E., 1991. A risk-probability map for millet production in south–west Niger. In: Sivakumar, M.V.K., Wallace, J.S., Renard, C., Giroux, C. (Eds.), Soil Water Balance in the Sudano-Sahelian Zone. IAHS Publ. No. 199, IAHS Press, Wallingford, pp. 571–581. Boast, C.W., Robertson, T.M., 1982. A ‘‘microlysimeter’’ method for determining evaporation from bare soil: description and laboratory evaluation. Soil Sci. Soc. Am. J. 46, 469–696. Boffa, J.M., 1999. Agroforestry Parklands in Sub-Saharan Africa. FAO Conservation Guide 34. FAO, Rome. Cannell, M.G.R., Van Noordwijk, M., Ong, C.K., 1996. The central agroforestry hypothesis: the trees must acquire resources that the crop would not otherwise acquire. Agrofor. Syst. 34, 27–31. Cooper, P.J., Keatinge, J.D.H., Hughes, G., 1983. Crop evapotranspiration—a technique for calculating its components by field measurements. Field Crops Res. 7, 299–312. Daamen, C.C., Simmonds, L.P., Wallace, J.S., Laryes, K.B., Sivakumar, M.V.K., 1993. Use of micro-lysimeters to measure evaporation from sandy soils. Agric. For. Meteorol. 65, 159– 173. Daamen, C.C., Simmonds, L.P., Sivakumar, M.V.K., 1995. The impact of sparse millet crops on evaporation from soil in semi-arid Niger. Agric. Water Manage. 27, 225–242.
J.M. Kinama et al. / Agricultural and Forest Meteorology 130 (2005) 149–162 FAO/UNESCO, 1988. Soil Map of the World 1: 5,000,000. Revised Legends. FAO, Rome. Jackson, N.A., 2000. Measured and modelled rainfall interception loss from an agroforestry system in Kenya. Agric. For. Meteorol. 100, 323–336. Jackson, N.A., Wallace, J.S., 1999. Soil evaporation measurements in an agroforestry system in Kenya. Agric. For. Meteorol. 94, 203–216. Johnsson, K., 1995. Agroforestry in dry savanna areas in Africa: interactions between trees, soils and crops. PhD Thesis. Swedish University of Agricultural Sciences, Umea, Sweden. Kibe, J.M., Ochung, H., Macharia, P.N., 1981. Soils and vegetation of the ICRAF experimental farm (Machakos District). Ministry of Agriculture, National Agricultural Laboratories, Kenya Soil Survey, Detailed Soil Survey Report No. D 23. Kiepe, P., 1995. No Runoff, No Soil Loss: Soil and Water Conservation in Hedgerow Barrier Systems, Tropical Resource Management Papers 10. Wageningen Agricultural University, The Netherlands, 156 pp. Kilewe, A.M., Ulsaker, L.G., 1984. Soil physical characteristics and their application to agriculture. E. Afr. Agric. For. J. 44, 247– 255. Kinama, J.M., 1997. The effects of hedgerows (alley cropping) on microclimate, soil and water conservation and competition for sustainable land use on the sloping areas of Machakos, Kenya. PhD Thesis. University of Nairobi, Kenya. Kinama, J.M., Ng’ang’a, J.K., Ong, C.K., Stigter, C.J., Gichuki, F.N., 1999. Quantification of soil evaporation in cowpea/maize or grass strip systems in semi-arid eastern Kenya. Paper Presented at the National Soil and Water Management Conference, KARI, Nairobi, Kenya. Kinama, J.M., Stigter, C.J., Ong, C.K., Ng’ang’a, J.K., Gichuki, F.N., in press. Contour hedgerows and grass strips in erosion & runoff control in semi-arid Kenya, and their competitive effects. Resubmitted after revision to Agric. Ecosyst. Environ. Kinyamario, J.I., Trlica, M.J., Njoka, T.J., 1995. Influence of tree shade on plant water status, gas exchange, and water use efficiency on Panicum maximum Jacq. and Themeda triandra Forsk in a Kenyan Savanna. Afric. J. Ecol. 33, 114–123. Leuning, R., Condon, A.G., Dunin, F.X., Zegelin, S., Denmead, O.T., 1994. Rainfall interception and evaporation from soil below a wheat canopy. Agric. For. Meteorol. 67, 221–238. Lott, J.E., 1998. Resource capture and use in semi-arid overstorey agroforestry systems. PhD Thesis. University of Nottingham, UK. Marimi, P.M., 1979. The effects of some tillage methods and cropping systems in conserving rainfall in a semi-arid area of Eastern Kenya. In: Presented at a Soil and Water Conservation Workshop. Kabete. University of Nairobi, Nairobi, Kenya. Mbuvi, P.J., Van de Weg, R.F., 1975. Some preliminary notes on the soils of Katumani, Kampi-ya-mawe, Embu and Murinduko Agricultural Research Stations. Report No. P25, Kenya Soil Survey, Nairobi, Kenya. Mclntyre, B.D., Riha, S.J., Ong, C.K., 1996. Light interception and evapotranspiration in hedgerow agroforestry systems. Agric. For. Meteorol. 81, 31–40.
161
Muchena, F.N., 1986. Soil fertility constraints in improving cereal yields in soils of the arid and semi-arid areas of Kenya. In: Paper Presented at Organisation of African Unity (OAU) Drought Symposium, Kenyatta International Conference Centre, OAU, Nairobi, Kenya. Mugendi, D.N., Mochoge, B.O., Coulson, C.L., Stigter, C.J., Arap Sang, F.K., 1994. Decomposition of Cassia siamea loppings in semi-arid Machakos, Kenya. Arid Soil Res. Rehabil. 8, 363–372. Mungai, D.N., 1995. A micrometeorological approach to understanding maize yield performance in alley cropping in the semi-arid areas of Machakos District, Kenya. In: Stigter, C.J., Wangati, F.J., Ng’ang’a, J.K., Mungai, D.N. (Eds.), The TTMI-Project and the ‘‘Picnic’’ Model. An Internal Evaluation of Approaches and Results and of Prospects for TTMI-Units. Wageningen Agricultural University, The Netherlands, pp. 111–123. Mungai, D.N., Stigter, C.J., Coulson, C.L., Ng’ang’a, J.K., 2000. Simply obtained global radiation, soil temperature and soil moisture in alley cropping systems in semi-arid Kenya. Theor. Appl. Climatol. 65, 63–78. Mungai, D.N., Stigter, C.J., Coulson, C.L., Ng’ang’a, J.K., Netondo, G.W.S., Umaya, G.O., 2001. Understanding yields in alley cropping maize (Zea mays L) and Cassia siamea (Lam) under semi-arid conditions in Machakos, Eastern Kenya. J. Environ. Sci. 13, 291–298. Olufayo, A.A., Stigter, C.J., Baldy, C., 1998. On needs and deeds in agrometeorology in tropical Africa. Agric. For. Meteorol. 92, 227–240. Ong, C.K., Black, C.R., 1992. Complementarity in resource use in intercropping and agroforestry systems. In: Paper Presented at the 52nd International Easter School on Resource Capture by Crops, School of Agriculture, Sutton Bonington, UK. Ong, C.K., Black, C.R., Marshall, F.M., Corlett, J.E., 1996. Principles of resource capture and utilisation of light and water. In: Ong, C.K., Huxley, P. (Eds.), Tree–Crop Interactions: A Physiological Approach. CAB International, Wallingford, UK, pp. 73–185. Plauborg, P., 1995. Evaporation from bare soil in a temperate humid climate. Measurements using micro-lysimeters and time domain reflectometry. Agric. For. Meteorol. 76, 1–17. Rao, M.R., Westley, S.B., 1989. Agroforestry for the African semiarid zone. Agrofor. Today 1, 5–11. Sanchez, P.A., 1995. Science in agroforestry. Agrofor. Syst. 30, 5– 55. Soil Survey Staff, 1990. Keys to Soil Taxonomy, 4th ed. S.M.S.S. Technical Monograph 19. Virginia Polytechnic and State University, Blacksburg, VA, 422 pp. Stigter, C.J., 1994. Micrometeorology. In: Griffiths, J.F. (Ed.), Handbook of Agricultural Meteorology. Oxford University Press, (Chapter 5), pp. 59–72. Van Noordwjik, M., Ong, C.K., 1999. Can the ecosystem mimic hypothesis be applied to farms in African savannahs? Agrofor. Syst. 45, 131–158. Wallace, J.S., 1991. The measurement and modelling of evaporation from semi-arid land. In: Sivakumar, M.V.K., Wallace,
162
J.M. Kinama et al. / Agricultural and Forest Meteorology 130 (2005) 149–162
J.S., Renard, C., Giroux, C. (Eds.), Soil Water Balance in the Sudano-Sahelian Zone. IAHS Publ. No. 199, IAHS Press, Wallingford, pp. 131–145. Wallace, J.S., Gash, J.H.C., McNeil, D.D., Sivakumar, M.V.K., 1988. Evaporation from sparse dryland millet crops in Niger West Africa. In: Unger, P.W., Sneed, T.V., Jordan, W.R., Jensen,
R. (Eds.), Proceedings of the International Conference on Dryland Farming, Texas Agricultural Experiment Station, Armavillo, Bushland, TX, pp. 325–327. Wallace, J.S., Jackson, N.A., Ong, C.K., 1999. Modelling soil evaporation in an agroforestry system in Kenya. Agric. For. Meteorol. 94, 189–203.