Eucalyptus–wheat interaction on Ethiopian Nitosols

AGRICULTURAL SYSTEMS Agricultural Systems 80 (2004) 151–170 www.elsevier.com/locate/agsy Eucalyptus–wheat interaction on Ethiopian Nitosols Selamyihun Kidanu a, Tekalign Mamo b, Leo Stroosnijder c,* a c Debre Zeit Agricultural Research Centre, P.O. Box 32, Debre Zeit, Ethiopia b Winrock International Ethiopia, P.O. Box 2417, Addis Ababa, Ethiopia Department of Environmental Sciences, Chair Group Erosion and Soil & Water Conservation, Wageningen University, Nieuwe Kanaal 11, 6709 PA Wageningen, The Netherlands Received 22 October 2002; received in revised form 2 July 2003; accepted 2 July 2003 Abstract Over the past few years a single row of Eucalyptus globulus trees planted along the borders of cropland has come to dominate central highland agroforestry practices. Although evidence is scanty, there is a perception that this practice adversely affects crop productivity. An onfarm trial was therefore conducted at Ginchi to determine the biomass production potential of eucalypt boundaries and their effect on the productivity of the adjacent wheat crop (Triticum aestivum) on highland Nitosols. Three rotation cycles of 4 years each, two stand ages within each rotation, four field aspects and six yield strata perpendicular to the tree–crop interface were arranged in a split–split plot design with three replications. The annual wood production rate, which was 345–903 kg ha
1 yr
1 with two- and four-year-old stands in the first cycle, was increased more than 2-fold in the subsequent two rotation cycles. With these productivities, eucalypt boundaries on a hectare of land in the second and third cycle would satisfy about 70% of the annual biomass energy requirement of a rural household with a family size of five people for four consecutive years. However, adjacent wheat yields were substantially reduced because of the combined effects of water, light and nutrient competition. In the last two rotation cycles, significant yield depressions occurred over the first 16 m from the line of trees as opposed to only the first 8 m in the first cycle. The yield drop was 4.5– 8.1%, and 8.1–10.4% in the first and last two rotation cycles, respectively. Nevertheless, the benefit accrued from the tree component adequately compensated for this reduction in wheat yield and generated additional income. The implications of these results are discussed in the * Corresponding author. Tel.: +31-317-482446; fax: +31-317-486103/484759. E-mail address: leo.stroosnijder@wur.nl (L. Stroosnijder). 0308-521X/$ - see front matter Ó 2003 Elsevier Ltd. All rights reserved. doi:10.1016/j.agsy.2003.07.001 152 S. Kidanu et al. = Agricultural Systems 80 (2004) 151–170 context of the suitability of the practice in the region and its role in relieving the increasing pressure on indigenous forest and woodland. Ó 2003 Elsevier Ltd. All rights reserved. Keywords: Eucalyptus globulus; Agroforestry; Tree–crop interface; Wheat 1. Introduction Eucalyptus trees are widely grown in the East African highlands as exotics to meet the ever-increasing demand for fuelwood and other wood products. In Ethiopia alone, Eucalyptus globulus plantations cover more than 100,000 ha (Pohjonen and Pukkala, 1990). Under most of the conditions prevailing in the Ethiopian highlands, eucalypts convert energy and available water into biomass more efficiently compared with exotic coniferous tree species (Pohjonen and Pukkala, 1990). However, to satisfy the biomass energy demand of the country, 6% of the total utilizable land area would have to be under tree plantations by 2014; this would entail a major shift in land use (Bojo and Cassells, 1995). Hence, introducing short-maturing multipleproduct tree species that can be combined with annual crops may be of considerable importance in addressing the current biomass energy crisis in Ethiopia. Traditional agroforestry practices in Ethiopia involve planting trees in various spatial patterns to meet the wood, fuel and fodder requirements of the farmers. In recent years, however, a single row of E. globulus trees planted along the borders of croplands has come to dominate the central highland landscape. Eucalyptus globulus trees, unpalatable to cattle, sheep and goats (Pohjonen and Pukkala, 1990), have a distinct advantage as a boundary planting in the Ethiopian highlands where the protection of privately-planted trees on farmland is difficult because of dry season free grazing practices. Trees are usually planted 1 m apart in line and are aligned in an east–west or north–south direction. The former orientation (80%) is predominant on bottomlands. In this environment, eucalypt boundaries produce harvestable trees within 4–5 years of planting. Although evidence is scanty, there is a perception that this practice adversely affects crop productivity. In a sub-humid, subtropical climate, Khybri et al. (1992) recorded a 41–61% wheat yield reduction in a unilateral open alley system with 100 trees ha
1 of unpruned Eucalyptus hybrid. Similarly, with Eucalyptus in a semi-arid region, Malik and Sharma (1990) reported a wheat yield reduction of 47% in the most active zone of competition. It is inadvisable to extrapolate such data to Ethiopian highland conditions, because the interaction effects between the tree and agricultural crops are affected by the cropsÕ spatial configuration (Ong et al., 1996) and are highly influenced by growth environment and management systems (Verinumbe, 1987; Sanginga and Swift, 1992; Onyewotu et al., 1994). There is, therefore, a need to determine the extent to which crop yields are influenced by eucalypt boundaries in the current production environment of Ethiopia. This paper presents the results of a study aimed at evaluating the biomass produc- S. Kidanu et al. = Agricultural Systems 80 (2004) 151–170 153 tion potential of eucalypt boundaries and their competitive effects on wheat (Triticum aestivum), as a function of rotation, stand age and aspect, under farmersÕ production circumstances on highland Nitosols in Ethiopia. 2. Materials and methods 2.1. The study area The study was conducted in the Ginchi watershed in the central highlands of Ethiopia (2200 m a.s.l., 38°E and 9°N), 90 km southwest of Addis Ababa. As is the case with most of the Ethiopian highlands, population pressure has substantially changed the land use/cover. In the last 40 years, the area under annual crops has increased from 34% to 61%, while areas under pasture and woodland have shrunk by more than half (JVP, 1995). The watershed has a sub-humid climate with an average annual rainfall of 1200 mm, 30% of which falls outside the main cropping season (Fig. 1) and under the traditional small cereal-based cropping system, 40–50% of the seasonal rainfall is lost as runoff with considerable risk of soil erosion (Tekelu et al., 1999). For these reasons, effective rainfall for crop production is therefore appreciably lower than the annual amounts. Pellic Vertisols and Nitosols, in that order, are the two most important soils in the watershed. On both soils, the cropping system is characterized by a three-year cropping sequence: a grain legume followed by 2 years of small-grain cereal production. The average farm size per household is about 2.5 ha, 0.5 ha of which is used to carry the livestock during the wet season (June–September) or to produce hay to 300 Rainfall (mm) 250 200 150 100 50 0 J F M A M J J A S O N D Months Fig. 1. Rainfall distribution during the 2000/2001 growing season compared with long-term (1995–2001) mean rainfall data for Ginchi, Ethiopia. 154 S. Kidanu et al. = Agricultural Systems 80 (2004) 151–170 supplement the dry season feed. Fallow land accounts for less than 1% of the total land holdings on both soil types. Livestock is herded on communal pasture, private grazing lands or kept in sheds, depending on the season. During the dry season (February–early June) animals are traditionally allowed to roam freely on private land. Within this system, animal manure is dried for use as cooking fuel, thereby precluding its use as organic fertilizer. 2.2. Experimental design and analysis On the Nitosols, which dominate the upper part of Ginchi watershed, eucalypt boundaries at three different rotation cycles (first, second and third) were identified based on stand uniformity. ÔSecondÕ means second generation, i.e., trees that have grown after coppicing the originally planted trees. ÔThirdÕ means grown after a second coppicing. Within each rotation cycle, two stand ages (2 and 4 years) were selected from line of trees oriented either north–south or east–west. North–southoriented lines of trees divide the farmlands into east–west aspects while the east– west-oriented lines of trees divide the farmlands into north–south aspects. The test crop was ET-13, a commonly grown bread wheat variety. Wheat plots were fertilized with 40 kg N ha
1 and 10 kg P ha
1 at time of planting. According to the National Fertilizer and Input Unit (1991), these rates are considered optimal for wheat production on Ethiopian highland Nitosols. The wheat crop followed faba beans. Crop yield data were taken from 2–4, 4–8, 8–12, 12–16, 16–20 and 20–30 m strips which represent uniform yield strata (Rao and Coe, 1991) perpendicular to the tree–crop interface. At crop maturity, 48 m2 were harvested from each strip. After threshing, the grain yield was measured and adjusted to 12% moisture content. Root distribution was studied by excavating roots down to 1 m depth by digging profile pits 1.5 m wide and 1.5 m deep. The pits were opened at both sides of eightyear-old E. globulus stands (second cycle at an age of 4 years) aligned in the east– west direction at distances of 5, 10, 15 and 20 m from the tree line. Wood production potential of E. globulus boundary plantings was estimated from stand height and diameter (at breast height) measurements using the fresh mass equation of Pukkala and Pohjonen (1989), which was then multiplied by a factor of 0.52 to obtain the dry mass of the stem as suggested by Pukkala and Pohjonen (1989). In fuel wood production the mass of branches and leaves are also important characteristics. Thus, 10% of the stem dry mass was added to account for the branches and leaves in calculating the total dry matter production per tree. At intervals of 15 days throughout the growing period, soil moisture dynamics across the tree–wheat interface were monitored gravimetrically at 10 cm depth intervals over the top 1 m rooting depth. These measurements were made at both sides of the tree line at distances of 0, 5, 10, 15, 20 and 30 m from the tree line in three replications. The stands were four-year-old in their second rotation cycle and aligned in the east–west direction, as for the root distribution quantification. Available soil water for plant use was calculated as the difference between the amount of water in the top 1 m soil and that retained at permanent wilting point (1.5 MPa). The soil moisture at 1.5 MP was determined by the pressure plate tech- S. Kidanu et al. = Agricultural Systems 80 (2004) 151–170 155 nique on five core samples (100 mm in diameter and 79-mm long) taken from each point where gravimetric soil moisture measurements were made using thin walled steel rings. Bulk density values were determined simultaneously on the same core samples using the core method (Peter, 1965). Crop yield data were analyzed using the MSTAT-C statistical package (MSTATC, 1991). Data were analyzed using a model for four-factor plus split–split plot design. Rotation cycle constituted the main factor, the sub-factor was stand age, and factorial combinations of field aspect and distance representing the uniform yield strata were assigned to sub–sub-plots. Multiple range subprograms within the MSTAT-C package were used to differentiate treatment effects and interaction means. 3. Results and discussion 3.1. The effect of eucalypt boundaries on wheat yield All treatment variables and their first- and second-degree interactions significantly influenced crop yield across the tree–crop interface (Table 1). Wheat yields across the tree–crop interface were significantly depressed over a considerable distance in comparison with wheat yields in the open field (Table 3). Nevertheless, the tree influence was not large enough to induce significant yield differences across the tree–crop interface beyond a distance of 8 and 16 m from the tree line in the first and the last two Table 1 Analysis of variance for grain yield of wheat planted adjacent to eucalypt boundaries on Nitosols at Ginchi, Ethiopia Source of variation Degree of freedom Mean square values Rotation cycle (R) Residual Stand age within the rotation (S) RS Residuals Field aspect (A) RA SA RSA Distance from the tree line (D) RD SD RSD AD RAD SAD RSAD Residuals 2 4 1 2 6 1 2 1 2 5 10 5 10 5 10 5 10 132 22,606,228 4666 348,566 187,923 7403 919,503 4342 789 12,772 8,733,790 252,199 84,316 45,325 94,865 48,396 4172 4272 5800 * Statistically significant at P 6 0:001. 156 Table 2 Grain yield (kg ha
1 ) of wheat as a function of rotation cycle (R), stand ages (S), field aspect (A) and distance (D) from the eucalypt boundaries on Nitosols at Ginchi, Ethiopia Distance (m) Rotation cycle and field aspect First rotation y y Interaction effects Second rotation Third rotation RAD NS EW NS EW NS EW NS EW AD SD 2 02–04 04–08 08–12 12–16 16–20 20–30 1107c 1307b 1547a 1513a 1561a 1529a 666e 1157c 1521a 1524a 1568a 1567a 300i 583g 1006d 1370b 1563a 1531a 158j 350i 746f 1150c 1486a 1527a 260i 477f 1147c 1394b 1544a 1581a 158j 354gh 707e 1134c 1555a 1574a 556g 789e 1233c 1426b 1556a 1547a 327h 620f 991d 1269b 1536a 1556a 495g 717f 1117d 1415b 1558a 1551a 441g 705e 1113c 1348b 1547a 1552a 4 02–04 04–08 08–12 12–16 16–20 20–30 752e 941d 1184c 1515a 1588a 1544a 355g 553f 1007c 1521a 1560a 1570a 282i 484h 833e 1383b 1545a 1566a 174j 298i 655g 1104c 1520a 1553a 267hi 511f 937d 1317b 1549a 1556a 208ij 365g 669e 1162c 1545a 1606a 434i 645f 1001d 1405c 1561a 1564a 264j 605h 810e 1262c 1541a 1576a 286h 513g 901e 1266c 1533a 1566a 340h 526f 906d 132b 1545a 1566a 2 1341A 4 1184B 2 981B 4 954B 2 991B 4 975B Stand age Rotation cycle * 1281A 964B 982B Values within the same rotation cycle followed by the same letter are not significantly different (P 6 0:05). Values within the same row followed by the same letter are not significantly different (P 6 0:05). *** Mean values within the same factor interactions followed by the same letter are not significant (P 6 0:05). NS and EW represent the pooled yield data on north and south, and east and west aspects, respectively. ** S. Kidanu et al. = Agricultural Systems 80 (2004) 151–170 Stand age (years) Distance (m) Rotation cycle and stand age (years) First rotation Second rotation 2 4 a 02–04 04–08 08–12 12–16 16–20 20–30 Loss (ha
1 ) a a 2 Third rotation 4 2 4 NS EW NS EW NS EW NS EW NS EW NS EW 29 16 0 0 0 0 57 27 2 0 0 0 51 40 24 2 0 0 77 64 35 2 0 0 80 63 35 12 0 0 89 77 52 26 4 0 82 69 46 11 0 0 89 80 57 29 2 0 83 69 26 10 0 0 89 77 54 27 0 0 82 76 39 15 0 0 86 76 57 25 0 0 10.3 10.0 10.4 4.7 4.5 8.1 7.7 9.7 10.0 8.8 0 NS and EW represent the pooled yield data on north–south and east–west aspects, respectively. 8.1 S. Kidanu et al. = Agricultural Systems 80 (2004) 151–170 Table 3 Grain yield reduction (%) as a function of rotation cycle, stand age, field aspect and distance from eucalypt boundaries on Nitosols at Ginchi, Ethiopia 157 158 S. Kidanu et al. = Agricultural Systems 80 (2004) 151–170 rotation cycles, respectively (Table 2). The mean crop yield obtained with the second and third rotations (964 vs. 982 kg ha
1 ) were of the same magnitude as, and significantly lower than the crop yield (1292 kg ha
1 ) obtained in the first rotation (Table 2). In Fig. 3, crop yield data as measured along the six uniform yield strata expressed as percent of yield in the open field (1556 kg ha
1 ), were plotted against distance from the tree line. Because in the last two rotations, the effects of stand age and rotation were not significant, the yield in the respective strata was averaged over rotation and stand age. In this way, each data point came from the mean of 24 independent measurements. In the first rotation, as data for the two stands were plotted separately every point represents a mean of 12 independent yield measurements. The relationship between crop yield and distance from the tree line was curvilinear in the first rotation cycle and linear in the second and third rotation cycles. In both cases, the functions explained about 98% of the yield variability across the tree–crop interface. 3.2. Biomass production of eucalypt boundaries Eucalypt boundaries in the same rotation cycle were relatively uniform in terms of tree height, diameter and inter-row spacing (Table 4). The annual wood production rates of two-year-old stands were 345, 923 and 765 kg ha
1 yr
1 in the first, second and third rotation cycles, respectively. At the age of four years, the annual wood production rate, which was only 903 kg ha
1 yr
1 in the first cycle, increased by 2-fold in the second and third rotations (Table 4). This high wood production rate in the second and third rotations was attributed to increased stand density rather than increased growth rate of individual trees (Table 4). In the second and third rotation cycles, farmers tend to maintain as many sprouting stems as possible per stump until the age of 2 years. Thereafter, they thin these down to two to three stems per stump. The number of these sprouting stems ranged from 4 to 8 with a mean of 5 and 4 in the second and third rotation cycles, respectively (Table 4). Although the intention of the thinning practice is to reduce competition between actively growing young shoots, it is also a good source of fuelwood for the farmers who grow eucalypt boundaries on their farmland. The stems remaining after thinning are allowed to grow for two more years before they are ready for the final harvest. Thus the stem density harvested at the end of the third and second rotation cycles, respectively, was two and three times higher than the stand density harvested in the first rotation cycle. 3.3. Root distribution of eucalypt boundaries After 8 years, fine roots (<10 mm diameter) accounting for more than 80% of the total root mass per unit area mostly extend less than 20 m into the adjacent crop area (Table 5). The greatest proportion of eucalyptus roots in cropped areas were in the 0–60 cm horizon and consisted mainly of root fibres with a diameter in the 1–5 mm class, rootlets in the 5–10 mm class, and a few roots in the 10–20 mm classes. Total Rotation cycle Stand age (years) No. farms sampled Height (m) Diameter (cm) No. stems per stump Stem volume (dm3 ) First 2 4 12 (10)a 12 (10) 2.90  0.08 6.80  0.06 5.0  0.27 12.5  0.27 – – 11 60 345 903 Second 2 4 12 (20) 12 (20) 3.92  0.02 7.9  0.02 4.8  0.16 13.0  0.29 5.0 2.0 8 58 923 1760 Third 2 4 12 (20) 12 (20) 3.09  0.01 6.98  0.01 4.2  0.12 11.0  0.15 4.0 2.0 6 52 765 1841 a The number in brackets is the number of trees sampled per farm. The number behind the  sign is the standard error. Wood production (kg ha
1 yr
1 ) S. Kidanu et al. = Agricultural Systems 80 (2004) 151–170 Table 4 Mean tree height, diameter at breast height, number of stems per stump and wood production of E. globulus boundary plantings at different stand ages and rotation cycles on Nitosols at Ginchi, Ethiopia 159 160 Depth (cm) Distance from the line of trees 5m 0–10 10–20 20–40 40–60 60–100 0–100 All diameters 10 m 15 m 20 m a b c d a b c d a b c d a b c d 0 0 0 7 3 10 0 0 1 21 62 85 0 15 37 92 3 147 32 62 105 61 0 260 0 0 0 3 13 16 0 0 4 3 2 9 0 25 17 12 0 54 38 33 0 21 0 90 0 0 0 0 3 3 0 0 1 1 2 4 0 5 32 9 3 49 22 12 68 0 0 92 0 0 0 0 0 0 0 0 0 0 0 0 0 5 0 0 0 5 0 2 5 0 0 7 502 169 148 12 S. Kidanu et al. = Agricultural Systems 80 (2004) 151–170 Table 5 Root number (m
2 ) in relation to distance from the line of E. globulus and soil depth for four diameter classes: a ¼> 20 mm; b ¼ 10–20 mm; c ¼ 5–10 mm; d ¼ 1–5 mm on Nitosols at Ginchi, Ethiopia S. Kidanu et al. = Agricultural Systems 80 (2004) 151–170 161 root number at a distance of 10, 15 and 20 m declined by 66%, 70% and 98%, respectively, compared with the total root number at distance of 5 m form the tree line. Nevertheless, the proportion of roots in the 1–5 and 5–10-mm diameter class did not change markedly with distance from the tree line. The fine roots in the 1–5 and 5–10 mm diameter class accounted for 50–63% and 29–33% of the total root number at each sampling point, respectively. In contrast, the proportion of roots in the 10–20 mm diameter class, which was 17% at a distance of 5 m, accounted for only 5% and 2% of the total root mass at 10 and 15 m distance, respectively. 3.4. Soil moisture use The rainfall in the 2000/2001 cropping season was not markedly different from the long-term rainfall pattern (Fig. 1). The seasonal soil moisture balance showed that Eucalyptus roots within the tree rows are capable of removing soil water close to the 1.5 MPa level. Nevertheless, in spite of the extensive lateral root spread, throughout the growing period this moisture extraction pattern did not in many cases extend as far as 10 m and in no case as far as 15 m, into the adjacent crop (Fig. 2). The soil moisture beyond a distance of 5 m was two to three times higher than the soil moisture content encountered within the tree rows. As a result, substantial quantities of soil moisture along the tree–crop interface remained in the upper 1 m of the soil when wheat was harvested in October. 3.5. Economic impact of boundary plantings -1 Soil moisture (mm m ) Economic impacts of eucalyptus boundary planting for smallholder farming can be viewed from various angles. It is, however, a short-term outcome that is of primary importance to resource-poor farmers who can tolerate only limited risk and rely heavily on crops for survival. Thus our analysis was based on the effects on 120 80 40 15-Jul 30-Aug 15-Oct 30-Jul 15-Sep 30-Oct 15-Aug 30-Sep 0 0 5 10 15 20 25 30 Fig. 2. Available soil moisture in the top 1 m of wheat fields as a function of the distance from eucalyptus in Ginchi, Ethiopia. S. Kidanu et al. = Agricultural Systems 80 (2004) 151–170 162 Grain yield (%) 100 80 60 40 NS 20 EW 0 4 8 12 16 20 Distance from the tree line (m) 24 Second and third cycle Grain yield (%) 100 80 60 40 NS 20 EW 0 4 8 12 16 20 24 Fig. 3. Wheat yield across the tree–crop interface expressed as a percentage of wheat yield in the open field on highland Nitosols at Ginchi, Ethiopia. (NS ¼ north–south facing plots, EW ¼ east–west facing plots.) the crop component as compared with production from a sole wheat cropping system. In the highland Nitosols environment, the benefits from eucalypt boundaries and trade-offs in crop yield start at the beginning of the fourth year onwards. Records of farms in the study revealed that the income from the sole wheat crop production system (grain + straw) averaged US$ 300 ha
1 yr
1 in contrast to US$ 283, 273 and 270 ha
1 yr
1 from wheat in the agroforestry system in the first, second and third rotation cycles, respectively. When the entire above-ground biomass is harvested as farmers normally do in the highland Nitosols, 18 kg nitrogen, 6 kg phosphorus and 12 kg potassium in the first cycle and 37 kg nitrogen, 13 kg phosphorus and 27 kg potassium in each subsequent rotation cycle were removed preferentially across the tree–crop interface. If these nutrients are replenished from inorganic fertilizer sources, they cost about US$ 1.5 and 2.5 ha
1 yr
1 in the first and in the last two rotation cycles, respectively. S. Kidanu et al. = Agricultural Systems 80 (2004) 151–170 -1 -1 Cost incured and Tree sales ($US ha y ) Wheat 163 Eucalyptus 50 40 30 20 10 0 1 2 3 Fig. 4. Cost incurred due to crop yield loss and soil nutrient extraction versus additional income generated through the sale of wood and wood products from eucalypt boundaries at Ginchi, Ethiopia. On the other hand, eucalyptus wood and wood products can make a significant contribution to the economics of the system and to the diversity of farm production. In cases where fuel is scarce, such as in Ginchi, the tree has value as fuel. Traditionally, farmers distinguish between the various parts of the tree because these products have different values. In this study, the price used in estimating the returns was derived from the average value of a tree by aggregating the price prevailing in the market for different tree ages. With this assumption, the income from the tree component was estimated at US$ 19, 46 and 52 ha
1 yr
1 in first, the second and third rotation cycles, respectively. Thus across all rotation cycles the additional income generated from the tree component of the agroforestry system was higher than the total cost incurred because of crop yield loss and the nutrients used to produce wood and wood products (Fig. 4). Along with a slight increase in crop yield loss and nutrients fixed in eucalypt biomass, the net benefits from the trees increased more than 10-fold in the second and third rotation cycles, compared with what it was in the first cycle. 4. Discussion of results In the present study, the greatest proportion of eucalypt fine roots occurred in the 0–60 cm horizon. Their symmetrical distribution with respect to the mid-point of the soil profile facilitates the ability of Eucalyptus roots to take up water and nutrients. This is an important aspect of the detrimental tree–crop interaction of eucalypt boundaries, where the roots of both components are distributed within the same soil volume and hence compete for the same resources. In contrast, roots in the 10–20 and >20 mm diameter classes were few in number and their distribution was skewed to the bottom horizons. This is probably because of the low plasticity of thicker 164 S. Kidanu et al. = Agricultural Systems 80 (2004) 151–170 roots to respond to favourable soil volumes, as pointed out by Eissentat (1992), or in response to repeated cultivation at shallow depth. Fast-growing tree species such as E. globulus deplete soil nutrient reserves particularly when grown in short rotations. However, if the leaf fraction was re turned back to the system, the differential nutrient depletion rate across the tree–crop interface can be reduced considerably. With few exceptions the leaf fraction of eucalyptus trees, although it accounts for less than 5–10% of the total biomass, contains more than 50% of the biomass store of nutrients (Holgen and Svensson, 1990). In addition, in the Ethiopian highlands, fuelwood production would displace huge amounts of dung fuel in favour of its higher value as fertilizer and part of this could be applied preferentially across the tree–crop interface to complement the nutrient recycling through leaf additions. When young, eucalyptus trees are not as competitive as even the most common fast-growing multipurpose tree species (Jama et al., 1998). In the long term, the main concerns are, therefore, competition for water and possible allelopathic effects (Lisanework and Michelsen, 1993), particularly when mulches are used for soil improvements. As allelopathic effects are transient process and only limited to freshly decomposing leaf materials (Reversat, 1999), application of well-decomposed eucalypt leaves rather than fresh leaf mulch may be necessary for safe nutrient recycling. In addition, well-decomposed leaf material can be applied at the time of crop planting in contrast to fresh leaf mulch, which requires a sufficiently long period of time with a favourable environment for decomposition between application and crop planting. Because of a high C:N ratio, eucalyptus leaf mulches decompose slowly and nutrient immobilization phase are known to persist for several months after their application (Sanginga and Swift, 1992). It seems probable that although the trees supplemented deep subsoil water with water from nearer the soil surface, most of the transpiration by eucalypt boundaries concerns water drawn from below the cropÕs rooting zone. Hence competition for water occurs close to the tree line, particularly during early crop establishment stages. However, later in the season when the crops are at sensitive growth stages, stored soil moisture can buffer the short dry spells that can occur between two rainfall events. The soil moisture gradient that one would expect to occur at a distance of 10 m and beyond was not detected, because of the confounding effect of periodic soil moisture recharge (by rain) throughout the growing season. Thus beyond this distance, competition for water between eucalypt boundaries and the wheat crop may have little impact on crop yield in years with normal seasonal rainfall distribution. However, in drought years, the area of tree–crop water competition across the tree–crop interface could potentially extend over the entire distance explored by the fine roots of eucalypt boundaries (15 m). The probability of drought in the study area is estimated as two years out of 10. The wood production rate of boundary plantings is high compared with the maximum production rate reported by Pohjonen and Pukkala (1989) for 4-year-old stands of E. globulus woodlot plantations in the highlands. Elsewhere, crops grown in tree plantations under different soil and climatic conditions have been reported to have a beneficial effect on tree growth (Dhyani and Tripathi, 1999). In this particular S. Kidanu et al. = Agricultural Systems 80 (2004) 151–170 165 study, the high wood production rates of eucalypt boundaries can be attributed to site differences, low site competition for growth resources (light, nutrients and water) and good management, particularly at early seedling establishment stage, when tree growth is more sensitive to weed competition (FAO, 1985). In addition, through their lateral roots, eucalypt boundaries may have access to plant nutrients applied to associated crops. Contrary to the practice in woodlot plantations, farmers who grow eucalypt boundaries on their farmland commonly apply 2–3 kg farmyard manure per pit at the time of planting eucalypt boundaries. Towards the end of the rainy season they also mulch the ground under individual trees with trash and crop residues after a shallow cultivation. Most farmers believe that mulch is very effective in conserving soil moisture, while the manure and in situ decomposed plant residues are seen by farmers as an important source of nutrients for the young Eucalyptus trees. Significant interaction between rotation cycle and stand ages (Table 1) indicated that the competitive ability of eucalyptus boundaries, as measured in terms of crop yield loss along the tree–crop interface at the two stand ages were more intense in the second and third rotations than it was in the first cycle (Table 2). Similarly, significant interaction effects between rotation and distance (Table 1) revealed that the area of tree influence in which significant crop yield reductions occurred was doubled in the second and third rotations compared with the 8 m distance encountered in the first cycle (Table 2). The significant interaction effect between distance and aspect (Table 1) was attributed to differences in the intensity and duration of shading, which is obviously higher on east–west facing plots than on north–south facing plots and declines as the distance from the tree line increases. As a result, in the last two rotation cycles the crop yields measured within the first 16 m from the tree lines were significantly higher on the north–south facing plots compared with yields in east–west facing plots (Table 2). On the other hand, this trend was apparent only within the first 8 m (2–4 and 4–8 m strips) in the first cycle (Table 2). When averaged across all rotations and stands ages, crop yields in north–south facing plots were significantly higher (1138 vs. 1011 kg ha
1 ) than in east–west facing plots, although the yield difference was modest. The interaction effects of aspect with either rotation or stand age were not significant (Table 1), probably because the above-ground competition between the two components has a second-order importance in determining crop yield. Because of the combined effect of light and root competition, wheat yields on 2–4 and 4–8 m strips was declined by 29% and 16% on north–south and 57% and 27% on east–west facing plots, respectively, in two-year-old stands in the first cycle, in comparison with wheat yields in the open field (Table 3). The relationships between wheat yield and distance from the tree line (Table 3) can be used to estimate crop yield losses due to eucalypt boundaries from distance measurements across the tree–crop interface. The shift from a curvilinear to a linear relationship in the last two rotations probably reflects the time needed for the fine roots of eucalypt boundaries to penetrate to a more favourable part of the soil profile and exploit it to become more competitive. 166 S. Kidanu et al. = Agricultural Systems 80 (2004) 151–170 5. General discussion Population pressure in the Ethiopian highlands has led to changes in land use/ land cover with the aim of increasing agricultural production. The indigenous forest coverage was 87%, which is now reduced to below 4% (IUCN, 1990). With remaining forest and woodland cover estimated to be diminishing at a rate of 50,000– 200,000 ha yr
1 , the need to increase biomass by significant volumes in the near future is critical (EFAP, 1993). The establishment of woodlots and plantations to satisfy demand for forest produce has long been advocated as a strategy for relieving pressure on indigenous forest and woodland. In this context, Eucalyptus globulus boundary plantings, although they are a more time-intensive activity compared with woodlots plantations, have great potential to satisfy the ever-increasing demands of wood and wood products without inducing major land-use shifts. This has advantages for land-constrained smallholder farmers who cannot allocate land for block plantations. If the current per capita biomass energy consumption, is estimated at 0.75 m3 (EFAP, 1993), remains constant, Eucalyptus globulus boundary stands on a hectare of land harvested at age four years (first rotation) would satisfy about 35% of the annual biomass energy consumption of rural households with a family size of five people for four consecutive years. In the second and third rotations, however, the thinned-out stands at 2 years of age alone can easily meet about 15% of the annual biomass energy requirement of the household for two consecutive years. In addition, boundary stands harvested at the end of the second and third rotations each would satisfy about 55% of the annual biomass energy requirement of the household for four consecutive years. In other words, eucalyptus boundaries on a hectare of land in the second and third rotation cycles could deliver 70% of the annual biomass requirements, replacing much dung fuel, which currently accounts for as much as 81% of total household energy consumption per annum. This shift would promote the use of dung as fertilizer. In addition, preservation of indigenous woodland and biodiversity may be achieved when substitutes for indigenous forest products are established. Approximately 95% of the total demand for wood and woody biomass in rural Ethiopia is for fuelwood (EFAP, 1993). On Ethiopian highland Nitosols, integration of Eucalyptus into the agricultural system may offer considerable potential for exploiting the off-season rainfall, which accounts for more than 30% of the annual rainfall. In addition, reduction in runoff resulting from the physical barrier presented by tree component and/or increased soil moisture abstraction during the rainy season would increase the proportion of annual rainfall retained for transpiration and hence the overall productivity of the system. However, in the eucalyptus–wheat association the yield losses incurred because of eucalypt boundaries over the two stand ages were on average 8.2% on north–south and 8.7% in east–west facing plots for three rotation cycles, with somewhat lower values for the first rotation (6.3) and somewhat higher values for the second and third rotations (9.5) (Table 3). Although the eucalypt stands were of similar height, and hence had a similar shading effect, in the last two rotations wheat yields over the S. Kidanu et al. = Agricultural Systems 80 (2004) 151–170 167 first 8 m from the tree line declined by more than 60% in comparison with wheat yields obtained over the same distance in the first rotation (Table 3). This, in conjugation with the lack of significant yield differences between the two stand ages except in the first rotation cycle, indicates that tree root competition was more detrimental to the yield of wheat than light competition. Since competition for water is unlikely beyond a distance of 10 m given the results discussed above, it follows that the tree component affects the yield of an adjoining wheat crop primarily by altering the availability of soil nutrients. An extensive lateral distribution of tree roots in the subsoil, particularly in areas of high rainfall such as Ginchi, can take up nutrients that leach below the rooting depth of annual crops. In a spatial association of trees and crops, however, a large lateral spread of tree roots in the topsoil may intensify the competition with adjacent crops for nutrients and water (Van Noordwijk et al., 1996). The effective range of root competition assessed in this study is comparable to that found with Eucalyptus by Onyewotu et al. (1994) and Zohar (1985), who reported a range of 20 m for rainfed millet and irrigated cotton, respectively. Because trees have an established root system, they have an advantage compared with the adjacent wheat crop, particularly early in the growing season. It is therefore difficult to fertilize only the wheat crop without losing some of the fertilizer to the eucalypt boundaries. Thus, a larger fertilizer application along the tree–crop interface may offset the effects of root competition to a certain extent. However, the application of fertilizer increases root concentration near the soil surface (Campbell et al., 1994) and this may result in increased root competition. Hence it may be best to apply fertilizer in small doses several times during the active growing stages of the wheat crop. Repeated deep ploughing between trees and crops during the cropping season may further reduce the competition from superficial and re-growing tree roots. Currently, farmers leave untilled strips of 2–3 m between the line of trees and the annual crop, because they believe that this is the zone of the most active competition. Zero tillage in Leucaena leucocephala hedgerow intercropping has been reported to lead to yield depression because of root competition (Ssekabembe, 1985), but 30 cm deep ploughing along the hedgerow increased the soil water content under the crop and boosted the crop yield (Korwar and Radder, 1994). Lines of trees laid out from east to west, i.e., along the sunÕs path, maximize the penetration of direct sunlight in north–south facing plots. Tree trunks and crowns cast dappled shade early in the morning and late in the evening on the west- and east-facing plots, respectively. The main source of radiation in shaded areas has been intercepted and transmitted, perhaps several times, by the foliage of the Eucalyptus canopy before it reaches the understorey wheat crop. This process is known to deplete the photosynthetically active light in tree shaded zones (Ong et al., 1996). In shaded zones, the grain yield advantage of north–south facing plots over east–west facing plots was as high as 41%. Lyles et al. (1984) reported a reduction of winter wheat yield over a distance twice the height of the trees, because of the combined effect of light and root competition. Onyewotu et al. (1994) noted a 50% millet yield reduction over a distance of 18 m from the tree line. The reduction in the wheat yield 168 S. Kidanu et al. = Agricultural Systems 80 (2004) 151–170 in our study occurred over a shorter distance than that reported by Lyles et al. (1984), at least in the first rotation cycle. The root distribution (Table 5) showed that root competition did not extend beyond 20 m. 6. Conclusions Eucalypt boundaries produce large volumes of biomass within a short time without inducing a major land-use shift. This has advantages for land-constrained smallholder farmers who cannot spare land for block plantations. The greater availability of wood may reduce the demand for dung and crop residues as fuel sources, and thus may contribute to improved soil management on croplands. In addition, by providing substitutes for indigenous forest products, eucalypt boundary plantations could help preserve indigenous woodland and biodiversity. Eucalypt boundaries affect the yield of adjacent agricultural crops by modifying water, light and soil nutrient conditions. The root competition between the two sub-systems is intense, not only because the Eucalyptus roots colonize the same soil horizon as the wheat roots but also because of the deep lateral spread of the tree roots into the adjacent crop area. Although we consider root competition the most critical factor determining crop yield, shading extending up to 8–12 m also contributed to reduced yields. Under the prevailing environmental conditions of Ethiopian highland Nitosols, the practice of eucalypt boundaries seems economically viable, at least over the study period considered. Farmers actually sell Eucalyptus products at any time the year, whenever they encounter a cash shortage. Thus, the practice can be considered as a smoothing mechanism to help avoid income risk and increase security. Its adaptability and fast growth usually makes E. globulus the first tree of choice in the Ethiopian highlands. However, E. globulus may have allelopathic effects (May and Ash, 1990; Lisanework and Michelsen, 1993) and deplete soil nutrients (Michelsen et al., 1993), which over time may have far-reaching implications for sustainable land use. Further research should therefore be done to assess the possible allelopathic effects and the extent to which long-term soil fertility is influenced by eucalypt boundaries, especially given the interest in converting land under block plantations to crop cultivation, and for the integration of Eucalyptus species into agroforestry systems. References Bojo, J., Cassells, D., 1995. Land degradation and rehabilitation in Ethiopia: a reassessment. AFTES Working Paper No. 7. World Bank, Washington, DC. Campbell, C.D., Mackie-Dawson, L.A., Reid, E.J., Pratt, S.M., Duff, E.I., Buckland, S.T., 1994. Manual recording of mini-rihizotron data and its application to study the effect of herbicide and nitrogen fertilizer on tree and pasture root growth in a silvopastoral system. Agroforestry Systems 26, 75–87. Dhyani, S.K., Tripathi, R.S., 1999. Tree growth and crop yield under agrisilvicultural practices in northeast India. Agroforestry Systems 44, 1–12. S. Kidanu et al. = Agricultural Systems 80 (2004) 151–170 169 EFAP, 1993. Ethiopian forestry action program: Final report. Ministry of Natural Resource Development and Environmental Protection. Addis Ababa, Ethiopia. Eissentat, D.M., 1992. Costs and benefits of constructing roots of small diameter. Journal of Plant Nutrition 15, 763–782. FAO, 1985. The ecological effects of Eucalyptus. FAO Forestry Paper No. 59, Rome, Italy. Holgen, P., Svensson, P., 1990. Loss of inorganic nutrients by whole tree utilization for firewood in Ethiopia. Working Paper No. 150, Swedish University of Agricultural Sciences International Rural Development Center. IUCN, 1990. Workshop Report on Human Population Dynamics and Resource Demand; 30 Nov. to 1 Dec., 1990. The World Conservation Union 18th General Assembly, Perth. Australia. Jama, B., Buresh, R.J., Ndufa, J.K., Shephred, K.D., 1998. Vertical distribution of roots and soil nitrate: tree species and phosphorus effects. Soil Science Society of America Journal 62, 280–286. JVP (Joint Vertisols Project) 1995. Resource management for improving and sustaining crop and livestock production on the highland Vertisols of Ethiopia. Report No. 2. Addis Ababa, Ethiopia. Khybri, M.L., Gupta, R.K., Sewa, R., Tomar, H.P.S., 1992. Crop yields of rice and wheat grown in rotation as intercrops with tree species in the outer hills of Western Himalayas. Agroforestry Systems 17, 193–204. Korwar, G.R., Radder, G.D., 1994. Influence of root pruning and cutting interval of Leucaena hedgerows on performance of alley cropped rabi sorghum. Agroforestry Systems 25, 95–109. Lisanework, N., Michelsen, A., 1993. Allelopathy in agroforestry systems: the effects of leaf extracts of Cupressus lustanica and three Eucalyptus spps. on four Ethiopian crops. Agroforestry Systems 21, 63– 74. Lyles, L., Tatarko, J., Dickerson, J.D., 1984. Windbreak effects on soil water and wheat yield. Trans. ASAE 27, 69–72. Malik, R.S., Sharma, S.K., 1990. Moisture extraction and crop yield as a function of distance from a row of Eucalyptus tereticornis. Agroforestry Systems 12, 128–187. May, F., Ash, 1990. An assessment of the allelopathic potential of Eucalyptus. Australian Journal of Botany 38, 245–254. Michelsen, A., Lisanework, N., Friis, I., 1993. Impacts of tree plantation in the Ethiopian highland on soil fertility, shoot and root growth, nutrient utilization and mycorrhizal colonization. Forest Ecology and Management 61, 299–324. MSTAT-C 1991. A microcomputer program for design, management and analysis of agronomic research experiments. Michigan State University, East Lansing, MI, USA. National Fertilizer and Input Unit 1991. Results of fertilizer trials conducted on major cereal crops. Working Paper No. 34, NFIU/MOA, Addis Ababa, Ethiopia. Ong, C.K., Black, C.R., Marshall, F.M., Corelett, J.E., 1996. Principles of resource capture and utilization of light and water. In: Ong, C.K., Huxely, P. (Eds.), Tree–Crop Interactions. A Physiological approach. CAB International, Wallingford, UK, pp. 73–158. Onyewotu, L.O.Z., Ogigirigi, M.A., Stigter, C.J., 1994. A study of the competitive effects between a Eucalyptus camaldulensis shelterbelt and adjacent millet (Pennisetum typhoides) crop. Agriculture, Ecosystems and Environment 51, 281–286. Peter, D.B., 1965. Water availability. In: Black, C.A. (Ed.), Methods of Soil Analysis. Part I. Agronomy Monograph No. 9. Madison, WI, pp. 279–285. Pohjonen, V., Pukkala, T., 1990. Eucalyptus globulus in Ethiopian Forestry. Forest Ecology and Management 36, 19–31. Pukkala, T., Pohjonen, V., 1989. Forest Inventory and Management Planning in the fuelwood plantations of Ethiopia. Faculty of Forestry, Joensuu University, Finland. Rao, M.R., Coe, R.D., 1991. Measuring crop yields in on-farm agroforestry studies. Agroforestry Systems 15, 275–289. Reversat, F.B., 1999. The leaching of Eucalyptus hybrids and Acacia aurculiformis leaf litter: laboratory experiments on early decomposition and ecological implications in congolese plantations. Applied Soil Ecology 12, 251–261. 170 S. Kidanu et al. = Agricultural Systems 80 (2004) 151–170 Sanginga, N., Swift, M.J., 1992. Nutritional effects of Eucalyptus litter on the growth of maize (Zea mays). Agriculture, Ecosystems and Environment 41, 55–65. Ssekabembe, C.K., 1985. Perspectives on hedgerow intercropping. Agroforestry Systems 3, 339–356. Tekelu, E., Selamyihun, K., Tekalign, M., Mesfin, A., 1999. Effects of land preparation methods on runoff and soil loss on Vertisols at Ginchi (Ethiopia). Ethiopian Journal of Natural Resources 2, 1–15. Van Noordwijk, M., Lawson, G., Soumare, A., Groot, J.J.R., Hairiah, K., 1996. Root distribution of trees and crops: competition and/or complementary. In: Ong, C.K., Huxely, P. (Eds.), Tree–Crop Interactions. A Physiological approach. CAB International, Wallingford, UK, pp. 319–364. Verinumbe, I., 1987. Crop production on soil under some forest plantations in the Sahel. Agroforestry Systems 5, 185–188. Zohar, Y., 1985. Root distribution of a Eucalyptus shelterbelt. Forest Ecology and Management 12, 305– 307.