Venant Rutunga & Henri Neel

Yield trends in the long-term crop rotation with organic and inorganic fertilisers on Alisols in Mata (Rwanda)

Résumé

Un système de rotation constitué de plusieurs espèces de cultures a fait l’objet d’expérimentation sur des « Alisols » sous prairie à Mata, versant oriental de la Crête Zaïre-Nil (CZN) au Rwanda. Des fumures minérales et organiques ont été apportées dans différentes parcelles arrangées suivant un dispositif en blocs complétement aléatoires en trois répétitions. Les données de rendements des cultures ont été mesurées chaque saison pendant une période de neuf ans. Les résultats ont montré que les rendements des cultures étaient faibles ou nuls dans les parcelles n’ayant pas reçu de fumures. En parcelles fumées, le rendement a augmenté en général mais est resté relativement bas. Seules quelques espèces ou variétés adaptées aux conditions écologiques de Mata telles les tubercules et racines tubéreuses ainsi que l’éleusine ont mieux réagi à la fumure. Le chaulage était indispensable pour obtenir tout effet positif de l’engrais minéral NPK sur la production des cultures. La forte dose de bon fumier avait un effet notoire et ses trois applications successives durant quatre saisons avaient un arrière-effet bénéfique jusqu’à quatre saisons après son application. Le compost de Mata (C/N>25, 0,3g P.kg^-1) produisait par contre un effet négligeable dans l’amélioration des rendements des cultures. Une, quatre et demi, et huit tonnes de chaux appliquée trois fois en huit ans ont augmenté le pH (eau) du sol mais pas jusqu’à la valeur de 6,5. Toutes ces constatations montrent que pour maîtriser la production performante des cultures à la CZN, la sélection des espèces et variétés adaptées aux conditions écologiques et l’amendement des sols constituent deux stratégies obligées plutôt complémentaires, mais pas séparées.

Abstract

A crop rotation system with various species was established on « Alisols » at Mata grassland site, oriental side of Zaire-Nile Watershed Divide (CZN), Rwanda. Inorganic and organic fertilisers were applied in various plots led in randomised complete blocs with three replicates. Crop yield data were each season recorded over a 9-year period. Results showed that there was very low or no harvest in plots without fertilisers. Where fertilisers were applied, the yield generally increased but remained relatively low: only few crops and varieties adapted to the Mata ecology such as potatoes and finger millet responded well to fertilisers. Liming was absolutely necessary to get any acceptable crop yield improvement with NPK. High rate of rich farmyard manure was efficient alone and its effect was seen up to 4 seasons after its four regular seasonal applications. Mata compost (C/N>25, 0,3 g P.kg^-1) had little beneficial effect. One, four and half, and eight tons of lime per ha applied 3 times in 8 years increased soil pH (in water) but not up to 6,5. It is then concluded that to succeed improving food production at the CZN area, selection of crops and varieties to fit ecological conditions and amending soils to fit crops be combined, but not opposed.

Inhoudstafel

1. Introduction

1Rwanda is a very high densely populated country and this has led to arable land scarcity. On average, the farms are smaller than 1 ha (Clay, 1996). Farmers are forced to do continuous cropping and to use soils of low fertility status for food production. Among the Rwanda less-productive soils are those low in bases, very acid and with high content in exchangeable Al, generally classified as Ferralsols, Acrisols and Alisols (World Reference Base). Crop yield obtained on such soils is very low when no fertiliser is applied (Driessen et al., 2002). Alisols dominate three agro-ecological zones i.e. Zaïre-Nile Watershed Divide (CZN), Buberuka and Ndiza Highlands (Birasa et al., 1992) that are situated at an altitude between 1900 and 2500 m from the sea level. They are also found in the « plateau central » zone attaining to the former zones and lying between 1500 and 1900 m of altitude. Approximately, Acrisols and Alisols occupy 25% of the Rwanda area (Verdoodt, van Ranst, 2003). Many farmers live in this area and their most important traditional crops are sweet potatoes (Ipomea batatas (L.) Lam), Irish potatoes (Solanum tuberosum L.), peas (Pisum sativum L.), beans (Phaseolus vulgaris L.), wheat (Triticum aestivum L.), maize (Zea mays L.) and sorghum (Sorghum vulgare Pers.) (Delepierre, 1974). Frankart et al. (1974) reported that crop yields on soils where farmers have applied organic farm manure for many times and during a long period (anthropic soils) are medium to good while they are too low or zero on soils under grassland with Eragrostis sp. as dominant species. Soil chemical and biological properties were improved under anthropic soils (low or no exchangeable Al, water pH above 5 and relatively high nutrient content and microbial activity). This technology for improving soils is efficient but requires long period and is then not appropriate for farmers who need high production in the short time for surviving.

2Field research was carried out from 1971 to 1980 in order to identify the most appropriate inorganic and organic fertilisers to apply for improving crop yields on these Alisols and Acrisols in a short time. These long-term field experiments contribute to a better quantitative knowledge of influences of soil properties, weather conditions and management system on the value of soil for crop production (Haans, Westerveld, 1970). Rutunga et al. (1998) reported results on non-humiferous Alic ferralsols in Rwanda « plateau central ». The current paper provides the results on humiferous Alisols at an altitude of 1650 m from the sea level.

2. Material and methods

3From 1971 to 1980 a rotational experiment with inorganic and organic fertilisers was established in Mata, Gikongoro region where the secondary forest had been cleared in the 1900’s. After the nutrients from forest biomass were exhausted through non-sustainable continuous cropping the site was abandoned and became public grassland dominated with Eragrostis blepharoglumis K. Schum. The climate at the site is Cw2 (ILACO, 1985) with a dry season occurring from June to August. The monthly rainfall and average temperature during the experimental period are given in table 1. Soil temperature regime is udic isothermic (Verdoodt, van Ranst, 2003). The slope at the site is 6 to 25 % but was 5 % in the experimental plots. Soil chemical properties at the beginning of the experiment are shown in table 2. A pot fertiliser trial with finger millet (Eleusina coracana Goertn) as test plant revealed high N, P and K deficiency and low to medium Ca and Mg deficiency in the soil (ISAR, 1972).

4The experiment was carried out in 4 benches: benches 1 to 3 comprised 16 treatments (Table 3 a, b and c) while bench 4 was for validating in few treatments 2 types of fertiliser combination i.e., lime with NPK and manure with phosphorus. Fertilisers tested were burnt lime (39.2% CaO, 4.0% MgO), mineral NPK fertilisers i.e. urea (46% N), triple super-phosphate (19% P) and various NPK formula (20-4-8; 15-4-12; 15-9-8 and 10-11-8), Mata compost (1.3g N, 0.3g P and 15g K.kg^-1, prepared from Setaria sphacelata (Schum.) Moss and Eragrostis blepharoglumis) and ISAR farmyard manure (FYM: 6g N, 2.5g P and 14.2g K.kg^-1). The bench 1 was for a crop response to mineral NPK rates on limed plots up to Feb 74; in Sep 74 the bench was converted into a factorial trial as shown in table 3a. The bench 2 was similar to bench 1 but was not limed up to Feb 74; it was also converted into a factorial trial in Sep 74 (Table 3b). The bench 3 received three rates of lime (1.0, 4.5 and 8.0 Mg.ha^-1) and various rates and formula of NPK during the 9 year-experimental period (Table 3c). The bench 4 was for testing liming, NPK fertilisers and farmyard manure rates and their period of application (Table 3d). The various treatments and the fertiliser-detailed rates over time are shown in the tables 3. Treatments in each bench were laid out in complete randomised block design with 3 replicates. Plot size for each treatment was 16 m² (4 m by 4 m).

5Test plants used in the rotation comprised traditional crops and 2 new crops introduced by ISAR i.e. sunflower (Helianthus annuus L.) and soya (Glycine max (L.) Merr.) (Table 4). Crop husbandry was manually done according to the ISAR recommendations (ISAR husbandry guides for various crops) summarised as follows:

6Land preparation: the land was manually tilled with a hoe and Digitaria scalarum (Schweinf.) Chiov. roots uprooted and removed from the field. This 1^st digging was followed by a 2^nd cultivation for soil levelling before planting.

7Fertiliser application and soil sampling: lime was broadcast by hand after the 1^st soil tillage, 15 days before the planting period and immediately incorporated in 0-15 cm soil depth, using a hoe. Mineral NPK, compost and FYM fertilisers were broadcast by hand after soil levelling just before planting and incorporated in 0-15 cm depth, using a hoe. One year after the 3^rd liming, soil samples were taken from the limed plots, bench 3, at 0-15 cm soil depth and the pH in a suspension of soil/water of 1/2.5 was measured with a pH-meter in order to identify the pH improvement due to liming.

8Crop management: cereal, legume and sunflower grains, potato tubers and sweet potato vines were provided by ISAR and were established as follows:

9– wheat and soya seeds sown continuously in the rows, distance of 40 cm between two rows, manually weeding, and harvest at 138 days after planting (DAP);

10– maize sown at 3 grains/hole on the rows with 50 cm between two holes and 60 cm between two rows, thinning to 2 seedlings one month after planting, weeding two times and harvest after 217 DAP;

11– finger millet seeds broadcast continuously in the rows, distance of 30 cm among rows, manually weeding, and harvest at 150 DAP;

12– sunflower sown at 2 seeds/hole on the rows with 30 cm between two holes and 60 cm between rows, thinning to 1 seedling one month after planting, weeding and harvest after 150 DAP;

13– peas sown at 1 grain/hole in the rows with a spacing of 5 cm by 40 cm, manually weeding, harvest after 112 DAP;

14– non climbing beans sown at 1 grain/hole in the rows with a spacing of 10 cm by 20 cm, manually weeding, harvest after 98 DAP;

15– potato tubers planted in rows, distance of 30 cm between two plants and 60 cm between two rows, weeding, spreading of dithane M45 for mildew control, harvest at 122 DAP;

16– sweet potato vine cuttings planted on ridges of 30 cm diameter, two vine cuttings of 30 cm length hole, distance of 30 cm between two holes and ridges of 80 cm-diameter, manually weeding, harvest at 240 DAP.

17Except for dithane M 45, no other crop protection chemicals were used.

18At harvest, fresh cereal, legume and sunflower top biomass were cut near ground level, removed from the plots, sun dried, and then thresholds separated from grains. Grains were weighed at 12 % moisture content. For Irish and sweet potatoes, fresh tubers, storage-roots and vines were immediately weighed. Data recorded from the four benches were statistically analysed, using two and three-way analyses of variance. Factorial analysis was performed for data obtained from the benches 1 to 3. The yield means were compared, using the least significant difference (LSD) method at 0.05 and 0.01 probability level. For factorial treatments with equidistant rates the contrast method was used for knowing the nature of interactions (linear, quadratic). All calculations were performed with the SAS statistic program (SAS Institute, 1988).

3. Results

19Data from soil analysis showed that liming increased soil pH (Table 5). Crop harvest obtained from the 4^th to 9^th year is shown in tables 6 to 9. With data from bench 1 (Table 6) it appears that:

20– fresh liming (year 4, year 5 and year 7), when supplemented with a seasonal NPK application and «residual lime-NPK» effect, (RFL2) significantly increase yields for all crops;

21– crop production was significantly improved with fresh NPK added every season to the « residual lime-NPK » effect (RF). This would indicate that lime applied in year 1 and year 2 still had an efficient residual effect but such a long residual effect cannot be explained. NPK fertiliser impact for crop yield improvement was increasing as was NPK rates. However, from 400 kg (R2F2) to 600 kg of NPK ha^-1 (R3F3) there was little or no yield increase for many crops;

22– compared to the crop yield obtained with the « residual lime-NPK » effect alone (R), the compost (RH) produced crop yield increase, but little. Relatively acceptable yield increase was recorded only with potatoes and finger millet. The effect of compost on yield improvement was generally inferior to that of NPK (RH<RF);

23– among crops finger millet showed the best response to the 2^nd and 3^rd residual effects of all fertilisers.

24Data from bench 2 (Table 7) lead to similar observations as those in table 6:

25– fresh liming (year 4, year 5 and year 7) was superior to the residual NPK effect and absolute control (R0+0) in improving crop yield (RL2>R=R0+0);

26– NPK fertiliser applied alone had little residual effect (R) on crop yield improvement and this had been already reported by Neel (1974). Only the residual NPK at a rate of 600 kg.ha^-1 (R3) produced a slightly significant yield increase for some crops. Mata compost (RH) produced a slightly higher crop yield increase than the residual NPK effect (R) and absolute control;

27– compost and fresh lime had beneficial direct and residual effects particularly on finger millet (1^st and 3^rd residual effect) production;

28– two-ton lime and compost when applied in the same plot (RL2) generally showed an additional and beneficial effect on crop yield improvement.

29With data from bench 3 (Table 8) the following comments are drawn:

30– one-ton lime applied alone three times (year 2, year 4, year 7) had very low and no significant direct and residual effects on crop yield increase (F0L1 versus F0L0);

31– NK fertiliser without P is not efficient (F0L1 versus F2L1): P is highly needed to improve crop yield (F0L1=F2L1<F3L1);

32– used with direct and residual lime effects, the direct or residual effect of P2 and that of NP2 on crop yield increase were statistically equivalent but less efficient than that of NKP2 (F6L8=F7L8<F3L8). Crop yield obtained with KP2 was approaching that with NKP2 (F3Lb and F5Lb);

33– direct and residual effects of CaO rates are observed with all crops but such an effect was higher with beans and wheat before year 4 (Neel, 1974). For more than a half of crops there is no yield increase due to lime rate higher than 4.5 Mg.ha^-1;

34– Lime+NPK has significant direct and residual effects on crop yield improvement. However, there was P rate effect only with some crops (beans and condea potato) and high effect had been observed with peas and wheat in the beginning, from year 1 to year 4 (op. cit.).

35Neel (1974) had reported that there was no effect of mineral NPK fertilisers applied alone (without prior liming) and no effect of P associated with 35 Mg of fresh FYM.ha^-1. In addition data from bench 4 (Table 9) indicate that in Mata experiment, even with liming every season, there was no interest in increasing N, P and K to more than 60 kg N, 33 kg P and 42 kg K.ha^-1 with the used crop varieties. Another observation is that four seasonal continuous applications of 35 Mg of fresh FYM ha^-1 had a beneficial residual effect on P availability and crop yield improvement for two years.

36In summary, without fertiliser application, most food crops gave very low or no harvest at all on Mata Alisols. Stunted crop roots and shoots or low rate of germination of irish potato from the soil were commonly observed. Liming supplemented with NPK and/or compost had the best effect on crop yield when compared to the treatments with lime alone or NPK or compost applied alone. Mata compost had low effect in improving crop production. Farmyard manure at high rate (35 Mg.ha^-1, fresh weight) showed direct and residual effects on crop production improvement. In general crop yields remained low to medium with the best fertiliser treatments except for finger millet. Diseases and attacks from pests were also observed on some crop varieties.

4. Discussion

37Soils in Mata had low pH, high level of exchangeable Al (>60% Al saturation) and of organic C and N, and very low available P and echangeable K (Table 2). They are classified as Alisols in the World Reference Base (Driessen et al., 2002). High Al saturation leads to high soil P fixation (van der Eijk, 1997) and is toxic to many annual plant species (Adams, Hathcock, 1984; Marschner, 1986). Crops indeed require a minimum level of nutrients to develop and give yield (Fageria, 1992) and such a level of nutrients was not attained in the Mata control plots. Low or no crop harvest obtained in the unfertilised Alisols is well known especially for the less acid-tolerant food crops (Lal, Stewart, 1995; Splett et al., 1992) and fodder species (Compère et al., 1994).

38In Mata soils the pH increase recorded in the 15 cm depth of the plots limed three times in 4 years (Table 5) was not high. Soil pH is the result of a large number of soil properties and processes and these do not change immediately upon liming. This is in accordance with Benites and Valverde (1982) who reported that the lime quantity determined by Al-neutralisation method is less than that needed to raise the soil pH at 6.5. The pH in Mata soils was not a permanent value (data not shown). It was higher some few days after liming, especially in plot limed with 8 Mg.ha^-1, and then progressively decreased with time as Ca was used by plants, mixed with soil of deep layer through tillage and/or lost through leaching and erosion. High rainfall and cool temperature at Mata (Table 1) and slope of 5% are conditions that favour leaching and erosion rather than evapo-transpiration. The current pH value only indicates the level of the Ca influence remaining after the 2^nd season following liming and will continue decreasing with time if no other lime is applied to the soil. This was quite different in warmer area of Rubona where 2 Mg of lime ha^-1 applied every two years led to over-liming after eight-year period (Rutunga et al., 1998).

39Al toxicities and soil P-fixing capacity are alleviated or reduced through liming that neutralises exchangeable Al (Fageria et al., 1990), increases P availability from fixed-P fraction and provide Ca (280 kg/ton) and some Mg (24 kg/ton in this study) to crops (Smith, Sanchez, 1980). This was confirmed in Mata experiment where significant crop yield performance was recorded in treatments « lime + NPK ». Lime alone had low effect on crop production, showing that either low amount of available P is released from soil fixed-P fraction or enough P is available but cannot have effect because of K deficiency naturally present in the soil (Table 2) or induced by excess of Ca provided by lime. Large amount of lime can lead to high Ca concentration in soil and this may increase P fixation capacity of the soil through adsorption and precipitation processes (Salingar, Kochva, 1994; Olsen, Watanabe, 1957) and may also induce K imbalance. This may probably be the case for the treatments with 8 Mg of lime ha^-1 where for most crops no yield increase was observed in comparison to the yield with 4.5 Mg lime ha^-1 (Table 8). If a crop tolerates the Al saturation level of 10% and its rooting depth is either 0-15 or 0-30 cm, theoretic calculation for neutralising 40 mmol (+) Al.kg^-1 of soil with 1 Mg.m^-3 bulk density and within 1ha gives 4.0 and 8.0 Mg of the Mata-used lime. This result of calculation indicates that the inefficiency of 8 Mg of lime ha^-1 tested in this study might also be due to its application in 0-15 cm-soil depth (see material and methods). Such shallow lime incorporation was done because it was not possible to well mix lime in 0-30 cm depth with a hoe.

40Adding mineral NPK fertiliser alone to Mata soils was not useful for crop production. When P water soluble fertilisers are applied to acid soils, P reacts with Al and Fe compounds (Sample et al., 1980). Sanchez and Uehara (1980) identified application of large amount of P or small amount of P placed near the root zone as means of managing P-fixing soils. A portion of added P is fixed by soils and neutralise exchangeable Al and its toxicities while the remaining P is available to plants. This was not the case in the current experiment since P fertiliser was broadcast and its rate was not very high. Either all added P was just used to alleviate exchangeable Al and toxicities and nothing else to spare or it was totally fixed by the soil but insufficient to eliminate all Al stresses. The former hypothesis leads to low or no availability of P as a limiting-growth factor and according to Janssen et al. (1990) such lack of P set the limit to the uptake of the non-limiting nutrients. In the latter assumption the crops can take up no P and nutrients since P is unavailable and plants cannot develop sufficient roots in the presence of Al toxicity. The P fixation and/or Al toxicity constraints joined with the probably losses by erosion and leaching also explain why application of NPK alone had little residual effect on crop yields.

41After liming, when required nutrients (NPK) were added, crop yields were significantly improved (Table 6, 8 and 9). The optimum mineral fertiliser rate was somewhere around 60 kg N, 35 kg P and 32 kg K.ha^-1 (significant quadratic component). However at this optimum rate yield for most crops was still low to medium. This may be related to low potential of crop varieties influenced by unfavourable climatic conditions and inappropriate plant health status.

42High organic C and N in soils, as the case of Mata, may be due to the poor status of other nutrients in the soil and strong organic matter complexation with Al and Fe oxides, preventing sufficient microbiological degradation as suggested by Hiemstra (Tjisse Hiemstra, personal communication, Wageningen University, 05 Jan. 2005). This changed after liming and addition of K and P, and thus some N from soil was provided to crop, for instance some production was obtained with F5Lb, table 8.

43For fresh biomass, Chikowo et al. (2004) stressed that application of poor quality manure or compost is not enough in itself to overcome nutrient deficiencies in very depleted soils and for making high yield. The crop production was effectively fairly improved when the low quality compost of Mata was applied, even into the limed plots. Liming generally does increase the biomass decomposition rate in the soil (Ducheaufour, 1977), but for sufficient mineralisation rate enough P level in the applied biomass (Blair, Boland, 1978) or in soil (Kabba, Aulakh, 2004) is needed. Thus in presence of lime and low P availability Mata compost either quickly or slowly decomposed but the final result was probably few nutrients released.

44Four consecutive seasonal application of farm-yard manure (35 Mg.ha^-1 season^-1, fresh weight) significantly increased crop yield (Neel, 1974) and had even a lasting beneficial residual effect (Table 9). Same observation was done on non-humiferous Alic Ferralsols at Butare (Rutunga et al., 1998). Large amount of FYM can provide enough various nutrients (Janssen, 2000; Van der Eijk, 1997) and may temporary reduce acidity constraints in the soil (Hue et al., 1986). Organic manure, only when containing more than 3 g of P kg^-1, facilitates P availability by blocking P sorption sites of soils (McDowell, Sharpley, 2004). Mata experiment showed that regular application of 35 Mg of good quality FYM ha^-1 did require no P supplement and no liming for ensuring acceptable to high yield. The problem is that such an amount is too large to be available on Rwanda farms.

45Due to all various difficulties (incorporating lime in 30 cm soil depth, FYM unavailability, uncertain or low crop response to mineral NPK fertilisers, nutrient imbalance, etc.) in Rwanda agricultural context, the beneficial use of both organic and inorganic fertilisers together as already reported through the literature (Lal, Stewart, 1995) was confirmed. Low amount of fresh FYM such as 10 Mg combined with 2.0 Mg of lime every two years and seasonal application of 50-21-41 kg of N, P and K elements ha^-1 was a rate and formula proposed for improving crop production on acid soils of Mata area.

46Finally, before making the large investment highly needed for nutrient depleted soils other agricultural techniques such as use of appropriate varieties, minimising losses from pests and diseases and appropriate water management should be optimised (van Reuler, Prins, 1993). Most of crop varieties used in Mata experiment were traditional and had low response to fertilisers (ISAR, 1976); some were even not adapted to the Mata zone (i.e. bataaf bean was a variety screened for warm and dry land). Most of times the performance of Annett and condea potatoes, romany wheat and kyondo peas was also reduced by severe attack of fungus, bacterial and viral diseases. Rusenya s. potato showed better and more stable response to fertiliser than did nsasagatebo in the Mata growth conditions but the yield was still low because many ISAR crop cultivars were developed in areas with fertile soils, therefore under low or zero selection pressure for nutrient use efficiency and toxicity tolerance. Coracana finger millet is the one of the few crops that performed well in Mata fertilised plots. There are two reasons for this: a) finger millet responds to low fertiliser application such as 20-90 kg N, 4-43 kg P and 6-83 kg K.ha^-1 (de Geus, 1973b) the crop is well adapted to Mata climatic conditions where it is traditionally cropped with soil-sod chunk turning over and burning method (Rachie, Peters, 1977) in order to provide through the ashes, the minimum nutrients needed for a small harvest on infertile soils. Ash has direct effects as NPK fertiliser and as liming material (van Reuler, 1996).

5. Conclusion

47Crop development and high yield depend on soil properties, climatic conditions, plant potential, disease and pest control and optimal crop management. If one of these factors is limiting, nutrient use efficiency reduced and crop yield declined. Application of fertilisers is a mean for correcting soil nutrient deficiencies and toxicities. Mata experiment confirmed that and this for Alisols: lime is essential but must be combined with N, P and K nutrients to be provided by organic manure or/and inorganic fertilisers in order to get adequate crop production. In Mata conditions, effect of two-ton lime ha^-1 remained significant up to three-four seasons, that of “35 Mg fresh FYM ha^-1 times four consecutive seasons” four seasons and that of mineral NPK fertiliser one season. Although Mata cropping had other limiting factors than soil properties, results over nine years were enough consistent for proposing 2 Mg of lime and 10 Mg of FYM every two years combined to a seasonal application of 50 N-21 P-41 K kg.ha^-1 as a start in improving crop yield on high altitude Alisols. Before April 1994 such a fertiliser combination was significantly increasing crop yields in many on-farm experiments carried out by agricultural projects (Ndindabahizi, Ngwabije, 1991). However, for Rwanda to succeed a high and sustainable improvement of food production, it remains essential that selection of crops to fit soil conditions, appropriate crop management and parasite control, and amending soils to fit crops be combined, but not opposed.

48Acknowledgement

49The authors are grateful to the ISRIC management for helping Dr Venant Rutunga to pursue his scientific work at International Soil Reference and Information Centre, Wageningen University and Research Centre. Prof Dr Bert Janssen helped in the statistic evaluation.

Bibliographie

Om dit artikel te citeren:

Over : Venant Rutunga

Over : Henri Neel