INTRODUCTION
Cassava (Manihot esculenta Crantz) is one of the key food crops worldwide (Burns et al., Reference Burns, Gleadow, Cliff, Zacaria and Cavagnaro2010) and is valued for food security as well as household income generation in many tropical countries (Agwu and Anyaeche, Reference Agwu and Anyaeche2007). In Côte d'Ivoire, it is currently the second most important food crop next to the yam (Dioscorea spp), contributing 109.7 kg per capita and per year to the food supply, with an annual production of ca. 2.5 million tons (FAO, 2014). Cassava cultivation is widespread in the country, but is predominantly a subsistence activity for smallholder farmers in the southern region, where it serves as the main staple food for the local population. In this region, as in many tropical agro-ecological zones, cassava is grown on highly weathered, poor to marginal soils (Cadavid et al., Reference Cadavid, El-Sharkawy, Acosta and Sanchez1998; Howeler, Reference Howeler1991) and under continuous cultivation conditions. More fertile land is devoted to lucrative cash crops such as rubber and oil palm. This practice stems from the perception that cassava more easily adapts to poor soil conditions compared to other crops and yields well even in risk-prone environments (Fasinmirin and Reichert, Reference Fasinmirin and Reichert2011).
However, poor weed management and lack of suitable and profitable agronomic practices are major constraints which limit cassava production. Current cropping systems are hampered by the scarcity of drought- and disease-resistant cassava varieties, the lack of (mineral) fertilizer inputs and, particularly, by the premature harvest of cassava storage roots, i.e. several months before optimal yield (which occurs 12 to 24 months after planting (MAP)). This particular practice is, on one hand, a result of a strong increase in the demand for cassava coupled with high land pressure. Both result from a high population growth rate, which puts upward pressure on demand for this staple crop and limits the amount of land available for cultivation in the region. On the other hand, the urgent need for cash, e.g. to send children to school offers farmers often no other option, but uprooting the cassava to get immediate cash. Intensification of production is hampered by poor land use resilience under these conditions and soil nutrient depletion. Hence, more sustainable methods of cassava cropping intensification are required (Vanlauwe et al., Reference Vanlauwe, Coyne, Gockowski, Hauser, Huising, Masso, Nziguheba, Schut and Van Asten2014). Smallholder farmers, due to poor resource endowments, should be encouraged to use methods which help to ensure efficient use of scarce external inputs (Tittonell and Giller, 2013). Advocated approaches include biological N2 fixation (BNF), through legume rotation or intercropping (Leihner et al., Reference Leihner, Rupenthal, Hilger and Castillo1996; Nyi et al., Reference Nyi, Mucheru-Muna, Shisanya, Lama, Mutuo, Pypers and Vanlauwe2014), and/or the combined use of moderate mineral fertilizers and organic amendments (Amanullah et al., Reference Amanullah, Alagesan, Vaiyapuri, Sathyamoorthy and Pazhanivelan2006; Anyaegbu et al., Reference Anyaegbu, Ezeibekwe, Amaechi and Omaliko2009). To address these choices, the concept of ISFM was implemented in southern Côte d'Ivoire. ISFM is a set of practices, adapted to local conditions, which aim to maximize the agronomic efficiency of applied nutrients and improve crop productivity. It focuses particularly on the use of fertilizers, organic inputs (e.g. N-rich legume residues) and improved germplasm (Vanlauwe et al., Reference Vanlauwe, Bationo, Chianu, Giller, Merchx, Mokwunye, Ohiopkehai, Pypers, Tabo, Shepherd, Smaling, Woomer and Sanginga2010).
Two on-farm experiments were carried out during the 2012–2013 cropping season in two localities (Dabou and Bingerville), characterized by widespread cassava production, in southern Côte d'Ivoire following an ISFM demonstration and negotiation phase with local farmers. The overall objective of this research was to evaluate, using multi-location trials, the performance and degree of acceptability of two selected ISFM systems by resource-poor smallholder farmers. More specifically, we aimed to determine the productivity and total profitability of both ISFM systems compared to traditional systems. This analysis will identify the most viable ISFM systems for sustainable intensification of local cassava production.
MATERIAL AND METHODS
Selection of ISFM technologies
Before the multi-location trials, two experiments were carried out in the study area during the previous 2012–2013 cropping season. A set of ISFM technologies were demonstrated and assessed with local stakeholders. The aim was to optimize the agronomic efficiency of the scarce external inputs, thereby using modest amounts of locally available nutrient sources in combination with disease resistant cassava varieties and legume intercropping. The technical components included (1) cassava genotypes (disease resistant varieties Yavo and Bocou 1, vs. a local variety Yace), (2) legume intercropping using cowpea and groundnut, (3) modified cassava spacing (S1: 1 × 1 m, i.e. 1 m between cassava rows and 1 m between plants in a row vs. S2: 2 × 0.5 m, i.e. 2 m between cassava rows and 0.5 m between plants in a row), and (4) modest application rates of organic (chicken manure at 5 t ha−1) or mineral fertilizer (50 kg NPK 15-15-15 ha−1). The manure and mineral fertilizer dose were selected based on their affordability and availability. The experiment was carried out as a split-split-plot design in three randomized blocks. Blocks were composed of three sub-blocks. Each sub-block contained one cassava variety grown at both spacings and included eight treatments (further referred to as T0 – T7). The treatments were T0: control (no inputs), T1: chicken manure applied at recommended dose in Côte d'Ivoire (10 t ha−1) further referred to as ‘reference dose’, T2: cowpea intercrop, T3: groundnut intercrop, T4: cowpea intercrop + NPK (50 kg ha−1), T5: groundnut intercrop + NPK (50 kg ha−1), T6: cowpea intercrop + half manure reference dose (5 t ha−1) and T7: groundnut intercrop + half manure reference dose (5 t ha−1).
Disease-free cassava stem cuttings of 20 cm length were planted upright on flat land after manual tillage at a density of 10 000 plants ha−1, both in spacing 1 × 1 m and 2 × 0.5 m, at the beginning of the rainy season in May 2012. The manure was applied 10 days after planting in a circular ditch of 10 cm radius around each cassava plant that was covered with soil. The mineral fertilizer was applied 20 days after planting, in trenches at 10 cm distance on either side of the cassava rows that were covered with soil. These methods were chosen over the local practice of broadcasting on the fields in order to prevent losses. The intercropped legumes were sown 30 days after cassava planting. Two and four rows of legume were intercropped between two rows of cassava in S1 and S2, respectively. The legume plant density was 0.2 × 0.2 m in S1 and 0.4 × 0.2 m in S2, with 0.4 m distance between cassava lines and legume rows for both spacings. Four months after the cassava was planted, the first legume crop was harvested for both spacings. The fresh aboveground residues from the harvest were cut and applied as surface mulch between casava plants. A second legume crop was sown 1.5 months after the first legume harvest in S2, but only within the two centre rows in order to avoid the shading effect from the cassava canopy. The second legume crop was harvested 8 months after the planting of the cassava, and residues were again applied as surface mulch.
The experimental unit plot size was 6 × 4 m from which a 3 × 3 m area was used for yield estimation. The unit plots were inspected for cassava mosaic disease (CMD) symptoms at 6 MAP, and the disease severity scores were recorded for each cassava stand, by rating plants on a 1 to 5 scale, as described by Hahn et al. (Reference Hahn, Terry and Leuschner1980). Cassava was harvested 9 MAP for fresh tuber yields in accordance with the local practices. The experiments were closely followed-up using a participatory rural appraisal with a broad number of farmers from different villages in the regions (Dabou and Bingerville) and local extension officers from ANADER (Agence Nationale d'Appui au Développement Rural). During this process, two ISFM systems were selected for multi-location testing.
Multi-location experiments
The study area:
Multi-location trials were carried out for one cassava cropping season from May 2013 to February 2014 in southern Côte d'Ivoire, in two locations near the cities Dabou and Bingerville, where cassava cultivation is the main subsistence agricultural activity. A total of 10 farmers were selected from 4 villages in each location, namely Debrimou (05°21′484″ N, 04°22′518″ W, 37 m asl), Kpass (05°19′493″ N, 04°24′337″ W, 51 m asl), Opoyou (05°23′167″ N, 04°32′386″ W, 36 m asl) and Lopou (05°25′340″ N, 04°27′310″ W, 39 m asl) in Dabou; and Anan (05°19′711″ N, 03°53′002″ W, 8 m asl), Bregbo (05°18′446″ N, 03°49′646″ W, 16 m asl), Eloka (05°18′258″ N et 03°45′140″ W, 54 m asl) and Santaï (05°22′196″ N, 03°54′014″ W, 40 m asl) in Bingerville, based on their geographical position to take into account variation in soil conditions. The rainfall pattern in the area is bimodal, with a long rainy season from March to June and a short one from October to November, followed by a short and long dry season, respectively. Rainfall collected from the meteorological station of the local CNRA (National Center for Agronomic Research) during the growing season was 1032.0 and 961.7 mm in Dabou and Bingerville, respectively, most of which occurred in May and June. Annual average temperatures during the cropping season were 27.0 and 27.7 °C, respectively. Soils are described as acidic, sandy (84 to 96% sand) Ferralsols (USS Working Group WRB, 2014) low in organic matter, CEC and nutrients (see further).
Experimental setup:
Twenty cassava smallholder farmers were selected based on their involvement in the previous on-farm ISFM demonstration trials. Prior to the experiments, farmers were given technical training on the implementation of the selected ISFM systems. This support included technical operational procedures and estimation of labour and other input cost and yields. Two ISFM systems were implemented along with a current traditional cassava cropping system as control. Two contrasting CMD tolerant varieties, one early maturing, drought tolerant variety (Yavo) harvested at 9 MAP, and a late maturing variety (Bocou 1) with the ability to shade-out weeds harvested at 12 MAP, were used together with cowpea intercropping and the application of chicken manure (5 t ha−1 dry matter) and an adjusted dose (100 kg ha−1) of compound fertilizer 15-15-15 (in terms of N-P2O5-K2O), as further detailed below. The applied mineral fertilizer dose was adjusted to 100 kg−1 in the multi-location experiments for the low soil fertility. The traditional cultivation system (C) was characterized by the use of local cassava varieties without legume intercropping and with a variable cassava planting distance. In the ISFM mono-varietal system (S), the CMD resistant cassava variety Yavo was grown at 2 × 0.5 m spacing, combined with cowpea intercropping and the application of chicken manure (5 t ha−1). In the ISFM mixed-varietal system (M), both phenotypically different CMD resistant cassava varieties (Yavo and Bocou 1) were planted in alternating lines at 2 × 0.5 m spacing, combined with cowpea intercropping and both manure (5 t ha−1) and mineral NPK fertilizer (100 kg ha−1) application. Plant materials for the two improved varieties were provided to farmers by the national extension service (ANADER). The three systems were implemented in both locations by individual farmers under the supervision of extension officers from ANADER at the onset of the rains in May 2013. Research staff was involved during the harvest of legumes and cassava on ‘farmers’ day’ for data collection. In the fields, each ISFM system was assigned one block (50 × 14 m in size) in order to streamline the field operations and to allow estimation of labor input per system. The spacing between blocks was 3 m and complete field size was 50 × 50 m. The blocks for the ISFM systems were composed of eight lines of cassava. Blocks were further divided into three sub-blocks of 14 × 16 m in size, used as repetitions. For the ISFM systems, cassava stem cuttings of 20 cm length were slanted planted on flat land.
Description of treatments
The manure (5 t ha−1) was applied in a circular ditch of 10 cm radius around cassava plants 10 days after planting. In the mixed-varietal ISFM system, mineral fertilizer (100 kg NPK ha−1) was applied together with the manure. The nutrients added were equivalent to 74 kg N ha−1, 28 kg P ha−1 and 21 kg K ha−1 for manure and 15 kg N ha−1, 6.5 kg P ha−1 and 12.5 kg N ha−1 for mineral fertilizer. Cowpea was intercropped 45 days after cassava planting based on previous experience. Four rows of legumes (40 × 20 cm) were sown between two lines of cassava. The spacing between cassava lines and external cowpea rows was 40 cm (Figure 1). In the mixed-varietal system, the lines of cassava alternated between the two varieties. In the traditional system, stem cuttings of variable length, from one or several local cassava varieties were planted on ridges or flat land with a variable cassava planting distance.
Cassava and legume spacings for the ISFM mono-varietal (S) and mixed-varietal (M) systems; X = cowpea plant; ○= cassava variety Yavo; ⊗= cassava variety Bocou 1.
Methods of soil sampling and analyses
Before the start of the experiments, the fields were cleared by burning. Soils of the different fields were sampled (0–20 cm depth) immediately after clearing using an auger. The composite samples, composed of all soil cores of the same field, were air dried, grounded and sieved (2 mm) before analytical processing. The pH was measured using electrode glass in soil/water (1/2.5) solution. The content of organic carbon (C) was determined by colorimetric measurement of the amount of Cr2O7 not reduced by C, according to ‘Walkley and Black’ (Pansu and Gautheyrou, Reference Pansu and Gautheyrou2003). Total N was measured according to the Kjeldhal method, which mineralizes the organic matter at 300 °C. The available P was determined by a modified Olsen method using FNH4 0.5 N + CO3NaH 0.5 N at pH 0.5 for extraction and the amount of extracted P was determined colorimetrically at 882 nm. Ammonia acetate 1 M at pH 7 was used for extraction of exchangeable cations (Ca2+, Mg2+, K+, Na+) and the extract was analyzed by atomic absorption spectrometry (AAS). The properties of the soils are shown in Table 1.
Chemical soil properties (0–20 cm depth) for fields in Dabou and Bingerville; different small letters per column indicate significant difference for fields and different capital letters per column indicate significant difference for sites at p < 0.05; values between brackets are standard deviations; *CV (%) = coefficient of variation.
Data collection
Weeding frequency (i.e. single, double or triples weeding) for individual fields was recorded by the farmers. Cowpea was harvested at 3 months after cowpea-planting and their residues were applied as mulch between cassava plants. Cassava was harvested at 9 MAP, in accordance with local practices. Cassava and cowpea plants harvested from the centre of each sub-block were used for yield estimation. Cassava root yield was estimated on a fresh weight basis. Cowpea yield was determined on a sun-dried grain basis. The economic return or profitability of a given system was calculated by subtracting the production cost from the total income generated (Profitability = (Yield × Price – Production cost), based on farm gate prices of 0.07 and 0.6 Euro kg−1 for fresh cassava roots and sun-dried cowpea, respectively. The calculation of total cost of production included labour costs for all field operations (land clearing, tillage, planting, fertilizers and/or manure applications, weeding, harvesting) and an input cost of 0.6 and 0.03 Euro kg−1 for mineral fertilizer and manure, respectively. The cost of labour for the specific field operations was calculated by the national extension service, and for the study areas was set at 0.52 Euro hour−1.
Statistical analysis
Data on yields and revenue and profitability were subjected to statistical analysis using SPSS statistics tools (SPSS version 21). A General Linear Model (GLM) and least square difference (LSD) post-hoc analyses were used to compare factors’ main effects and their interactions at p ≤ 0.05. First, crop yield variability between replicated farmer's fields (n = 10) per location was assessed using the three sub-blocks per field as replicate. Treatment effect on crop yields, revenue and profitability were assessed using average farmer's field data per location (n = 10) as replicate. Furthermore, a stepwise multiple linear regression model and Pearson's correlation were used to determine the relationship between soil parameters, weed control and yield variability for each system.
RESULTS
Agronomic performance
There was a significant (p = 0.000) field variability for cowpea and cassava productions for the different cultivation systems in both locations (Table 2). The cowpea grain yields for the ISFM system ranged from 1.3±0.2 to 4.3±0.3 t ha−1 and 1.4±0.1 to 4.6±0.1 t ha−1 for the mono-varietal and mixed-varietal systems, respectively, in Dabou; and ranged from 1.2±0.1 to 4.7±0.6 and 1.5±0.1 to 4.9±0.6 t ha−1, respectively, in Bingerville. The cassava fresh root yields ranged from 3.1±1.7 to 13.7±0.5, 8.8±0.3 to 31.7±11.0 and 7.0±1.3 to 27.0±5.8 t ha−1 for the local, ISFM mono-varietal and mixed-varietal systems, respectively, in Dabou; and ranged from 2.3 ± 0.6 to 8.5±1.3, 6.8±0.8 to 17.1±5.4 and 6.6±2.0 to 19.0±9.4 t ha−1, respectively, in Bingerville.
Variability of cowpea dry grain and fresh cassava root yields (t ha−1) for local system (C), ISFM mono-varietal (S) and mixed-varietal (M) systems across fields; different small letters per column indicate significant difference for fields at p < 0.05 and values between brackets are standard deviations; *CV (%) = coefficient of variation.
There were highly significant (p = 0.001) differences in cassava fresh root yields between the different cultivation systems (Table 3). Cassava yield increased for the ISFM systems compared to the traditional system in both locations. A statistically similar cassava yield was observed for Both ISFM systems in both locations. Around Dabou, the average cassava fresh root yield was 6.5±3.4, 14.7±7.0 and 14.2±6.0 t ha−1 for the local, ISFM mono-varietal and mixed-varietal systems, respectively. Around Bingerville, the cassava fresh root yield was 5.7±2.4, 13.2±3.4 and 12.0±3.7 t ha−1, respectively. There were no significant (p > 0.05) differences in cowpea yields between the ISFM system S and M in both locations (Table 3). The average cowpea yield observed for both ISFM systems was 3.3±1.0 and 2.8±1.3 t ha−1 in Dabou and Bingerville, respectively.
Average cowpea yield and revenue, cassava yield and revenue, total system revenue and profitability for the local system (C), ISFM mono-varietal (S) and mixed-varietal (M) systems; different small letters per column indicate significant difference between systems at p < 0.05 and values between brackets are standard deviations; *CV (%) = coefficient of variation.
n.a. not applicable.
Economic performance
The average revenue and profitability for the different systems are shown in Table 3. There were significant (p = 0.000) differences between the cultivation systems for cassava revenue, total revenue and profitability in both locations. The profitability increased considerably for the ISFM systems compared to the local system. The average profitability in Dabou was 155±233, 2586±962 and 2653±901 Euro ha−1 for the local, ISFM mono-varietal and mixed varietal systems, respectively; and was 116±168, 2300±753 and 2307±785 Euro ha−1, respectively, in Bingerville. On average, 66% of profitability was generated by cowpea.
Factors driving yield variation
Soil properties showed variability between fields in both locations (Table 1). In general, soils are sandy (70 to 96% sand), acid (pH ranging from 3.3 to 5.4), low in organic matter content and cation exchanged capacity (CEC < 10 cmol kg−1). Comparing soil fertility parameters for both locations, fields in Dabou showed significantly higher values compared to Bingerville. The average values were 16.2 and 11.3 g C kg−1, 1.2 and 0.8 g N kg−1, 88.0 and 62.0 mg P kg−1, 0.05 and 0.03 cmol K kg−1, and a CEC of 7.0 and 5.0 cmol kg−1, in Dabou and Bingerville, respectively. The results of stepwise multiple linear regression analyses to describe cassava and cowpea yield via soil properties and weed control for the different cultivation systems in Dabou and Bingerville are shown in Table 4. In Dabou, cassava yield increased with CEC (β = 0.42) and weed control (β = 0.58) for the local system. For the ISFM mono-varietal system, cassava yield increased with CEC (β = 2.11) and SOC (β = 0.42), and for the ISFM mixed-varietal system with CEC (β = 1.61) and weed control (β = 0.42). In Bingerville, cassava yield was only increased by weed control for all cultivation systems. The regression coefficients were β = 0.74, β = 0.54 and β = 0.64 for the local, ISFM mono-varietal and mixed-varietal systems, respectively. Cowpea yield increased with SOC (β = 1.57) and weed control (β = 0. 97) in Dabou, and with soil N (β = 0.41) in Bingerville.
Results of a stepwise multiple linear regression between weed control, soil parameters, and cassava yield for the local system (C), ISFM mono-varietal (S) and mixed-varietal (M) systems in Dabou and Bingerville; *significance at p < 0.05; **significance at p < 0.001.
DISCUSSION
Selection of ISFM technologies
The ISFM technologies tested in the multi-location experiments were based on participatory research with farmers in Dabou and Bingerville from which the results are shown in Table S1 and S2 (see supplementary material online at http://dx.doi.org/10.1017/S0014479716000041). The criteria used in technologies selection included yields and profitability of the systems, resilience to pest and disease (CMD) and appreciation of cooked or processed cassava by the farmers, and the availability of resources. The most viable technologies were ‘groundnut + half manure dose’ in Dabou and ‘groundnut + NPK’ in Bingerville, which increased significantly both legume and cassava yields, in particular when variety Yavo was used. Overall, Yavo was found to be the best-bet cassava variety, and groundnut the best intercrop for cassava. However, groundnut intercropping was discarded for the multi-location trials by farmers due to the animal pest which is associated with its cultivation. Cowpea was instead selected as the intercrop for cassava, due to its improved grains and residue biomass production compared to groundnut. The cassava variety Bocou 1 was also selected by farmers for its vigorous growth with a good ability to shade-out weeds. Hence, a mixed-varietal system, including contrasting cassava varieties Yavo and Bocou 1, were used in the multi-location experiments.
Performance of ISFM technologies
The cassava yield improvement observed for both ISFM systems as compared to the local system could, to some extent, be related to the CMD resistant trait of improved cassava varieties (Yavo and Bocou 1) and susceptibility of local varieties. This could, however, not be quantified as it was not recorded by the farmers during the multi-location trials. Secondly, the intercropping of cowpea, the addition of organic or mineral fertilizer and the application of cowpea residues to the soil, together likely increased soil fertility and consequently increased cassava production. Third, the improved performance of ISFM systems compared to traditional methods could also be explained by the early maturity of cassava variety Yavo, which was a component of both ISFM systems. The ISFM mixed-varietal system, which combined both organic and mineral fertilizers and hence was expected to increase cassava production more, gave similar average yields compared to the mono-varietal system, where only 5 t ha−1 of manure was applied. The full production potential of this mixed-varietal system was likely not achieved because of the premature harvest of the late maturing variety Bocou 1. The premature harvest was a farmer's decision, which was unfortunately not controlled by the experiment. The early harvest of roots is a real challenge for improving cassava production in the study area. On one hand, this practice is driven by the increase in demand for cassava to feed the local population, as cassava is a staple food, and also people in the densely populated nearby city of Abidjan (economic capital). On the other hand, premature harvest is also driven by the scarcity of land available to expand cultivation. The high land pressure is a result of competition from industrial and more lucrative crops, such as oil palm, rubber and coconut, which occupy a large part of agricultural land and pushes cassava cultivation to small plots of land on mostly marginal soils. Nevertheless, when applied to regions with relatively low land pressure and allowing full maturity of cassava, this agricultural intensification system is expected to achieve higher cassava production and also to reduce labour, through the weed shade-out ability of variety Bocou 1 and by allowing that cassava harvest is made in 2 stages, between 9 and 15 MAP for variety Yavo and Bocou 1, respectively. This potential arises from the possibility to harvest the early maturing variety Yavo at 9 months and postpone the harvest of the late maturing variety Bocou 1 to between 12 and 15 MAP. This will not only help to alleviate the seasonal shortage of cassava roots, but will allow the Bocou 1 variety to reach maturity and to benefit more from the improved soil fertility provided by the intercropped cowpea.
Comparing the economic returns, the ISFM systems gave higher profitability compared to the traditional system, mainly due to the high market value of the cowpea (0.6 Euro kg−1). The dependence of the system's profitability on cowpea means that it could vary with the market value of legumes, which depends on the period and on demand. Cowpea intercropping could also have the benefit of allowing farmers to postpone cassava harvest (traditionally at 9 MAP in this area) by providing early revenue, i.e. 5 months before cassava harvest, thus solving one of the main obstacles for higher cassava productivity. In contrast, manure and mineral fertilizer that were needed for increasing cassava productivity are expensive commodities that are difficult to access by farmers. This could limit a broad scale application by smallholder farmers.
The large variation of cassava and cowpea yields between fields could be attributed to differences in inherent soil fertility and farmer management (weed control). Soils in the area are acid, low in organic matter content and CEC, and with high levels of P. The clearance of fields by burning, in accordance with the local practices, could have contributed to the high levels of phosphorus in the soil through ash depositions. Moreover, the level of P is also determined by the method of P extraction (modified Olsen method) used. CEC and weed control significantly influenced cassava productivity for all the systems, as revealed by regression and correlation analyses. Iyagba (Reference Iyagba2010) reported a 48 to 90% yield loss in cassava production as a result of weed infestation. Consequently, productivity of the cultivation systems and, in turn, farmers’ income is largely driven by inherent soil fertility and weed control. The low CEC of soils in the area likely restricted the release of potassium, which is required for building up the storage roots. The importance of potassium for the productivity of root crops such as cassava is well known and can be explained by its prominent role in the synthesis and translocation of carbohydrates (Abd El-Baky et al., Reference Abd El-Baky, Ahmed, El-Nerm and Zaki2010). The lack of significant correlations with soil parameters for all systems in Bingerville could be related to the very poor soil fertility exacerbated by poor weed control for many fields, which may have obscured any influence of soil properties on cassava productivity.
Cassava root yield in all the ISFM treatments was much higher than the 6 t ha−1 average cassava yield reported for the country (FAO, 2014). However, in the majority of fields and for both ISFM systems, cassava yields were lower than those previously reported for soils with similar characteristics to those in this study (texture with 80% sand, pH of 4.5, CEC of 3.8 cmol kg−1 and a total N of 0.09 %). The study by Anyaegbu et al. (Reference Anyaegbu, Ezeibekwe, Amaechi and Omaliko2009) observed yields of 56.4 to 69 t ha−1 fresh roots when cassava-groundnut intercropping was combined with 10 t ha−1 poultry manure application. High cassava fresh root yield (e.g. 32.2 t ha−1) was also observed by Amanullah et al. (Reference Amanullah, Alagesan, Vaiyapuri, Sathyamoorthy and Pazhanivelan2006) in cassava-cowpea intercropping combined using poultry manure (10 t ha−1). However, yields in both aforementioned studies were achieved at full maturity with different cassava varieties and higher doses of manure. For the same cassava varieties, our yields were lower than those achieved in the same region by cassava mono-cropping: i.e. 40 t ha−1 for Yavo and 27 t ha−1 for Bocou 1 (ANADER, 2014). However, these results were again achieved with a later harvest, i.e. 12 to 15 MAP. The profitability of the above-mentioned cassava mono-cropping system, i.e. 4962 Euro ha−1 for variety Yavo and 2500 Euro ha−1 for variety Bocou 1 was improved compared to the ISFM technologies which gave on average 2460 Euro ha−1, and with on average 66% of this total generated by cowpea. The improved economic returns of the cassava mono-cropping system was mainly due to the observed yield increases and improved market for cassava (0.1 Euro kg−1), and also included the revenue from cassava stems. This lower productivity of the improved varieties in the multi-location experiments can be attributed firstly to the premature harvesting of roots, and secondly to the competition for nutrients between cassava, cowpea and weeds in these low fertility soil. Furthermore, cassava productivity for the ISFM system could also have been hampered by N immobilization during the first 2 months after cowpea residue incorporation into the soil, as observed in incubation experiments (data not shown). Since the residue from the intercropped cowpea was applied as mulch at 4.5 MAP, N immobilization may have affected cassava production between 4.5 and 6.5 MAP.
CONCLUSION
The tested ISFM systems showed improved agronomic and economic performance over traditional systems, with an increase in cassava productivity and a large increase in profitability mainly via the revenue generated by the intercropped cowpea. The between field variation of cassava productivity was mainly explained by differences in soil CEC and weed control. The latter suggests that effectiveness of the tested ISFM technologies to increase cassava production also depends on local soil fertility, farmer's management and resource endowment to purchase fertilizer. The improved cassava production of the ISFM technologies over local systems was related to the CMD resistant trait of the component varieties, and to legume residues and manure or mineral fertilizer additions to counteract the low soil fertility. Moreover, potential productivity of the ISFM systems was likely largely hampered by the premature harvest of cassava and even a competition with cowpea. Nevertheless, implementation of the tested ISFM systems could be a key strategy to achieve sustainable agricultural intensification in the study area as yields and especially farmer's income could be spread and increased. However, only long-term application of such ISFM system could alleviate soil fertility reduction and prevent premature cassava harvest.
Acknowledgments
We would like to express our gratitude to the VLIR-UOS who supported the funding of this project (Own initiative, ZEIN2008SEL04) and provided an additional PhD grant of one year to Jean-Baptiste Gnélié Gnahoua.
SUPPLEMENTARY MATERIAL
For supplementary material for this article, please visit http://dx.doi.org/10.1017/S0014479716000041.
