Malawi is a country with an agriculture-based economy.

In  2015,  agriculture  accounted  for  30%  of   the   gross domestic product (GDP) and 80% of the export earnings (Malawi Government, 2015). In 2013, agriculture employed 64.1% of the work force. The country has 2.4 M-ha of under cultivation, mostly by smallholder farmers who cultivate an average of 0.64 ha of land. Of the agriculture GDP, 70% is from smallholder farmers (Malawi Government, 2016).  Agricultural production is almost fully dependent of rain-fed cultivation. There is one rainy season of 3 to 5 months per annum. Climate variability, particularly in the form of erratic rainfall is one of the major biophysical constraints to agricultural productivity (Challinor et al., 2007). Climate projections for Southern Africa to 2050 suggest an average increase in temperature by 2.12°C, a delay in the onset of the rains and more intense and widespread of droughts and floods (Cairns et al., 2013). CIMMYT (2013) noted that 40% of the area under maize in sub-Saharan Africa experiences drought stress, which causes yield loss of 10 to 25%. The effects of drought increase the risk of crop failure which becomes a strong disincentive to farmers to invest in chemical fertilizers which are widely known to have positive influence on crop productivity. Other main constraints to crop production include poor and declining soil fertility (Zambezi et al., 1993; Kumwenda et al., 1997; ICRISAT/MAI, 2000; Blackie and Mann, 2005; MoAIFS 2005) and insects’ pests, parasitic weeds and diseases (Kabambe et al., 2008; Kabambe et al., 2014; MoAIWD, 2012). For example, phosphorus levels range from sufficient to low with widespread deficiencies in nitrogen and organic carbon ranging from 0.8 to 1.5% on Malawian smallholders fields (Snapp, 1998). Thus, to overcome the widespread problems of soil fertility decline a more integrated soil fertility management (ISFM) approach is required. These include long term rehabilitation to build up soil fertility before crops respond to efficient use of applied nutrients (Tittonell et al., 2007). A major national intervention to redress the poor soil fertility problem has been the Farm Input Subsidy Program (FISP), which has been making fertilizers available at very low prices (GoM, 2012). The FISP also includes a component of legume seeds. 

In Malawi, grain legumes are increasingly growing in importance. The national export strategy identified groundnuts to be among the four crops in priority area one for the export of oil seed products (GOM, 2013). As a green manure source, grain legumes are an important climate adaptation intervention as they help retain soil water (Tisdale et al., 1985). They contribute directly to household food security, and to the household cash income. Legume systems can positively contribute to the nitrogen economy of soils through biological nitrogen fixation, BNF (Snapp, 1998; Nyemba and Dakora, 2010). Recent studies in Malawi indicate that  groundnut  can  fix between 21 and 124 kg/ha of N (Njira et al., 2012; Mhango, 2011). In Kenya, Ojiem et al. (2007) reported N fixation of 41 kg/ha under low rainfall and 124 kg/ha under high rainfall. Turner and Rao (2013) noted that while systems that apply N fertilizer have higher yields, they will be more impacted and have larger reductions in yields from climate change. However, Turner and Rao (2013) reported that even if impacted by periods of drought, these higher yields would still be higher than yields without fertilizer or with low inputs.  In a study involving maize planting dates, cultivars and crop nutrient management under low and high rainfall environments in Zimbabwe, Rurinda et al. (2013) reported that nutrient management had an overriding effect on crop production, suggesting that nutrient management is the priority option for adaptation in rain-fed smallholder cropping systems. 

Balaka is one of the districts in Malawi that are vulnerable to climate shock, particularly drought (GOM, 2006). The intensification of legumes in smallholder farming systems therefore has the benefits of diversifying food and income sources as well as the potential to increase soil N and increase water available. The studies in this report were therefore aimed at assessing the productivity of groundnut-maize rotations system in drought prone Balaka district, Southern Malawi.

Study sites and design

A two-year legume-maize rotation study was conducted in Ulongwe Extension Planning Area (EPA) in Balaka district in Machinga Agricultural Development Division (ADD) in southern region of Malawi. Specifically, experimental sites were located in four sections of the EPA, namely Chibwana Nsamala, Hindahinda, Chitseko, and Mulambe. Being a field rotation study, field plots were established in 2015/2016 as the first season. In this season, pure stands of groundnuts were planted in fields of famers designated as lead (0.1 ha) or follower (0.05 ha) farmers and a designated density of 8.88 plants m^-2. The farmers were provided with basic seed and trained on good agricultural practices. The groundnut variety used was CG7 which has maturity period of 130 to 150 days and yield potential 2,500 kg/ha. Farmers were trained and supervised to ensure that recommended planting geometry of 75 cm between ridges, 15 cm between station, and 1 plant per station (MoAFS, 2012) were followed and that residues were incorporated. In the second season (2016/2017), five fertilizer treatments were imposed as shown in Table 1. These fertilizer rates and packages represented choices available and recommended to farmers based on the fertility of their area (MoAIFS, 2012). In addition, this was aimed at improving the teaching value of the studies.  Plots had 4 rows and 6 m × 0.75 m apart (18 m²), giving an expected density of 5.33 plants m^-2. Yield, plant count data were recorded from the two middle rows plot. All five treatments were randomly laid out in one field. The design was thus a randomized block with a farmer as a replicate.

Data collection and number of farmers involved   

In the first year, 132 famers hosted groundnut plots, also designated as Learning Centres (LCs) in four sections. In each village, there was one lead famer and ten follower farmers. Of these 10 famers, all lead famers and 3 follower famers were sampled, for a targeted total of 48 farmers. From these famers, data were collected on soil, yield and stover to provide a basis for understanding the year two results. In year two, maize was grown as rotation crop on the same 48 farmers. This study design was randomized complete design with each farmer as replicate. Grain moisture content was recorded at harvest, and maize grain yields reported were adjusted to 12.5% storage moisture levels.

Data analysis

For the baseline year, data was summarized into means and standard deviations for each section and by legume crop type. For the year two data, analysis of variance was done on yields from legume maize plots and continuous maize separately using the structure sections×fertilizer level for each section.  

Gross margin analysis computations

The computation for gross margin analysis involved determination of the difference between gross income from sales and production costs. The gross income was based on produce sales quoted at farm gate with negligible marketing cost. This is the actual situation in Malawi whereby small or big traders mount buying points in rural areas. The cost related to marketing is packaging which comprise a new sack for each 50 kg of harvest. The labour costs for groundnuts included land preparation,   planting, weeding, stripping, shelling and grading while for maize these have included land preparation, ridging, planting, shelling, cleaning and packaging. These are basic components described by several authors (Dzanja, 2008; Ngulube et al., 2001; Takane, 2008). Dzanja (2008) estimated the total labour requirement to be 240 man-days for groundnut and 139 man-days for maize and these were used in the calculations. However, Ngulube et al. (2001) estimated the labour requirement for groundnut to be 637 man-days, while Takane (2008) estimated labour requirement for maize to be 176 man-days. Tables 2 and 3 show the total costs for groundnuts and maize at the five rates of fertilizer. The costs of inputs  and  labour  use  were those of the 2017/2018 cropping season in Malawi. At this time the exchange rate of the Malawi Kwacha to US $ was 1: 733 (June 2018 Newspapers). The labour was costed based on the minimum wage daily rate of MK 962,00 for the time of computation (June 2018). Calculations were made for different price and output scenarios. Breakeven yield was determined by dividing total costs of production by the price level. 

First year soils, groundnut stover and grain yield baseline results

Results of rainfall from four measuring points are shown in Table 4. The total rainfall, ranging from 326 mm at Chombe village in Chibwana Msamala village in Chibwana Msaamala section of 527 mm at Kalembo 1 village in Chitseko was much less than normal rainfall. The long term annual rainfall for the EPA is 840 to 1000 mm. The rainfall was erratic, with up to 3 dry spells (periods with at least 10 days of no rain or trace rainfall) recorded in two sections, and 2 dry spells in another.  Kalembo 1 village in Chitseko section had no dry spells as well as the highest rainfall. 

Results of soil properties, stover and grain yield are shown in Tables 5 and 6. The results largely show poor soil fertility status, based on Chilimba and Mkosi (2014) thresholds. From the raw data (data not shown) all soil pH_w values were above 6.0 (neutral category) in all sections except Mulambe where values just below 6.0 were observed. All soils were very low in phosphorus and potassium with values <8.0 ug/g and <5.0 (very low categories). In terms of organic matter, soils were mostly in low (<2.1%) to medium (2.1 to 3.9%) category. For nitrogen, soils belonged to very low (<0.08%), medium (0.08 to 0.12%), low or medium (0.12 to 2.0) categories. All soils were very high in zinc (>3.0 ug/g). According to Chilimba and Mkosi (2014), these soils would  require  40 kgha^-1 of P₂O₅, 30 to 60 kgha^-1 of K₂O, and 46 to 92 kgha^-1 of N. However, sulphur, a potential important element, not determined.  While all soils were low in fertility, the yields were quite variable. Average grain yield was the highest (1016 kgha^-1) in Hindahinda section and lowest (303 kgha^-1) in Mulambe.  Of the individual plot grain yield (raw data not shown), the results showed that 27% of the 43 Learning Centres (LCs) studied obtained very low yields of >300 kgha^-1, while 24% obtained yields >1,000 kgha^-1 (Table 7). The target plant density in the study was 8.88 plants m^-2.  However, up to 22% of the 46 LCs had ≤50% of targeted plant stand of 8.88 plants m^-2, 37% achieved a plant density of between 50 and 60% of the desired plant stand suggesting poor establishment. This was most likely due to dry spells. There could be other soil factors too, particularly those linked to water holding capacity of soils. The reasons for poor stand were not studied. Table 8 shows the close association between grain and stover yield with the plant density categories.

Maize results in year two

Results on maize yield response to fertilizer rates in rotation with groundnuts or under continuous maize are shown in Tables 9 to 12. Significant treatment differences were detected in all the sections. The pattern of response was linear in all cases (Table 13). The incremental benefits due to groundnut rotation and residues incorporation varied according to the section. The benefits were highest in Chitseko section (range 233 to 732 kg/ha), followed by Chibwana Nsamala section (range 31 to 253). In Hindahinda, the results varied with some negative differences as well. There were no records obtained from Mulambe section.

Gross margin and break even yields

The gross margin for groundnuts were determined at four levels of production (300, 500, 100 and 1,500 kg/ha) and thrice price scenarios (MK 250, 350 and 500) to reflect the actual level obtained in the study. The results in Table 14, as expected, show that positive gross margins were only found and at yield levels of 1000 to 1,500 kg/ha and the K350 or K500 price scenarios. These yield levels were associated with plots that had high crop establishment (Table 8).  Using  a  median  total  variable cost value of K334,120, the break even yield was 1336, 955 and 668 kg/ha for the MK250, 350 and 500 price scenario. For maize, positive gross margins were only possible at yields equal or above 2,500 which were associated with a rate of 46:23:0:4 or higher (Table 15). The break even yields were 1467, 1698, 1817, 1938, and 2058 for nil to 92:23:0:4 rate, respectively.

This suggests that nutrient and non-nutrient factors were important, such as slope of land (not monitored), planting dates. There was significant regression relation between groundnut density and grain yield. Most of the fields recorded plant stand much lower than the expected stand of 8.88 plants m^-2. Thus, low establishment could be the

main reason for low yields of groundnuts. The poor establishment is most likely due to dry spells experienced in the area. Being a large seeded crop, groundnut requires good soil moisture for establishment (MoAIWD, 2012). The result on relationship between plant stand and yield is in agreement with Mhango et al. (2017) who reported that plant density was one of the drivers of biological nitrogen fixation in groundnuts. Most of the variation may be explained by variation between fields. Edmeades et al. (2000) reported that for fields varying in  topography, texture and thickness of top soil, yields may vary ten-fold. 

For groundnuts, poor plant establishment was a key driver for yield. In this study, treated basic seed of groundnuts was provided and used. Hence, the reasons for poor establishment are likely to be the dry spells. Timing of planting relative to planting rains is important and this may be improved through provision of rainfall forecasting services and skills to determine moisture adequacy in soil. Possible ways to improve establishment include adoption of in-situ rain water harvesting practices such mulching, box ridges, and manure. It is recommended that all possible options to increase establishment should be tested and rolled out.

While there are no recommendations for nutrient application in groundnuts in Malawi, several studies have shown responses of P fertilizer application in groundnuts (Tarawali and Quee, 2014; Dakora, 1984). Mhango et al. (2017) reported that P was a key driver to BNF. It is recommended that further studies should be conducted to determine the role of P and other elements to increase yields of groundnuts in the area conducted. In a review, Chianu et al. (2011) highlighted several factors, including high soil temperatures, soil moisture stress, and P deficiency as important for groundnut yield.

Year two maize results

While responses to fertilizer were significantly different with or with residue incorporation, the yield levels of ≤4 tha^-2 are still low as compared to potential of 5 to 10 t/ha for expected farmers’ fields (MoAIWD, 2012). The low yields are expected as the soil analysis results showed that the soils were low in P and K. Higher rates would be required for higher yields (Chilimba and Mkosi, 2012). These soils would require 40 kg/ha phosphorus, 30 to 60 kg/ha of potassium and 46 to 92 kg/ha of nitrogen (Chilimba and Mkosi, 2012). The incremental benefits due to legume residue incorporation varied with section, with the highest benefits noted of 200 to 730 kg/ha recorded from Chitseko section. While the contribution legume residues to subsequent is well documented (Ngwira et al., 2012; Mwato et al., 1999; Mhango, 2011). Inconsistencies in maize response to legume rotations have previously been reported (Ngwira et al., 2012).

Gross margin analysis and breakeven yields

The gross margins determinations showed that profitability is higher at higher yield levels, which were associated with higher plant establishment in groundnuts and higher fertilizer rates in maize. Famers may reduce cost of groundnut seed by recycling their original certified seed. However, the value of insecticide or fungicide seed treatment with purchased seed may be lost. The best ways to increase yield remain good agricultural practices, such as timely planting and weeding, and proper plant density as discussed earlier.

The results have shown that general low yields in both groundnuts and maize are common and a constraint to profitability. For groundnuts, poor plant establishment was a key driver for yield. It is recommended that all possible options to increase establishment should be tested and rolled out. Further options to increase yields in groundnuts should be investigated, including application of P fertilizers. As the results have shown a linear response to fertilizer application in maize, an agronomic optimum could not be determined. While current fertilizer recommendation can be maintained, further studies on N response and their interaction with P should be added.

The authors have not declared any conflict of interests.

The authors thank the Norwegian Government for funding this work through its funding to the Capacity Building For Managing Climate Change program at Lilongwe University of Agriculture and Natural Resources in Malawi. They also thank staff and farmers of Balaka District Agriculture Office and Ulongwe Extension Planning Area for collaboration in this project.