Land degradation, population growth, and chronic poverty in Eastern and Southern Africa challenge the sustainability of livelihoods for smallholder farmers. These farmers often manage soils depleted of nutrients, apply limited amounts of mineral fertilizer, and take decisions about their cropping systems that involve multiple trade-offs. The rotation of cereals with legumes bears agronomic and ecological merit; however, the socio-economic implications of the cereal-legume rotation require a deeper understanding. This study explores the yield, labor, profit, and risk implications of different legume and mineral fertilizer practices in maize-based cropping systems in central Malawi. Our method involves coupling crop modeling and an agricultural household survey with a socio-economic analysis. We use a process-based cropping systems model to simulate the yield effects of integrating legumes into maize monocultures and applying mineral fertilizer over multiple seasons. We combine the simulated yields with socio-economic data from an agricultural household survey to calculate indicators of cropping-system performance. Our results show that a maize-groundnut rotation increases average economic profits by 75% compared with maize monoculture that uses more mineral fertilizer than in the rotation. The maize-groundnut rotation increases the stability of profits, reduces the likelihood of negative profits, and increases risk-adjusted profits. In contrast, the maize-groundnut rotation has a 54% lower average caloric yield and uses more labor than the maize monoculture with mineral fertilization. By comparing labor require- ments with labor supply at the household scale, we show for the first time that the additional labor requirements of the maize- groundnut rotation can increase the likelihood of experiencing a labor shortage, if this rotation is undertaken by farm households in central Malawi. We demonstrate that risk and labor factors can be important when examining trade-offs among alternative cropping systems. . . . . . Keywords Agricultural households Cropping systems Groundnut Maize Synergies Trade-offs 1 Introduction soil fertility. Focusing on improving the productivity of cropping systems is a long-standing, though still relevant, ap- Maize is the most commonly grown staple crop in Eastern and proach to improve the livelihoods of farmers who face declin- Southern Africa. Historically, cropping systems for smallhold- ing soil fertility (Tittonell and Giller 2013). Combining le- er farmers (hereafter farmers) in this region often included gumes and mineral fertilizer (hereafter fertilizer) can help long fallows, which allowed soils to replenish their nutrients maintain farmer productivity and profits (Chianu et al. 2012; and in turn maintain crop productivity. In densely populated Onduru and Du Preez 2007; Chianu et al. 2011). In Malawi, rural areas such as in central Malawi, high population pressure maize-groundnut rotations with fertilizer (Fig. 1) are often has strongly reduced the use of fallow, and farmers often prac- more productive than maize monocultures (Thierfelder et al. tice the continuous cropping of maize (Thierfelder et al. 2013; Snapp et al. 2010; Ngwira et al. 2012). 2013). This practice of maize monoculture in turn has reduced Despite the often-observed productivity benefits of using alternative practices related to legumes and fertilizer in maize- based cropping systems in Eastern and Southern Africa * Adam M. Komarek (Droppelmann et al. 2017), multiple factors can influence their attractiveness. One factor is the availability of agricultural labor to meet the labor requirements for these practices International Food Policy Research Institute, Washington DC, USA (Ngwira et al. 2012). Practices are often developed and tested 32 Page 2 of 10 Agron. Sustain. Dev. (2018) 38:32 Fig. 1 Malawian maize-groundnut rotation. Source: https://flic.kr/p/c7f7Vj. Photo credit: T. Samson/CIMMYT at the field scale, although farmers often encounter constraints affect profits. For example, Ngwira et al. (2012) and Ngwira to using these practices at the farm or household scale. This et al. (2014) used average grain prices to compare profits in recognition of constraints has led to different initiatives, such conventional and conservation agriculture systems. A cross- as the Soil Health Consortia for Eastern and Southern Africa, sectional study in Malawi showed that diversification of maize seeking to identify the socio-economic feasibility of different monoculture into a maize-legume system and using reduced technologies under a range of agro-ecological conditions. tillage can increase crop yields and reduce downside risks Weather and price variability are other, risk-related, factors (Kassie et al. 2015). Integrating legumes into maize monocul- that farmers encounter. tures and using fertilizer can increase productivity and partial Some labor and risk studies related to legume and fertilizer profitability (Ngwira et al. 2012; Snapp et al. 2010). Ortega et practices in maize-based cropping systems exist in Eastern al. (2016) urged additional research on risk and labor in the and Southern Africa. For example, Rusinamhodzi (2015) context of maize-legume systems. clustered farmers in Zimbabwe based on resource endow- Based on the above-mentioned studies, we see scope ments, including labor availability, and showed how digging to improve our understanding of the economic, risk, and planting basins (a reduced tillage method) increased labor de- labor effects of planting legumes and applying fertilizer mands. This increase seems in contrast with the desire to find in maize-based cropping systems. Our study aims to pro- technologies that simultaneously reduce labor demands and vide useful insights into the potential labor, economic, improve soil fertility. In Malawi, Thierfelder et al. (2015a) and risk effects of changes in cropping practices for and Ngwira et al. (2012) focused on labor requirements, rather farmers in central Malawi. Many of these previous stud- than labor availability, to highlight the labor-saving effects of ies use a partial economic budgeting approach, which alternative practices that fall under conservation agriculture, often only considers the gross value of production and which includes growing legumes, such as cowpea or pigeon associated financial costs. We complement the existing pea, in rotation with maize. Ortega et al. (2016)used choice literature by considering the opportunity cost of labor experiments in central Malawi to show that labor demands are in our economic profit calculations. We supplement stud- a major constraint to legume adoption. Studies have also ies that use cross-sectional household survey data shown that maize can have a greater labor-use efficiency than by including a risk analysis based on variability in both legumes such as groundnut (Franke et al. 2014), and that con- grain yields and prices over time, thus adding a temporal servation agriculture can also raise labor-use efficiency com- dimension to trade-off analysis. We complement the pared with conventional agriculture (Thierfelder et al. 2015b). above-mentioned studies on labor-use efficiency by ex- However, simultaneously examining labor supply and de- amining labor balances. Our study aims to answer two mand for different cropping systems can help highlight possi- questions: ble trade-offs associated with maize-legume integration at the household scale. For risk, studies considering weather vari- 1. How do different legume and mineral fertilizer practices ability in Malawi have shown that systems with legumes can affect productivity, labor use, profit, and risk? increase yield stability (Ngwira et al. 2012,2014). Some of 2. Do farmers access enough agricultural labor to sustain a maize-groundnut rotation? these studies have shown how maize-legume rotations can Agron. Sustain. Dev. (2018) 38:32 Page 3 of 10 32 2Methods These systems matched those in a Golomoti field experi- ment reported in Smith et al. (2016). To match the protocol in 2.1 Background the field experiment, each simulated system omitted the appli- cation of manure and removed 70% of crop residues from the Malawian farmers typically grow maize monoculture (contin- field. The four cropping systems reflect a mix of current and uous cropping of maize) often rotated or intercropped with alternative farmer practices and government recommenda- legumes or sometimes rotated or intercropped with cassava tions. For example, the Malawian government recommends −1 or cash crops. Maize consistently occupies over 70% of culti- applying 69 kg [N] ha to maize (Mungai et al. 2016). vated land in Malawi, with groundnut the most commonly We simulated yearly crop yields over time to help account grown legume (FAO 2017). Fallows are rare in Malawi for temporal weather variability and the cumulative effects of (Mungai et al. 2016). Farmers often apply fertilizer to their on-farm practices on productivity. We calibrated DSSAT with fields (Mungai et al. 2016). Farmers have a limited ability to local data on soils, crop cultivar characteristics, and manage- use manure as an organic source of nitrogen because they keep ment of the crop(s) for the conditions of the study area. Model minimal livestock. calibration data were taken directly from Smith et al. (2016). Our study focused on the Golomoti Extension Planning Soil data included, among others, soil texture (% sand, silt, Area of central Malawi. This area is in the lakeshore zone and clay), soil % carbon and nitrogen, soil pH, Bray P (ppm), within the Dedza district and is approximately 500 m above plant available water capacity (mm), and fraction of organic sea level. Seasonal precipitation averaged 734 mm from 1989 carbon in microbial biomass for each standard soil profile to 2010, with a coefficient of variation (C.V., defined as the layer depth (0–15 centimeters (cm), 15–30, 30–60, 60–90, ratio between the standard deviation and average) of 0.16 and 90–120 cm). Calibration data for crop cultivar and crop suggesting a relatively stable precipitation. We characterized management included crop residue use, planting density, households in Golomoti with household survey data collected planting depth, fertilizer applied, and rotation. Genetic as part of the Africa Research In Sustainable Intensification growth coefficients for our DSSAT simulations of maize for the Next Generation (Africa RISING) program (IFPRI variety SC403 included P1 195.4, P2 0.852, P5 809.1, G2 2015), which included 121 households. The survey was con- 607.2, G3 8.11, and PHINT 31.74. Ruane et al. (2015)sup- ducted in the summer of 2013, with data referring to crops plied the daily weather data. grown between October 2012 and May 2013. The survey The performance of DSSAT has been previously evaluated collected data on family size, grain yields, areas cultivated, in maize-based systems in Eastern and Southern Africa and inputs and labor used for cropping activities (at the (Ngwira et al. 2014). To evaluate DSSAT in our study, we household-crop-activity level). compared simulated grain yields with observed grain yields reported in Smith et al. (2016) from the 2012–2013 and 2013– 2014 growing seasons. This included experimental data on 2.2 Cropping system simulations maize yields in MM0, MM69, and MG, and groundnut yields in MG. We calculated the normalized root mean squared er- We used the Decision Support System for Agro-technology ror—a quantitative measure of the deviation of simulated data Transfer (DSSAT) model v4.5 (Jones et al. 2003)to simulate from observed data. crop yields for 22 years from 1989 to 2010 to generate data on grain yields and nitrogen-use efficiency. Indicators (discussed 2.3 Cropping systems indicators in section 2.3) were calculated based on simulated data, household survey data, and price and cost data from secondary In this study, cropping system refers to either maize monocul- sources. We simulated four cropping systems, the first three ture or the maize-groundnut rotation. The field scale refers to are the continuous cropping of maize with differing nitrogen the cropping system simulated in DSSAT on a per hectare [N] fertilizer application rates: basis. The household scale refers to the cropping system sim- ulated in DSSAT on a per-farm-household basis. At the house- 1. Maize monoculture with no fertilizer applied (MM0— hold scale, we allocated the available arable land to each of the unfertilized); simulated cropping systems with the total area planted equal −1 2. Maize monoculture with 35 kg [N] ha of urea fertilizer to the observed area the household planted to maize and le- applied (MM35—moderately fertilized); gumes. Household data in Golomoti suggest that farmers al- −1 3. Maize monoculture with 69 kg [N] ha of urea fertilizer locate approximately 80% of their arable land to either maize applied (MM69—intensely fertilized); and or legumes (Mungai et al. 2016). The indicators we calculated −1 4. Maize-groundnut rotation with 35 kg [N] ha of urea for each system (discussed in sections 2.3.1 and 2.3.2)related −1 fertilizer applied to maize and 12 kg [N] ha of urea to caloric yields, nitrogen-use efficiencies, labor balances, fertilizer applied to groundnut (MG). profits, and risks. We calculated all the indicators at the 32 Page 4 of 10 Agron. Sustain. Dev. (2018) 38:32 field-scale level, except for labor balance, which we calculated time for fertilizer application and the quantity of fertilizer ap- at the household scale. Groundnuts can be sold either shelled plied to calculate the household-specific time taken to apply or unshelled, and shelling increased both labor use and sales 1 kg of fertilizer. The time spent per kilogram was then used to price. Thus, we considered five systems from an economic calculate labor use for the differing fertilizer quantities, which perspective: (1) MM0, (2) MM35, and (3) MM69 defined in varied by system. Household data are for shelled groundnuts, section 2.2, and MG for (4) shelled groundnuts (MGS), and the common practice for sales, although a market does exist (5) unshelled (MGUS) groundnuts. for unshelled groundnuts. To differentiate labor used for shelled or unshelled groundnuts, shelling required −1 20 days ton of groundnut grain (Waddington et al. 2007). 2.3.1 Productivity and labor indicators Equation (2) defines labor supply at the household scale for each season (LS ). h,s −1 We simulated grain yields (kg ha ) from each crop in each system and calculated the caloric yield of each system MA þ LA h h LS ¼ LF ðÞ 30 3 −OFL ð2Þ −1 −1 h;s h;s h;s (kcal ha ). Maize contained 357 kcal 100 g and groundnut TA −1 contained 549 kcal 100 g (FAO 1968). Nitrogen-use effi- In Eq. (2), labor supply was calculated by multiplying the ciency for maize was calculated as the ratio of grain yield to reported available family labor per household (persons aged > nitrogen fertilizer applied, which measures the partial factor 15 and < 65 years, LF ) by the days available to allocate to productivity of nitrogen fertilizer. h,s the simulated cropping system. Each working family member We used household survey data to calculate agricultural −1 had30daysmonth to allocate to either the simulated labor demand in each system and household labor available −1 −1 −1 cropping system, other farm activities, or off-farm work (both in days season household ). Equation (1)defines (OFL ). We calculated the days each household allocated to agricultural labor demand for each cropping system at the s off-farm work, based on reported off-farm income and local household scale (LD ). Season refers to one of the four h,s wages. The days spent on other farm activities were propor- (meteorological) seasons: summer (December, January, and tional to the area allocated to crops other than maize and February), autumn (March, April, and May), winter (June, legumes, i.e., 19% of the average household’s total arable land July, and August), and spring (September, October, and (TA ). The remaining days in each month (30 minus other November). farm activities minus off-farm work) were available for each A C worker to allocate to the simulated cropping system. The dif- LD ¼ ∑ ∑ L ðÞ MA þ LA ð1Þ h;s a;c;s;h h h a c ference between family labor availability (LS )and totalla- h,s bor demand (LD ) provides an insight into which systems h,s In Eq. (1), L is the reported days spent on each a,c,s,h might have a labor deficit. cropping activity (a), for each crop (c), in each season (s)by −1 each household (h) in person-days ha . The observed house- hold area (in ha) of maize is MA and legumes is LA .The 2.3.2 Economic and risk indicators h h combined maize and legume area was, on average, 81% of the −1 total 1.06 ha of reported arable land (TA ). Thus, we had an h We calculated two measures of field-scale profit (US $ ha ) indicator of labor demand for the five economic systems at the in each year (y) for each system: financial profits (FP )shown household scale in each season (LD ). The household survey h,s in Eq. (3) and economic profits (EP )shown in Eq.(4). collected data on labor use for crop activities (a) including land preparation, weeding, herbicide application, fertilizer ap- FP ¼ ∑ GY P −ðÞ ðÞ QF FþðÞ QS Sð3Þ y c;y c;y c c plication, organic matter application, pest control, and harvest- A C ing and post-harvest activities. The survey asked labor de- EP ¼ FP −∑ ∑ L w ð4Þ y y a;c mands for harvesting and post-harvest activities as a single a c value. Labor activities for maize and groundnut occur in dif- ferent months. For maize, June–July: incorporation of resi- Financial profits equaled the value of grain production, a dues (clearing), August–October: incorporation of residues multiplication of the grain yield of c in y (GY ) and its market c,y and ridging, November–December: ridging, planting, price (P ), minus associated financial costs. Financial costs c,y weeding, and fertilizing, January–February: weeding and fer- equaled the sum of the unit cost of fertilizer (F) and fertilizer tilizing, and March–April: harvest. For groundnut, May–July: quantity applied (QF ), plus the sum of the unit cost of seed harvesting and clearing, August–September: post-harvest, (S ) and the quantityofseedused(QS ). Our study used grain c c November–December: planting, and January–February: prices in Golomoti markets from 1989 to 2010, supplied by weeding. To differentiate labor used in each simulated system the Malawian Agricultural Market Information System (IFPRI −1 by the amount of fertilizer applied, we used farmer-reported 2013). Nitrogen fertilizer cost 0.67 US $ kg (Franke et al. Agron. Sustain. Dev. (2018) 38:32 Page 5 of 10 32 2014). Seed costs were calculated from the household survey receive to forgo an uncertain profit. We calculated certainty −1 −1 and were 0.44 US $ kg for maize and 0.43 US $ kg for equivalents with the method in Lehmann et al. (2013). groundnut. Economic profits were financial profits minus the opportunity cost of labor, the latter defined as the implicit 1 r CE ¼ EP− V ð5Þ value of labor computed based on labor used—L from Eq. a,c 2 EP −1 (4)—and the daily wage (w). The wage was 1.33 US $ day In Eq. (5), r is the Arrow-Pratt relative risk aversion coef- (Franke et al. 2014). We maintain that the household first used family labor to meet labor demand and, if the labor balances in ficient and V is the variance of field-scale economic profits, and here, variance is the square of the standard deviation. The section 2.3.1 identified a negative labor balance, the house- hold hired labor to meet the deficit. Hired labor had a 20% analysis of certainty equivalents only considers systems with a higher wage than the local wage, to account for transaction positive average economic profit. Equation (5) implies con- costs. Price and cost data were adjusted for inflation with the stant relative risk aversion. We computed the certainty equiv- Malawi Consumer Price index for a base year 2013 with an alent of each system with calculated values for expected profits and their variance, similar to Lehmann et al. (2013). exchange rate of 1 US $ = 150 Malawian Kwacha. Therefore, grain prices varied each year, whereas seed costs, fertilizer We calculated the certainty equivalent for r equal to zero (in- difference to risk) and r equal to one (moderate risk aversion). costs, and wages were fixed over time in real inflation- adjusted US $. We calculated the net present value of econom- ic profits in each system from 1989 to 2010 using a discount rate of 6% per year. Net present values incorporate both the 3 Results and discussion timing and magnitude of economic costs and benefits, which is important if profits change over time. We calculated four indicators for economic risk in each 3.1 Household characterization system: the stability of profits, the probability of returning a positive economic profit, the average of the lowest 10% of Table 1 summarizes the Golomoti household survey data. Households had on average 1.06 ha of arable land of which profits (Conditional Value at Risk), and the certainty equiva- lent. The C.V. was used to measure the stability of profits. 53% was maize and 28% was legumes, with the remaining land planted to a variety of crops such as cotton and sweet potato. Next, we calculated the probability of a system generating a positive economic profit. Third, the Conditional Value at Risk The legume area (as a percentage of all arable land) was com- parable to the 30% at the national level (FAO 2017). Maize of the lowest 10% of possible economic profits was calculated −1 to measure the downside risk of extreme loss associated with yields averaged 1.6 t ha (C.V. = 0.7) compared to the national −1 average of 2.1 t ha in 2013 (FAO 2017). Maize yields in unfavorable events. The Conditional Value at Risk is the av- erage of the lowest 10% of economic profits for all the simu- Malawi take on a wide range. For example, Tamene et al. −1 lated years. Finally, we used Eq. (5) to calculate the certainty (2016) report maize yields to range between 0.4 and 12 t ha in Dedza district. About 91% of households used fertilizer. For equivalent (CE) of each system. The certainty equivalent is a risk-adjusted measure of profits, defined as the difference be- maize, the fertilizer application rate ranged from zero to −1 157.9 kg [N] ha , with average rates in Table 1 similar to rates tween expected profit (EP) and a risk premium (RP), i.e., CE = EP − RP (Antle 1987). Here, EP is the average of yearly in Mungaietal. (2016). Limited off-farm earnings and live- stock assets were reported, buttressing calls to improve crop economic profits (EP ). The certainty equivalent represents the smallest amount of certain money a farmer is willing to productivity as a livelihood improvement strategy. Table 1 Summary of household Indicators Average C.V. Minimum Maximum survey data −1 Total arable land (ha ) 1.06 0.64 0.10 3.24 −1 Maize area (ha ) 0.56 0.69 0.039 2.12 −1 Legume area (ha ) 0.30 0.94 0 1.62 −1 Maize yield (kg ha ) 1553.3 0.68 159.4 5208.3 −1 Fertilizer applied among users (kg [N] ha ) 43.2 0.91 10.2 157.9 Off-farm income ($) 36.7 1.39 0 297.1 Tropical livestock units (number) 0.41 1.35 0 2.70 Data are reported at the household level and based on 121 households. N represents nitrogen. One tropical livestock unit equals a 250-kg liveweight ruminant. C.V. coefficient of variation 32 Page 6 of 10 Agron. Sustain. Dev. (2018) 38:32 −1 The average labor used (person-days ha ) was 217 rotation with groundnut (MG) had an average simulated yield −1 (C.V. = 0.62) for maize and 355 (C.V. = 0.70) for groundnut. of 3114 kg ha (C.V. = 0.14), which was approximately −1 −1 The observed labor used for maize and groundnut broadly 1tha more than the 2063 kg ha yield in the maize mono- −1 concurs with other calculations of labor use in central culture with 35 kg [N] ha (MM35). Two factors helped −1 Malawi, for example, Franke et al. (2014). Land preparation explain the 1 t ha yield benefit in the rotation. First, ground- was a major use of time for both crops. For groundnuts, the nut’s average biological nitrogen fixation rate of −1 −1 average household spent 91 days ha for weeding and 117kg[N] ha helped increase the nitrogen content of soil −1 87 days ha for harvesting and post-harvest activities, where- in the rotation. Despite, in general, much of the nitrogen fixed −1 as, for maize, weeding used 61 days ha and harvesting and by legumes being removed from the system in high-protein −1 post-harvest activities used 31 days ha . seed, a net residual contribution of fixed nitrogen to the nitro- gen content of soil often exists. Our simulated nitrogen fixa- 3.2 Productivity and labor indicators tion rate for groundnut was within the range reported in other Malawi studies (Mhango et al. 2017), which were broadly Comparing our simulated grain yields for two maize mono- comparable to our study. Second, in our study, we retained 30% ofcropresidues(section 2.2), including the green resi- cultures (MM0 and MM69) and the rotation (MG) with ob- served yields reported for these systems in Smith et al. (2016) dues from groundnut. Green residues from groundnut often produced a normalized root mean squared error of 14%. contain large amounts of nitrogen at harvest and can supply −1 Simulated grain yields averaged 4447 kg ha in the maize more nitrogen for subsequent crops and other grain legumes −1 monoculture with 69 kg [N] ha (C.V. = 0.13) (Fig. 2), with such as soybean. Consequently, the simulated nitrogen uptake −1 −1 yields at least six times lower in the unfertilized maize mono- by maize averaged 77 kg [N] ha y in the rotation (MG) −1 −1 culture. Simulated grain yields in the intensely fertilized maize and58kg[N] ha y in the moderately fertilized monocul- ture (MM35). monoculture (MM69) were almost double the national aver- age, mainly because farmers across Malawi use, on average, In Golomoti, groundnut prices exceeded maize prices, and −1 maize prices had a higher C.V. than groundnut prices (Fig. 2). less than the 69 kg [N] ha applied in the simulations. Because farmers often apply some fertilizer (Table 1), their Shelled groundnut prices exceeded unshelled groundnut prices. The C.V. for maize prices was 0.43, exceeding the yields exceeded the simulated yields of unfertilized maize −1 monoculture (MM0). Maize with 35 kg [N] ha grown in C.V. of the most variable groundnut price, 0.23. The −1 Fig. 2 Simulated grain yields and inflation-adjusted maize and groundnut groundnut rotation (MG) had 35 kg [N] ha of urea applied to maize −1 prices in Golomoti from 1989 to 2010. Fertilized maize monocultures had (green diamond) and 12 kg [N] ha of urea applied to groundnut (blue either 35 (MM35, black plus sign) or 69 (MM69, red hollow triangle) −1 square) kg [N] ha of urea applied. N represents nitrogen. The maize- Agron. Sustain. Dev. (2018) 38:32 Page 7 of 10 32 correlation coefficient between maize grain price and ground- had the highest percentage of households with a negative la- nut (shelled) price was 0.24, and the correlation coefficient bor balance during summer and spring. For example, if each was − 0.08 between maize grain price and groundnut surveyed household used one of the maize monocultures, 5% (unshelled) price. of households would have a labor deficit in spring, compared 6 −1 The simulated caloric yield of 15.9 × 10 kcal ha in the with 12% of households who used the maize-groundnut rota- intensely fertilized maize monoculture (MM69) was 54% tion. Labor dynamics add another complexity to the econom- 6 −1 greater than the 10.3 × 10 kcal ha yieldinthe maize- ics of integrating legumes into maize systems. Ngwira et al. groundnut rotation (MG), and was greater than the unfertilized (2012), Franke et al. (2014), and Thierfelder et al. (2015b) 6 −1 monoculture (2.6 × 10 kcal ha ) and the moderately fertiliz- calculated labor-use efficiency to better understand the pro- 6 −1 er monoculture (MM35) (7.4 × 10 kcal ha ). The maize- ductivity of maize-legume systems. Ngwira et al. (2012)and −1 groundnut rotation used more labor (281 days ha )than the Thierfelder et al. (2015a) found that maize monoculture can −1 monocultures (< 223 days ha ), mainly because labor used use slightly less labor than a maize-legume system. Ngwira et for groundnut exceeded that for maize (section 3.1). Nitrogen- al. (2012) also showed maize monocultures can have a higher −1 use efficiency was highest in the maize-groundnut rotation labor productivity (kg grain per day worked )than maize- −1 (89 kg grain kg [N] fertilizer ), and declined as fertilizer legume systems. Here, we add to these productivity-focused −1 application rates rose (65 kg grain kg [N] fertilizer in studies by illustrating potential bottlenecks between labor re- MM69). quired and available at the household scale (Fig. 3). Depending on the system examined, some households had labor use exceeding family labor supply (Fig. 3). More house- 3.3 Economic and risk indicators holds incurred a labor deficit in the maize-groundnut rotations, compared with the maize monoculture (MM69) given that the Table 2 summarizes how different practices affected simulated former system required more labor than the latter (Section economic and risk indicators. Costs were the highest in the 3.1). Each year, the average household had 7 days in the maize intensely fertilized maize monoculture (MM69), attributable monoculture (MM69) and 17 days in the maize-groundnut to the cost of fertilizer. Malawian farmers are often sensitive to rotation (MGS) for which family labor supply was insufficient changes in the cost of fertilizer (Komarek et al. 2017), with to meet labor requirements. The maize-groundnut rotations cost changes posing a threat to the attractiveness of fertilizer Fig. 3 Seasonal labor use and availability for Golomoti farmers. Intensely shelled (shelled rotation). N represents nitrogen. Results reported for −1 fertilized maize monoculture (MM69) has 69 kg [N] ha of urea applied. two aggregated time periods: Spring and Summer (red cross) −1 The rotation (MGS) has 35 kg [N] ha of urea applied to maize and (September to February), and Autumn and Winter (blue hollow square) −1 12 kg [N] ha of urea applied to groundnut. Groundnuts were sold (March to August) 32 Page 8 of 10 Agron. Sustain. Dev. (2018) 38:32 Table 2 Economic and risk Indicator MM0 MM35 MM69 MGUS MGS indicators for the simulated cropping systems −1 Average financial input cost (US $ ha ) 6.76 37.9 68.1 32.1 32.1 −1 Average cost of labor (US $ ha ) 341.7 351.2 359.4 452.3 460.5 −1 Average financial profit (US $ ha ) 116.6 326.6 716.5 807.7 1086.7 Coefficient of variation of financial profit 0.37 0.53 0.50 0.23 0.26 −1 Average economic profit (US $ ha ) − 225.0 − 24.6 357.1 355.3 626.2 Standard deviation of economic profit 43.6 173.4 361.3 182.2 280.6 Coefficient of variation of economic profit . . 1.01 0.51 0.45 −1 Net present value (US $ ha ) − 2810 − 788 3246 3855 6733 Probability of positive economic profit (%) 0 31.8 90.9 100 100 −1 Conditional Value at Risk (US $ ha ) − 267.2 − 191.8 − 12.3 28.8 97.4 −1 Risk premium (RP) (US $ ha ) . . 182.8 46.7 62.9 −1 Certainty equivalent (CE) (US $ ha ) . . 174.3 308.6 563.3 MM0 indicates unfertilized maize monoculture. Fertilized maize monoculture with 35 (MM35) and 69 −1 (MM69) kg [N] ha of urea. Maize-groundnut (shelled) rotation (MGS) and maize-groundnut (unshelled) −1 −1 rotation (MGUS) had 35 kg [N] ha of urea applied to maize and 12 kg [N] ha of urea applied to groundnut. Groundnuts were sold shelled (MGS) or unshelled (MGUS). N represents nitrogen. RP and CE use an Arrow- Pratt relative risk aversion coefficient of 1 use. Shelling increased labor use for groundnut, which in- Our study had systems with differing practices (i.e., fertilizer creased the implicit cost of labor. The unfertilized maize rates), but the broad context was similar. In addition, we found monoculture (MM0) was the least profitable system, having that risk indicators differed across the systems. a negative economic profit in all years, indicating that the The maize-groundnut (unshelled) rotation (MGUS) had value of crop production was less than the total cost of fertil- similar average economic profits and approximately a 50% izer, seed, and implicit labor. The maize monocultures had lower standard deviation in economic profit (calculated as greater downside risks than the maize-groundnut rotations, variability over simulated years) compared with intensely fer- with the Conditional Value at Risk negative in the maize tilized maize monoculture (MM69) (Table 2). The maize- monocultures but positive in the maize-groundnut rotations. groundnut rotation had a lower variance of economic profits The most profitable system (MGS in Table 2)produced less partly because groundnut prices had a lower C.V. than maize calories than the intensely fertilized maize monoculture prices. In addition, the caloric yield of the maize-groundnut (MM69) (section 3.2). Profit in fertilized maize monocultures rotation had a slightly lower C.V. than the C.V. of intensely is often similar to or lower than profit in maize-legume sys- fertilized maize monoculture. Growing two crops as opposed tems, but highly context specific (Thierfelder et al. 2015b; to a monoculture reduced the C.V. of profits, given the mini- Ngwira et al. 2012). In our study, we examined average yearly mal level of correlation between maize and groundnut prices economic profits and also examined net present values. The (section 3.2). The unfertilized maize monoculture had nega- intensely fertilized maize monoculture (MM69) had a 19% tive profits in all 22 years and even MM35 had a 32% chance lower net present value than the maize-groundnut of negative profits, highlighting monoculture can be unprofit- (unshelled) rotation (MGUS), despite MM69 and MGUS hav- able—as also showninFrankeetal. (2014). The maize- −1 ing similar economic profits—357 US $ ha in MM69 and groundnut rotations had more years of positive profits −1 355 US $ ha in MGUS. The cumulative agronomic benefits (100%) than the intensely fertilized maize monoculture of integrating legumes into maize systems, combined with (MM69) (91%). When we accounted for risk aversion, both changes in prices, translated into these net present values. maize-groundnut rotations had a higher certainty equivalent Yield is only one indicator farmers consider when evaluating (risk-adjusted profit) than the intensely fertilized maize mono- alternative crop practices and, in our study, the system that culture. When the Arrow-Pratt relative risk aversion coeffi- produced the most calories generated less profit than the most cient was equal to one, the certainty equivalent of the maize- −1 profitable system. Franke et al. (2014) showed that maize groundnut (unshelled) rotation was US $ 309 ha and was US −1 produces a higher caloric yield than groundnut, which in turn $ 174 ha in the intensely fertilized maize monoculture has higher caloric yield than soybean. Franke et al. (2014) (MM69) (Table 2). However, economic profits in the rotation showed that replacing half of a simulated farm’s maize and were similar to profits in MM69 (Table 2). Including risk soybean area with groundnut can slightly increase caloric pro- aversion resulted in the maize-groundnut rotation providing duction, possibly because soybean has a lower caloric yield higher risk-adjusted profits (the certainty equivalent) than un- than groundnut. This replacement ultimately increased profits. der risk neutrality. Providing advice based on risk neutrality Agron. Sustain. Dev. (2018) 38:32 Page 9 of 10 32 PIM is in turn supported by the CGIAR Fund donors. The HarvestChoice generated different systems rankings than providing advice Project (Grant No. OPPGD1450), funded by the Bill and Melinda Gates based on risk aversion. Under risk neutrality, the intensely Foundation, also funded this study. The United States Agency for fertilized maize monoculture and the maize-groundnut International Development funded the collection of the household survey (unshelled) rotation had similar average profits. However, un- data, as part of the Africa RISING program. der risk aversion, the intensely fertilized maize monoculture Open Access This article is distributed under the terms of the Creative generated a lower certainty equivalent compared to the maize- Commons Attribution 4.0 International License (http:// groundnut (unshelled) rotation. This lower certainty equiva- creativecommons.org/licenses/by/4.0/), which permits unrestricted use, distribution, and reproduction in any medium, provided you give appro- lent finding complements Gandorfer et al. (2011), who report- priate credit to the original author(s) and the source, provide a link to the ed that risk analyses rarely compare the certainty equivalents Creative Commons license, and indicate if changes were made. for risk-averse and risk-neutral strategies. Taken together, our results suggest that maize-groundnut rotations can reduce eco- nomic risks compared with maize monocultures. References 4 Conclusions Antle JM (1987) Econometric estimation of producers’ risk attitudes. Am J Agric Econ 69(3):509–522. https://doi.org/10.2307/1241687 We examined how simulated changes in crop practices Chianu JN, Nkonya EM, Mairura FS, Chianu JN, Akinnifesi FK (2011) Biological nitrogen fixation and socioeconomic factors for legume altered the performance of maize-based cropping systems production in sub-Saharan Africa: a review. Agron Sustain Dev with data from central Malawi. Four cropping systems 31(1):139–154. https://doi.org/10.1051/agro/2010004 were simulated: unfertilized maize monoculture, two fer- Chianu JN, Chianu JN, Mairura F (2012) Mineral fertilizers in the farm- tilized maize monocultures, and a maize-groundnut rota- ing systems of sub-Saharan Africa. A review. Agron Sustain Dev tion. Our study demonstrated some of the trade-offs and 32(2):545–566. https://doi.org/10.1007/s13593-011-0050-0 Droppelmann KJ, Snapp SS, Waddington SR (2017) Sustainable intensi- synergies that existed among indicators of different fication options for smallholder maize-based farming systems in cropping systems related to caloric yields, labor balances, sub-Saharan Africa. Food Security 9(1):133–150. https://doi.org/ profits, and risks. Overall, the maize-groundnut rotation, 10.1007/s12571-016-0636-0 relative to the intensely fertilized maize monoculture, FAO (1968) Food composition tables for use in Africa. Food and Agriculture Organisation Rome, Italy/U.S. Department of Health, yielded a higher nitrogen-use efficiency, was more profit- Education and Welfare, Bethesda, MD, USA, p 306 able, had a higher net present value, and had reduced FAO (2017) FAOSTAT. Food and Agriculture Organization of the United risks. However, the rotation produced lower caloric yields Nations, Rome and required more labor, relative to the intensely fertilized Franke AC, van den Brand GJ, Giller KE (2014) Which farmers benefit maize monoculture. Because of differences in labor re- most from sustainable intensification? An ex-ante impact assess- ment of expanding grain legume production in Malawi. Eur J quirements, more households incurred a labor deficit in Agron 58:28–38. https://doi.org/10.1016/j.eja.2014.04.002 the maize-groundnut rotation, compared with those under Gandorfer M, Pannell D, Meyer-Aurich A (2011) Analyzing the effects of the maize monocultures. risk and uncertainty on optimal tillage and nitrogen fertilizer inten- Hybrid studies, such as ours, that combine secondary sity for field crops in Germany. Agric Syst 104(8):615–622. https:// data from field experiments with primary household sur- doi.org/10.1016/j.agsy.2011.06.004 IFPRI (2013) 2013 assessment: Malawi agricultural market information vey data and simulation modeling are complements—not system. International Food Policy Research Institute, Washington substitutes—to field experiments. Simulations can help D.C identify possible trade-offs and synergies associated with IFPRI (2015) Malawi Africa Research in Sustainable Intensification for counterfactuals of changes in crop practices on system the Next Generation (Africa RISING) baseline evaluation survey. indicators, which might be a challenge to demonstrate in Harvard Dataverse. https://doi.org/10.7910/DVN/28557 Jones JW, Hoogenboom G, Porter CH, Boote KJ, Batchelor WD, Hunt an experimental setting. Highlighting these trade-offs and LA, Wilkens PW, Singh U, Gijsman AJ, Ritchie JT (2003) The synergies helps build evidence for policy dialogs on how DSSAT cropping system model. Eur J Agron 18(3–4):235–265. to support the sustainable development of cropping sys- https://doi.org/10.1016/S1161-0301(02)00107-7 tems for labor- and resource-limited farmers. Kassie M, Teklewold H, Marenya P, Jaleta M, Erenstein O (2015) Production risks and food security under alternative technology choices in Malawi: application of a multinomial endogenous Data availability statement The replication materials, includ- switching regression. J Agric Econ 66(3):640–659. https://doi.org/ ing datasets and code, for the current study are available in the 10.1111/1477-9552.12099 Dataverse repository, https://doi.org/10.7910/DVN/ILVSXJ. Komarek AM, Drogue S, Chenoune R, Hawkins J, Msangi S, Belhouchette H, Flichman G (2017) Agricultural household effects Funding information This work was undertaken as part of, and funded of fertilizer price changes for smallholder farmers in central Malawi. by, the CGIAR Research Program on Policies, Institutions, and Markets Agric Syst 154:168–178. https://doi.org/10.1016/j.agsy.2017.03. (PIM) led by the International Food Policy Research Institute (IFPRI). 016 32 Page 10 of 10 Agron. Sustain. Dev. (2018) 38:32 Lehmann N, Finger R, Klein T, Calanca P, Walter A (2013) Adapting Smith A, Snapp S, Dimes J, Gwenambira C, Chikowo R (2016) Doubled- crop management practices to climate change: modeling optimal up legume rotations improve soil fertility and maintain productivity solutions at the field scale. Agric Syst 117:55–65. https://doi.org/ under variable conditions in maize-based cropping systems in 10.1016/j.agsy.2012.12.011 Malawi. Agric Syst 145:139–149. https://doi.org/10.1016/j.agsy. 2016.03.008 Mhango WG, Snapp S, Kanyama-Phiri GY (2017) Biological nitrogen Snapp SS, Blackie MJ, Gilbert RA, Bezner-Kerr R, Kanyama-Phiri GY fixation and yield of pigeonpea and groundnut: quantifying response (2010) Biodiversity can support a greener revolution in Africa. Proc on smallholder farms in northern Malawi. Afr J Agric Res 12(16): Natl Acad Sci 107(48):20840–20845. https://doi.org/10.1073/pnas. 1385–1394 Mungai LM, Snapp S, Messina JP, Chikowo R, Smith A, Anders E, Tamene L, Mponela P, Ndengu G, Kihara J (2016) Assessment of maize Richardson RB, Li G (2016) Smallholder farms and the potential yield gap and major determinant factors between smallholder for sustainable intensification. Front Plant Sci 7:1720. https://doi. farmers in the Dedza district of Malawi. Nutr Cycl Agroecosyst org/10.3389/fpls.2016.01720 105(3):291–308. https://doi.org/10.1007/s10705-015-9692-7 Ngwira AR, Aune JB, Mkwinda S (2012) On-farm evaluation of yield Thierfelder C, Chisui JL, Gama M, Cheesman S, Jere ZD, Bunderson and economic benefit of short term maize legume intercropping WT, Eash NS, Rusinamhodzi L (2013) Maize-based conservation systems under conservation agriculture in Malawi. Field Crop Res agriculture systems in Malawi: long-term trends in productivity. 132:149–157. https://doi.org/10.1016/j.fcr.2011.12.014 Field Crop Res 142:47–57. https://doi.org/10.1016/j.fcr.2012.11. Ngwira AR, Aune JB, Thierfelder C (2014) DSSAT modelling of con- servation agriculture maize response to climate change in Malawi. Thierfelder C, Bunderson WT, Mupangwa W (2015a) Evidence and les- Soil Tillage Res 143(0):85–94. https://doi.org/10.1016/j.still.2014. sons learned from long-term on-farm research on conservation agri- 05.003 culture systems in communities in Malawi and Zimbabwe. Onduru DD, Du Preez CC (2007) Ecological and agro-economic study of Environments 2(3):317–337. https://doi.org/10.3390/ small farms in sub-Saharan Africa. Agron Sustain Dev 27(3):197– environments2030317 208. https://doi.org/10.1051/agro:2007003 Thierfelder C, Rusinamhodzi L, Ngwira AR, Mupangwa W, Nyagumbo Ortega DL, Waldman KB, Richardson RB, Clay DC, Snapp S (2016) I, Kassie GT, Cairns JE (2015b) Conservation agriculture in Sustainable intensification and farmer preferences for crop system at- Southern Africa: advances in knowledge. Renew Agric Food Syst tributes: evidence from Malawi’s central and southern regions. World 30(04):328–348. https://doi.org/10.1017/S1742170513000550 Dev 87:139–151. https://doi.org/10.1016/j.worlddev.2016.06.007 Tittonell P, Giller KE (2013) When yield gaps are poverty traps: the Ruane AC, Goldberg R, Chryssanthacopoulos J (2015) Climate forcing paradigm of ecological intensification in African smallholder agri- datasets for agricultural modeling: merged products for gap-filling culture. Field Crop Res 143:76–90. https://doi.org/10.1016/j.fcr. and historical climate series estimation. Agric For Meteorol 200: 2012.10.007 233–248. https://doi.org/10.1016/j.agrformet.2014.09.016 Waddington SR, Mekuria M, Siziba S, Karigwindi J (2007) Long-term Rusinamhodzi L (2015) Tinkering on the periphery: labour burden not yield sustainability and financial returns from grain legume–maize crop productivity increased under no-till planting basins on small- intercrops on a sandy soil in subhumid north central Zimbabwe. Exp holder farms in Murehwa district, Zimbabwe. Field Crop Res Agric 43(04):489 –503. https://doi.org/10.1017/ 170(0):66–75. https://doi.org/10.1016/j.fcr.2014.10.006 S0014479707005303
Agronomy for Sustainable Development – Springer Journals
Published: May 30, 2018
It’s your single place to instantly
discover and read the research
that matters to you.
Enjoy affordable access to
over 18 million articles from more than
15,000 peer-reviewed journals.
All for just $49/month
Query the DeepDyve database, plus search all of PubMed and Google Scholar seamlessly
Save any article or search result from DeepDyve, PubMed, and Google Scholar... all in one place.
Get unlimited, online access to over 18 million full-text articles from more than 15,000 scientific journals.
Read from thousands of the leading scholarly journals from SpringerNature, Elsevier, Wiley-Blackwell, Oxford University Press and more.
All the latest content is available, no embargo periods.
“Hi guys, I cannot tell you how much I love this resource. Incredible. I really believe you've hit the nail on the head with this site in regards to solving the research-purchase issue.”Daniel C.
“Whoa! It’s like Spotify but for academic articles.”@Phil_Robichaud
“I must say, @deepdyve is a fabulous solution to the independent researcher's problem of #access to #information.”@deepthiw
“My last article couldn't be possible without the platform @deepdyve that makes journal papers cheaper.”@JoseServera