BROADENING SMALLHOLDER FARMER OPTIONS THROUGH LEGUME ROTATIONAL AND INTERCROP DIVERSITY IN MAIZE-BASED CROPPING SYSTEMS OF MALAWI Chiwimbo Perseverance Mwika By 2020 A DISSERTATION Submitted to Michigan State University in partial fulfillment of the requirements for the degree of Crop and Soil Sciences – Doctor of Philosophy ABSTRACT BROADENING SMALLHOLDER FARMER OPTIONS THROUGH LEGUME ROTATIONAL AND INTERCROP DIVERSITY IN MAIZE-BASED CROPPING SYSTEMS OF MALAWI By Chiwimbo Perseverance Mwika Sustainability of rainfed cropping across southern Africa is undermined by maize (Zea mays L.) monocultures which are mostly cultivated on nitrogen (N) deficient soils. Smallholder farmers rarely achieve access to adequate quantities of inorganic fertilizers, and this limits crop productivity and negatively impacts food and nutritional security. Sustainable intensification with legumes has been proposed as a solution to address these challenges. Legumes such as groundnut (Arachis hypogaea L.) and pigeonpea (Cajanus cajan L.) potentially improve soil fertility and productivity of cereal crops grown in sequence through biological nitrogen fixation (BNF) and high-quality organic residues. However, successfully addressing smallholder farmer challenges requires understanding cropping system performance on-farm, in different environments. This is critical for site-specific agricultural technology recommendations that suit smallholder farmer goals. This dissertation consists of three studies on sustainable intensification with legumes in maize- legume cropping systems in central and southern Malawi. A participatory research approach of researcher designed, and farmer-managed trials were used to evaluate legume and maize production, the economic feasibility of cropping systems, BNF contributions and effect of residue quality and quantity on soil N dynamics. In the first chapter, four cropping systems in on-farm experiments, in five locations from three agroecologies in central Malawi were used to compare intercrop diversity and rotational diversity. The objective of the study was to determine suitable cropping systems for smallholder farmers in terms of grain production and economic viability. Above and belowground biomass was monitored to understand inter- and intra-specific competition of pigeonpea, and groundnut compared to the traditional maize/pigeonpea intercrop. Pigeonpea biomass was suppressed when intercropped with either groundnut or maize, whereas groundnut was not sensitive to interspecific competition. The second chapter is an evaluation of on-farm nitrogen dynamics, including a detailed assessment of BNF by plant tissue components of groundnut and pigeonpea within four legume-maize diversified cropping systems in central Malawi. The findings show that the net nitrogen balance with groundnut varied markedly, from deficit to a net benefit for subsequent crops, depending on growth and residue management. Another finding was that pigeonpea, especially when grown as part of a doubled- up legume system, provided substantial nitrogen inputs on rain-fed smallholder farms. The third chapter involved determining the effect of incorporating high- vs low-quality and quantity of crop residues on mineral N dynamics and subsequent maize yield in southern Malawi. Findings show the actual crop residue biomass quantity and quality that smallholder farmers are capable of producing depending on their biophysical environment. Each of the three studies highlights implications for on-farm sustainable intensification with legumes to address various farmer goals in different environmental context. This dissertation is dedicated to my wonderful husband Villeneuve and our children Eliana and Elihai. Thank you for your unwavering support, love and patience throughout my PhD journey. iv ACKNOWLEDGEMENTS I would like to sincerely thank my advisor Dr. Sieg Snapp for providing me this opportunity, for her guidance, encouragement, support, patience, understanding, help, mentorship, and collaboration. I am very grateful for the incredible support I received from my committee members. Dr. Regis Chikowo for his encouragement, mentorship, care, and collaboration. Dr. Nicky Mason-Wardell for her thorough and prompt feedback. Dr. Alvin Smucker for his care, encouragement, and feedback. The Africa RISING-Malawi team, smallholder farmers, extension agents, and USAID funding through Michigan State University made this research possible. I am thankful to Emmanuel Jambo for his tremendous help with my fieldwork, laboratory work, driving, teaching me Chichewa, and for his friendship. Dr. Vimbayi Chimonyo and Dr. Wezi Mhango for their guidance, mentorship, and care. Edward Mzumara, Hannah Livuza, and Jeckner Phiri for help with fieldwork. I will always be grateful to the Malawian smallholder farmers who welcomed me to their fields wholeheartedly, taught me Chichewa, helped with fieldwork and were generous with their time. I am thankful for Prof. Emmanuel Kaunda, Mrs. Celina Kaunda, and family for making me part of their family and seeing to my every need during my time in Malawi. I am grateful for our partners at the Lilongwe University of Agriculture and Natural Resources where I lodged and carried out my laboratory work with the help of Ladson and Kondwani Chirwa. v Many thanks to my friends and colleagues in the Snapp lab at Michigan State University. They made my PhD journey a rewarding experience (Princess Hayford, Alexia Witcombe, Alison Nord, Vanessa Thomas, Leah Mungai, Placid Mpeketula). Thanks to Brad Peter for the beautiful maps he made for me. I am also thankful for the mentorship I got from Dr. Isaac Kalumbu, Dr. Julie Libarkin, Dr. Jeno Rivera and Dustin Petty. The love, support, encouragement, prayers, and help with childcare I received from Pastor Dan, Lynn and Amber Ratliff, Leah, and Hope and Henry Akaeze made all the difference in my journey. I am also thankful for loving friends and family who kept me going- Pastor D. and Mrs. S. Kabaira, Pastor J.M and Mrs. A. Chikosi, Rachael Matiure, Rufaro Govere, Farayi Mwashita, Mrs. L. Chikosi, Mrs. S. Kaseke, Mrs J. Mkanganwi, Faith Kaunda, the Maweja family, Bonnie and Gene Albert, Pastor Eric and Corrine Mulanda, and Ted Halsted. I am thankful for my great parents Rev. C and Rev. P. Gwenambira for their love, prayers, support, encouragement, many sacrifices, and investment in my education. My siblings Ruzivo and Joshua for their love, support, encouragement, and prayers. The little hugs, laughs, and sweet smiles I got from my children Eliana and Elihai gave me the drive and enthusiasm to go on. I am forever grateful for my wonderful husband Villeneuve for his unwavering support, love, prayers, patience, and understanding and being there for me and our children during my whole PhD journey. Thank you! vi TABLE OF CONTENTS LIST OF TABLES ............................................................................................................................... ix LIST OF FIGURES ............................................................................................................................. xii 1. Chapter One: Legume rotational and intercrop diversity in maize-based cropping systems of Malawi ............................................................................................................................... 1 Abstract ........................................................................................................................................... 1 Introduction ..................................................................................................................................... 2 Materials and methods .................................................................................................................... 5 Study sites and field sites characterization ......................................................................................................... 5 Experimental design ............................................................................................................................................ 8 Management of cropping systems ...................................................................................................................... 8 Aboveground and belowground biomass and yield ............................................................................................ 9 Economic feasibility of cropping systems .......................................................................................................... 10 Statistical analysis .............................................................................................................................................. 11 Results ........................................................................................................................................... 11 Environment ...................................................................................................................................................... 11 Legume biomass ................................................................................................................................................ 12 Legume and maize grain yield ........................................................................................................................... 16 Interspecific vs. intraspecific competition ......................................................................................................... 17 Economic feasibility of cropping systems .......................................................................................................... 18 Discussion ...................................................................................................................................... 20 Legume biomass by cropping system and environment ................................................................................... 20 Maize yield response to legume rotation .......................................................................................................... 21 Interspecific vs. intraspecific competition in intercrops ................................................................................... 22 Economic feasibility and overall performance .................................................................................................. 23 Conclusions .................................................................................................................................... 25 REFERENCES ................................................................................................................................... 27 2. Chapter Two: Pigeonpea and groundnut productivity, biological nitrogen fixation, and residual benefits to maize on smallholder farms in central Malawi ....................................... 31 Abstract ......................................................................................................................................... 31 Introduction ................................................................................................................................... 32 Sustainable agriculture through BNF ................................................................................................................ 32 Malawi legume crops and nitrogen balance on-farm ....................................................................................... 32 Pigeonpea and groundnut BNF ......................................................................................................................... 33 Materials and methods .................................................................................................................. 35 Study sites and field sites characterization ....................................................................................................... 35 Experimental design .......................................................................................................................................... 36 Management of cropping systems .................................................................................................................... 37 Aboveground and belowground biomass and yield .......................................................................................... 38 Biological nitrogen fixation ................................................................................................................................ 39 Statistical analysis .............................................................................................................................................. 40 Results ........................................................................................................................................... 41 Environment ...................................................................................................................................................... 41 vii Crop biomass production .................................................................................................................................. 42 Percent N of groundnut and pigeonpea ............................................................................................................ 44 Total nitrogen fixed by groundnut and pigeonpea ............................................................................................ 47 Groundnut and pigeonpea BNF-N % ................................................................................................................. 50 Discussion ...................................................................................................................................... 53 N dynamics by species ....................................................................................................................................... 53 Cropping systems effect .................................................................................................................................... 55 Sustainability implications ................................................................................................................................. 57 Conclusions .................................................................................................................................... 59 REFERENCES ................................................................................................................................... 61 3. Chapter Three: Short-term interactions between legume and maize residues and mineral N fertilizer on nitrate-N dynamics and maize productivity in Malawi .................................... 64 Abstract ......................................................................................................................................... 64 Introduction ................................................................................................................................... 65 Crop residues ..................................................................................................................................................... 65 Effect of crop residues on soil C and N .............................................................................................................. 66 Materials and methods .................................................................................................................. 69 Study locations .................................................................................................................................................. 69 Experimental design .......................................................................................................................................... 71 Crop management ............................................................................................................................................. 75 Plant biomass assessments ............................................................................................................................... 75 Yield assessment and residue management ..................................................................................................... 76 Tissue quality analysis ....................................................................................................................................... 76 Soil sampling and analyses ................................................................................................................................ 77 Statistical analysis .............................................................................................................................................. 78 Results ........................................................................................................................................... 79 Environment ...................................................................................................................................................... 79 Aboveground biomass ....................................................................................................................................... 81 Tissue quality ..................................................................................................................................................... 83 Maize grain yield following incorporation of crop residues .............................................................................. 83 Soil NO3-N .......................................................................................................................................................... 87 Soil NH4-N .......................................................................................................................................................... 88 Discussion ...................................................................................................................................... 91 Biomass production and maize grain yield ........................................................................................................ 91 Crop response to nitrogen ................................................................................................................................ 93 Tissue quality ..................................................................................................................................................... 94 Soil inorganic N .................................................................................................................................................. 95 Conclusions .................................................................................................................................... 97 REFERENCES ................................................................................................................................... 99 viii LIST OF TABLES Table 1.1. Soil chemical and physical properties at research locations (0-20 cm) from marginal (Golomoti A and B), moderate (Kandeu) and mesic (Linthipe A and B) sites in central Malawi from sampling conducted in June of 2016. Values are means followed by standard deviations. ........ 12 Table 1.2. Analysis of variance for pigeonpea and groundnut shoot biomass in five locations across central Malawi during the 2016 growing season. Presented values are means followed by standard deviations. Cropping systems shown are sole pigeonpea (PP), groundnut/pigeonpea intercrop (GNPP), maize/pigeonpea intercrop (MZPP) and sole groundnut (GN). ...................... 13 Table 1.3. Analysis of variance for pigeonpea root biomass from the 0-60 cm soil depth, three intervals, from five locations across central Malawi in 2016. Presented values are means followed by standard deviations. Root biomass is shown in Mg/ha. Cropping systems shown are sole pigeonpea (PP), groundnut/pigeonpea intercrop (GNPP), maize/pigeonpea intercrop (MZPP) and sole groundnut (GN). .................................................................................................................... 15 Table 1.4. Analysis of variance for pigeonpea, groundnut and maize grain yield in five locations across central Malawi during the 2016 (legume) and 2017 (maize) growing seasons. Presented values are means followed by standard deviations. Cropping systems shown are sole pigeonpea (PP), groundnut/pigeonpea intercrop (GNPP), maize/pigeonpea intercrop (MZPP) and sole groundnut (GN). ........................................................................................................................... 17 Table 1.5. Economic feasibility of four cropping systems involving maize, pigeonpea, and groundnut across three agroecologies in central Malawi. Cropping systems shown are sole pigeonpea rotated with sole maize (PP-MZ), sole groundnut rotated with sole maize (GN-MZ), groundnut/pigeonpea intercrop rotated with sole maize (GNPP-MZ) and maize/pigeonpea intercrop rotated with maize/pigeonpea intercrop (MZPP-MZPP). Figures in italics show the total economic values of combined crops/ha for every cropping system assessed over two years. ... 19 Table 2.1. Soil chemical and physical properties at research locations (0-20 cm) from mesic (Linthipe A and B), marginal (Golomoti A and B) and moderate (Kandeu A and B) sites in central Malawi from sampling conducted in June of 2016. Values are means followed by standard deviations. .................................................................................................................................... 41 Table 2.2. Analysis of variance for groundnutf and pigeonpea total biomass including grain in six locations across central Malawi during the 2016 growing season. Presented values are means followed by standard deviations. Cropping systems shown are sole groundnut (GN), sole pigeonpea (PP), groundnut/pigeonpea intercrop (GNPP), and maize/pigeonpea intercrop (MZPP) ...................................................................................................................................................... 43 ix Table 2.3. Analysis of variance for groundnut and pigeonpea percent N at six locations in Malawi during the 2016 growing season. Presented values are means followed by standard deviations. ...................................................................................................................................................... 45 Table 2.4. Percent N and total N means and standards deviations of groundnut and pigeonpea in four cropping systems in central Malawi. Cropping systems are sole groundnut (GN), sole pigeonpea (PP), doubled-up groundnut/pigeonpea intercrop (GNPP), and maize/pigeonpea intercrop (MZPP) .......................................................................................................................... 46 Table 2.5. Analysis of variance for groundnut and pigeonpea total N at six locations in Malawi during the 2016 growing season. Presented values are means followed by standard deviations ...................................................................................................................................................... 49 Table 2.6. Analysis of variance for groundnut and pigeonpea BNF-N % at six locations in Malawi during the 2016 growing season. Presented values are means followed by standard deviations. ...................................................................................................................................................... 52 Table 3.1. Total annual precipitation in mm and day temperature means during the 2016, 2017, and 2018 growing seasons for the three locations. Precipitation based on CHIRPS and temperature is from NASA MODIS Land Surface Temperature. .................................................. 71 Table 3.2. Treatments structure and fertilization in the residue generation phases and in the residue quality and quantity assessment phases. Rationale for the treatment choices is also included. ....................................................................................................................................... 73 Table 3.3. Soil chemical and physical properties at research locations (0-20 cm) from marginal (Mtubwi), moderate (Nsanama) and mesic (Nyambi) sites in southern Malawi. Values are means followed by standard deviations. ................................................................................................. 80 Table 3.4. Crop residue biomass produced from nine treatments at three locations in 2017. For the species column, maize represents maize stover, GNPP-Senesced is a mixture of senesced groundnut and pigeonpea leaves, GNPP-Groundnut represents groundnut stover and GNPP- Pigeonpea is pigeonpea leaves and twigs. The incorporated columns show the total amount of incorporated crop residues, where treatments with 0R had no residues, 1R is 1x residues and 2R is 2x residues. Presented values are means, followed by standard deviations. ........................... 82 Table 3.5. Chemical composition of on-farm crop residues from Malawi which were incorporated ...................................................................................................................................................... 83 Table 3.6. The effect of residue quantity, residue quality and N fertilizer on maize yield in 2018. Means are followed by standard deviations for nine treatments at three locations in southern Malawi. Planned contrasts of six treatments across three locations are shown as well as the overall ANOVAS. ........................................................................................................................... 85 Table 3.7. Soil NO3-N in the 0-20 cm soil depth by treatment and three sampling times at three locations in southern Malawi. ...................................................................................................... 88 x Table 3.8. Soil NH4-N in the 0-20 cm soil depth by treatment and three sampling times at three locations in southern Malawi. ...................................................................................................... 90 xi LIST OF FIGURES Figure 1.1. Map of Malawi showing the location of the three research agroecologies with precipitation and temperature averages. Map courtesy of Brad Peter, Department of Geography, Michigan State University (2016) ................................................................................................... 6 Figure 1.2. Cumulative rainfall received in Linthipe, Kandeu and Golomoti during the 2015-2016 (Year 1) and 2016-2017 (Year 2) growing seasons ......................................................................... 7 Figure 2.1. Belowground biomass destructive sampling pit measuring 90 cm x 75 cm x 60 cm. Hand hoes were used to excavate roots from the pit. Roots were collected in increments from the 0-20, 20-40 and 40-60 cm soil depths. Sketch prepared by Kibale V. Mwika. ....................... 39 Figure 2.2. Total nitrogen in kg/ha by plant parts in four cropping systems in central Malawi. Cropping systems (PP), doubled-up groundnut/pigeonpea intercrop (GNPP), and maize/pigeonpea intercrop (MZPP). .................... 50 include are groundnut (GN), sole pigeonpea Figure 3.1. Map of Malawi showing the location of the three research agroecologies with 15-year temperature and normalized difference vegetation index (NDVI) averages. Temperature and NDVI are averaged across 2001-2015 growing seasons. Temperature measurements were sourced from the MODIS Land Surface Temperature. Map was prepared by Dr. Brad Peter, University of Alabama (2019). ...................................................................................................... 70 Figure 3.2. Maize grain yield from six treatments across three locations in southern Malawi. At all locations, treatments MZ-0N-0R and MZ-0N-2R were unfertilized while the rest were fertilized. ...................................................................................................................................................... 86 Figure 3.3. Cumulative soil NO3-N from the 0-80 cm soil depth during three sampling times at three locations in southern Malawi. From a two-way ANOVA, (model NO3-N= location|time), location (P=<.0001) and time (P= 0.0035) had an effect on soil NO3-N in 2017-2018. ............... 87 Figure 3.4. Cumulative soil NH4-N from the 0-80 cm soil depth during three sampling times at three locations in southern Malawi. From a two-way ANOVA, (model NH4-N = location|time), location (P<.0001) and time (P<.0001) had an effect on soil NH4-N. ........................................... 89 xii 1. Chapter One: Legume rotational and intercrop diversity in maize-based cropping systems of Malawi Abstract Sustainability of rainfed farming across east and southern Africa is undermined by the continuous presence of maize (Zea mays L.) across the vast majority of fields. Farmers have traditionally intercropped maize with food legumes, an important source of field and dietary diversity, but simplified cropping systems including sole cropped maize are common. An important agrodiversity question remains, is intercrop diversity sufficient for sustainability, or is rotational diversity more advantageous? A two-year, on-farm experiment was conducted in central Malawi at five sites along a marginal to a mesic gradient. Above and belowground biomass was monitored to understand inter- and intra-specific competition of pigeonpea (Cajanus cajun L.) and groundnut (Arachis hypogaea L.) compared to the traditional maize/pigeonpea intercrop (MZPP). Alternate systems studied included: pigeonpea rotated with maize (PP-MZ), groundnut rotated with maize (GN-MZ), and GNPP intercrop rotated with maize (GNPP-MZ). Across locations, PP-MZ produced the highest grain yield in the maize rotation phase at 5.51 Mg/ha. System performance in terms of gross margins ranged from $1407 to $1145/ha and in the order: GNPP-MZ=MZPP-MZPP>GN-MZ>PP-MZ. Pigeonpea biomass was suppressed when intercropped with either maize or groundnut, whereas groundnut was not sensitive to interspecific competition. The GNPP-MZ system met multiple goals including high amounts of biomass from pigeonpea for soil fertility replenishment, profitable groundnut grain, and high yields of maize 1 grown in sequence with minimal fertilizer application. The PP-MZ rotation was highly suitable for sites with infertile soils. However, pigeonpea grain alone might be insufficient to make the PP- MZ system attractive to farmers. Keywords: Groundnut, Intercropping, Legumes, Maize, On-farm, Pigeonpea, Rotation Introduction Efficient use of limited land and nutrient resources is key to viable farming systems in some densely populated Sub-Sahara African (SSA) countries. Agroecology principles can be harnessed to improve the sustainability of food production on smallholder farms in SSA. This includes the careful selection of diverse species, and planned mixtures, to enhance facilitation and complementarity (Malézieux et al., 2009; Raseduzzaman and Jensen, 2017). Maize-dominated smallholder farming in SSA countries is becoming intensified through the adoption of fertilizer and other inputs yet crop diversification has been overlooked as a means to sustainably intensify, and in some cases intercrop diversification has been actively suppressed through agricultural policies that promote sole cropping, with inadvertent consequences (Isaacs et al., 2016). The complementarity of legumes associated with nitrogen fixation and other soil fertility enhancing traits can be used to reinforce sustainable production of maize. This is critical in SSA where N limited soils are limiting maize’s genetic expression of yield potential (Giller and Cadisch, 1995). Legume food crops provide nutritious grain. The long-duration growth types, such as pigeonpea, 2 provide forage in the form of copious vegetation and deep root systems for soil rehabilitation (Odeny, 2007; Snapp and Silim, 2002). Farmers use both intercrops and rotational sequences to integrate legumes into their cropping systems. However, whether intercrops or rotations are more effective at supporting sustainable intensification on smallholder farms is debated. Intercropping of maize with grain legumes enhances land-use efficiency and productivity (Raseduzzaman and Jensen, 2017). About 50% of fields in Malawi are intercropped, which is consistent with farmers maximizing returns to land on smallholder farms of about 0.5 to 1.5 hectares (Silberg et al., 2017). There is conflicting evidence regarding the relative advantages of rotations vs. intercrops as a means to lower cost of production, improve soil fertility, increase total grain yield, and support yield stability (Bhuva et al., 2017; Malézieux et al., 2009; Raseduzzaman and Jensen, 2017). When inappropriate component plant densities or incompatible crop varieties are intercropped, there is often markedly reduced yields of one or both of the species (Dalal, 1974; Rusinamhodzi et al., 2012; Snapp and Silim, 2002). According to Tully et al. (2015), integrating legumes in rotations is an effective strategy in terms of water-use efficiency, increased nutrient availability, increased soil biological activity and C sequestration. Chimonyo et al. (2019) reported that rotating maize with legumes such as pigeonpea and groundnut results in high and stable maize grain yield. At the same time, a meta- analysis by (Raseduzzaman and Jensen, 2017), concluded that intercropping cereals with grain legumes generally improved yield stability compared to sole cropping. The environmental 3 context may influence the relative performance of rotational vs. intercrop diversity and needs to be considered. To address the knowledge gaps, this research involved an analysis of crop diversification through rotations and intercrops in different environmental contexts. The focus of this study was on two grain legumes that are highly suited to maize-based rainfed smallholder farms in SSA: pigeonpea and groundnut. Pigeonpea is a long duration grain legume crop that accumulates large biomass over a six to eight-month growth period. Under favourable environmental conditions, pigeonpea can biologically fix up to 200 kg/ha N but in unfavourable conditions, it might be as low as 6 kg/ha N (Adu-Gyamfi et al., 2007; Kumar Rao and Dart, 1987). Initial pigeonpea growth is slow, making it an ideal crop for intercropping with competitive crops such as maize or short early maturing legumes such as groundnut (Snapp and Silim, 2002). Groundnut is widely grown in sub-humid and semi-arid SSA as a high oil content cash or food crop, where it is often cultivated with maize, cassava (Manihot esculenta L.), and plantain (Musa paradisiaca L.) (Wendt and Atemkeng, 2004). Our study evaluated the performance of farmers’ traditional maize/pigeonpea intercrop and that of alternative maize cropping sequence strategies through rotation with groundnut and pigeonpea cultivated as sole crops and as a doubled-up legume system (intercrop of two legumes). Specifically, objectives were: 1) to document legume above and belowground biomass in sole and mixed cropping system patterns, for a gradient of marginal to mesic on-farm sites in central Malawi; 2) to document biomass, yield and economic performance of maize, groundnut, and pigeonpea in sole, intercropped and doubled-up legume maize systems relative to a 4 maize/pigeonpea intercrop farmer control; and 3) evaluate interspecific vs. intraspecific competition in sole and intercropped systems for pigeonpea and groundnut. Materials and methods Study sites and field sites characterization A stratified random approach was used to select the research sites to represent marginal to mesic potential areas (Mungai et al., 2016) in central Malawi (13.2543° S, 34.3015° E). The locations were in the Linthipe, Golomoti and Kandeu Extension Planning Areas (EPA) (Figure 1.1). Linthipe is a high potential, sub-humid tropical agroecology with well-distributed rainfall in most years. Kandeu is a medium agricultural potential, sub-humid tropical agroecology with a rainfall distribution that is above average in most years. Golomoti is a low potential, semi-arid to sub- humid tropical agroecology, at low altitude and with erratic rainfall; high temperatures increase evapotranspiration at this site. Linthipe soils are ferric luvisols, the Kandeu study site has a mixture of chromic luvisols and orthic ferralsols, while Golomoti soils are a mixture of eutric cambisols and eutric fluvisols (Mungai et al., 2016). The field experiments were conducted at five field sites (Table 1.1). Soil sampling was conducted in late June and early July of 2016, at 0– 20 cm depth, three composite samples per plot using a dutch auger (5.5 cm diameter). Soils were air-dried, sieved (2 mm) and soil texture was determined using the hydrometer method (Kellogg Biological Station/ Long-Term Ecological Research, 2008). Soil pH was determined in a 1:2 soil to water suspension (Kellogg Biological Station/ Long-Term Ecological Research, 2016). Total soil carbon and total soil N were determined, after grinding to a fine powder with a shatter mill, by 5 dry combustion with Costech ECS 4010 (Costech Analytical Technologies, Inc., Valencia, CA). To determine potential mineralizable NH4-N (PMN), soil samples were anaerobically incubated at 30 °C for seven days (Kane et al., 2015). After incubation, NH4-N was extracted using 2.66 M KCl. KCl extracts were analysed for inorganic N concentrations on a Thermo MultiskanTM 96-well plate reader (Kane et al., 2015). The cumulative rainfall from the two study years received in Linthipe, Kandeu and Golomoti during the 2015-2016 and 2016-2017 growing seasons is shown on Figure 1.2. Figure 1.1. Map of Malawi showing the location of the three research agroecologies with precipitation and temperature averages. Map courtesy of Brad Peter, Department of Geography, Michigan State University (2016) 6 Figure 1.2. Cumulative rainfall received in Linthipe, Kandeu and Golomoti during the 2015-2016 (Year 1) and 2016-2017 (Year 2) growing seasons 7 Experimental design Four cropping systems were tested and set up in a randomized complete block design with four replicates per location. The four cropping systems included in this study consisted of sole pigeonpea rotated with maize (PP-MZ), sole groundnut rotated with maize (GN-MZ), groundnut/pigeonpea doubled-up intercrop rotated with maize (GNPP-MZ), and the maize/pigeonpea intercrop system rotated with another maize/pigeonpea intercrop system in the second year (MZPP-MZPP). The traditional maize/pigeonpea intercrop (MZPP) system was included as the control. Management of cropping systems Field experiments were set up during the 2015/16 growing season. The productivity of legumes was assessed during year one with a rotational maize crop used to quantify the cropping system effect in year two. Crops were planted in December of 2015, harvested between May and July 2016, and a second year of the experiment was planted in 2016 and harvested in 2017. Hand hoes were used for land preparation and making ridges where crops are normally planted. Plot sizes were 5m x 5m and ridges were spaced at 75 cm apart. Fields were planted in December after the first effective rains. Planting at each of the sites was completed on each day. Sole pigeonpea was planted at a spacing of 0.9 m x 0.75 m, with three plants per planting station to achieve a plant population density of 44 000/ha. Sole groundnut within row spacing was 0.1 m to achieve a plant population density of 120 000 plants/ha. An additive intercropping design was used for the pigeonpea/groundnut intercrop to attain a total plant population density of about 164 000 plants/ha. In the pigeonpea/maize cropping system, both maize and pigeonpea were 8 planted at a spacing of 0.9 m x 0.75 m in an additive intercropping design. There were three plants per planting station for both pigeonpea and maize. Crops in all intercrop systems were planted in the same row. The doubled-up legume intercrops were fertilized just before planting with 23:21 N: P compound fertilizer at the rate of 11.5 kg ha-1 N and 10.5 kg ha-1 of P. The pigeonpea/maize intercrop was fertilized at the rate of 23 kg ha-1 N and 21 kg ha-1 P, with a side-dress application of urea at 100 kg ha-1 which provided 46 kg ha-1 N. The plots were weeded by hand hoes three times at each location. Plant population density assessment in all plots was done at harvesting to determine final plant population densities. Aboveground and belowground biomass and yield Destructive harvest timing varied with plant species and location to ensure sampling was done when plants were at physiological maturity. For maize and groundnut, it was from early-mid May 2016 and for pigeonpea, it was in July 2016. In 2017, maize was harvested in May and pigeonpea in July. Harvested net plots for groundnut destructive sampling were two ridges x 2m for a total area of 2.25 m2. Groundnut biomass was separated into stems, leaves, pods, and roots. Maize plants were harvested from three middle rows x 5m long on the plot to determine total biomass and cob weights from the total harvested area of 11.25 m2. For pigeonpea, four plants were randomly selected per plot, cut at ground level, chopped, and fresh biomass was determined for aboveground biomass separated into stems, twigs, leaves, and pods. The legume plant samples were oven-dried at 75 °C to constant weight, to determine dry weights. 9 All plants were cut at ground level and a soil volume was excavated by hand hoe to a depth of 0.60 m (0.405 m3 soil volume) in three increments, 0– 0.20 m, 0.20– 0.40 m and 0.40– 0.60 m. Large roots were removed through dry sieving (2-mm). Fine roots were hand-picked using tweezers. In intercrops, legume roots were identified through their characteristic smell – ‘the Gwezu smell technique” (Mapfumo et al., 2005), a method which was found to be straight forward to distinguish pigeonpea in a previous study (Gwenambira, 2015). The roots were weighed fresh, oven-dried at 75 oC to constant weight and reweighed for dry weight determination. Economic feasibility of cropping systems Crop input and output prices were obtained from farmer-reported values from an Africa RISING Malawi panel household survey which was conducted in 2015 (n=324) (Mungai et al., 2016; Snapp et al., 2018). Market prices we collected at our trial site locations in the same year (Snapp et al., 2018). To consider the value of legumes as fuelwood and forage sources, we used a Kenya study (Rao and Mathuva, 2000) to set values at $0.03/kg (pigeonpea stems for fuelwood, and groundnut haulms) and $0.08/kg (pigeonpea leaves). It is challenging to assign a value to these products, but it was important to do so because previous research in southern Malawi shows that farmers value pigeonpea stems for fuelwood (Orr et al., 2015). To address this, we conducted a sensitivity analysis to consider 30% lower values for fuelwood and forage and found a modest difference in the economic ranking of cropping systems. 10 The formulas used for estimating cropping systems economic feasibility (per hectare) were as follows, based on gross margin performance as a widely used metric (Snapp et al., 2018). Cost of production= Cost of fertilizer + Cost of seed Total revenue= Yield in kg/ha x Price/kg Gross margin= Total Revenue - Total cost of production Statistical analysis Data were analysed using the MIXED and GLIMMIX procedures in SAS 9.4 (SAS Institute, 2002) statistical package. A two-way analysis of variance (ANOVA) model of cropping system by location was used to evaluate the response of legume biomass and grain yield to sole crops vs. intercrops in 2016. The 2017 maize grain yield response to the different 2016 legume-based cropping systems was also determined using a two-way ANOVA of cropping system by location. The Least Significant Difference (LSD) at 5% level of significance was used to test mean differences. Results Environment The soil properties at the five locations are shown in Table 1.1. Overall, soils at three of the on- farm sites were moderately acidic as indicated by values that ranged from 5.4 to 5.9. The drier sites in Golomoti were close to neutral pH with values almost all from 6.2-6.4. The topsoil organic matter content was low at all but one site, as indicated by soil organic C levels of 0.9 – 1.5 %C for 11 four sites, and 3.2 %C at Linthipe A. Soil total N % was low with a range of 0.06 % in Golomoti A to 0.19 % in Linthipe A. Generally, the sites were sandy, with the highest clay content in the 0- 20 cm soil depth being 30 % at the Linthipe A location and the lowest clay content 12 % at Golomoti B (Table 1.1). Table 1.1. Soil chemical and physical properties at research locations (0-20 cm) from marginal (Golomoti A and B), moderate (Kandeu) and mesic (Linthipe A and B) sites in central Malawi from sampling conducted in June of 2016. Values are means followed by standard deviations. Location N (%) C (%) PMN (mg/kg soil) pH Clay (%) Sand (%) Linthipe A 0.19 ± 0.03 3.20 ± 0.61 0.67 ± 0.40 5.74 ± 0.17 30 ± 3.80 59 ± 5.84 Linthipe B 0.09 ± 0.01 1.46 ± 0.18 0.20 ± 0.16 5.41 ± 0.29 19 ± 7.35 73 ± 11.05 Kandeu 0.07 ± 0.01 0.85 ± 0.19 0.94 ± 0.25 5.88 ± 0.26 14 ± 3.07 81 ± 3.03 Golomoti A 0.06 ± 0.01 0.92 ± 0.21 0.22 ± 0.19 6.16 ± 0.19 17 ± 5.55 76 ± 6.04 Golomoti B 0.07 ± 0.02 0.87 ± 0.23 0.40 ± 0.31 6.38 ± 0.23 12 ± 1.60 83 ± 1.97 Legume biomass Cropping system had a marked effect on pigeonpea biomass, along with the location (Table 1.2). For pigeonpea aboveground biomass, Linthipe A had the lowest amount of biomass at 3.70 Mg/ha and biomass at Golomoti A was more than threefold higher, at 12.33 Mg/ha. Sole pigeonpea had the largest biomass, whereas pigeonpea biomass was lower in intercrops, with equivalent amounts produced in the groundnut/pigeonpea and maize/pigeonpea intercrops. There was a location and cropping system interaction effect on pigeonpea biomass. As for 12 groundnut biomass, there was a strong location effect but no cropping system effect on groundnut biomass. In contrast to pigeonpea, groundnut biomass in sole (GN) was equivalent to intercropped groundnut (GNPP) (Table 1.2). Table 1.2. Analysis of variance for pigeonpea and groundnut shoot biomass in five locations across central Malawi during the 2016 growing season. Presented values are means followed by standard deviations. Cropping systems shown are sole pigeonpea (PP), groundnut/pigeonpea intercrop (GNPP), maize/pigeonpea intercrop (MZPP) and sole groundnut (GN). Cropping system Linthipe A Linthipe B Kandeu Golomoti A Golomoti B Mg/ha PP (PP) 3.70 ± 1.58 8.44 ± 2.13 9.43 ± 2.61 12.33 ± 2.50 7.36 ± 2.74 GNPP (PP) 2.91 ± 2.03 3.53 ± 1.04 10.55 ± 2.27 6.70 ± 3.03 4.86 ± 1.48 MZPP (PP) 2.87 ± 0.73 6.19 ± 1.77 8.63 ± 3.93 8.04 ± 4.73 8.09 ± 3.96 GN (GN) 2.29 ± 0.92 4.88 ± 0.75 3.77 ± 0.59 3.02 ± 0.66 1.86 ± 0.63 GNPP (GN) 2.22 ± 0.37 4.90 ± 0.61 3.63 ± 1.11 2.55 ± 0.58 2.09 ± 0.55 ANOVAS Pigeonpea Groundnut Location Pr > F = <.0001* Pr > F = <.0001* Cropping system Pr > F = 0.0018* Pr > F = 0.6493 Location x cropping system Pr > F = 0.0341* Pr > F = 0.8015 *Significant at P = 0.05 13 There was a strong location effect on both pigeonpea, and groundnut root biomass excavated from the 0-20 cm soil depth. Cropping system had a significant effect on pigeonpea root biomass but not on groundnut roots. Pigeonpea root biomass from the topsoil layer (0-20 cm) followed a trend similar to shoot biomass with the lowest root biomass in Linthipe A at 0.72 Mg/ha for sole pigeonpea and 0.47 Mg/ha and 0.40 Mg/ha for intercropped pigeonpea (Table 1.3). Pigeonpea root biomass in Kandeu was almost twofold as that of Linthipe A with 1.42 Mg/ha for sole pigeonpea and 1.27 and 0.81 Mg/ha for intercropped pigeonpea (Table 1.3). Even though Golomoti A had good pigeonpea shoot biomass, root biomass in the topsoil layer was relatively low. Groundnut root biomass ranged from 0.05 Mg/ha (Golomoti B) to 0.20 Mg/ha (Linthipe B). There were no groundnut roots below the 20 cm depth at all locations. For pigeonpea, all locations had some roots in the 20-40 cm depth. However, some locations did not have pigeonpea roots in the 40-60 cm soil depth layer (Table 1.3). 14 Table 1.3. Analysis of variance for pigeonpea root biomass from the 0-60 cm soil depth, three intervals, from five locations across central Malawi in 2016. Presented values are means followed by standard deviations. Root biomass is shown in Mg/ha. Cropping systems shown are sole pigeonpea (PP), groundnut/pigeonpea intercrop (GNPP), maize/pigeonpea intercrop (MZPP) and sole groundnut (GN). Cropping Soil Depth Linthipe A Linthipe B Kandeu Golomoti A Golomoti B system (cm) Mg/ha PP GNPP (PP) MZPP (PP) GN GNPP (GN) PP GNPP (PP) MZPP (PP) PP GNPP (PP) MZPP (PP) 0-20 0-20 0-20 0-20 0-20 20-40 20-40 20-40 40-60 40-60 40-60 0.72 ± 0.43 1.00 ± 0.30 1.42 ± 0.47 1.25 ± 0.55 0.47 ± 0.32 0.52 ± 0.22 1.27 ± 0.30 0.94 ± 0.44 0.40 ± 0.07 0.67 ± 0.19 0.81 ± 0.32 1.05 ± 0.65 0.14 ± 0.02 0.19 ± 0.03 0.10 ± 0.02 0.07 ± 0.02 0.13 ± 0.01 0.20 ± 0.02 0.12 ± 0.03 0.06 ± 0.02 1.02 ± 0.24 0.68 ± 0.31 1.12 ± 0.57 0.05 ± 0.02 0.06 ± 0.01 0.05 ± 0.05 0.20 ± 0.14 0.33 ± 0.18 0.26 ± 0.15 0.05 ± 0.03 0.04 ± 0.06 0.26 ± 0.13 0.21 ± 0.09 0.04 ± 0.02 0.08 ± 0.04 0.20 ± 0.08 0.34 ± 0.21 0.19 ± 0.04 0.13 ± 0.10 0.30 ± 0.23 _ _ _ 0.06 ± 0.03 0.03 ± 0.02 0.02 ± 0.02 _ 0.02 ± 0.01 0.03 ± 0.01 0.02 ± 0.01 ANOVAS 0.03 ± 0.01 _ 0.02 ± 0.01 0.02 ± 0.01 0.02 ± 0.01 Pigeonpea Groundnut Pigeonpea Pigeonpea Depth 0-20 cm 0-20 cm 20-40 cm 40-60 cm Location Pr >F = <.0001* Pr >F =.0001* Pr >F = <.0001* Pr >F = 0.0048* Cropping System Pr >F = 0.0062 Pr >F = 0.5470 Pr >F = 0.0768 Pr >F = 0.4819 Location x C System Pr >F = 0.2823 Pr >F = 0.4921 Pr >F = 0.1135 Pr >F = 0.0615 *Significant at P = 0.05 15 Legume and maize grain yield Pigeonpea, groundnut and maize grain yields are reported in Table 1.4. Location had a strong influence on grain yield for both pigeonpea and groundnut. In contrast to the pattern observed for biomass, modest to nil grain yield was produced by pigeonpea (0.03 to 0.6 Mg/ha), and it was not influenced by cropping system. There was a large range of groundnut yields (0.5 to 1.8 Mg/ha) and biomass accumulation patterns. For example, aboveground biomass was markedly high at Linthipe B (4.9 Mg/ha), as was belowground biomass (0.2 Mg/ha). Linthipe B also produced high groundnut grain yield (1.7 to 1.8 Mg/ha). However, about one-half as much groundnut biomass was accumulated at Linthipe A and Golomoti B, which were also the low yielding sites at 0.5 to 0.9 Mg/ha (Table 1.4). Both location and cropping system had a significant effect on maize grain but with no interaction effect. In a comparison of all systems that were fertilized (69kg N/ha for sole maize and 35 kg N/ha for the MZPP intercrop), the performance of maize yield across locations in 2017 varied. Maize yield after sole pigeonpea produced the highest maize grain (5.51 Mg/ha), maize after sole groundnut was 5.01 Mg/ha, maize after the GNPP intercrop was 4.06 Mg/ha, and maize yield was lowest in the MZPP intercrop system at 3.05 Mg/ha. These findings were consistent across four of the five locations, the one exception was the lowest yielding site (Golomoti B) (Table 1.4). During the 2016 agricultural season, the sole groundnut cropping system produced the lowest shoot biomass across all sites. However, the system produced high maize grain yield in 2017 (Table 1.4). 16 Table 1.4. Analysis of variance for pigeonpea, groundnut and maize grain yield in five locations across central Malawi during the 2016 (legume) and 2017 (maize) growing seasons. Presented values are means followed by standard deviations. Cropping systems shown are sole pigeonpea (PP), groundnut/pigeonpea intercrop (GNPP), maize/pigeonpea intercrop (MZPP) and sole groundnut (GN). Cropping system Linthipe A Linthipe B Kandeu Golomoti A Golomoti B Mg/ha PP (PP) 0.12 ± 0.12 0.58 ± 0.60 0.34 ± 0.45 0.38 ± 0.48 0.03 ± 0.02 GNPP (PP) 0.12 ± 0.12 0.32 ± 0.32 0.13 ± 0.19 0.31 ± 0.22 0.04 ± 0.04 MZPP (PP) 0.09 ± 0.05 0.56 ± 0.45 0.29 ± 0.36 0.22 ± 0.43 0.07 ± 0.06 GN (GN) 0.54 ± 0.36 1.72 ± 0.47 1.34 ± 0.32 1.60 ± 0.30 0.87 ± 0.19 GNPP (GN) 0.45 ± 0.21 1.80 ± 0.32 1.28 ± 0.80 1.44 ± 0.32 0.93 ± 0.18 PP-MZ (MZ) 7.78 ± 0.47 8.14 ± 0.82 5.25 ± 1.33 4.54 ± 0.58 1.85 ± 0.33 GN-MZ (MZ) 6.48 ± 0.92 6.58 ± 0.58 4.60 ± 1.38 4.35 ± 1.66 3.02 ± 0.99 GNPP-MZ (MZ) 4.68 ± 0.82 6.18 ± 1.50 4.70 ± 0.23 2.87 ± 1.24 1.87 ± 1.16 MZPP-MZPP (MZ) 4.22 ± 0.59 5.01 ± 1.01 2.60 ± 0.59 2.27 ± 1.47 1.18 ± 0.14 ANOVAS Pigeonpea Groundnut Maize Pr > F = <.0001* Pr > F = <.0001* Pr > F = <.0001* Location Cropping system Pr > F = 0.1631 Pr > F = 0.7605 Pr > F = <.0001* Location x cropping system Pr > F = 0.6991 Pr > F = 0.9323 Pr > F = 0.2677 *Significant at P = 0.05 Interspecific vs. intraspecific competition Across locations, the PP-MZ rotation sequence produced the highest maize grain yield at 5.51 Mg/ha. The GN-MZ, GNPP-MZ and the MZPP-MZPP systems produced 9%, 26% and 45% less maize yield respectively than the PP-MZ system (p=<.0001, Table 1.4). Intercropping pigeonpea with maize and groundnut negatively impacted pigeonpea biomass (p=0.0018, Table 1.2). There was no cropping system on pigeonpea grain yield but there was a trend of lower grain yield in all intercrops. Compared to sole pigeonpea grain yield at 0.30 Mg/ha, there was a 16% drop in the 17 MZPP system and a 42% drop in grain yield in the GNPP system (Table 1.4). For shoots, with sole pigeonpea at 8.25 Mg/ha, there was an 18% drop in the MZPP system and a 31% drop in shoot biomass in the GNPP system. Compared to sole pigeonpea roots at 1.32 Mg/ha, there was a 23% drop in in the MZPP system and a 31% drop in roots in the GNPP system. For groundnut, there was no intercrop system effect; grain, shoot, and root biomass were not reduced relative to the sole crop (Tables 1.2-1.4). Economic feasibility of cropping systems Gross margins of the four cropping systems ranged from $1145 (PP-MZ) to $1407 (GNPP-MZ). The best two performing cropping systems in terms of monetary gain were the GNPP-MZ and the MZPP-MZPP systems with gross margins of $1404 and $1407, respectively. The cropping system with the highest cost of production was GNPP-MZ at $353 and the lowest was PP-MZ at $223 (Table 1.5). Overall, when legume stems and haulms are included with prices at $0.03 and $0.08, respectively, the order of technology system valuation is GNPP-MZ=MZPP-MZPP> GN-MZ> PP- MZ (Table 1.5). However, when the prices are 30% less, the order changes to GN-MZ> GNPP-MZ > MZPP-MZPP> PP-MZ. 18 Table 1.5. Economic feasibility of four cropping systems involving maize, pigeonpea, and groundnut across three agroecologies in central Malawi. Cropping systems shown are sole pigeonpea rotated with sole maize (PP-MZ), sole groundnut rotated with sole maize (GN-MZ), groundnut/pigeonpea intercrop rotated with sole maize (GNPP-MZ) and maize/pigeonpea intercrop rotated with maize/pigeonpea intercrop (MZPP-MZPP). Figures in italics show the total economic values of combined crops/ha for every cropping system assessed over two years. * Seed and fertilizer cost over two years # Economic returns included forage, fuelwood, and grain 19 Discussion Legume biomass by cropping system and environment The choice of sites was strategic, in that it allowed evaluation of cropping system performance along a gradient from marginal to mesic. A strong location effect was observed on legume shoots, roots and grain, with Kandeu providing conditions that supported high legume biomass above and belowground in 2016, whereas the Linthipe A site supported surprisingly modest pigeonpea biomass, presumably due to dry spells in this drought year (Figure 1.2). The drought is likely to have affected the overall performance and economic feasibility of the various intercrops and rotations. The cumulative rainfall in the 2015-16 agricultural season was 55% lower than in the 2016-17 season (Figure 1.2). A combination of drought and infertile sandy soils at Golomoti B could have contributed to the modest yields of pigeonpea and groundnut at this site. The susceptibility of pigeonpea to early cessation of rains has been observed previously (Wendt and Atemkeng, 2004). To the best of our knowledge, this is the first report of on-farm pigeonpea root, shoot biomass and grain yields in traditional and alternate legume diversified maize systems that include a doubled-up legume system. Pigeonpea roots in doubled-up legume intercrop were comparable to sole pigeonpea at 0.8 vs. 1.1 Mg/ha. Pigeonpea belowground biomass across all systems was about 15% of total biomass while groundnut was about 3%. These are conservative estimates given the challenges of recovering root systems in the field. This additional biomass has rarely been considered in previous studies and is important in terms of soil carbon accrual potential of 20 cropping systems (Kell, 2012). Overall, pigeonpea dry matter (6.95 Mg/ha) including shoot, root, and grain is similar to that observed by Kumar Rao and Dart (1987) where 11 pigeonpea cultivars averaged 6.86 Mg/ha. Pigeonpea grain yields were frequently under 0.4 Mg/ha, similar to on- farm values reported by Chirwa et al. (2003), Myaka et al. (2006) and Høgh-Jensen et al. (2007). Previous research from central Malawi indicates that livestock damage is contributing to these low yields, with about 44% of farmers reporting substantial goat damage (Waldman et al., 2017). Cropping system had no effect on groundnut shoots, roots and grain (Tables 1.2 and 1.3). This is important as farmers value groundnut. The presence of pigeonpea in an intercrop did not reduce groundnut biomass and yields, even at the site with substantial dry spells (Tables 1.2, 1.3 and 1.5). This is suggestive that the root system of pigeonpea exerts minimal competition for water in a groundnut system. Previous studies are consistent with this, as pigeonpea may enhance soil moisture in the topsoil through a hydraulic lift mechanism (Sekiya and Yano, 2004). Maize yield response to legume rotation Maize yield response to sole pigeonpea was positive across sites (Table 1.4). This might have been due to improved nutrient and water uptake from pigeonpea’s deep root system since there was no relationship of maize yield to vegetative legume biomass (r=0.2015). Based on model simulations, Ollenburger and Snapp (2014) indicate seasonal water recharge in pigeonpea-maize rotation sequences for a wide range of rainfall scenarios and soil types in Malawi. In Kenya, Tully et al. (2015) report on a multi-site study where maize grain yields were positively related to 21 legume biomass N only under sufficient rainfall conditions, and dry spells may have contributed to inconsistent maize response in our study. Even though groundnut biomass was lower than pigeonpea at all locations with a range from 1.86 to 4.9 Mg/ha (Table 1.2), this modest amount of biomass generally supported good maize growth in the subsequent year (Table 1.4). This could be due to a rotation effect that is independent of residue nutrient inputs (Tully et al., 2015). In an 11-yr experiment in northern Ghana, Agyare et al. (2006) reported that over the long term, maize rotations with crops including groundnut and soybean (Glycine max L.) were superior in terms of household food security potential than maize intercrops with the same crops, and a maize-groundnut rotation was the best performing combination with a 3 Mg/ha maize yield response that was comparable to our study. Interspecific vs. intraspecific competition in intercrops Pigeonpea biomass and grain yield in intercrops was much lower than sole pigeonpea, consistent with high sensitivity of this crop to interspecific competition. The drought could have exacerbated the competition. A field study of a pigeonpea-soybean intercrop system found similar results, and attributed pigeonpea biomass suppression to nutrient limitations (Ghosh et al., 2006). This sensitivity to competition within an intercrop indicates that if farmers are interested in the rehabilitation of degraded soils, or have livestock forage requirements, then sole pigeonpea is a high performer and should be considered as a substitute for current systems such as the maize- pigeonpea intercrop. According to a choice experiment conducted in Malawi, Waldman et al., 22 (2017) found that farmers value maize yield twice as much as pigeonpea yield, yet a substantial minority of farmers recognize pigeonpea’s ability to provide some grain combined with high biomass for soil fertility and grow pigeonpea primarily for its copious biomass. Groundnut root, shoot, and grain yields were not suppressed by the presence of pigeonpea in the doubled-up legume intercrop, thus this crop was not sensitive to interspecific competition in this mixed planting system. The slow growth of pigeonpea early in the growing season may minimize the competitive effects of this species on groundnut (Snapp and Silim, 2002). Economic feasibility and overall performance Considering economic feasibility is critical because farmers usually have multi-faceted goals and have to consider costs and returns associated with a cropping system before they adopt. The initial cost of production with all systems involving groundnut was high because groundnut seed is significantly more expensive than maize and pigeonpea. In India, Bhuva et al. (2017) reported similar results to ours, in that high groundnut seed expense did not reduce the attractiveness of groundnut-based systems as gross returns were high relative to other systems. A crop modeling study conducted in Central Malawi (Smith et al., 2016) combined with an economic analysis found 75% higher profits associated with a groundnut rotation compared to maize monocultures; due in large part to 50% reduced requirements for nitrogen fertilizer in the maize phase of the rotation (Komarek et al., 2018). The high gross margin associated with the traditional farmer MZ/PP intercrop system was not surprising as this shows farmer wisdom in preferring a system that is economically robust and which provides maize grain for subsistence. Of the alternatives tested here, the novel doubled-up GNPP intercrop rotated with maize was the only one to 23 perform as well economically as the farmer check. This is consistent with earlier findings of farmer preference for a highly diverse doubled-up rotational system (Snapp et al., 2018). A breadth of environmental and economic returns is required to compensate for opportunity costs associated with maize production limitations due to small farm sizes. Previous research has shown that the choices smallholder farmers in Malawi make are mostly due to a combination of factors (Hockett and Richardson, 2018) such as socio-economic status, environment, access to inputs and credit. In our study, environmental context and farmer goals both influence which system can be the overall ‘best bet’. At the mesic sites where groundnut grew well, sole groundnut or doubled-up groundnut/pigeonpea rotation systems were a good choice, in terms of highly nutritious food produced, high maize yield response, and economic profitability. At marginal sites, particularly where soil organic matter was low, the additional soil fertility benefits from pigeonpea systems could lead to these being the top choice. The value farmers place on vegetation (for fuelwood, fodder or soil fertility), matters in terms of system performance ranking. Rusinamhodzi et al. (2012) found that farmers prioritize cropping systems with multiple benefits such as food security, combined with cash income that can be generated with low input costs. This may be uniquely addressed by the doubled-up legume system where half-rates of fertilizer supported high maize grain in our study. What is the ‘best bet’ may change in the near future as a result of climate change. An adaptability study conducted in central Malawi on over a dozen maize-based systems provided evidence of superior maize yield stability in grain legume rotations relative to both continuous maize and 24 intercrops (Chimonyo et al., 2019a). This suggests there may be a resilience value associated with rotational systems that require attention as climatic risk increases. Conclusions This research is an important contribution to the scarce knowledge on the performance of legume-based intercrops vs. legume-maize rotations in rain-fed, on-farm trials from different environmental contexts. This study provides evidence that in a drought, pigeonpea is highly sensitive to interspecific competition. Therefore, maximizing growth for the production of pigeonpea grain, fuelwood, forage, and soil fertility benefits would require sole pigeonpea which could then be rotated with maize. At the same time which systems farmers prefer will depend on their goals, resources and the environment. If a farmer’s goal is to obtain maize and some legume grain every year, the ‘best bet’ system might be the currently popular maize/pigeonpea intercrop. Groundnut was not sensitive to interspecific competition. Therefore, if farmers have sufficient land for sole legumes, the novel doubled-up groundnut/pigeonpea system could help farmers meet multiple goals. Farmers would get nutritiously enriched groundnut grain along with environmental benefits of pigeonpea for a subsequent maize crop. The relative advantage of each component would depend on the site. This research guides recommendations for context- specific cropping systems which enhance farm productivity and food security in Malawi. 25 REFERENCES 26 REFERENCES Adu-Gyamfi, J.J., Myaka, F. a., Sakala, W.D., Odgaard, R., Vesterager, J.M., Høgh-Jensen, H., 2007. Biological nitrogen fixation and nitrogen and phosphorus budgets in farmer- managed intercrops of maize-pigeonpea in semi-arid southern and eastern Africa. Plant Soil 295, 127–136. https://doi.org/10.1007/s11104-007-9270-0 Agyare, W.A., Clottey, V.A., Mercer-Quarshie, H., Kombiok, J.M., 2006. Maize yield response in a long-term rotation and intercropping systems in the Guinea Savannah zone of Northern Ghana. J. Agron. 5, 232–238. https://doi.org/10.3923/ja.2006.232.238 Bhuva, H.M., Kumawat, P.D., Mehta, A.C., Chaudhari, N.N., Patel, P.R., 2017. Effect of groundnut ( Arachis hypogaea L.)-based intercropping systems on yield and economics under rainfed condition. Indian J. Agric. Res. 51, 448–452. https://doi.org/10.18805/ijare.a-4823 Chimonyo, V.G.P., Snapp, S.S., Chikowo, R., 2019. Grain Legumes Increase Yield Stability in Maize Based Cropping Systems. Crop Sci. 59, 1222. https://doi.org/10.2135/cropsci2018.09.0532 Chirwa, P.W., Black, C.R., Ong, C.K., Maghembe, J.A., 2003. Tree and crop productivity in gliricidia/maize/pigeonpea cropping systems in southern Malawi, Agroforestry Systems. 265-277. https://doi.org/https://doi.org/10.1023/B:AGFO.0000005227.69260.f9 Dalal, R.C., 1974. Effects of intercropping maize with pigeonpeas on grain yield and nutrient uptake. Exp. Agric. 10 83–90. https://doi.org/10.1017/S0014479700000454 Ghosh, P.K., Mohanty, M., Bandyopadhyay, K.K., Painuli, D.K., Misra, A.K., 2006. Growth, competition, yields advantage and economics in soybean/pigeonpea intercropping system in semi-arid tropics of India: II. Effect of nutrient management. F. Crop. Res. 96, 90–97. https://doi.org/10.1016/J.FCR.2005.05.010 Giller, K.E.E., Cadisch, G., 1995. Future benefits from biological nitrogen fixation: An ecological approach to agriculture. Plant Soil 174, 255–277. https://doi.org/10.1007/BF00032251 Gwenambira, C.P., 2015. Below- and aboveground pigeonpea productivity in on-farm sole nd intercrop systems in central Malawi. Michigan State Univ. Crop Soil Sci. Thesis. 1-75. Hockett, M., Richardson, R.B., 2018. Examining the drivers of agricultural experimentation among smallholder farmers in Malawi. Exp. Agric. 54, 45–65. https://doi.org/10.1017/S0014479716000673 27 Høgh-Jensen, Myaka, F.A., Sakala, W.D., Kamalongo, D., Ngwira, A., Odgaard R, Adu-Gyamfi, 2007. Yields and qualities of pigeonpea varieties grown under smallholder farmers’ conditions in Eastern and Southern Africa, African Journal of Agricultural Research. Isaacs, K.B., Snapp, S.S., Chung, K., Waldman, K.B., 2016. Assessing the value of diverse cropping systems under a new agricultural policy environment in Rwanda. Food Secur. 8, 491– 506. https://doi.org/10.1007/s12571-016-0582-x Kane, D.A., Snapp, S.S., Davis, A.S., 2015. Ridge Tillage Concentrates Potentially Mineralizable Soil Nitrogen, Facilitating Maize Nitrogen Uptake. Soil Sci. Soc. Am. J. 79, 81. https://doi.org/10.2136/sssaj2014.07.0273 Kell, D.B., 2012. Large-scale sequestration of atmospheric carbon via plant roots in natural and agricultural ecosystems: why and how. Philos. Trans. R. Soc. B Biol. Sci. 367, 1589–1597. https://doi.org/10.1098/rstb.2011.0244 Kellogg Biological Station/ Long-Term Ecological Research, 2016. Protocol - Soil pH [WWW Document]. URL https://lter.kbs.msu.edu/protocols/163 (accessed 9.6.19). Kellogg Biological Station/ Long-Term Ecological Research, 2008. Particle Size Analysis for Soil Texture Determination (Hydrometer Method) [WWW Document]. URL https://lter.kbs.msu.edu/protocols/108 (accessed 9.2.19). Komarek, A.M., Koo, J., Haile, B., Msangi, S., Azzarri, C., 2018. Trade-offs and synergies between yield, labor, profit, and risk in Malawian maize-based cropping systems. Agron. Sustain. Dev. 38, 32. https://doi.org/10.1007/s13593-018-0506-6 Kumar Rao, J.V.D.K., Dart, P.J., 1987. Nodulation, nitrogen fixation and nitrogen uptake in pigeonpea (Cajanus cajan (L.) Millsp) of different maturity groups. Plant Soil 99, 255– 266. https://doi.org/10.1007/BF02370872 Malézieux, E., Crozat, Y., Dupraz, C., Laurans, M., Makowski, D., Ozier-Lafontaine, H., Rapidel, B., Tourdonnet, S., Valantin-Morison, M., 2009. Mixing plant species in cropping systems: concepts, tools and models. A review. Agron. Sustain. Dev. 29, 43–62. https://doi.org/10.1051/agro:2007057 Mapfumo, P., Mtambanengwe, F., Giller, K.E., Mpepereki, S., 2005. Tapping indigenous herbaceous legumes for soil fertility management by resource-poor farmers in Zimbabwe. Agric. Ecosyst. Environ. 109, 221–233. https://doi.org/10.1016/J.AGEE.2005.03.015 Mungai, L.M., Snapp, S., Messina, J.P., Chikowo, R., Smith, A., Anders, E., Richardson, R.B., Li, G., 2016. Smallholder Farms and the Potential for Sustainable Intensification. Front. Plant Sci. 7, 1–17. https://doi.org/10.3389/fpls.2016.01720 28 Myaka, F.M., Sakala, W.D., Adu-Gyamfi, J.J., Kamalongo, D., Ngwira, A., Odgaard, R., Nielsen, N.E., Høgh-Jensen, H., 2006. Yields and accumulations of N and P in farmer-managed intercrops of maize–pigeonpea in semi-arid Africa. Plant Soil 285, 207–220. https://doi.org/10.1007/s11104-006-9006-6 Odeny, D.A., 2007. The potential of pigeonpea (Cajanus cajan (L.) Millsp.) in Africa. Nat. Resour. Forum 31, 297–305. https://doi.org/10.1111/j.1477-8947.2007.00157.x Ollenburger, M., Snapp, S., 2014. Model Applications for Sustainable Intensification of Maize- Based Smallholder Cropping in a Changing World. Pract. Appl. Agric. Syst. Model. to Optim. Use Ltd. Water 5, 375–398. https://doi.org/10.2134/advagricsystmodel5.c14 Orr, A., Kambombo, B., Roth, C., Harris, D., Doyle, V., 2015. Adoption of integrated food-energy systems: Improved cookstoves and pigeonpea in southern Malawi. Exp. Agric. 51, 191– 209. https://doi.org/10.1017/S0014479714000222 Rao, M.R., Mathuva, M.N., 2000. Legumes for improving maize yields and income in semi-arid Kenya. Agric. Ecosyst. Environ. 78, 123–137. https://doi.org/10.1016/S0167- 8809(99)00125-5 Raseduzzaman, M., Jensen, E.S., 2017. Does intercropping enhance yield stability in arable crop production? A meta-analysis. Eur. J. Agron. 91, 25–33. https://doi.org/10.1016/J.EJA.2017.09.009 Rusinamhodzi, L., Corbeels, M., Nyamangara, J., Giller, K.E., 2012. Maize–grain legume intercropping is an attractive option for ecological intensification that reduces climatic risk for smallholder farmers in central Mozambique. F. Crop. Res. 136, 12–22. https://doi.org/10.1016/j.fcr.2012.07.014 Sekiya, N., Yano, K., 2004. Do pigeon pea and sesbania supply groundwater to intercropped maize through hydraulic lift?—Hydrogen stable isotope investigation of xylem waters. F. Crop. Res. 86, 167–173. https://doi.org/10.1016/J.FCR.2003.08.007 Silberg, T.R., Richardson, R.B., Hockett, M., Snapp, S.S., 2017. Maize-legume intercropping in central Malawi: determinants of practice. Int. J. Agric. Sustain. 15, 662–680. https://doi.org/10.1080/14735903.2017.1375070 Smith, A., Snapp, S., Dimes, J., Gwenambira, C., Chikowo, R., 2016. Doubled-up legume rotations improve soil fertility and maintain productivity under variable conditions in maize-based cropping systems in Malawi. Agric. Syst. 145, 139–149. https://doi.org/10.1016/j.agsy.2016.03.008 29 Snapp, S.S., Grabowski, P., Chikowo, R., Smith, A., Anders, E., Sirrine, D., Chimonyo, V., Bekunda, M., 2018. Maize yield and profitability tradeoffs with social, human and environmental performance: Is sustainable intensification feasible? Agric. Syst. 162, 77– 88. https://doi.org/10.1016/J.AGSY.2018.01.012 Snapp, S.S., Silim, S.N., 2002. Farmer preferences and legume intensification for low nutrient environments. Plant Soil. https://doi.org/10.1023/A:1020658715648 Tully, K.L., Wood, S.A., Almaraz, M., Neill, C., Palm, C., 2015. The effect of mineral and organic nutrient input on yields and nitrogen balances in western Kenya. Agric. Ecosyst. Environ. https://doi.org/10.1016/j.agee.2015.08.006 Waldman, K.B., Ortega, D.L., Richardson, R.B., Snapp, S.S., 2017. Estimating demand for perennial pigeon pea in Malawi using choice experiments. Ecol. Econ. 131, 222–230. https://doi.org/10.1016/J.ECOLECON.2016.09.006 Wendt, J.W., Atemkeng, M.F., 2004. Soybean, cowpea, groundnut, and pigeonpea response to soils, rainfall, and cropping season in the forest margins of Cameroon. Plant Soil 263, 121–132. https://doi.org/10.1023/B:PLSO.0000047731.35668.e0 30 2. Chapter Two: Pigeonpea and groundnut productivity, biological nitrogen fixation, and residual benefits to maize on smallholder farms in central Malawi Abstract Smallholder farmers rarely achieve access to adequate quantities of inorganic fertilizers when nitrogen (N) is often the most limiting element for cereal grain production. Legumes such as groundnut (Arachis hypogaea L.) and pigeonpea (Cajanus cajan L.) fix N through biological nitrogen fixation (BNF) and potentially improve soil fertility and productivity of cereal crops grown in sequence. However, legume BNF benefits under smallholder, on-farm conditions are understudied. The objective of this study was to determine the effect of location and cropping system on yield and BNF of sole legumes, doubled-up legumes and cereal-legume cropping systems on rain-fed smallholder farms. On-farm experiments were conducted at six locations in in high, low and medium agro-ecological potential zones in central Malawi. The study evaluated four cropping systems: 1) sole groundnut 2) sole pigeonpea 3) groundnut/pigeonpea intercrop (referred to as a doubled-up legume system) and 4) maize/pigeonpea intercrop which is the farmer check system. All the systems were rotated with maize in the second year. Overall, total N input by cropping system was in the order groundnut/pigeonpea> sole pigeonpea> maize/pigeonpea> sole groundnut at 180, 130, 103 and 87 kg N/ha respectively. The mean proportion of BNF was 66% for groundnut and 52% for pigeonpea. Our study highlights that in maize-based systems, there is need for integrated soil fertility approaches that include legume intercrops, rotations, residue incorporation, and modest inorganic N inputs for sustainable maize 31 production. Keywords: Biological nitrogen fixation, Cropping system, Groundnut, Intercrop, On-farm, Pigeonpea, 15N natural abundance. Introduction Sustainable agriculture through BNF BNF is an important and significant part of the global N cycle. Annually, BNF is estimated to be around 175 million tones N, of which close to 79% is from terrestrial fixation. Sustainable agriculture depends on renewable resources like BNF which plays an important role in maintaining soil fertility through economically and ecologically sound means of reducing external N inputs and improving the quality and quantity of internal resources (Wani et al., 1995). BNF is dependent upon environmental, physical and biological factors. Inclusion of N fixing plants in agricultural systems does not automatically result in increased contributions to the soil N pool (Wani et al., 1995). However, in developing countries where fertilizers are expensive and inaccessible to many, legumes are the main source of N, through BNF (Giller and Cadisch, 1995). Malawi legume crops and nitrogen balance on-farm Smallholder farm sustainability is challenged in Malawi, as it is across much of sub-Saharan Africa, by the overwhelming presence of maize, a highly N-demanding cereal crop. Legume diversification has been studied in some detail as a means to improve sustainability of 32 smallholder maize-based farming. Although long-duration agroforestry and green manure systems show sustainable potential to redress the nitrogen deficit imposed by maize production, smallholder farm adoption of such systems has been almost nil (Sirrine et al., 2010), presumably due to associated risks and labor requirements. Farmer interest in short duration grain legume species such as common bean and groundnut is often much higher than long-duration indeterminant species, although medium to long duration cultivars of pigeonpea are in some cases a notable exception (Waldman et al., 2017). A number of grain legumes are grown in Malawi either as sole crops or in intercrops with cereals, particularly maize. The most commonly cultivated legumes in Malawi are common bean (Phaseolus vulgaris L.) and groundnut (Snapp et al., 2002). Pigeonpea is mostly grown in southern Malawi where it is traditionally intercropped with maize (Orr et al., 2015). A doubled-up legume system of complementary growth habits has been introduced in Malawi, whereby a short duration groundnut is grown as an under species to a shrubby, long duration pigeonpea species (Snapp et al., 2010). What is not well known is the extent to which this legume diversified system can address the N balance deficit on smallholder farms, within maize-based farming systems, and the interaction with environmental context. Pigeonpea and groundnut BNF Pigeonpea can fix substantial amounts of N that can reach up to 235 kg N/ha (Peoples et al., 1995). According to Kumar Rao and Dart (1987), pigeonpea has excellent nodulation and fixes most of its N from the atmosphere. Late maturing pigeonpea varieties have much to offer since they produce more biomass and fix more N. Pigeonpea sheds its leaves throughout the growing season and senesced pigeonpea leaves from late maturing varieties can contribute substantial 33 amounts of about 28–40 kg N (Kumar Rao and Dart, 1987). Kumar Rao & Dart (1987) reported that in India, pigeonpea fixed about 58 to 88 kg N/ha when grown as a sole crop and may provide 30 to 50 kg/ha of residual N in crop rotations with maize, wheat and sorghum. In a multi-country, on-farm study, Adu-Gyamfi et al. (2007) reported that in Malawi, pigeonpea contributed about 38 -117 kg/ha of N through BNF while in Tanzania it was about 6 -72 kg/ha. Groundnut has been shown to fix substantial quantities of N from the atmosphere under favorable conditions in the tropics ranging between 41 and 150kg N/ha (Giller et al. 1987; Toomsan et al., 1995; Okito et al., 2004). From a study in northeast Thailand, Toomsan et al. (1995) reported that groundnut fixed 72-77% of its N amounting to 150-200 kg N/ha. Giller et al. (1987) also reported that for groundnut, 86% to 92% of its plant nitrogen was derived from BNF. However, the net N balance associated with growing groundnut in a rotation sequence is often low as whole plants are often uprooted during harvesting, a practice that can remove roots and nodules from the field. The quantity of N that is contributed from groundnut BNF is variable in the literature. It can be as high as 86-99% N fixation as shown in a field study in India (Giller et al., 1987), an outdoor basin study in Sri Lanka (Senaratne et al., 1995) and an on-farm study in northern Malawi (Mhango et al., 2017). Other studies such as a field study from Thailand suggest modest rates ranging from 59-64% (Mcdonagh et al., 1993). This may well be a result of BNF being a function of local agro-ecological conditions and varieties (Giller et al., 1987). 34 Additionally, in most previous field experiments, for both pigeonpea and groundnut, belowground legume N fixation is not measured. To address these research gaps, detailed below and aboveground assessments of BNF in mixed cropping systems on smallholder farms were conducted. Overall, the objectives of this study were 1) To quantify the N fixation associated with sole pigeonpea, sole groundnut, pigeonpea/groundnut and pigeonpea/maize cropping systems 2) To determine if N fixation rate in pigeonpea and groundnut are reduced when grown in higher fertility soil or increased when grown in a maize intercrop due to competition for N by the maize which stimulates BNF in pigeonpea. The hypotheses of this study were 1) Cropping system will have an effect on total N and will be in the order: groundnut/pigeonpea> sole pigeonpea> sole groundnut> maize/pigeonpea cropping systems and 2) Groundnut and pigeonpea BNF will decrease with increasing agro-ecology crop production potential and soil fertility. Materials and methods Study sites and field sites characterization A stratified random approach was used to select the research sites to represent marginal to mesic potential areas (Mungai et al., 2016) in central Malawi (13.2543° S, 34.3015° E). The locations were in the Linthipe, Golomoti and Kandeu Extension Planning Areas (EPA). Linthipe is a high potential, sub-humid tropical agroecology with well-distributed rainfall in most years. Kandeu is a medium agricultural potential, sub-humid tropical agroecology with decent rainfall distribution in most years. Golomoti is a low potential, semi-arid to sub-humid tropical agroecology, at low altitude and with erratic rainfall; high temperatures increase evapotranspiration at this site. 35 Linthipe soils are ferric luvisols, the Kandeu study site has a mixture of chromic luvisols and orthic ferralsols while Golomoti soils are a mixture of eutric cambisols and eutric fluvisols (Mungai et al., 2016). The field experiments were conducted at six locations (Table 2.1). Soil sampling was conducted in late June and early July of 2016, at 0– 20 cm depth, three composite samples per plot using a dutch auger (5.5 cm diameter). Soils were air-dried, sieved (2 mm) and soil texture was determined using the hydrometer method (Kellogg Biological Station/ Long-Term Ecological Research, 2008). Soil pH was determined in a 1:2 soil to water suspension (Kellogg Biological Station/ Long-Term Ecological Research, 2016). Total soil carbon and total soil N were determined after grinding to a fine powder with a shatter mill, by dry combustion with Costech ECS 4010 (Costech Analytical Technologies, Inc., Valencia, CA). To determine potential mineralizable NH4- N (PMN), soil samples were anaerobically incubated at 30 °C for seven days (Kane et al., 2015). After incubation, NH4-N was extracted using 2.66 M KCl. KCl extracts were analysed for inorganic N concentrations on a Thermo MultiskanTM 96-well plate reader (Kane et al., 2015). Experimental design Four cropping system treatments were set up in a randomized complete block design with four replicates per location. The four cropping systems included in this study consisted of sole pigeonpea (PP), sole groundnut (GN), groundnut/pigeonpea doubled-up intercrop (GNPP), and the maize/pigeonpea intercrop system (MZPP). The traditional maize/pigeonpea intercrop (MZPP) system was included as the control. 36 Management of cropping systems Field experiments were set up during the 2015/16 growing season. Crops were planted in December of 2015, and the productivity of the legumes was assessed between May and July 2016. During the following cropping season 2016/17, the residual effects of the systems was assessed by planting maize in all plots. Hand hoes were used for land preparation and making ridges because the local practice in this farming system is to make ridges on which rows of crops are planted. Plot sizes were 5m x 5m and ridges were spaced at 75 cm apart. Fields were planted in December after the first effective rains. Planting at each of the sites was completed on each day. Sole pigeonpea was planted at a spacing of 0.9 m x 0.75 m, with three plants per planting station to achieve a plant population density of 44 000/ha. Sole groundnut within row spacing was 0.1 m to achieve a plant population density of 120 000 plants/ha. An additive intercropping design was used for the pigeonpea/groundnut intercrop to attain a total plant population density of about 164 000 plants/ha. In the pigeonpea/maize cropping system, both maize and pigeonpea were planted at a spacing of 0.9 m x 0.75 m in an additive intercropping design. There were three plants per planting station for both pigeonpea and maize. In intercrop systems, the associated crops were planted on the same ridge. The doubled-up legume intercrops were fertilized just before planting with 23:21 N: P compound fertilizer at the rate of 11.5 kg/ha N and 10.5 kg/ha of P. The pigeonpea/maize intercrop was fertilized at the rate of 23 kg ha-1 N and 21 kg/ha, with a side-dress application of urea at 100 kg/ha which provided 46 kg/ha N. The plots were weeded by hand hoes three times at each location. Plant population density assessment in all plots was done at harvesting to determine 37 final plant population densities. Aboveground and belowground biomass and yield Destructive harvest timing varied with plant species and location to ensure sampling was done when plants were at physiological maturity. For groundnut, it was from early-mid May 2016 and for pigeonpea, it was in July 2016. Harvested net plots for groundnut destructive sampling were two ridges x 2m for a total area of 2.25 m2. Groundnut biomass was separated into stems, leaves, pods, and roots. For pigeonpea, four plants were randomly selected per plot, cut at ground level, chopped, and fresh biomass was determined for aboveground biomass separated into stems, twigs, leaves, and pods. The legume plant samples were oven-dried at 75 °C to constant weight, to determine dry weights. For belowground biomass determinations, destructive root sampling was conducted during harvesting. Pigeonpea plants were cut at ground level and roots were excavated using hand hoes. Soil pits were dug to a depth of 60 cm and roots were collected in three increments from the 0- 20 cm, 20-40 cm and 40-60 cm soil layers. Roots from each layer were collected by first completing excavating the 20 cm (depth) x 90 cm (intrarow spacing) x 75 cm (interrow spacing) layer (Figure 2.1). When all roots were collected from the three layers from the 0-60 cm soil depths, the total soil volume was 0.405 m3. Large roots were removed through dry sieving (2 mm sieve). Fine roots were hand-picked using tweezers. In intercrops, legume roots were identified through their characteristic smell – ‘the Gwezu smell technique” (Mapfumo et al., 2005), a method which was found to be straight forward to distinguish pigeonpea in a previous study 38 (Gwenambira, 2015). The roots were weighed fresh, oven-dried at 75 oC to constant weight and reweighed for dry weight determination. Figure 2.1. Belowground biomass destructive sampling pit measuring 90 cm x 75 cm x 60 cm. Hand hoes were used to excavate roots from the pit. Roots were collected in increments from the 0-20, 20-40 and 40-60 cm soil depths. Sketch prepared by Kibale V. Mwika. Biological nitrogen fixation Shoot, root and plant residue components were ground to pass through a 1mm sieve, and the amount of nitrogen biologically fixed by pigeonpea was determined using the 15N natural abundance method. The stable isotope analysis was conducted at the UC Davis stable isotope facility. Maize was used as the non-fixing reference crop. The calculation that was used for BNF is: 39 % N2-fixation = 100 x d15N (reference crop: maize) - d15N (legume N)/ (d15N (reference crop) – B) (Hogberg, 1997) Statistical analysis Data were analyzed using the MIXED procedure in SAS 9.4 (SAS Institute, 2002) statistical package. Two-way analysis of variance (ANOVA) tests were used to determine the combined effect of cropping system and location on the percent N, total N and BNF-N percent of groundnut and pigeonpea shoot and root biomass, where sole-cropped and intercropped groundnut and pigeonpea cropping systems were compared. The Least Significant Difference (LSD) at 5% level of significance was used to test mean differences. 40 Results Environment The soil properties at the six locations are shown in Table 2.1. Overall, soils at four of the sites were moderately acidic as indicated by values that ranged from 5.4 to 5.9. The drier sites in Golomoti were close to neutral pH with values all from 6.2-6.4. The topsoil organic matter content was low at all but one site, as indicated by soil organic C levels of 0.43 – 1.5 %C for five sites, and 3.2 %C at Linthipe A. Soil total N % was low with a range of 0.04 % in Kandeu B to 0.19 % in Linthipe A. Potential mineralized nitrogen was low at all six locations and ranged from 0.22 at Golomoti A to 0.94 mg/kg soil at Kandeu A. Generally, the sites were sandy, with the highest clay content in the 0-20 cm soil depth being 30 % at the Linthipe A location and the lowest clay content 12 % at Golomoti B and Kandeu B (Table 2.1). Table 2.1. Soil chemical and physical properties at research locations (0-20 cm) from mesic (Linthipe A and B), marginal (Golomoti A and B) and moderate (Kandeu A and B) sites in central Malawi from sampling conducted in June of 2016. Values are means followed by standard deviations. Location N (%) C (%) PMN (mg/kg soil) pH Clay (%) Sand (%) Linthipe A Linthipe B Golomoti A Golomoti B Kandeu A Kandeu B 0.19 ± 0.03 0.09 ± 0.01 0.06 ± 0.01 0.07 ± 0.02 0.07 ± 0.01 0.04 ± 0.01 3.20 ± 0.61 1.46 ± 0.18 0.92 ± 0.21 0.87 ± 0.23 0.85 ± 0.19 0.43 ± 0.08 0.67 ± 0.40 0.20 ± 0.16 0.22 ± 0.19 0.40 ± 0.31 0.94 ± 0.25 0.38 ± 0.22 5.74 ± 0.17 5.41 ± 0.29 6.16 ± 0.19 6.38 ± 0.23 5.88 ± 0.26 5.54 ± 0.31 30 ± 3.80 19 ± 7.35 17 ± 5.55 12 ± 1.60 14 ± 3.07 12 ± 4.12 59 ± 5.84 73 ± 11.05 76 ± 6.04 83 ± 1.97 81 ± 3.03 83 ± 4.84 41 Crop biomass production Total groundnut and pigeonpea biomass is presented in Table 2.2. For groundnut, total biomass is defined as the total sum of stem, leaf, grain and root biomass. For pigeonpea, total biomass is defined as the total sum of stem, twig, leaf and pod biomass. Groundnut total biomass in the sole groundnut cropping system ranged from 1.86 Mg/ha in Golomoti B to 4.88 Mg/ha in Linthipe B. For the groundnut/pigeonpea intercrop, groundnut produced biomass which ranged from 2.09 Mg/ha in Golomoti B to 4.91Mg/ha in Linthipe B. Location had a significant effect on groundnut biomass production (Pr>F = <.0001) but cropping system did not. Pigeonpea total biomass including grain ranged from 4.11 Mg/ha to 12.33 Mg/ha in the sole pigeonpea cropping system, 2.91 Mg/ha to 10.55 Mg/ha for groundnut/pigeonpea, and from 2.85 Mg/ha to 8.63 Mg/ha for the maize/pigeonpea intercrop. Location and cropping system had significant effects on pigeonpea total shoot biomass but there was no location x cropping system interaction. Sole cropped pigeonpea biomass was greater than pigeonpea grown in an intercrop (Table 2.2). 42 Table 2.2. Analysis of variance for groundnutf and pigeonpea total biomass including grain in six locations across central Malawi during the 2016 growing season. Presented values are means followed by standard deviations. Cropping systems shown are sole groundnut (GN), sole pigeonpea (PP), groundnut/pigeonpea intercrop (GNPP), and maize/pigeonpea intercrop (MZPP) Cropping System GN GNPP (GN) PP GNPP (PP) MZPP (PP) Linthipe A 2.65 ± 0.55 2.22 ± 0.30 4.11 ± 1.57 2.91 ± 1.60 2.87 ± 0.65 Linthipe B Golomoti A Golomoti B Kandeu A Kandeu B 4.88 ± 0.53 4.91 ± 0.55 8.44 ± 2.15 3.53 ± 1.08 6.19 ± 0.98 Mg/ha 3.02 ± 0.65 2.55 ± 0.13 12.33 ± 1.88 6.69 ± 3.24 8.05 ± 5.01 1.86 ± 0.65 2.09 ± 0.53 7.37 ± 1.74 4.85 ± 1.23 8.09 ± 3.20 3.77 ± 0.41 3.63 ± 0.80 9.43 ± 1.02 10.55 ± 2.09 8.63 ± 2.14 3.24 ± 0.38 2.93 ± 0.75 4.14 ± 1.06 3.64 ± 1.05 2.85 ± 0.50 Groundnut Pigeonpea ANOVAS Location Cropping System Location x Cropping System Pr>F = <.0001* Pr>F = 0.3341 Pr>F = 0.8626 fgroundnut biomass includes roots; * Significant at P=0.05 Pr>F = <.0001* Pr>F = 0.0080* Pr>F = 0.1573 43 Percent N of groundnut and pigeonpea Tissue N concentration is presented in Table 2.3. Percent N of groundnut stems ranged from 1.26% (Golomoti A) to 2.38% (Linthipe A), groundnut leaves from 2.63% (Kandeu A) to 3.31 % (Linthipe A), groundnut grain from 3.78% (Kandeu A) to 5.02% (Linthipe A) and roots from 1.76% (Golomoti A) to 2.96% at the Linthipe A location (Table 2.3). Location had an effect on the percent N of groundnut stems, leaves and roots but not on grain. Linthipe A was consistently associated with high %N. Cropping system had an effect on groundnut leaves. The concentration in leaves from sole cropped groundnut was significantly higher than intercropped groundnut (Table 2.4). There was no location x cropping system interaction on any groundnut plant parts (Table 2.3). Pigeonpea stems percent N ranged from 1.03% (Golomoti) to 1.16% (Kandeu), leaves from 2.82% (Linthipe A) to 3.43% (Kandeu A), pods from 2.43% (Kandeu B) to 3.11% (Kandeu A), and roots from 0.89% (Linthipe B) to 1.40% (Golomoti B). Location had an effect on the percent N of pigeonpea leaves, pods and roots but not stems. The medium potential, Kandeu A location was consistently associated with high %N. Cropping system had an effect on stems N concentration which was in the order maize/pigeonpea>groundnut/pigeonpea>sole pigeonpea at 1.19, 1.16 and 0.93% respectively. In addition, there was a location x cropping system effect on pigeonpea stems, leaves, and roots (Table 2.3). 44 Table 2.3. Analysis of variance for groundnut and pigeonpea percent N at six locations in Malawi during the 2016 growing season. Presented values are means followed by standard deviations. *Significant at P=0.05; **Significant at P=0.10 45 Table 2.4. Percent N and total N means and standards deviations of groundnut and pigeonpea in four cropping systems in central Malawi. Cropping systems are sole groundnut (GN), sole pigeonpea (PP), doubled-up groundnut/pigeonpea intercrop (GNPP), and maize/pigeonpea intercrop (MZPP) Cropping Leaves % N Leaves Total Pods % N Pods Total N Roots % N Roots Total N system GN PP N kg/ha kg/ha kg/ha 2.99 ± 0.35 8.86 ± 6.82 4.10 ± 1.46 48.96 ± 27.71 2.21 ± 0.50 2.41 ± 1.26 3.08 ± 0.24 23.94 ± 13.71 2.87 ± 0.39 9.47 ± 9.18 1.21 ± 0.30 1.21 ± 0.30 MZPP (PP) 3.09 ± 0.24 21.31 ± 12.43 2.86 ± 0.55 8.73 ± 8.85 1.14 ± 0.31 1.14 ± 0.31 GNPP (GN) 2.78 ± 0.33 5.95 ± 5.08 4.68 ± 0.37 55.29 ± 27.75 2.19 ± 0.51 2.41 ± 1.10 GNPP (PP) 3.05 ± 0.29 17.19 ± 8.84 2.87 ± 0.37 7.49 ± 7.52 1.09 ± 0.28 1.09 ± 0.28 GNPP (GN+PP) 23.14 62.78 3.50 46 Total nitrogen fixed by groundnut and pigeonpea The total N that was contained in groundnut stems ranged from 11.65 kg/ha (Golomoti A) to 38.74 kg/ha (Linthipe B), leaves from 2.85 kg/ha (Linthipe A) to 15.57 kg/ha (Golomoti B), and roots from 0.91 kg/ha (Kandeu B) to 3.82 kg/ha (Golomoti B). The protein contribution of the groundnut cropping systems in terms of total N contained in the grain ranged from 22.34 kg/ha (Golomoti A) to 78.73 kg/ha (Golomoti B). Location had an effect on all groundnut plant parts while cropping system had an effect on groundnut leaves only. The low potential, hot and dry Golomoti B location was consistently associated with high total N. Total N contained in groundnut leaves was higher in sole groundnut compared to intercropped groundnut (Table 2.4). There was no location x cropping system effect on any groundnut plant parts (Table 2.5). Total N contained in pigeonpea stems ranged from 31.48 kg/ha (Kandeu B) to 109.42 kg/ha (Kandeu A), 13.00 kg/ha (Linthipe A) to 35.82 kg/ha (Linthipe B) for leaves, 1.38 kg/ha (GB) to 15.54 (Kandeu B) for pods, and 3.61 kg/ha (Golomoti B) to 14.86 kg/ha (Kandeu A) for roots. Location had an effect on all pigeonpea plant parts and cropping system had an effect on stems, leaves, and roots but not pods. Total N contained in pigeonpea leaves, pods and roots was in the order sole pigeonpea>maize/pigeonpea>groundnut/pigeonpea (Table 2.4) There was a location x cropping system interaction on leaves and roots (Table 2.5). The total N contained in different groundnut and pigeonpea plant parts in four cropping systems is shown on Figure 2.2. Overall, groundnut total fixed N was 87.01 kg/ha in the sole groundnut cropping system. Pigeonpea total fixed N was 129.54 kg/ha in the sole pigeonpea cropping system and 102.69 kg/ha in the maize/pigeonpea cropping system. When combined, groundnut 47 and pigeonpea total fixed N was a 179.64 kg/ha in the groundnut/pigeonpea cropping system. Therefore, total N by cropping system was in the order groundnut/pigeonpea>sole pigeonpea>maize/pigeonpea>sole groundnut (Figure 2.2). 48 Table 2.5. Analysis of variance for groundnut and pigeonpea total N at six locations in Malawi during the 2016 growing season. Presented values are means followed by standard deviations. f Pigeonpea stems include twigs; *Significant at P=0.05; **Significant at P=0.10 49 Figure 2.2. Total nitrogen in kg/ha by plant parts in four cropping systems in central Malawi. Cropping systems (PP), doubled-up groundnut/pigeonpea intercrop (GNPP), and maize/pigeonpea intercrop (MZPP). include are groundnut (GN), sole pigeonpea Groundnut and pigeonpea BNF-N % The proportion of fixed N that was located in groundnut stems ranged from 48.65% (Linthipe A) to 89.24% (Kandeu B), leaves from 39.99% (Kandeu A) to 83.42% (Linthipe B), grain from 39.52% (Kandeu B) to 79.37% (Linthipe B), and roots from 25.72% (Kandeu B) to 97.86% at the Linthipe B location (Table 2.6). Location had an effect on the BNF-N % of all groundnut plant parts. The high potential Linthipe B location was consistently associated with high BNF-N %. Cropping system had an effect on leaves at P=0.10 and there was a location x cropping system interaction on leaves (Table 2.6). Intercropped groundnut had higher BNF-N% than sole cropped groundnut. Overall, the mean proportion of groundnut BNF was 65.83% (± 21.99). 50 For pigeonpea stems, BNF-N % ranged from 75.74% (Golomoti) to 78.99% (Linthipe), leaves from 33.21% (Linthipe A) to 91.81% (Kandeu A), pods from 31.35% (Kandeu A) to 60.48% (Golomoti A), and roots from 13.13% (Kandeu A) to 51.81% at the Linthipe B location (Table 2.6). Location had an effect on the BNF-N % in pigeonpea leaves, pods and roots but not stems. Cropping system had an effect on pigeonpea leaves and pods. For pigeonpea leaves, BNF-N% was in the order groundnut/pigeonpea>sole pigeonpea=maize/pigeonpea and for pods it was maize/pigeonpea =groundnut/pigeonpea>sole pigeonpea. There was a location x cropping system interaction on pigeonpea leaves and pods as well (Table 2.6). Overall, the mean proportion of pigeonpea BNF was 52.01% (± 29.03) for pigeonpea. 51 Table 2.6. Analysis of variance for groundnut and pigeonpea BNF-N % at six locations in Malawi during the 2016 growing season. Presented values are means followed by standard deviations. -Data not available; *Significant at P=0.05; **Significant at P=0.10 52 N dynamics by species Discussion This is one of the first detailed reports of N allocation to legume plant tissues within intercropped and sole cropped systems on smallholder farms. This study is unique as it documented BNF contributions from not only aboveground biomass of groundnut and pigeonpea but belowground biomass as well. Additionally, the study assesses nitrogen allocation of groundnut and pigeonpea within maize-legume cropping systems at multiple locations which ranged from high to low agricultural potential. We hypothesized that cropping system would have an effect on total N, it would be in the order: groundnut/pigeonpea> sole pigeonpea> sole groundnut> maize/ pigeonpea cropping systems and that BNF would decrease with agro-ecology crop production potential. Cropping system had a significant effect on total N (Figure 2.2) but was in the order: groundnut/pigeonpea> sole pigeonpea> maize/pigeonpea> sole groundnut. However, BNF did not follow a particular trend in terms of agro-ecology potential. This might have been due to the effect of the 2016 drought on plant biomass production and ultimately BNF. The 2016 cumulative rainfall across locations was 55% less than the norm. Overall, the nitrogen accumulated by groundnut and pigeonpea in our study (87 to 180 kg/ha) was in a range similar to that of previous studies (Giller et al., 1987; Mcdonagh et al., 1993; Mhango et al., 2017; Okito et al., 2004; Senaratne et al., 1995; Toomsan et al., 1995). Using the 15N isotope dilution method in a field study in India, Giller et al. (1987) reported 86 to 92% of 53 groundnut N fixation rates amounting to 100 to 153 kg N/ha. Senaratne et al., (1995) also reported high groundnut N fixation of 85% from an outdoor basin study using the isotope dilution method in Sri Lanka. The groundnut BNF-N% mean of 66% in our study was much closer to the 59-64% range reported by Mcdonagh et al., (1993) from a field study in Thailand where the 15N isotope dilution method was used. From researcher managed field trials on farmers’ fields in Thailand, Toomsan et al. (1995) reported 72-77% of fixed N amounting to 150-200 kg N/ha in groundnut using the 15N isotope dilution method. We used the 15N natural abundance method and the sole groundnut cropping system in our study produced much less total N at 87 kg/ha compared to (Toomsan et al., 1995). Mhango et al. (2017) also used the 15N natural abundance method in an on-farm study in northern Malawi and reported total N fixed means of 75% for groundnut and 76% for pigeonpea. However, the authors reported a very wide range of 22-99% N fixation from BNF for pigeonpea and very low total N (below 35 kg N/ha). In a field study on a research station in Brazil using the 15N natural abundance method, Okito et al. (2004) also reported low groundnut contribution from BNF at 40.9 kg N/ha. A unique finding in our study was that groundnut roots had higher %N, %BNF and total N than pigeonpea roots by about 50% or more. This may be related to groundnut roots being located near pods, so N that is assimilated up to the leaves is cycled back down to the pods near the roots when the growing season is ending. In pigeonpea, when N is cycled from leaves to pods, they are not near roots so N might be expected to be lower in pigeonpea roots. Species performance varied by site, in that groundnut grew notably well at the high potential site (Linthipe B), and pigeonpea was more resilient in that it grew well and accumulated N at marginal sites such as 54 Golomoti A. Similarly, in an earlier report on legume-based cropping systems in on-farm trials in central Malawi, it was observed that pigeonpea based systems perform well in marginal environments (Chimonyo et al., 2019b). Crop modeling results support these findings as well, where simulation scenarios conducted at these sites in central Malawi predicted that groundnut responds to adequate moisture with high growth and grain yield, whereas pigeonpea performs well over a wide range of environments including marginal sites such as Golomoti (Snapp et al., 2018) Cropping systems effect There was very limited cropping system effect overall. Groundnut leaf BNF-N% was higher within the doubled-up groundnut/pigeonpea intercrop compared to sole groundnut and there was a significant effect of cropping system (Table 2.6). This was not due to %N or total N being higher in leaves of groundnut in the groundnut/pigeonpea intercrop as actually the opposite was observed. Groundnut leaf %N and total amount of N was higher in sole cropped groundnut than in the groundnut/pigeonpea intercrop. Shading in the doubled-up groundnut/pigeonpea intercrop might have led to a reduction in groundnut N uptake. However, since N fixation was not suppressed, the shaded groundnut plants growing under pigeonpea might rely more on BNF. In an alfalfa/walnut agroforestry study in France, Querné et al. (2017) reported that shading reduced alfalfa yield but significantly increased BNF which they attributed to probable N competition and mineralization inhibition under walnut trees. However, from a review on BNF in 55 cereal-legume intercrops, Fujita et al. (1992) reported that mutual shading of component crops by tall cereals reduces both legume yield and BNF. Cropping system effect on pigeonpea stem N concentration might have been due to higher stem biomass in the sole crop compared to intercropped pigeonpea. The total N contained in pigeonpea stems was high because of high stem biomass compared to leaves, pods and roots. Our findings are similar to Maskey et al. (2001) who reported that total N fixed in various legume crops was primarily influenced by crop growth, crop duration and rainfall. However, N contained in the stems could be discounted in systems N inputs as many smallholder farmers use pigeonpea stems as firewood (Orr et al., 2015). BNF-N% was surprisingly low in the sole pigeonpea cropping system. This could be due to the 2016 drought in Malawi which might have impacted BNF, although the multiple sites reported on this study did vary in terms of rainfall. Previous studies have reported the effect of drought on BNF. From a Thailand research station field study, using the N-difference method, Pimratch et al. (2008) reported reductions in both groundnut biomass production and BNF with increasing levels of drought stress. Mhango et al. (2017) reported that total N differed with cropping system in a drought where intercropped pigeonpea produced lower total N compared to sole pigeonpea. According to Mhango et al. (2017), interspecific competition is one of the drivers of BNF in a pigeonpea-based intercrop especially in a drought. Similar to Katayama et al. (1995), our study shows overall higher BNF-N% in leaves of intercropped pigeonpea (78.37 ± 27.12 in groundnut/pigeonpea) compared to sole pigeonpea (64.47 ± 31.15). Katayama et al. (1995) attributed that to pigeonpea’s increased dependency on 56 BNF when intercropped with cereals in soils with low soil N. However, in our study, this was true when pigeonpea was intercropped with both maize and groundnut. Sustainability implications The proportion of BNF by cropping system is important to consider in terms of sustainability as this indicates reduced reliance on N fertilizer, an external input. Our results showed that sole groundnut produced the least amount of total N fixed compared to sole pigeonpea, maize/pigeonpea, and the groundnut/pigeonpea intercrops. Similar to Njira et al.(2012), the doubled-up groundnut/pigeonpea intercrop was effective at enhancing most nitrogen per unit area (Figure 2.2) compared to sole pigeonpea and sole groundnut. On the one hand, groundnut is highly preferred by many smallholder farmers who rotate it with a maize crop because of the possibility of both groundnut and maize grain (Okito et al., 2004). However, from a simple N balance calculation, Okito et al. (2004) indicated that the groundnut- maize sequence would, in the long term, deplete soil N reserves. In Ghana, a comparison of N fixation and growth of legume rotation and intercrop systems in Northern and Southern Guinea savanna illustrated that groundnut was associated with a net negative N balance at one site but not the other (Kermah et al., 2018). Our results are similar, consistent with groundnut rotation as a potentially N depleting system yet showing potential to contribute N fertility under some conditions. One sustainability challenge highlighted here with groundnut is that roots and nodules are often removed during harvest when plants are pulled out and removed from the field with the shoots, impacting total N and soil fertility (Toomsan et al., 1995). Our finding that 57 root N concentration was high in groundnut further highlights the potential for a nitrogen deficit with this crop. In our study, this could also be the reason why subsequent maize yield after sole groundnut was the lowest (4.06 Mg/ha) compared to maize after groundnut/pigeonpea intercrop (5.00 Mg/ha) and sole pigeonpea (5.51 Mg/ha). The doubled-up legume system which provided farmer-preferred crop of groundnut grain with combination of high total N from both pigeonpea and groundnut. Mcdonagh et al. (1993) demonstrated that positive N balances in groundnut-based systems can be attained if farmers return and incorporate legume residues to the soil. Further in an on-farm Ghana study, only longer duration varieties of groundnut provided a net positive N balance (Kermah et al., 2018).Taken together with our study, these findings highlight that appropriate residue management and incorporation of longer-duration legume species are both key if N sustainability is to be pursued in maize-legume diversified cropping systems without much access to external N inputs. This study demonstrates a clear advantage in terms of soil fertility of the doubled-up groundnut/pigeonpea system is the N accumulation in plant parts, especially stems and leaves and roots (Figure 2.2). Leaf total N was significantly higher in both groundnut and pigeonpea in the doubled-up legume system compared to other cropping systems thereby reinforcing that there is more plant N accumulated in the doubled-up system. Considering that grain and stems harvested for firewood (pigeonpea) are included, the total amount of N fixed in this study, conducted in a relatively dry year, was modest (ranged from 87 to 180 kg/ha). This means that 58 after crop harvesting, the remaining soil N is insufficient to support high productivity in subsequent maize rotation sequences without additional N sources (Njira et al., 2012). Legume rotational systems when integrated with modest doses of fertilizer have been shown to enhance N fertilizer efficiency and stability of maize rotation response on smallholder farms (Snapp et al., 2010). This study extends the earlier investigations of single and double legume technologies within smallholder maize-based systems, through detailed assessment of nitrogen fixation and allocation by plant tissue type as well as species. Conclusions Interspecific legume-based cropping systems alone might not sustain maize productivity and hence the need for integrated soil fertility approaches. Modest amounts of inorganic N combined with legume residue incorporation have the potential to maintain positive N balances. Therefore, there is need for extension messages to reinforce this and educate farmers on long-term sustainability aspects. Additionally, N contributions from belowground biomass have been under appreciated, and the potential negative impacts of groundnut root removal during harvest is highlighted in this study and should be reflected in extension education. The doubled-up legume cropping system shows potential as a means to consistently obtain a positive N balance when grown in rotation with maize based systems. 59 REFERENCES 60 REFERENCES Adu-Gyamfi, J.J., Myaka, F. a., Sakala, W.D., Odgaard, R., Vesterager, J.M., Høgh-Jensen, H., 2007. Biological nitrogen fixation and nitrogen and phosphorus budgets in farmer- managed intercrops of maize-pigeonpea in semi-arid southern and eastern Africa. Plant Soil 295, 127–136. https://doi.org/10.1007/s11104-007-9270-0 Chimonyo, V.G.P., Snapp, S.S., Chikowo, R., 2019. Grain legumes increase yield stability in maize based cropping systems. Crop Sci. 59, 1222–1235. https://doi.org/10.2135/cropsci2018.09.0532 Fujita, K., Ofosu-Budu, K.G., Ogata, S., 1992. Biological nitrogen fixation in mixed legume.cereal cropping systems, Plant and Soil. Giller, K.E., Nambiar, P.T.C., Rao, S.B., Dart, P.J., Day, J.M., 1987. A comparison of nitrogen fixation in genotypes of groundnut (Arachis hypogaea L.) using 15N-isotope dilution *. Biol Fertil Soils 5, 23–25. Giller, K.E.E., Cadisch, G., 1995. Future benefits from biological nitrogen fixation: An ecological approach to agriculture. Plant Soil 174, 255–277. https://doi.org/10.1007/BF00032251 Gwenambira, C.P., 2015. Below- and aboveground pigeonpea productivity in on-farm sole nd intercrop systems in central Malawi. Michigan State Univ. Crop Soil Sci. Thesis. 1-75. Kane, D.A., Snapp, S.S., Davis, A.S., 2015. Ridge Tillage Concentrates Potentially Mineralizable Soil Nitrogen, Facilitating Maize Nitrogen Uptake. Soil Sci. Soc. Am. J. 79, 81. https://doi.org/10.2136/sssaj2014.07.0273 Katayama, K., Ito, O., Matsunaga, R., Adu-Gyamfi, J.J., Rao, T.P., Anders, M.M., Lee, K.K., 1995. Nitrogen balance and root behavior in four pigeonpea-based intercropping systems. Fertil. Res. 42, 315–319. https://doi.org/10.1007/BF00750523 Kellogg Biological Station/ Long-Term Ecological Research, 2016. Protocol - Soil pH [WWW Document]. URL https://lter.kbs.msu.edu/protocols/163 (accessed 9.6.19). Kellogg Biological Station/ Long-Term Ecological Research, 2008. Particle Size Analysis for Soil Texture Determination (Hydrometer Method) [WWW Document]. URL https://lter.kbs.msu.edu/protocols/108 (accessed 9.2.19). Kermah, M., Franke, A.C., Adjei-Nsiah, S., Ahiabor, B.D.K., Abaidoo, R.C., Giller, K.E., 2018. N2- fixation and N contribution by grain legumes under different soil fertility status and 61 cropping systems in the Guinea savanna of northern Ghana. Agric. Ecosyst. Environ. 261, 201–210. https://doi.org/10.1016/j.agee.2017.08.028 Kumar Rao, J.V.D.K., Dart, P.J., 1987. Nodulation, nitrogen fixation and nitrogen uptake in pigeonpea (Cajanus cajan (L.) Millsp) of different maturity groups. Plant Soil 99, 255– 266. https://doi.org/10.1007/BF02370872 Mapfumo, P., Mtambanengwe, F., Giller, K.E., Mpepereki, S., 2005. Tapping indigenous herbaceous legumes for soil fertility management by resource-poor farmers in Zimbabwe. Agric. Ecosyst. Environ. 109, 221–233. https://doi.org/10.1016/J.AGEE.2005.03.015 Maskey, S.L., Bhattarai, S., Peoples, M.B., Herridge, D.F., 2001. On-farm measurements of nitrogen fixation by winter and summer legumes in the hill and terai regions of nepal. F. Crop. Res. 70, 209–221. https://doi.org/10.1016/S0378-4290(01)00140-X Mcdonagh, J.F., Toomsan, B., Limpinuntana, V., Giller I~’, K.E., 1993. Estimates of the residual nitrogen benefit of groundnut to maize in Northeast Thailand, Plant and Soil. Mhango, W.G., Snapp, S.S., Kanyama-Phiri, G.Y., 2017. Biological nitrogen fixation and yield of pigeonpea and groundnut: Quantifying response on smallholder farms in northern Malawi. African J. Agric. Res. 12, 1385–1394. https://doi.org/10.5897/ajar2017.12232 Mungai, L.M., Snapp, S., Messina, J.P., Chikowo, R., Smith, A., Anders, E., Richardson, R.B., Li, G., 2016. Smallholder Farms and the Potential for Sustainable Intensification. Front. Plant Sci. 7, 1–17. https://doi.org/10.3389/fpls.2016.01720 Njira, K.O.W., Nalivata, P.C., Kanyama-Phiri, G.Y., Lowole, M.W., 2012. Biological nitrogen fixation in sole and doubled-up legume cropping systems on the sandy soils of Kasungu , Central Malawi. J. Soil Sci. Environ. Manag. 3, 224–230. https://doi.org/10.5897/JSSEM12.056 Okito, A., Alves, B.J.R., Urquiaga, S., Boddey, R.M., 2004. Nitrogen fixation by groundnut and velvet bean and residual benefit to a subsequent maize crop. Pesqui. Agropecu. Bras. 39, 1183–1190. https://doi.org/10.1590/s0100-204x2004001200004 Orr, A., Kambombo, B., Roth, C., Harris, D., Doyle, V., 2015. Adoption of integrated food-energy systems: Improved cookstoves and pigeonpea in southern Malawi. Exp. Agric. 51, 191– 209. https://doi.org/10.1017/S0014479714000222 Peoples, M.B., Herridge, D.F., Ladha, J.K., 1995. Biological Nitrogen-Fixation - an Efficient Source of Nitrogen for Sustainable Agricultural Production. Plant Soil 174, 3–28. https://doi.org/10.1007/BF00032239 62 Pimratch, S., Jogloy, S., Vorasoot, N., Toomsan, B., Patanothai, A., Holbrook, C.C., 2008. Relationship between biomass production and nitrogen fixation under drought-stress conditions in peanut genotypes with different levels of drought resistance. J. Agron. Crop Sci. 194, 15–25. https://doi.org/10.1111/j.1439-037X.2007.00286.x Querné, A., Battie-laclau, P., Dufour, L., Wery, J., Dupraz, C., 2017. Effects of walnut trees on biological nitrogen fixation and yield of intercropped alfalfa in a Mediterranean agroforestry system. Eur. J. Agron. 84, 35–46. https://doi.org/10.1016/j.eja.2016.12.001 Senaratne, R., Liyanage, N., Soper, R., 1995. Nitrogen fixation of and N transfer from cowpea, mungbean and groundnut when intercropped with maize, Fertilizer Research. Kluwer Academic Publishers. Sirrine, D., Shennan, C., Snapp, S., Kanyama-Phiri, G., Kamanga, B., Sirrine, J.R., 2010. Improving recommendations resulting from on-farm research: Agroforestry, risk, profitability and vulnerability in southern Malawi. Int. J. Agric. Sustain. 8, 290–304. https://doi.org/10.3763/ijas.2010.0471 Snapp, S.S., Blackie, M.J., Gilbert, R.A., Bezner-Kerr, R., Kanyama-Phiri, G.Y., 2010. Biodiversity can support a greener revolution in Africa. Proc Natl Acad Sci U S A 107, 20840–20845. https://doi.org/10.1073/pnas.1007199107 Snapp, S.S., Grabowski, P., Chikowo, R., Smith, A., Anders, E., Sirrine, D., Chimonyo, V., Bekunda, M., 2018. Maize yield and profitability tradeoffs with social, human and environmental performance: Is sustainable intensification feasible? Agric. Syst. 162, 77– 88. https://doi.org/10.1016/J.AGSY.2018.01.012 Snapp, S.S., Rohrbach, D.D., Simtowe, F., Freeman, H.A., 2002. Sustainable soil management options for Malawi: Can smallholder farmers grow more legumes? Agric. Ecosyst. Environ. 91, 159–174. https://doi.org/10.1016/S0167-8809(01)00238-9 Toomsan, B., Mcdonagh, J.E., Limpinuntana, V., Giller, K.E., 1995. Nitrogen fixation by groundnut and soyabean and residual nitrogen benefits to rice in farmers’ fields in Northeast Thailand, Plant and Soil. Kluwer Academic Publishers. Waldman, K.B., Ortega, D.L., Richardson, R.B., Snapp, S.S., 2017. Estimating demand for perennial pigeon pea in Malawi using choice experiments. Ecol. Econ. 131, 222–230. https://doi.org/10.1016/J.ECOLECON.2016.09.006 Wani, S., Rupela, O., Lee, K., 1995. Sustainable agriculture in the semi-arid tropics through biological nitrogen fixation in grain legumes, Plant and Soil. Kluwer Academic Publishers. 63 3. Chapter Three: Short-term interactions between legume and maize residues and mineral N fertilizer on nitrate-N dynamics and maize productivity in Malawi Abstract Smallholder farmers in Sub-Saharan Africa (SSA) produce food crops for subsistence on nitrogen- deficient soils. This limits crop productivity and negatively impacts food and nutritional security. The use of inorganic fertilizers and organic resources from high quality crop residues have been proposed as a solution to the dilemma. However, not much is known about how different on- farm crop residues interact with inorganic N fertilizer in different agro-ecologies. To address these research gaps, two-year on-farm trials were implemented in three locations in southern Malawi that are typical of low, medium and high agro-ecological potential. During Year 1, maize and doubled-up legume rotation (groundnut/pigeonpea) plots were optimally managed to generate both grain and biomass. At harvest, 0X, 1X and 2X residues were incorporated and maize was planted in these plots in Year 2 to test the effect of residue quality and quantity. From the three locations, on-farm biomass in year one from maize (low quality residues) ranged from 1.67 to 10.42 Mg/ha and groundnut/pigeonpea (high quality residues) ranged from 2.39 to 11.96 Mg/ha. In year two, location and treatment effect on maize grain yield ranged from 1.06 to 8.27 Mg/ha. However, treatment effects were only due to fertilizer and not residue quality or quantity. Residue % N content was in the order: green pigeonpea leaves > senesced groundnut leaves > senesced groundnut stover > green pigeonpea twigs > senesced pigeonpea leaves > 64 maize stover. Residue quality and quantity did not have an effect on inorganic soil N from the 0- 80 cm soil depth. Our study shows the actual crop residue biomass quantity and quality that smallholder farmers are capable of producing depending on their biophysical environment. As no residue quality or quantity effects were detected on maize yield or soil N this shows the importance of integrated nutrient management over a long term for soil health and crop production. Agricultural development policies focus on access to inorganic fertilizer sources, and indeed nitrogen fertilizer has clear benefits as shown here. Keywords: Inorganic nitrogen, Legumes, Maize, Residues Crop residues Introduction Crop residues are often the only available source of organic amendments used by smallholder farmers to improve and conserve soils, given the limited access to livestock and green manures. Many soils on smallholder farms in Sub-Saharan Africa (SSA) have low soil organic matter (SOM) content (Smaling et al., 1997), which results in poor crop productivity. Building SOM requires improving both cereal and legume crops primary productivity through mineral fertilizers and retaining the associated crop residues on the cropped lands. These residues decompose to provide mineral N to crops grown in sequence, as well as being an important source for SOM capitalization (Beedy et al., 2010). Residues of legume crops have a narrow C/N ratio which supports N availability to the rotational crop, whereas residues of maize, which have a wide C/N 65 ratio, promote N immobilization. Therefore, high quality legume residues are important for integrated nutrient management in SSA smallholder farming systems (Snapp et al., 1998). Soil C inputs from crop residues play a key role in soil organic carbon (SOC) sequestration and stabilization (Kong et al., 2005). Studies have shown that increases in SOC are often related to the incorporation of fresh organic materials to the soil such as from legumes, cover crops, and manure, if biochemical residue quality is sufficient to support soil C accrual,. Six et al. (2002) found that the light fraction and particulate organic matter pools are mainly made up of recently derived, partially decomposed plant residues. Effect of crop residues on soil C and N According to Chikowo et al. (2003), the chemical composition or quality of residues determines the release of mineral-N in soil from plant residues. Under leaching conditions, Chikowo et al. (2003) conducted an incubation study of different types of litter for 120 days and found that sesbania litter released more mineral N than acacia litter. Sakala et al. (2000) conducted a leaching tube incubation study with different textured soils and investigated the interactions of pigeonpea and maize residues during decomposition and N mineralization. The authors reported that the C:N ratio of senesced pigeonpea leaves was 24, while maize residues were at 75. Even though there eventually was N mineralization in the senesced pigeonpea treatment, there was an initial immobilization period that ranged from two to four weeks in all soil types. The authors also observed that the N immobilization period in the maize residue and in the mixed (maize and pigeonpea residues) treatments were longer but similar, lasting 18 weeks. Sakala et al. (2000) concluded that mixing pigeonpea and maize residues did not significantly reduce N 66 immobilization. However, in intercropping systems, residues of varying quality are mixed and decompose concurrently, resulting in N mineralization processes which differ from those of residues in sole cropping systems (Sakala et al., 2000). McDaniel et al. (2014) conducted a meta-analysis of 122 studies to examine crop rotation, crop types and management practices effects on total soil C and N in different soil types and climates. The authors found that increasing crop diversity by adding one or more crops in a rotation to a monoculture increased total N by 5.3% and total soil C by 3.6% and when rotations included a cover crop total N increased by 12.8% and total C by 8.5%. In a different study by Tiemann et al. (2015), increasing the quantity, quality and chemical diversity of residues through high diversity rotations sustained soil microbes, soil organic matter and soil fertility. Additionally, incorporating residues of mixed qualities in diverse cropping systems resulted in soil organic nitrogen mineralization. The quality of residues determines their rate of decomposition, N release and eventually how much subsequent crops benefit (Sakala et al., 2000). Therefore, research on residue quality and management of organic inputs is of paramount importance. Sustainability is difficult to assess in cropping systems where residue inputs from above and belowground biomass remain uncharacterized, leaving uncertain this important determinant of soil C and N trends over time. Malawi farming systems and farmer nutrient management Farmland cultivation in Malawi is labor intensive as manual labor is used for land preparation and 67 hand hoe weeding. The most common cropping systems include continuous maize, maize- groundnut rotations and maize/pigeonpea intercrops (Mungai et al., 2016). Mungai et al. (2016) established that more than 80% of farmers applied mineral fertilizer in one or more of their fields. Many farmers combine residue incorporation, or compost application, with fertilizer use in almost half of their fields. Farmers were also reported to remove or burn legume residues from their fields. Mungai et al. (2016) also found that timing of residue incorporation varied within and across locations where some farmers incorporated residues soon after crop harvesting while others incorporated residues during land preparation. Malawi smallholder farming systems provide an important opportunity to evaluate agricultural practices for performance under variable and degraded soil conditions (Li et al., 2017). Mineral fertilizer use has grown markedly in recent years in Malawi due to policies that improve farmer access, yet maize yields are not always following the expected upward trend (Messina et al., 2017). Therefore, there is need to determine the practices under which crop and soil management practices support sustainable intensification (Snapp et al., 2018). This study investigated the effect of incorporating high- and low-quality crop residues on soil inorganic N status and maize yield on different soils and across a climate gradient in southern Malawi. Our specific objectives were 1) To determine the effect of incorporating high- and low-quality crop residues on the effect of incorporated residue quality and quantity on mineral N dynamics during different times in the cropping season and 2) To determine the effect of inorganic fertilizer combined with high- and low-quality residues on subsequent on-farm maize yield. The 68 hypotheses were 1) High residue quality will enhance soil inorganic N pools whereas high residue quantity will be associated with temporary immobilization of inorganic N and 2) Organic and inorganic N sources (high N-content residues + fertilizer) will act as additive sources of N, both being associated with high maize yields. Study locations Materials and methods This study is part of the Africa RISING project, which was conducted between 2016-2018 in Machinga, in the southern region of Malawi, at three locations representing low yield production potential associated with hot and dry conditions (Mtubwi), moderate yield production potential with decent rainfall distribution in most years (Nsanama) and high yield production potential associated with the warm, mesic rainfall site of Nyambi (Figure 3.1). These sites were chosen based on net primary productivity potential through remote sensing data to obtain a range of yield production potential sites and validated through local agricultural land use systems. Mtubwi is in the Shire Valley, a low elevation zone that historically receives about 600 mm of rainfall annually and is the hottest site. Nsanama and Nyambi are at medium-high elevation and historically receive about 800-1200 mm of rainfall per year. The total annual rainfall and day temperature averages for the three locations are shown in Table 3.1. 69 Figure 3.1. Map of Malawi showing the location of the three research agroecologies with 15-year temperature and normalized difference vegetation index (NDVI) averages. Temperature and NDVI are averaged across 2001-2015 growing seasons. Temperature measurements were sourced from the MODIS Land Surface Temperature. Map was prepared by Dr. Brad Peter, University of Alabama (2019). 70 Table 3.1. Total annual precipitation in mm and day temperature means during the 2016, 2017, and 2018 growing seasons for the three locations. Precipitation based on CHIRPS and temperature is from NASA MODIS Land Surface Temperature. Location Mtubwi (Low potential) Cumulative precipitation (mm) 2016 704 35.8 Day temperature mean (°C) Nsanama (medium potential) Cumulative precipitation (mm) 678 Day temperature mean (°C) Nyambi (high potential) Cumulative precipitation (mm) Day temperature mean (°C) 35.3 735 34.4 2017 2018 839 33.9 811 33.6 897 32.6 775 35.6 770 34.1 904 33.4 During the 2016-17 agricultural season, there were three dry spells at each of the three locations and four in 2017-18. There were two flooding events at the Mtubwi site in 2017-18. Experimental design This study investigated maize productivity and soil mineral/inorganic N dynamics associated with inorganic and organic nutrient sources, comparing fertilizer effects, and quality and quantity of residues associated with continuous maize (ContMz) versus a doubled-up legume rotation (DLR) with maize. 71 Two year on-farm trials were implemented in three sites in southern Malawi that are typical of low, medium and high agro-ecological potential. During Year 1, maize and doubled-up legume rotation (groundnut/pigeonpea) plots were optimally managed to generate both grain and biomass. There were six ‘treatments’ that were essentially similar in Year 1 to generate maize biomass and three other doubled-up legume treatments (Table 3.2). At harvest, three levels of quantity of residues was achieved through a design that included 0X (shoot residues removed), 1X (shoot residues retained in situ) and 2X (doubled shoot residues through addition of residues from the 0X plot) (Table 3.2). Quality of residues were at two levels, based on the two cropping systems studied (continuous maize and doubled-up legume). Nitrogen fertilizer was applied in treatments 4-9 to assess inorganic N response in combination with quality and quantity of residues, relative to maize residues only (Treatments 1-3). The experiment was repeated in time, as a legume phase was established in 2018/19 and maize yield benefits were assessed during 2019/20, however only the initial study with years 1 and 2 are reported here. The research was carried out at three locations, using a randomized complete block design (RCBD) with three replications at all locations. As described in Table 3.2, the farming systems rationale for these systems include testing various scenarios. In addition, this incomplete factorial allowed testing of the effect of quality, quantity and fertilizer. Treatments MZ-0N-0R, MZ-0N-2R and MZ-0N-1R received no fertilizer in year two while the other six treatments received 35 kg/ha N (23 kg NP-N and 12 kg Urea-N). To avoid P being limiting, 9 kg P was applied to all treatments as single super phosphate. All plots in all locations were under sole maize in the second year of the experiment, where intensive soil monitoring was carried out 72 over 2017/18 in six of the treatments (MZ-0N-0R, MZ-0N-2R, MZ-35N-0R, MZ-35N-2R, GNPP- 35N-0R and GNPP-35N-2R) at the three sites. 73 Table 3.2. Treatments structure and fertilization in the residue generation phases and, in the residue, quality and quantity assessment phases. Rationale for the treatment choices is also included. 74 Crop management Crops were planted in late November or early December, after the first effective rains. Maize was planted at 75 cm spacing between ridges and 25 cm within ridge spacing to target a plant population of 53,000 plants/ha (Chimonyo et al., 2019a). The DLR with maize system had a population density 174,000 plants/ha for groundnut and 44,000 plants/ha for pigeonpea in the doubled-up legume phase of the rotation, and 53,000 plants/ha in the maize rotation phase. This followed Malawi government recommendation for DLR, which is an additive design, combining the recommended sole crop population density for groundnut and pigeonpea (Mungai et al., 2016). The planting system arrangement was that groundnut is spaced at 10 cm along a row, and pigeonpea planted at 4 seeds at planting stations every 90 cm along the row, thinned to 3 plants at 3 weeks after germination. All fields were kept weed free through hand-hoe cultivation. Plant biomass assessments In May of 2017, maize and groundnut crops from a whole plot of 26.25 m2 per plot (7 ridges x 5 m x 0.75 m) were harvested and weighed. A randomly selected sample of approximately 3 kg of maize residues was collected by compositing multiple subsamples from a well-mixed, chopped up sample (approximately 8 cm) using a machete in the field. In July of 2017 after weighing pigeonpea biomass, stems were separated by hand from the leaves, pods, and twigs. All the dry and senesced pigeonpea and groundnut leaf mixtures were collected from the groundnut/pigeonpea plots and weighed. A random hand sampling technique was used to obtain representative fresh pigeonpea, senesced pigeonpea and groundnut residues, which were chopped and dried to constant weight in an oven. 75 Yield assessment and residue management Maize and groundnut crops were harvested at physiological maturity in April-May 2017, and pigeonpea in July 2017. In groundnut/pigeonpea plots, pigeonpea pods were harvested by hand and weighed. Crop residues from year 1 were incorporated as described in Table 3.2. For maize systems, MZ-0N-2R, MZ-0N-1R, MZ-35N-2R and MZ-35N-1R (Table 3.2), maize residues were sun- dried in the field, placed in furrows, and ridges were made to incorporate these residues into the soil in June 2017. Groundnut residues were sun-dried in the field until pigeonpea physiological maturity. After pigeonpea was harvested in July 2017, mixtures of fresh pigeonpea leaves, twigs and flowers, senesced pigeonpea and groundnut leaves and groundnut stover were laid out in the furrows and ridges were made to incorporate the residues in the soil. The ridges were made following Malawi farmer practice using the hand hoe, to about 25 cm in height, and spaced at 75 cm. All plots were cropped with maize during the next cropping season that commenced in November 2017, to determine the effect of residue quality and quantity on maize productivity at maize physiological maturity in April-May 2018. In both years, crops were harvested on a whole plot basis. Tissue quality analysis Sub-samples of legume mixtures were ground using a laboratory mill (1 mm). Ground tissue samples were packed in tins, and C and N content was determined at the University of California Davis Stable Isotope Facility, by an elemental analyzer and isotope ratio mass spectrometer (https://stableisotopefacility.ucdavis.edu/13cand15n.html). For lignin and acid detergent fiber (ADF) %, crop residue samples were sent to the Dairyland Laboratories for analysis, where a 76 modified AOAC Official Method 973.18 was used (https://www.dairylandlabs.com/feed-and- forage/methods). Soil sampling and analyses Soil samples were collected from the ridges before incorporation of residues in June of 2017. Randomly collected soil samples were composited (three per plot) to represent the 0-20 and 20- 40 cm soil depths, using a dutch auger (5.5 cm diameter). Soils were air-dried, sieved (2 mm) and soil texture was determined using the hydrometer method (Kellogg Biological Station/ Long-Term Ecological Research, 2008). Total soil C and N % were determined after grinding to a fine powder with a shatter mill, by dry combustion with Costech ECS 4010 (Costech Analytical Technologies, Inc., Valencia, CA). Soil samples were sent to the A and L Great Lakes laboratory for pH (1:1 soil to water ratio), phosphorous (P), and potassium (K) analyses (Mehlich-3 ICP method). A subset of six treatments (MZ-0N-0R, MZ-0N-2R, MZ-35N-0R, MZ-35N-2R, GNPP-35N-0R, and GNPP-35N- 2R) were used for detailed soil inorganic N monitoring. Soil samples were collected from the 0- 20, 20-40, 40-60 and 60-80 cm soil depths at three times over the 2017-18 season. The first soil sampling for inorganic N determination was carried out at maize planting (December 2017), thirty days after planting (January 2018), and at maize harvest (May 2018). Inorganic N was extracted using 2M KCl. Soil samples of 10 g were weighed into a 100 ml plastic centrifuge cups and 40 ml of 2M KCl was added. The cups were shaken for one hour on a reciprocal shaker at approx. 180 strokes per minute. After shaking, cups were allowed to settle for 15 mins. The supernatant was filtered through a Whatman No. 1 filter paper (GE Healthcare 77 Bio-Sciences, Pittsburg, PA) and poured into small plastic vials. The samples were frozen until they were analyzed. KCl extractants were analyzed for inorganic N (NO3-N and NH4-N) on a Thermo MultiskanTM 96-well plate reader (Kane et al., 2015). Gravimetric soil moisture content was determined by weighing approximately 10 g of soil into tins or bags, and oven-drying the soil at 105 °C for 48 hours or until constant weight. The soil moisture content was then determined by the formula: % soil moisture = ((wet soil wt) – (dry soil wt – tin/bag wt.) / (dry soil wt. – tin/bag wt.)) * 100 (KBS LTER, 2019) Statistical analysis In 2017, the research design was an RCBD with fertilized sole maize or groundnut/pigeonpea intercrops as treatment factors. In 2018, the maize response year, the research design was an RCBD with residue quality (maize stover or groundnut/pigeonpea residues), residue quantity (0x, 1x and 2x) and inorganic fertilizer N levels (0 and 35 kg/ha N) as treatment factors. Maize grain yield response to treatment (residue quality and quantity) and location (low, medium and high yield potential gradient) effects were analyzed by a two-way analysis of variance (ANOVA) in SAS 9.4 (SAS Institute, 2002) using PROC MIXED to compare differences across treatments and locations. The ANOVA model that was used to test for maize yield response was (Maizegrainyield= Location|Treatment). Nitrate dynamics over the three sampling times were analyzed using the repeated measures ANOVA in SAS with the model (N=treatment time/ddfm=kr). Planned contrasts were used to determine residue quality and quantity effects 78 on inorganic N and maize grain yield. The planned contrast models included block nested within location and treatment by block as random effects. Environment Results The soil properties at the three locations are shown in Table 3.2. Overall, soils at all locations were close to neutral or slightly alkaline, with pH values of 6.8-7.1. The topsoil content of organic matter was low at all sites as indicated by soil organic C levels of 0.36 %C at Mtubwi, 0.47 %C at Nsanama and 1.23 %C at Nyambi. Soil total N % did not vary and was about 0.06 % at all three sites. Total inorganic N in the 0-20 cm soil depth ranged from 1.95 mg/kg soil (Nyambi) to 2.16 mg/kg soil (Nsanama). Soil P ranged from 36 ppm (Mtubwi) to 113 ppm (Nsanama) while soil K ranged from 67 ppm to 186 ppm (Nyambi). Generally, the sites were sandy, with the clay content in the 0-20 cm soil depth varying from 13% (Nsanama) to 19 % (Nyambi) (Table 3.3). 79 Table 3.3. Soil chemical and physical properties at research locations (0-20 cm) from marginal (Mtubwi), moderate (Nsanama) and mesic (Nyambi) sites in southern Malawi. Values are means followed by standard deviations. 80 Aboveground biomass Across treatments and locations, maize biomass production in 2017 ranged from 1.67 Mg/ha in Mtubwi to 10.42 Mg/ha in Nsanama (Table 3.4). Maize biomass in 2018 was lower at all sites than in 2017. Total aboveground biomass from groundnut and pigeonpea ranged from 2.39 Mg/ha in Mtubwi to 11.96 Mg/ha in Nsanama (Table 3.4). Legume biomass included senesced groundnut leaves, groundnut stover, senesced pigeonpea leaves, fresh pigeonpea leaves, and twigs. 81 Table 3.4. Crop residue biomass produced from nine treatments at three locations in 2017. For the species column, maize represents maize stover, GNPP-Senesced is a mixture of senesced groundnut and pigeonpea leaves, GNPP-Groundnut represents groundnut stover and GNPP-Pigeonpea is pigeonpea leaves and twigs. The incorporated columns show the total amount of incorporated crop residues, where treatments with 0R had no residues, 1R is 1x residues and 2R is 2x residues. Presented values are means, followed by standard deviations. 82 Tissue quality The chemical composition of incorporated maize, groundnut and pigeonpea residues are shown in Table 3.5. As expected, maize stover had the lowest N content of 0.87 % and green pigeonpea leaves had the highest N content of 3.06 %. The lignin content of senesced pigeonpea leaves (24.35 %) was much higher than that of maize stover (5.82 %) and senesced groundnut leaves (3.44 %). Senesced groundnut leaves had the lowest ADF % (19.30) whereas senesced pigeonpea leaves had the highest (56.17). Table 3.5. Chemical composition of on-farm crop residues from Malawi which were incorporated Residue type Maize stover Senesced groundnut leaves Senesced groundnut stover Senesced pigeonpea leaves Green pigeonpea leaves Green pigeonpea twigs N % 0.87 2.97 1.62 1.50 3.06 1.59 Lignin % ADF % Lignin:N 5.82 3.44 9.07 24.35 15.84 11.27 51.28 19.30 44.51 56.17 29.14 42.59 12.38 1.16 5.60 16.23 5.18 7.09 Maize grain yield following incorporation of crop residues There was a location and treatment effect on maize grain yield but no interaction (Table 3.6). Across locations and treatments, maize grain yield ranged from 1.06 Mg/ha in Mtubwi to 8.27 Mg/ha in Nyambi. From all treatments, the average maize grain yield at the Mtubwi location was 1.96 Mg/ha, 5.20 Mg/ha at Nsanama, and 7.20 Mg/ha at Nyambi. The treatment effect was due to fertilizer, as residue quality or quantity had no effect on maize grain yield as shown by planned contrasts (Table 3.6). The fertilizer effect (P=0.0634) was found at all locations, as shown by planned contrasts of fertilized vs. unfertilized treatments (Table 3.6). Across locations, 83 unfertilized treatments had lower maize grain yield on average (4.00 Mg/ha) compared to fertilized treatments (5.18 Mg/ha). When ANOVAs were done separately by location, the fertilizer effect was even greater at the Nyambi location (P= 0.0034) and the Mtubwi location (P= 0.0215). Figure 3.2 shows maize grain yield from unfertilized and fertilized treatments from the six treatments which were studied in detail. 84 Table 3.6. The effect of residue quantity, residue quality and N fertilizer on maize yield in 2018. Means are followed by standard deviations for nine treatments at three locations in southern Malawi. Planned contrasts of six treatments across three locations are shown as well as the overall ANOVAS. Treatment Mtubwi Nsanama Nyambi Mg/ha MZ-0N-0R MZ-0N-2R MZ-0N-1R MZ-35N-0R MZ-35N-2R MZ-35N-1R GNPP-35N-0R GNPP-35N-2R GNPP-35N-1R 1.54 ± 0.83 1.06 ± 0.41 1.52 ± 0.80 2.54 ± 0.88 2.31 ± 1.09 2.12 ± 1.43 1.97 ± 0.34 2.23 ± 0.22 2.38 ± 0.71 4.85 ± 1.04 5.97 ± 1.53 4.52 ± 0.48 5.53 ± 1.04 4.04 ± 1.69 6.95 ± 1.24 5.57 ± 1.38 8.07 ± 1.52 5.92 ± 1.68 6.84 ± 0.64 5.52 ± 0.30 6.96 ± 1.49 5.53 ± 1.06 8.03 ± 0.51 5.91 ± 0.60 8.27 ± 0.76 4.94 ± 2.39 8.16 ± 0.88 Planned Contrasts No residue vs. low quality residue treatments Pr > F = 0.7956 No residue vs. high quality residue treatments Pr > F = 0.8537 Low vs. high quantity residue treatments Pr > F = 0.9635 Fertilized vs. unfertilized treatments Pr > F = 0.0634** ANOVAS Location Treatment Location x Treatment Pr > F = <.0001* Pr > F = 0.0195* Pr > F = 0.9647 *Significant at P = 0.05 **Significant at P = 0.10 85 Figure 3.2. Maize grain yield from six treatments across three locations in southern Malawi. At all locations, treatments MZ-0N-0R and MZ-0N-2R were unfertilized while the rest were fertilized. 86 Soil NO3-N Soil NO3-N was at moderate levels at all sampling times and was analyzed over time using a repeated measures model (NO3-N=treatment time/ddfm=kr). Treatment had no effect on cumulative soil NO3-N from the 0-80 cm depth (P=0.9266). Sampling time influenced soil NO3-N (P=0.0103), as NO3-N was low during the 2018 harvest (Figure 3.3). From a two-way ANOVA (model NO3-N= location|time), cumulative soil NO3-N was also influenced by location (P=<.0001), time (P= 0.0035) and by location x time (P= 0.0506) (Figure 3.3). At two locations (Nsanama and Nyambi), soil NO3-N followed the same trend where the highest soil NO3-N level was in December of 2017 at crop planting time and was lower over time. However, at the Mtubwi location, soil NO3-N levels were highest in January 2018 and lowest in June 2018 (Figure 3.3). Figure 3.3. Cumulative soil NO3-N from the 0-80 cm soil depth during three sampling times at three locations in southern Malawi. From a two-way ANOVA, (model NO3-N= location|time), location (P=<.0001) and time (P= 0.0035) had an effect on soil NO3-N in 2017-2018. 87 Location had an effect on NO3-N (P= 0.0006) so the average topsoil (0-20 cm) NO3-N is shown by location, and sampling time in Table 3.7. Treatment residue quantity or quality had no effect on topsoil soil NO3-N (Table 3.7). Across locations and treatments, soil NO3-N ranged from 1.17 mg/kg soil to 2.20 mg/kg soil (Table 3.7). For the Mtubwi location, at the 0-20 cm soil depth only, time had an effect on soil nitrate N (P=<.0001, Table 3.7). Table 3.7. Soil NO3-N in the 0-20 cm soil depth by treatment and three sampling times at three locations in southern Malawi. Nsanama Nyambi MZ-0N-0R MZ-0N-2R MZ-35N-0R MZ-35N-2R GNPP-35N-0R GNPP-35N-2R Time - Dec Time - Jan Time - Jun Mtubwi Soil NO3-N (mg/kg soil) 1.23 1.26 1.20 1.27 1.17 1.17 1.29 1.38 0.98 1.22 1.23 1.32 1.18 1.54 1.04 1.48 1.15 1.13 1.87 1.51 1.91 1.57 1.50 2.20 2.30 1.51 1.47 ANOVAS 0.5682 Treatment Time <.0001* Treatment x Time 0.7982 *Significant at P = 0.05 0.4301 0.1949 0.5051 0.8137 0.2693 0.7672 Soil NH4-N Cumulative soil NH4-N was analyzed using a repeated measures model (NH4-N =treatment|time/ddfm=kr). Both treatment (P=0.0185) and sampling time (P=<.0001) had an effect on cumulative soil NH4-N from the 0-80 cm depth but there was no interaction effect (P=0.8886). At two locations (Mtubwi and Nsanama), cumulative inorganic soil NH4-N from the 88 0-80 cm soil depths followed the same trend, where the highest soil NH4-N level was in January of 2018 at 30 days after planting, lower in December of 2017 at crop planting time, and much lower at crop maturity in June of 2018. However, at the Mtubwi location, soil NH4-N levels were highest in December 2017 and lowest in June 2018 (Figure 3.4). Figure 3.4. Cumulative soil NH4-N from the 0-80 cm soil depth during three sampling times at three locations in southern Malawi. From a two-way ANOVA, (model NH4-N = location|time), location (P<.0001) and time (P<.0001) had an effect on soil NH4-N. The average 0-20 cm NH4-N is shown by location, and sampling time in Table 3.8. Treatment had an effect on soil NH4-N at the Mtubwi location only (P=0.0031, Table 3.8). Soil NH4-N ranged from 0.91 mg/kg soil in the MZ-0N-0R treatment to 3.76 mg/kg soil in the GNPP-35N-2R treatment (Table 3.8). Planned contrasts in Table 3.8 show that treatment effects on NH4-N were due to fertilizer and not residue quality or quantity. Sampling time had a significant effect on soil NH4-N at the Mtubwi (1.13 to 4.91 mg/kg soil range) and Nyambi (0.44 to 3.90 mg/kg soil range) locations (Table 3.8). 89 Table 3.8. Soil NH4-N in the 0-20 cm soil depth by treatment and three sampling times at three locations in southern Malawi. MZ-0N-0R MZ-0N-2R MZ-35N-0R MZ-35N-2R GNPP-35N-0R GNPP-35N-2R Time - Dec Time - Jan Time - Jun Mtubwi 0.91 1.10 3.54 3.14 2.46 3.76 1.41 4.91 1.13 ANOVAS Nsanama Nyambi Soil NH4-N (mg/kg soil) 1.01 1.01 2.18 2.11 2.38 2.23 3.90 1.12 0.44 0.99 0.38 2.21 1.01 1.35 1.71 1.40 1.39 1.04 Treatment Time Treatment x Time 0.0031* <.0001* 0.1584 PLANNED CONTRASTS No residue vs. low quality residue treatments 0.9738 No residue vs. high quality residue treatments 0.5340 Low vs. high quantity residue treatments 0.6897 0.0033* Fertilized vs. unfertilized treatments 0.1612 0.7057 0.8820 0.2236 <.0001* 0.4070 0.1453 0.8918 0.3269 0.0206* 0.8499 0.9542 0.8799 0.0639** *Significant at P = 0.05 *Significant at P = 0.10 90 Discussion Biomass production and maize grain yield This study shows on-farm biomass production quantities ranging from 1.67 to 10.42 Mg/ha for maize and 2.39 to 11.96 Mg/ha for the groundnut/pigeonpea intercrop. Similar to our study, Makumba et al. (2006) reported maize stover biomass amounts ranging from 1.7 Mg/ha (unfertilized sole maize) to 8.2 Mg/ha in a fertilized gliricidia/maize intercrop from an 11-year gliricidia/maize intercrop study in southern Malawi. For high quality legume biomass in SSA, different studies have reported incorporated gliricidia prunings amounting to 4.6 Mg/ha in Malawi (Makumba et al., 2006), 2.5-5.5 Mg/ha in Togo (Vanlauwe et al., 2005) and 5.7 Mg/ha in Nigeria (Kang et al., 1999). According to Snapp et al. (1998) about 7 Mg/ha to 10 Mg/ha of crop residues are needed to support soil organic matter gains in tropical sub-humid conditions, depending with the quality of residues. This suggests that in some cases, sufficient biomass can be produced to ameliorate soil organic matter and help maintain soil health on smallholder farms. In 2017, biomass production of crops did not quite follow the anticipated pattern for the productivity gradient the experiment was established along, from low potential (Mtubwi) to high potential (Nyambi). The low potential Mtubwi location did have the lowest biomass, however the medium potential Nsanama location had twice the biomass of the high potential Nyambi site (Table 3.4). There is clearly high year to year variation in tropical rain-fed agricultural systems, particularly in Sub-Saharan Africa (Rockström et al., 2003) 91 The average maize grain yield from all treatments in 2018 followed the expected pattern along the productivity gradient, from low potential Mtubwi to high potential Nyambi (Figure 3.1). Mtubwi had the lowest maize grain yield (1.96 Mg/ha), followed by Nsanama (5.20 Mg/ha) and finally high potential Nyambi (7.20 Mg/ha). The low biomass and yield at the Mtubwi location (Tables 3.4 and 3.6), that remote sensing NDVI indicated was low potential, we expect is related to high evapotranspiration and soil properties, including sandy soil texture and low soil organic matter status (Figure 3.1, Tables 3.1 and 3.3). Further, Mtubwi was also water-logged in 2018 (personal observations). Low maize grain yield in 2018 at the Mtubwi and Nsanama locations could also have been due to some army worm damage (personal observations). Higher maize grain yield in 2018 at the Nsanama and Nyambi locations compared to the Mtubwi location could be attributed to lower evapotranspiration, higher NDVI and higher SOC. The soils at the Nsanama and Nyambi locations could have also responded better to inorganic nitrogen fertilizer than the Mtubwi soils. Similar to our results, Kafesu et al. (2018) reported that response to mineral N increased with increasing SOC from a maize productivity study on sandy soils in Zimbabwe. According to Rusinamhodzi et al. (2011), when adequate fertilizer is available, rainfall is an important determinant of yield in southern Africa in rain-fed systems. However, these observations show that other determinants need to be considered, as total rainfall was similar at all sites. It should be kept in mind that soil quality, rainfall distribution, and temperature may all play a role (Smaling et al., 1997; Rockström et al., 2003) . 92 Crop response to nitrogen Neither high quality residues nor large amounts of residues influenced maize yield. This result is somewhat surprising. Other studies have reported a positive impact of high-quality residues on maize yield. For instance, Makumba et al. (2006) reported that maize yields were roughly doubled with incorporating gliricidia prunings in an 11-year study. However, Makumba et al. (2006) might have been able to detect gliricidia residue quality effects on maize yield because it was a long-term study and the residues were highly N-enriched. In fact, in the first season of their study, Makumba et al. (2006) did not find any difference between maize yield from gliricidia/maize and sole maize cropping systems. However, cultivating gliricidia trees for 10 or more years involves considerable land and labor investment, which might not be feasible for most SSA smallholder farmers. Our study highlights that residues produced on-farm may not be as high quality as would be ideal for an integrated organic matter technology (Table 3.5). Maize grain yield response was influenced by inorganic nitrogen fertilizer only and not residue quality or quantity (Table 3.6). The short two-year duration of this study could be the main reason why no residue quality and quantity effects were found on maize grain yield. Similarly, Nziguheba et al. (2005) did not detect a maize yield response to organic treatments in a 2.5 year study in western Kenya as there was no significant difference in cumulative maize yield from five seasons between organic residues and fertilizers. From a Zimbabwe study, Mtambanengwe and Mapfumo (2006) also reported no significant differences in maize grain yield among high quality, organic N sources from legumes (C. juncea and C. calothyrsus), manure and mineral N fertilizer at low application rate. 93 Nitrogen fertilizer and its appropriate use has been proposed as the fourth principle for conservation agriculture on smallholder farms, due to the widespread requirement in cereal based farming on N-deficient soils (Vanlauwe et al., 2014). Fertilizer use can substantially increase crop productivity and thus organic residue availability in smallholder farms in SSA where farmers lack adequate organic resources for nutrient management (Vanlauwe et al., 2014). Other studies have shown the importance of integrated nutrient management on degraded soils. A two- year study on degraded alfisols in Pakistan showed that having 50% N from organic and 50% from inorganic sources is the best management practice for sustainable production on (Ahmad et al., 2017). From a study in Malawi, Kalasa et al. (2018) also reported that maize grain yield can be improved by adding legume organic residues supplemented with low N inputs from mineral fertilizer. Kafesu et al.(2018) reported consistently higher maize grain yields when integrated nutrient management was practised by combining organic and inorganic N sources compared to either just organic or inorganic N sources. From a 14-year study in Malawi, Beedy et al. (2010) also confirmed the importance of combining organic N sources (gliricidia residues) and N fertilizers for soil recapitalization and maize production. At the same time, our study shows that producing sufficient organic high-quality residues on-farm is a challenge and short-term positive impacts from these may not be feasible. Tissue quality The N content of pigeonpea residues (Table 3.5) was similar to that reported by Thippayarugs et al. (2008) and Sakala et al. (2000). Sakala et al. (2000) reported an N content of 1.9% for senesced pigeonpea leaves and 3.2% for green pigeonpea leaves. The maize stover N content, 94 lignin and ADF % were also similar to those reported by Sakala et al. (2000). However, the lack of treatment differences between maize grain yield from high- and low-quality residue could be attributed to other aspects of residue quality other than N and C % which impact N release. From an 8-week laboratory experiment, Palm and Sanchez (1991) concluded that some leguminous plant materials (including pigeonpea leaves) may not be readily available sources of N due to high polyphenolic contents despite their high % N values. However Sakala et al. (2000) reported that high amounts of lignin and not polyphenols in pigeonpea leaves could have contributed to less N mineralization than expected. This might be the case in our study as high-quality pigeonpea residues had higher N content and also much higher lignin content than maize. Soil inorganic N Inorganic N status (Figures 3.3 and 3.4) varied over time and sampling time had an effect on soil inorganic N. Similar to our results, Mtambanengwe and Mapfumo (2006) conducted soil sampling for inorganic N seven different times and managed to pick some differences in soil N status throughout the season. The differences in soil inorganic N over time may well be attributed to plant N use and precipitation patterns over time. No response of cumulative NO3-N was observed to treatment. However, treatment influenced NH4-N. Similar to maize grain yield, NH4-N response to treatment was due to the fertilizer effect as planned contrasts (Table 3.8) showed that for NH4-N, only fertilized and unfertilized were significantly different. Therefore, over the two seasons of this study, organic residue quantity and quality did not influence soil inorganic N. Nziguheba et al. (2005) found similar results with 95 potentially mineralizable N (PMN), where PMN did not corelate with any residue quality value. Nziguheba et al. (2005) also reported that even though high quality organic residues from calliandra, senna and tithonia improved some soil parameters, their effect were rarely greater than that of inorganic fertilizers. Contrary to our results, a research station study from Zimbabwe depicted that soil NH4-N was significantly influenced by residue quality (Mtambanengwe and Mapfumo, 2006). In our study, the extreme heat in Machinga and the long dry season between harvest/incorporation of residues and the rotational crop could degrade the quality of doubled- up legume residues and be the overriding factor. Mtambanengwe and Mapfumo (2006) noted that most of the organic resources used by smallholder farmers are usually of lower quality than the C. juncea and C. calothyrsus they used at the research station. Additionally, Snapp and colleagues (1998) reported that research stations in Malawi and surrounding areas usually have 50% more soil carbon and about 30% more soil N than smallholder farms have. Additionally, in the short term, under field conditions, it is common to get a lot of noise in responses. However, in the medium to long term, the systems tested will not be overly driven by N fertilizer effects alone, rather, over time it might be possible to detect the effect of accumulated residues on both the soil chemical and physical properties. Therefore, for on-farm soil N differences to be detected from organic residues, our study highlights that multiple rotation cycles are needed. 96 Conclusions Nitrogen fertilizer benefits to maize yield were clear in our study, consistent with previous studies, and why farmers and governments invest in inorganic fertilizer. After one rotation cycle, we did not detect any residue quality or quantity effect on soil inorganic N or maize grain yield. Therefore, more rotation cycles are required for further studies on residue quality and quantity effects on soil N and crop yields in smallholder farming systems. Additionally, farmers may well need to be supported through policies that provide support for sustainable intensification practices such as high-quality residue retention. This is because it often takes multiple years to see an appreciable effect when such practices are implemented. This stands in contrast to current policies that often focus on access to inorganic fertilizer sources, with almost no attention to complementary investments in farmer access to organic, high quality amendments from legume residues. 97 REFERENCES 98 REFERENCES Ahmad, W., Khan, F., Shah, Z., & Khan, M.J., 2017. Changes in quality and yield potential of moderately degraded Alfisol with types and levels of nutrient input and legumes in crop rotation. Pedosphere 2017. https://doi.org/10.1016/S1002-0160(17)60364-3 Beedy, T.L., Snapp, S.S., Akinnifesi, F.K., Sileshi, G.W., 2010. Impact of Gliricidia sepium intercropping on soil organic matter fractions in a maize-based cropping system. Agric. Ecosyst. Environ. 138, 139–146. https://doi.org/10.1016/j.agee.2010.04.008 Chikowo, R., Mapfumo, P., Nyamugafata, P., Nyamadzawo, G., Giller, K.E., 2003. Nitrate-N dynamics following improved fallows and maize root development in a Zimbabwean sandy clay loam, in: Agroforestry Systems. pp. 187–195. https://doi.org/10.1023/B:AGFO.0000005219.07409.a0 Chimonyo, V.G.P., Snapp, S.S., Chikowo, R., 2019. Grain Legumes Increase Yield Stability in Maize Based Cropping Systems. Crop Sci. 59, 1222. https://doi.org/10.2135/cropsci2018.09.0532 Kafesu, N., Chikowo, R., Mazarura, U., Gwenzi, W., Snapp, S., Zingore, S., 2018. Comparative fertilization effects on maize productivity under conservation and conventional tillage on sandy soils in a smallholder cropping system in Zimbabwe. F. Crop. Res. 218, 106–114. https://doi.org/10.1016/j.fcr.2018.01.014 Kalasa, J.B., Mhango, W.G., Kanyama-Phiri, G.Y., Lowole, M.W., 2018. Maize response to doubled-up legumes, compost manure and inorganic fertiliser on smallholder farms in Ntcheu district of Malawi. African J. Agric. Res. 13, 1101–1110. https://doi.org/10.5897/ajar2017.12917 Kane, D.A., Snapp, S.S., Davis, A.S., 2015. Ridge Tillage Concentrates Potentially Mineralizable Soil Nitrogen, Facilitating Maize Nitrogen Uptake. Soil Sci. Soc. Am. J. 79, 81. https://doi.org/10.2136/sssaj2014.07.0273 Kang, B.T., Caveness, F.E., Tian, G., Kolawole, G.O., 1999. Longterm alley cropping with four hedgerow species on an Alfisol in southwestern Nigeria - Effect on crop performance, soil chemical properties and nematode population. Nutr. Cycl. Agroecosystems. https://doi.org/10.1023/A:1009757830508 Kellogg Biological Station/ Long-Term Ecological Research, 2019. Protocol - Gravimetric Soil Moisture [WWW Document]. URL https://lter.kbs.msu.edu/protocols/24 (accessed 4.7.20). 99 Kellogg Biological Station/ Long-Term Ecological Research, 2008. Particle Size Analysis for Soil Texture Determination (Hydrometer Method) [WWW Document]. URL https://lter.kbs.msu.edu/protocols/108 (accessed 9.2.19). Kong, A.Y.Y., Six, J., Bryant, D.C., Denison, R.F., van Kessel, C., 2005. The Relationship between Carbon Input, Aggregation, and Soil Organic Carbon Stabilization in Sustainable Cropping Systems. Soil Sci. Soc. Am. J. 69, 1078. https://doi.org/10.2136/sssaj2004.0215 Li, G., Messina, J.P., Peter, B.G., Snapp, S.S., 2017. Mapping Land Suitability for Agriculture in Malawi. L. Degrad. Dev. 28, 2001–2016. https://doi.org/10.1002/ldr.2723 Makumba, W., Janssen, B., Oenema, O., Akinnifesi, F.K., Mweta, D., Kwesiga, F., 2006. The long- term effects of a gliricidia-maize intercropping system in Southern Malawi, on gliricidia and maize yields, and soil properties. Agric. Ecosyst. Environ. https://doi.org/10.1016/j.agee.2006.03.012 Messina, J.P., Peter, B.G., Snapp, S.S., 2017. Re-evaluating the Malawian Farm Input Subsidy Programme. Nat. Plants 3. https://doi.org/10.1038/nplants.2017.13 Mtambanengwe, F., Mapfumo, P., 2006. Effects of organic resource quality on soil profile N dynamics and maize yields on sandy soils in Zimbabwe. Plant Soil 281, 173–191. https://doi.org/10.1007/s11104-005-4182-3 Mungai, L.M., Snapp, S., Messina, J.P., Chikowo, R., Smith, A., Anders, E., Richardson, R.B., Li, G., 2016. Smallholder Farms and the Potential for Sustainable Intensification. Front. Plant Sci. 7, 1–17. https://doi.org/10.3389/fpls.2016.01720 Nziguheba, G., Merckx, R., Palm, C.A., 2005. Carbon and nitrogen dynamics in a phosphorus- deficient soil amended with organic residues and fertilizers in western Kenya. Biol. Fertil. Soils 41, 240–248. https://doi.org/10.1007/s00374-005-0832-0 Palm, C.A., Sanchez, P.A., 1991. Nitrogen release from the leaves of some tropical legumes as affected by their lignin and polyphenolic contents. Soil Biol. Biochem. 23, 83–88. https://doi.org/10.1016/0038-0717(91)90166-H Rockström, J., Barron, J., Fox, P., Kijne, J.W., Barker, R., Molden, D., 2003. Water Productivity in Rain-fed Agriculture: Challenges and Opportunities for Smallholder Farmers in Drought- prone Tropical Agroecosystems. Water productivity in agriculture: Limits and opportunities for improvement. CAB Int. 85, 1–8. Rusinamhodzi, L., Corbeels, M., Van Wijk, M.T., Rufino, M.C., Nyamangara, J., Giller, K.E., 2011. A meta-analysis of long-term effects of conservation agriculture on maize grain yield under rain-fed conditions. Agron. Sustain. Dev. https://doi.org/10.1007/s13593-011-0040-2 Sakala, W.D., Cadisch, G., Giller, K.E., 2000. Interactions between residues of maize and 100 pigeonpea and mineral N fertilizers during decomposition and N mineralization. Soil Biol. Biochem. 32, 679–688. https://doi.org/10.1016/S0038-0717(99)00204-7 Six, J., Conant, R.T., Paul, E. a, Paustian, K., 2002. Stabilization mechanisms of soil organic matter: Implications for C-saturatin of soils. Plant Soil 241, 155–176. https://doi.org/10.1023/A:1016125726789 Smaling, E.M.A., Nandwa, S.M., Janssen, B.H., 1997. Soil fertility in Africa is at sake. Replenishing Soil Fertil. Africa. SSSA Spec., 47–62. Snapp, S.S., Grabowski, P., Chikowo, R., Smith, A., Anders, E., Sirrine, D., Chimonyo, V., Bekunda, M., 2018. Maize yield and profitability tradeoffs with social, human and environmental performance: Is sustainable intensification feasible? Agric. Syst. 162, 77–88. https://doi.org/10.1016/J.AGSY.2018.01.012 Snapp, S.S., Mafongoya, P.L., Waddington, S., 1998. Organic matter technologies for integrated nutrient management in smallholder cropping systems of southern Africa. Agric. Ecosyst. Environ. 71, 185–200. https://doi.org/10.1016/S0167-8809(98)00140-6 Thippayarugs, S., Toomsan, B., Vityakon, P., Limpinuntana, V., Patanothai, A., Cadisch, G., 2008. Interactions in decomposition and N mineralization between tropical legume residue components. Agrofor. Syst. 72, 137–148. https://doi.org/10.1007/s10457-007-9062-9 Vanlauwe, B., Aihou, K., Tossah, B.K., Diels, J., Sanginga, N., Merckx, R., 2005. Senna siamea trees recycle Ca from a Ca-rich subsoil and increase the topsoil pH in agroforestry systems in the West African derived savanna zone, in: Plant and Soil. https://doi.org/10.1007/s11104-004-0599-3 Vanlauwe, B., Wendt, J., Giller, K.E., Corbeels, M., Gerard, B., Nolte, C., 2014. A fourth principle is required to define Conservation Agriculture in sub-Saharan Africa: The appropriate use of fertilizer to enhance crop productivity. F. Crop. Res. 155, 10–13. https://doi.org/10.1016/j.fcr.2013.10.002 101