Using improved varieties and fertility enhancements for increasing yield of common beans (Phaseolus vulgaris L.) grown by small-landholder farmers in Uganda

Productivity of common bean (Phaseolus vulgaris L.) in Uganda is less than 30% of the yield of improved varieties grown on research stations. This yield gap has been attributed mainly to low soil fertility and susceptibility of local varieties saved by farmers to pest and disease infestations. This study evaluated the impact of four improved varieties and soil fertility improvement on bean yields on small-landholder farms in three agro-ecological zones in Uganda. Yields of common bean on-farm without fertilization were on average 523 kg/ha. Enhancing soil fertility on-farm with cattle manure (10 t/ha), P (60 kg/ha), or manure (5 t/ha) + P (30 kg/ha) led to average yields of 631, 615, and 659 kg/ha, respectively. On average, improved varieties produced more yield than the local farmer-saved variety, with or without soil fertility improvement. Improved variety K131 yielded 807 kg/ha, on average, in response to manure application, which was 54% greater than the yield of the local variety. P intensification up to 180 kg/ha per season, however, did not increase bean yields significantly at any of three research stations. These results confirm the yield advantage of growing improved varieties on small-landholder farms. The combination of improved genetics and fertility intensification alone, however, did not eliminate the yield gap between on-farm and potential bean yields.

with beans accounting for 7% of the national agricultural gross domestic product (CIAT, 2008).Despite the importance of beans as a dietary staple for rural households and agricultural product, average bean yields were estimated at 406 kg/ha in 2011 -less than 30% of the potential yield of improved varieties grown under optimum conditions (FAOSTAT, 2013).
The low productivity of the bean crop is attributed partly to widespread reliance on local varieties, which farmers save and sow year after year.According to the Uganda Census for Agriculture 2008/9, approximately 92% of rural households use saved seed; only 31% used improved or hybrid seed (UBOS, 2010).The widespread dependence on saved seed is attributed to limited access to improved seed due to poor distribution of formal seed outlets across the country, reliance on a few commercial varieties, poor marketing and marketing information systems, very narrow product range or low value addition, storage constraints, and high cost of certified seed.As part of its mandate, National Crops Resources Research Institute (NaCRRI) has released several improved varieties tolerant to several biotic and abiotic constraints in recent years.But these varieties are not widely available to farmers.These improved varieties have a greater yield potential and thus their adoption could result in significant increases in bean productivity.Field trials of the improved varieties in Uganda showed a yield advantage averaging 37% over local farmers' varieties (CIAT, 2008).A survey of farmers in areas of Uganda where improved varieties had been formally introduced showed high potential for adoption (David et al., 2000).Yet, limited accessibility to improved varieties by smallscale farmers and dietary preferences for local varieties continue to limit their use (Buruchara et al., 2011).
The most important abiotic factor limiting bean productivity in Uganda is poor soil fertility (Wortmann and Kaizzi, 1998;Wortmann et al., 1998;Bekunda et al., 2004;Lubanga et al., 2012).Phosphorus and nitrogen are the most limiting soil nutrients in bean producing areas of eastern Africa (Wortmann et al., 1998).Uganda has one of the lowest fertilizer usage in the region, estimated at only 1.8 kg/ha annually (Benson et al., 2012) compared to an average 7.1 kg/ha across sub-Saharan Africa (Druilhe and Barreiro-Hurl´e, 2012).In 2008/2009, the Uganda Bureau of Statistics estimated only 1% of the farming households in Uganda use chemical fertilizers, and only 6.8% apply manure, due to lack of availability and high cost of purchasing and maintaining animals (UBOS, 2010).
Integrating varieties that have superior yield potential with a soil fertility management program using locally available fertilizers might provide an affordable and sustainable option for small-landholder farmers to overcome these primary yield constraints (Okalebo et al., 2006;Lubanga et al., 2012).To this end, we conducted a series of on-farm and experiment station trials in three agro-ecological zones in Uganda where soils are severely depleted of nutrients.The overall goal was to identify methods for increasing bean yield on small-scale farms in these zones.The study had three specific objectives: (1) To evaluate the performance of improved varieties on nutrient depleted soils under typical on-farm conditions; (2) To determine the impact of increased nitrogen from manure and phosphorus application on bean yields on these nutrient-depleted soils under farmer conditions, and (3) To determine whether intensive phosphorus fertilization could improve bean yields.

On-farm experiments
Three on-farm experiments were conducted in Butansi and Bugulumbya sub-counties located in Kamuli District (00°55'N, 33°06'E) in southeastern Uganda.The district is located 1100 m above sea level and typically experiences two rainy seasons with peaks in March to June (season A) and August to November (season B).The average annual rainfall is 1350 mm and the average monthly temperature varies from 19 to 25°C.
The trials involved three farmer groups from Butansi sub-county and three farmer groups from Bugulumbya sub-county that were actively involved in the establishment and evaluation of the trials.Activities performed by farmer-cooperators included site selection, brush clearing, hand hoeing, planting, weeding, harvesting, drying, threshing, and measuring seed yield.Throughout the trials, farmers were trained on standard agronomic management of beans and guided as needed by technical staff from NaCRRI and Volunteer Efforts for Development Concerns (VEDCO), an indigenous nongovernmental development organization.Individual trials in each season were hosted by a farmer and managed by the farmer group to which they belonged.
The on-farm study consisted of three experiments.In the first experiment conducted in seasons A and B of 2009 and season A of 2010, farmers tested the performance of four improved varieties released by NaCRRI (K131, K132, NABE4 and NABE6) and a local variety, Kanyebwa, without soil amendments (Table 1).This involved a total of 54 on-farm trials, 18 in each season.
The second set of trials tested the response of the same varieties to 10 t/ha application of cattle manure just prior to planting.The manure was measured on a dry weight basis.The manure was evenly spread on top of the soil and incorporated by hand hoeing.The bean varieties were evaluated in two groups to reduce the size of the trial at each site and the workload for the farmer cooperators.Group 1 included K131, NABE4 and Kanyebwa; Group 2 included K132, NABE6 and Kanyebwa.A total of 36 on-farm trials, 12 in each of the three seasons, were conducted in season A and B in 2009.Plot area was 5 m × 5 m arranged in a randomized complete block design with two replicates.
A third trial evaluated the response of three varieties selected by the farmer cooperators (K131, NABE4, Kanyebwa) to cattle manure, phosphorus, and a combination of manure and phosphatic fertilizer.The fertilizer treatments were 60 kg/ha P, 10 t/ha manure, and 30 kg/ha P + 5 t/ha manure.The phosphorus source was triple super phosphate (46% P2O5, Lebanon Chemical Company S.A.L, Beirut, Lebanon).All fertilizer treatments were applied just prior to planting.Phosphatic fertilizer was applied manually and mixed into the soil with a hoe along the sowing line.Plot area was 3 m × 3 m arranged in randomized complete block design with two replicates.The experiment was conducted during seasons A and B in 2010 At all three locations, the soils were acidic ferralsols with an average pH below the optimum of 6.5 to 7.0 for dry bean yields (Table 2).The average organic matter content (3.1%) was in a range considered to be moderate (Okalebo et al., 2002).Analyses also confirmed the soils were low in nitrogen, phosphorus and potassium.Thung (1991) reported the critical level for soil P ranged between 8 to 15 mg/kg (Bray 2) for beans.
The bean varieties were Kanyebwa, a landrace commonly grown in Uganda, and NABE4, an improved variety released by NaCRRI.Both varieties were sown at a spacing of 50 cm × 10 cm.giving a target plant population of 20 plants/m 2 .Plot size was 5 × 5 m.The source of phosphorous and nitrogen was triple super phosphate (46% P205) and urea (46% N), respectively.P was applied at 0, 60, 120 and 180 kg P/ha.The dose of N was 25 kg N/ ha to all plots according to NaCRRI recommendations.In addition, all seeds were inoculated with Rhizobia TALL 899 strain (Makerere University, Kampala, Uganda; [4 × 10 6 rhizobia/g] with one 250-g packet sufficient to inoculate 15 kg seed).Bean plots were rotated with maize variety 'Longe 5' grown with similar levels of P and N fertility.The same plot locations were used repeatedly over three seasons for each phosphorous level.Thus, bean plots treated with 0, 60, 120 and 180 kg P/ha received the same fertilizer treatment for each of three successive seasons.At each site, the eight factorial treatment combinations were arrayed in a randomized complete block design (RCBD) with three replicates.

Soil analysis
Composite soil samples obtained from the surface 20 cm were collected from each site before land preparation.Soil and manure analyses for the on-farm trials were conducted at the National Agricultural Research Laboratories at Kawanda (NARL Kawanda), while those from on-station trials were analyzed at the Makerere University soil analysis laboratory following procedures by Okalebo et al. (2002) and Analytical procedures were similar at both laboratories, except extractable P was determined using the Melich 3 method at NARL Kawanda, while the Bray 1 method was used at Makerere University.

Yield and yield components
Harvested plant population (plants/m 2 ), pods/plant, seeds/pod, 100seed weight (g) and seed yield (kg/ha) were estimated at physiological maturity.Harvested plant population was estimated from the number of harvested plants divided by the harvest area.
The number of pods/plant was calculated from 20 randomly chosen plants.A pod was counted if it contained at least one mature seed.Number of seeds/pod was determined from 20 randomly chosen pods from these 20 plants.Beans were threshed in the customary way with sticks after air-drying the pods.Seed moisture content was determined using a moisture meter (Steinlite SL95, Atchison, Kansas, USA).The seed yield and 100-seed weight are calculated at 13% moisture content.

Statistical analyses
The data were subjected to analysis of variance using Proc Mixed in SAS 9.3 (2007).Variety and fertility treatments were considered fixed factors.Location and replication were considered random effects.Significant differences between means were determined by least significant differences (LSD) at p = 0.10.

Soil chemical and physical analyses
The majority of the soils in Kamuli district are classified as orthic ferralsols (FAO and UNESCO, 1977).These soils have good drainage, but are severely weathered and tend to have a low cation exchange capacity.Most soils at the trial sites were acidic ranging in pH from 4.4 to 6.5 (Table 2).While beans generally tolerate slightly acidic soils, alkaline conditions above pH 7 decrease availability of micronutrients such as iron and zinc (Schwartz et al., 2004).Results from 116 soil samples indicated 92% of the on-farm plots had less than critical phosphorus (P), 62% were deficient in nitrogen (N), and 44% were low in organic matter (OM) (data not shown).Similar nutrient deficiencies for Kamuli soils have been reported by Wortmann and Kaizzi (1998) and Tenywa et al. (1999).
The average values for soil OM, N and P were 3.15%, 0.2% and 1.85 mg/kg respectively (Table 2).As macronutrients required for proper plant growth and development, P and N deficiencies will negatively impacts seed yield.Soil deficiencies in P and N observed in these Kamuli trials certainly confirm earlier reports of low soil fertility as a primary constraint to increased yields in sub-Saharan Africa (Okalebo et al., 2006).Organic matter plays an important role in improving nutrient availability by increasing cation and anion exchange capacity, increasing water retention and improving soil structure (Johnston et al., 2009).The poor soil fertility observed was attributed to negative nutrient balances from the common small-landholder practice of continuous cultivation without replenishment of nutrients extracted by the crops (Shepherd et al., 1996;Bekunda et al., 1997;Sanchez, 2002).According to the 2008/9 Uganda Census on Agriculture, the majority of small-scale farmers do not use any form of inorganic or organic fertilizer (Okoboi and Barungi, 2012).

Yield of improved varieties on nutrient depleted soils
Adoption of improved varieties has great potential as a sustainable way to improve yields among resource poor farmers (David et al., 2000;Maredia et al., 2000).The objective of our on-farm trials was to engage farmers in direct comparisons of improved varieties -K131, K132, NABE4 and NABE6 against a popular local variety Kanyebwa under typical farming conditions.On average, the improved varieties yielded 3 to 25% more than the local bean variety (Table 3).K131, in particular, consistently produced significantly greater yield than the other four varieties.The local variety grown from farmersaved seed generally yielded the least, averaging approximately 464 kg/ha.The superior performance of K131 was attributed to a combination of greater number of pods/plant and a greater number of harvested plants/m 2 .K131 and K132 also yielded 65 and 35% more than Kanyebwa in on-farm field trials on fairly mineral-rich nitosol soils in the Mbale district of Eastern Uganda (David et al., 2000).The small seeded varieties, K131 and NABE6, produced significantly more pods/plant and seeds/pod than the large-seeded varieties Kanyebwa, K132, and NABE4.
Harvest plant population was an important determinant of crop performance across seasons and locations.It was generally far less than the recommended optimum plant population of 20 plants/m 2 , which was the target planting population for all varieties (Table 3).K131 retained the most plants throughout harvest and generally produced the best yield.Kanyebwa typically had far fewer plants survive until harvest and often was the poorest yielding variety.De Brum Piana et al. ( 2007) also associated yield loss with a failure to maintain plant population.Failure of common bean varieties to maintain an optimum number of plants throughout the season could reflect poor seed germination, poor vigor, or mortality under stressful field conditions.Studies relating temporal loss of plants with occurrence of biotic and abiotic stresses during the growing season are needed to target management strategies that stabilize plant population density.
There were significant effects of variety, location, and location × variety interactions on yield, 100-seed weight, pods/plant, and seeds/pod (Table 3).In addition, location and location × variety interaction effects were significant for the number of plants harvested per unit area.Evidently, the improved varieties and farmer-saved variety responded differently to local soil, weather, and/or farm-management conditions.This outcome underscores the importance of active variety testing and selection for local conditions.It clearly indicates the need for farmer access to a greater number of improved bean varieties with genetic potential for yield under their local growing conditions.
Indeed, even the yields of the improved varieties were well below the potential yields observed under controlled conditions at the national research stations.Based on data provided by NaCRRI (Table 1), average productivity for the improved varieties tested was only 19 to 28% of the potential yields for NABE6, K131, NABE4, and K132.Thus, simply introducing improved varieties on smalllandholder farms alone was not sufficient to bridge the yield gap.These results led us to address another major constraint to productivity, low soil fertility, in an attempt to achieve the significant yield improvements expected of these improved varieties.

Response to manure
As a soil improvement strategy, we tested the response of common bean varieties K131, K132, NABE4, NABE6 and Kanyebwa to 10 t/ha of locally sourced cattle manure applied at planting time.Local manure was chosen because it is a renewable resource and an economical means of soil improvement, especially for resource poor farmers.The manure used averaged 14 g N/kg, 1.6 g P/kg, and 7.2 g K/kg on a dry weight basis.Based on this analysis, 10 tons of manure per hectare would provide 140 kg N/ha, 16 kg P/ha, and 72 kg K/ha.The amount of phosphorus provided by manure was far short of the 60 kg P/ha recommended for common beans by the National Agricultural Research Laboratories at Kawanda.Manure from communally grazed cattle has been reported to contain low amounts of nitrogen, phosphorus and potassium (Palm et al., 1997;Tenywa et al., 1999;Materechera, 2010).Nonetheless, these fertility treatments were expected to improve on-farm yields, as soils in all trial sites were very low in nutrients particularly N, P and OM (Table 2).Averaged across all varieties and locations, manure application significantly increased yields by 26% (Table 4).Comparison between varieties, however, revealed significant yield increases only for K131 and Kanyebwa, which were 49 and 50%, respectively.The observed increase was associated with greater seed weight, and increased number of pods/plant and seeds/pod.Application of manure significantly increased the seed size of the large seeded K132 and NABE4.The smaller seeded NABE6 showed a significant increase in number of pods/plant in response to manure.Although these results generally indicated that improved soil fertility management would have a positive impact on bean yields, there were significant variety x treatment and location x treatment interactions for the yield response to manure application.This interaction might have been due to variation in rates of manure decomposition, which depends on local soil and environmental conditions (Eghball, 2000;Eghball et al., 2002).Because trial sites were different each season, the long-term benefit of manure decomposition would not have been realized.

Response
to manure, phosphorus, and a phosphorus-manure combination While there were some positive results from the initial onfarm variety and fertility trials, it was evident that improved varieties and manure alone were not sufficient to bridge the yield gap between on-farm and potential yield.Soil chemical analyses indicated that phosphorus was consistently the most limiting macronutrient among the field sites.Phosphorus is particularly important for promoting nitrogen fixation, which is often limiting in common beans (Vance et al., 2003).Therefore, we tested whether a combination of inorganic phosphorus and manure could improve yield of common beans on smallholder farms.The yield response to manure, phosphorus fertilizer, and the combination of phosphorus and manure varied by variety and treatment (Table 5).The yield of K131 was not improved by application of phosphorus or the combination of manure and phosphorus.Combined application of manure and phosphorus, however, increased the yield of the farmer variety, Kanyebwa.The yield advantage resulted from an increase in pods/plant and greater plant population at harvest (Table 5).NABE4 yields also responded significantly to manure, phosphorus fertilizer, and manure + phosphorus fertilization.The increase in yield from manure also was associated with greater plant population at harvest, while phosphorus + manure increased seeds/pod.Yields from application of phosphorus alone (60 kg/ha) often matched those from application of manure (10 t/ha) confirming the importance of phosphorus as a yield-limiting nutrient in farmers' fields.Further, combined application of phosphorus and manure led to significant increase in yield for both Kanyebwa and NABE4.
Although smallholder farmers generally recognize the potential benefits of manure application, it is not a common practice because of its bulkiness.Likewise, P application is limited because of its expense.The results of this study suggest such deterrents to widespread use of these yield enhancers could be overcome by combining them at modest levels without sacrificing yield response.Complementary use of inorganic and organic fertilizers has been suggested as an effective practice of soil fertility management to ensure plant nutrition while regenerating the soil (Palm et al., 1997;Lubanga et al., 2012).A combined application approach could be quite effective as a long-term strategy to increase nutrient availability in the depleted soils of Kamuli.

Effect of P intensification on bean yield
Lack of available phosphorus is often reported as the most limiting nutrient for greater bean production (Sanchez and Logan, 1992;Wortmann et al., 1998;Lunze et al., 2007).Several authors have proposed Because of the high cost of commercial fertilizers (Okoboi and Barungi, 2012), there are few options to improve soil fertility for resource limited farmers.Incorporating animal manure, crop residues as green manure, agroforestry trees and legumes into the land management system have proven beneficial.These options, however, often are required in large quantities, are labor intensive, and generally are low in nutrient quality and concentration (Okalebo et al., 2006).Widespread use of these approaches requires significant changes in local farming systems such as adopting mixed agriculture to generate adequate supplies of animal manure.An integrated approach that naturally couples these options withinorganic sources of nutrients has proven beneficial (Sauer and Tchale, 2009).To restore the depleted soils encountered in our study, nutrient additions will have to be applied for several years.Soil replenishment and logistical support for distribution of manure should be part of the development policy (Smale et al., 2013), so that there is deliberate government investment aimed at building soil nutrients as agricultural capital.
Even when grown under well-managed conditions, the yields of improved varieties were typically less than 50% of the potential yields NaCRRI has reported (Table 1).In nearly all cases, harvest populations were well below the recommended optimum of 20 plants/m 2 (in some cases 50% less than the planted population).The reasons for the extensive plant loss are not known, but much of the loss is likely attributable to poor seed quality resulting in low percent germination, poor vigor, and mortality under stressful field conditions (Ochilo et al., 2013).Documenting the temporal loss of plants during the growing season might reveal management strategies to help farmers stabilize plant population density.But changes in crop management will not overcome the widespread lack of high quality seed, which remains a major yield limitation for small landholder farmers.One promising approach to increase farmer access to quality seed is community-based seed production of Quality Declared Seed (QDS) (Takoutsing et al., 2012).The Ministry of Agriculture and Animal Industry (MAAIF) oversees this component of the Informal Seed Sector in Uganda and recognizes QDS as commercially acceptable for crop production.
MAAIF programs that link community-based seed producers with public breeders and the agricultural extension system could accelerate the production of high quality bean seed, and help small landholder farmers in Uganda overcome the gap between potential and realized bean yield.

Table 1 .
Characteristics of six common beans varieties grown in Uganda.Information compiled from the NASECO Seed Company, Kampala, and the National Crops Resources Research Institute, Namulonge, Uganda.
Agricultural Research Institute in Southwestern Uganda (1430 masl, average rainfall 915 to 1020 mm/annum).Hereafter, these locations are referred to as Nakabango, Namulonge and Mbarara representing three agro-ecological zones where beans are commonly grown.Field trials were conducted in the first (A) and second (B) season of 2011 and the first (A) season of 2012.Typically, season A rains occur from March to June; season B rains occur from September to December.

Table 2 .
Characteristics of soils sampled from trials sites in Bugulumbya, Butansi, Nakabango, Namulonge, and Mbarara, Uganda.The values are averages of samples taken at each location.
*Critical valued according to Soil and Plant Analysis Laboratory -NARL Kawanda.

Table 3 .
Average yield, yield components, and harvested plant populations for five common bean varieties evaluated in on-farm trials.Data are pooled for three growing seasons and 60 locations in 2009 and 2010.
*Means in columns followed by the same later are not significantly different at p = 0.10.

Table 4 .
Yield and yield component responses of common bean varieties K131, K132, NABE4, NABE6, Kanyebwa to 10T/ha manure applied at planting.Data are the mean of 12 locations pooled for two growing seasons in 2009.
*Means in columns followed by the same letter within each variety are not significantly different at p= 0.10.

Table 5 .
Response of yield and yield components of three common bean varieties to phosphorus (60 kg/ha), manure (10 t/ha) and phosphorus (30 kg/ha) + manure (5 t/ha) fertilizer treatments.Data are the mean of 22 locations pooled for three seasons in 2010 and 2011.Within each variety, fertility treatment means followed by the same later are not significantly different at p= 0.10. *

Table 8 .
Estimate of phosphorus fixing capacity by soils at the Nakabango, Namulonge, and Mbarara trial sites.