Enhancing crop productivity and its economic farm profitabilty of smallholder farmer through the use of green manures from Alnus acuminata

Abstract

Decline in soil fertility is a major threat to land productivity and food security in the East African highlands. This calls for the application of nutrient inputs to improve crop productivity. A study was conducted in Nyabihu District of Rwanda from 2013 to 2016 to assess the effect of Alnus acuminata green manure (AGM)─applied in situ, through biomass alone, or combined with inorganic fertilizer─on the productivity and profitability of maize (Zea mays), beans (Phaseolus vulgaris), and potato (Solanum tuberosum). The treatments included application of AGM, inorganic fertilizer, combination of AGM and inorganic fertilizer, and unfertilized plot as a control (except for potato). These treatments were compared in two seasons and on two local soil fertility levels (medium and high) as defined by the farmer’s knowledge and experience. There was a convergence between farmers’ criteria and soil analysis in the soil fertility evaluation. Crops yields were analyzed using a linear mixed model while for other parameters, descriptive statistics were applied. The combination of AGM and inorganic fertilizer recorded the highest increment in maize (44%) and bean (46%) yields compared with inorganic fertilizer while it increased up to 87% compared with the unfertilized control. The financial analysis showed that AGM + inorganic fertilizer recorded the highest value-to-cost ratio (VCR) of 24.6 for potato and a significantly lower VCR (2.9) for maize and beans. The high VCR highlights a significant potential contribution of AGM + inorganic fertilizer to increase incomes of resource-constrained potato farmers in the Rwandan highlands. However, low crop prices possibly make this practice less attractive for those cultivating maize and beans.

Introduction

Since the area under agriculture cannot be expanded indefinitely to meet the rising demand for food, there is a need to increase crop productivity sustainably on available agricultural land, especially in densely populated countries (Van Ittersum et al. 2016). In spite of the commitment of most African countries to increase the use of inorganic fertilizer from 8 to 50 kg ha⁻¹ (Wanzala 2011), crop productivity has remained unsatisfactory due to various factors, among which is the low purchasing power of smallholder farmers (Sheahan and Barrett 2017). In Rwanda, crop production gaps (margin between potential production based on attainable yield and actual production) are still important for major crops. Such gaps range from 60.7, 45.9, 36.3, 71.7, 64.0, to 76.4% for maize, wheat, rice, bean, cassava, and Irish potato, respectively (IFDC 2014). Across the globe, application of inorganic fertilizers is widely used to overcome nutrient deficiencies, leading to dramatic increases in yields of several crops (Huang et al. 2010). However, their use remains very low in sub-Saharan Africa (SSA) (including the Rwanda highlands) and ineffective in sustaining crop production and maintaining soil fertility since they are not affordable and accessible to poor-resource farmers (Kaho et al. 2011). Moreover, continuous use of inorganic fertilizers breeds soil acidity, cation loss, and macronutrient depletion, by way of losses in gaseous form and leaching, especially in cases of noncompliance to recommended doses (Huang et al. 2010). As farmers face financial constraints that prevent them from purchasing inorganic fertilizers (due to the high price situation exacerbated by the Ukraine and Russia war, as well as the low profit margins), alternative affordable and environment-friendly options must be found. However, there is very little evidence and models as to the use of alternative resources such as agroforestry fertilizer alone or in combination with inorganic fertilizer (Akinnifesi et al. 2008).

Combining mineral and organic fertilizers is widely accepted as a viable approach to address soil fertility problems and sustain crop productivity in SSA (Chivenge et al. 2011; Gram et al. 2020). The approach is largely described as part of the integrated soil fertility management (ISFM) framework. Organic fertilizers may be supplied through various forms, including animal manure, household compost, crop residues, leguminous cover crops and non-leguminous trees, and shrubs. Agroforestry-fertilizing trees may potentially provide substantial quantity of organic resources as compared with other sources such as animal manure. The practice has been reported to not only increase crop productivity but to also contribute to optimizing agronomic efficiency (AE), among other merits (Sanginga and Woomer 2009; Vanlauwe et al. 2015). The combined application of mineral and organic fertilizers has been shown to play an effective role in increasing AE of N and productivity for a number of crops, including maize (Nziguheba et al. 2002; Chivenge et al. 2011; Abdelzaher et al. 2017). Benefits of the practice include the likelihood of immediately making both minor and major nutrients available to crops and, with organic fertilizer use, increase soil organic matter content and improve soil biological properties and soil moisture content (Sanginga and Woomer 2009; Gram et al. 2020).

While organic resources have largely been reported to boost crop yields, variations in yield response stemming from application of organic fertilizers are observed and these are mainly linked to the quality of the organic resource applied. The quality of resource material is linked to availability of N and lignin and polyphenol content (Palm et al. 2001).

In addition to the quality aspect, the beneficial effects of applying combined organic and mineral fertilizer also depend on climate conditions and soil environment. Climate relates to weather conditions prevailing during the cropping season that are directly linked to the availability/absence of rains over the season. With regard to climate factors, Tian et al. (2017) noted a higher decomposition rate and subsequent nutrient release with increase in quality of organic material in wetter climates. However, in drier climates, decomposition and nutrient release were faster with low-quality organic material than with high-quality organic inputs.

In the Rwanda context, weather conditions differ across seasons. The two seasons differ in terms of amount and frequency of rainfall across the seasonal period. While the long rainy season is characterized by moderate rainfall distributed over almost 4 months, the short season is characterized by heavy rain running for a shorter period of time, from March to May. This may significantly influence the decomposition of organic fertilizer and nutrient release to crops. Also, many crops have different nutrient requirements, thus the need to apply different types and amounts of fertilizers at specific stages of plant growth.

Farming is carried out under different soil and climate conditions. Crop responses to organic fertilizer are likely to differ, depending on specific contexts. This necessitates assessing the performance of various organic resources under specific farming situations.

Agroforestry fertilizer trees have high N and carbon contents in their biomass, with potential to complement available fertilizers at the farm level (manure and crop residues) and provide a way to add organic carbon to improve soil health (Ngetich et al. 2012). Combined application of organic materials and inorganic fertilizers was found to enhance carbon storage in soils and reduce emissions from N fertilizer use while contributing to high crop productivity in agriculture (Abbasi and Yousra 2012).

Alnus acuminata is a globalwide common fertilizer shrub (Bare and Ashton 2016). East African highlands, including Rwanda, are no exception to that trend (Ndoli et al. 2017). It is worth noting that the use of AGM is adopted by 80% of farmer households in the northwestern part of Rwanda (Mukuralinda et al. 2016). A gap was observed in reporting macronutrient (NPK) AE (Gitari et al. 2018) as well as in financial metrics such as VCR (Morris et al. 2007) with regard to green manure (strictly referred to here as fresh biomass from fresh leaves), particularly in Africa (Bolwig et al. 2009; Rhodes et al. 2012).

Cyamweshi et al. (2023) carried out tests to assess the effects of A. acuminata fresh biomass on potato and wheat under a controlled environment. Siriri et al. (2010) used biomass directly below the pruned trees in a controlled experiment (on-station basis). For their part, Okorio et al. (1994) and Ndoli et al. (2017) focused on the effect of shaded leaves of A.acuminata on crop yield. In the last three decades, almost no study has focused on the effect of AGM on crop yield and profitability in farmers’ settings. The knowledge gap remains significant, as A. acuminata biomass is often not used in degraded land (Kuria et al. 2019). Moreover, the use of inorganic fertilizers remains poor, while the rate of adoption depends on the specific crop involved (Mugabo et al. 2020). Testing in farm environments allows closing such knowledge gaps and contextualizing related recommendations. Sinclair and Coe (2019) suggest that participatory research should define the contextual variation, as well as the critical contextual variables, to address the complexity of farmer livelihoods.

The main objective of this study is to assess the extent to which AGM could be used to improve crop productivity and profitability in smallholder farms in Rwanda using a participatory approach. The key hypotheses were: that (1) maize, bean, and potato yields are increased by application of AGM with and without adding inorganic fertlizer; (2) nutrient (NPK) AE is enhanced when combined with organic and inorganic sources of nutrients; and (3) financial profitability varies by level of soil fertility, inputs applied, and crop type. The specific objectives are as follows: (1) to evaluate the effects of AGM with or without inorganic fertilizer combination on the yields of maize (Zea mays), beans (Phaseolus vulgaris), and potato (Solanum tuberosum); (2) to determine the nutrient (NPK) AE from inputs; and (3) to assess the use of AGM alone or in combination with inorganic fertilizer as it relates to financial profitability from growing these three major crops in contrasting soil fertility classes.

Material and methods

Study area

Farmer participatory trials were conducted from September 2013 to September 2016 in Kadahenda cell, Karago sector, Nyabihu District in the western province of Rwanda (latitude 1° 39′ 46.53″–1° 37′ 31.43″ S; longitude 29° 29′ 48.57″–29° 31′ 27.13″ E) (Fig. 1). The site is located in the Congo-Nile divide agro-ecological zone (Verdoodt and van Ranst 2003), with altitude ranging from 2300 to 2600 m above sea level. The mean annual temperature is 13.2 °C and average annual rainfall stood at 1550 mm, distributed over two cropping seasons: the ‘short rains’ running from mid-February to mid-July (season B), and the ‘long rains’ running from September to January (season A). The highest amounts of precipitation were recorded in April for season B and in October and November for season A (Fig. 1). As reported by the Soil Information Systems (1983–1993), the soil is dominantly of volcanic parental materials mixed with moderately acidic debris mainly from granites (Verdoodt and van Ranst 2003). The main crops grown in Nyabihu District include potato, maize, and climbing beans (NISR 2015).

Fig. 1

Location of the study area (Kadahenda cell) and daily rainfall and temperature trends during the study period (Musanze Meteorological Station)

The annual mean temperature varied between 14.0 and 15.5 °C during the 3 years of experimentation. Maximum and minimum monthly temperatures recorded were 25.0 and 10.2 °C, respectively. The annual rainfall ranged from 1450 to 1600 mm during the study period (Fig. 1).

Participatory assessment of farm characteristics

A participatory trial design (ParTriDes) workshop was used to support farm characterization with regards to farm socioeconomic and biophysical factors, as well as local soil fertility management and classification (Barrios et al. 2015). To better understand the context of soil fertility decline using participatory methodological tools (Barrios et al. 2012), farmers were invited to define soil fertility indicators that were relevant to their farms. The most common criteria used by local farmers were soil color and texture. Farmers associated high soil fertility with soft soils of dark color and medium fertility with soft soils of either brown to reddish color or sandy soils of dark color.

Classification of soil fertility management

The classification of soil fertility by farmers was compared with four common soil fertility properties: (1) water pH (pH_w), (2) available phosphorus (P_av), (3) exchangeable potassium (K_ex), (4) total N (Ntot), and (5) soil organic carbon (OC).

These soil properties were then scored on a scale of 0 (very constraining) to 1 (very favorable). The scores were aggregated in two indexes. The first index, called SFI (soil fertility index), is an aggregation of individual scores into one additive and unit-less index by summing the individual score for each soil property and dividing by the total number of soil properties considered and multiplied by 10 (Mukashema 2007) (Eq. 1). To get a second aggregating index, MLFI (most limiting factor index), the approach adopted by George (2005) and by (Verdoodt and van Ranst 2003) was followed (Eq. 2).

$$SFI=\frac{\sum_{i=1}^{N}Sci}{N}\times 10$$

(1)

$$MLFI = min\sum\nolimits_{i = 1}^{N} {Sc_{i} }$$

(2)

where Sc_i is the score of each of the soil properties (pH_w, OC, P_av, Ntot, and K_ex) and N is the number of soil properties. Min refers to minimum value.

Farmers’ selection in participatory experimental trials

Forty-one farmers were selected on the basis of their willingness, land ownership, and capacity to conduct the experiment before hosting the experimental trials to assess the challenges associated with soil fertility and crop production. The number of farmers involved in the participatory trials, referred to as replicates, were 15 for maize (eight farmers in 2014 season A and seven additional farmers in 2015 season B), 16 farmers for beans (12 farmers in 2015 season A and additional seven in 2015 season B, with three farmers participating in the two seasons but on different plots), and 11 farmers for potato (seven farmers in 2015 season B and additional seven farmers in 2016 season A, with three farmers participating in the two seasons but on different plots). The variation in the number of farmers, by season and crop, was related to the number of farmers who volunteered to participate (offering their land and biomass for use).

Experimental layout and management

The experiment consisted of four treatments laid out in a randomized complete block design per crop from September 2013 to September 2016. The treatments were plots with an application of AGM alone, mineral fertilizer (MF) alone, AGM + MF, and a fertilizer-free control (C) for potato and common beans. Maize trials involved three treatments (MF, AGM, and AGM + MF). Although the control did not receive MF or AGM, it did get the same crop management as did the other treatments (crop variety, land preparation, weeding, and pesticide application for potato). These treatments were tested under two local soil fertility levels (medium and high) as defined by the farmer’s knowledge and experience (see Sect. “Classification of soil fertility management”) and under two seasons (seasons A and B).

A 6 × 6-m plot was used to evaluate the responses of Pool 9 maize (Z. mays L.), Kirundo potato (S. tuberosum) variety, and Gasilda climbing beans (P. vulgaris) varieties to green manure (GM), MF, and their combination during cropping seasons. Maize was planted at inter- and intrarow spacing of 30 and 80 cm (83,333 plants ha⁻¹), respectively. Beans were planted with a row spacing of 20 cm and within-row spacing of 50 cm, accounting for a plant density of 200,000 plants ha⁻¹, while one potato tuber seed was sown per hole at 0.80 × 0.30 m (41,666 plants ha⁻¹).

A. acuminata leaves are soft (lignin: 35%), N rich (N: 2%) (Encalada et al. 2010) with a small C-N ratio (< 17) (Aceñolaza and Lancho 1999). Its highest rate of mineralization happens in the first 14 days (Iñiguez‐Armijos et al. 2016). Green manure strictly refers to fresh biomass from fresh leaves; these were collected from trees planted by farmers in the previous years (from mature trees). The A. acuminata leaf biomass (GM) collected in the same farm was broadcast and incorporated into the plough layer (10–20 cm) soil 1 week before planting the seed at a rate of 8 tons ha⁻¹ for maize and beans, respectively, and 2 tons ha⁻¹ dry matter equivalent (applied fresh) for potato, while inorganic fertilizers were applied in the hole for each crop in each season 1 day before sowing. The national recommended rates of 300 kg NPK ha⁻¹ (17:6:14.2) was applied for potato, while application rates for maize consisted of basal fertilizer at 100 kg ha⁻¹ of (AU: what? How many days?) days after planting (DAP) (18:16:0) and 50 kg urea ha⁻¹ (46% N) and top dressing of 50 kg urea ha⁻¹ (46% N) within 45 days after planting. The control had no application of GM and inorganic fertilizers. Weeding was done at the same time for all farmers, twice for each cropping period at 45 and 90 DAP for maize and at 30 and 45 DAP for beans and potato.

Soil and plant sampling and analysis

Composite topsoil samples were collected from every participating farm with the use of a soil auger at 0–20 cm depth before the establishment of the participatory trials. After air drying, the soil was crushed and passed through a 2 mm sieve before analyses of pH water, available P, and exchangeable K content were done. Further, other portions of the composite samples were ground and filtered through a 0.5 mm sieve for SOC and N_tot analyses.

Soil pH in water was measured at a 2.5:1 ratio of water to soil suspension using a method described by Van Reeuwijk (1993), while N_tot was measured using the Kjeldahl method (Van Reeuwijk 1993). P_av was analyzed using Bray 1 method (Bray and Kurtz 1945), where P was extracted by NH₄F 1N before using a UV spectrophotometer to determine its concentration in the soil samples (Van Reeuwijk 1993). Exchangeable K was analyzed using a flame photometer after extraction with NH₄OAc 1N at pH 7. Soil organic content was measured using the (Walkley and Black 1934) method which serves titration and colorimetric measurements. Three random samples of A. acuminata green leaves were collected at each farm plot before incorporation into the soil, dried at 72 °C for 48 h. Total N and P were extracted in a single digestion of dried plant samples with hydrogen peroxide and sulfuric acid solution. Selenium and salicylic acids were used as catalysts in the reaction. Total N was then determined by distillation and titration, and total phosphorus was assessed by the molybdenum-blue method (Okalebo et al. 2002). Potassium was extracted by sulfuric acid (1N) analyzed using a flame photometer (Okalebo et al. 2002).

Crop yield measurement

To establish the effect of treatments on maize yield, the number of cobs, shoots, and dry weight of the grains were determined for each farm plot. Of the 36 m² of planted area, 25 m² (5 × 5 m) were considered as harvested area by clearing 1 m at each side to avoid edge effects. Plant population per plot (5 × 5 m) was determined before excising them at the ground level. Composite plant samples were taken to determine the moisture content in the grain. The latter was air-dried until it reaches 14% moisture content and then weighed.

Grain dry weight was determined after shelling the cobs. Total grain dry weight for all plants in each plot was determined by multiplying the grain dry weight to cob fresh weight ratio for the sub-sample by the total fresh weight of cobs from all plants in the same sampling plot. The same plot size was used for beans and potato. For potato and beans, fresh weight in the harvest area was used to estimate yield (tons ha⁻¹). In the case of common beans, fresh grain weight at maturity was measured after removal from the pods.

Nutrient use efficiency indices and profitability analysis

The N, P, and K from AGM were considered in the computation of the nutrients’ AE, together with the nutrients contained in inorganic fertilizers, as per the treatment applied. NPK AE was determined, and financial analyses were performed using the formulas given in Table 1. The financial analysis was simplified to account only for the cost of major inputs, namely GM, MF, seeds, pesticides, and labor for tilling, leaf biomass collection and application, planting, weeding, spraying pesticides, and harvesting. The benefit is represented by the value of the main products of the crops (maize grain, climbing beans dry grain, and fresh potato tubers). The costs of seed, fertilizer, and labor (tillage, weeding, spraying pesticide, and harvesting) per season were 40 000, 196 000, and 194 000 Rwf ha⁻¹ for maize, respectively. The respective input costs for climbing beans per season were 100 000, 153 000, and 141 000 Rwf ha⁻¹. Those for potato were 225 000, 153 000, and 247 000 Rwf ha⁻¹. Additionally, fetching and application of AGM cost 1650 Rwf kg⁻¹ of dry matter applied. The prices of outputs used during the period under study were estimated using 2015 local prices with 180, 250, and 150 Rwf kg⁻¹ for dry maize grain, beans grains, and fresh potato tubers, respectively. The exchange rate to USD was 727 Rwf USD⁻¹ on average between 15 September 2014 and 15 February 2016 (690–764 Rwf USD⁻¹).

Table 1 Nutrient agronomic efficiency and profitability calculation formulas

Statistical analysis

The yield of maize grain, beans grains, and fresh potato total weights were analyzed using the linear mixed model effect (LME) in R software version 2.2.3. The fixed variables consisted of season, soil fertility class, and treatment. The random variable consisted of farmer’s plot. All possible models were fitted, and the best ones were chosen based on the least Akaike information criterion. Treatment means were compared using the least significant difference (LSD) at p < 0.05, using the predict mean R package. Descriptive statistics were used to analyze the data on agronomic use efficiency and option profitability factors.

The AE trend was compared with that of the fertilizer response curve by analyzing highly variable responses to treatments through the use of boundary lines fitted to the maximum response values across the range of yields for a range of yield-determining factors. Descriptive statistics of financial metrics presented in Table 1 were used to analyze the options’ profitability and AE. The financial metrics used for each recommendation treatment were estimated through a partial budgeting approach.

Results

On-farm soil fertility characterization

The indices used (SFI and MLFI) validated soil fertility classes previously defined by farmers in the study area. The average value of SFI (5.8) was similar for both soil fertility classes (medium and high soil fertility) defined by farmers (Table 2). However, the MLFI indicated that all medium-fertility soils had at least one of the soil properties scoring 4.0, which was not the case in high-fertility soil classes (average 4.7). This description of soil fertility classes was used to discriminate among groups of farms in the analysis of the effect of AGM biomass and inorganic fertilizers.

Table 2 Values of soil properties defining SFI per soil class (average ± standard deviation)

Considering the threshold and classification for each of the soil properties, the most limiting parameter was P (8.4–25.6 mg kg⁻¹, scored 0–0.4 on a scale of 0–1), while the most favorable parameters were OC (1.2–3.4%, scored 0.6–0.8 on a scale of 0–1), and pH (5.5–7.1, scored 0.6 to 1 on a scale of 0–1). Phosphorus content of the soil varied from 8.4 to 18 mg kg⁻¹ (very low to low) in medium soil fertility classes; whereas it varied from 10.3 to 25.6 mg kg⁻¹ (low to moderate) for the high soil fertility class. Potassium was very low to moderate (0.05–0.63 cmol_c kg⁻¹) for the medium-fertility soils and low to very high (0.09–3.72 cmol_c kg⁻¹) for the high- fertility class.

Crop yield response to different management options

Crop yields varied from 0.5 to 4.1, from 5.3 to 56.7, and from 2.7 to 10.8 t ha⁻¹ for beans, fresh potato, and maize, respectively. The combination of GM and MF significantly increased maize, beans, and fresh potato yield compared with MF in the two soil fertility classes (medium and high) (p < 0.05). It was also observed that the same treatment (inorganic fertilizer + AGM) increased maize yield by 24–55% during seasons B and A, respectively. Furthermore, it helped raise yield of climbing beans by 24–44% during seasons A and B, respectively, as compared with the plot where inorganic fertilizer was applied alone. The same trend was observed for fresh potato yield: AGM + inorganic fertilizer increased it by 8–6% during seasons B and A, respectively, as compared with the control. Treatments with sole application of MF or AGM did not exhibit a significant increase in production as compared with the control plot from the two soil fertility classes.

Bean yield under AGM + MF in high-fertility soils in season 2015 B was significantly higher (0.7 t ha⁻¹ increment on average, p < 0.05) compared with all other treatments in both seasons and soil fertility classes. The control in high-fertility soils in season 2015 B was significantly higher (0.7 t ha⁻¹, p < 0.05) than all treatments in season 2015 A, except for the AGM + inorganic fertilizer treatment. A significant interaction (p < 0.05) between treatment and season was observed for maize but not with soil fertility classes or between season and soil fertility classes. Due to this interaction, AGM + inorganic fertilizer had significantly higher yield in season 2014 A (1.5 t ha⁻¹ increment on average, p < 0.05) compared with season 2015 B and induced significantly higher maize yield (p < 0.05) than did inorganic fertilizer only in that favorable season. AGM + inorganic fertilizer had higher potato yield compared with the control for all the seasons and all the soil fertility classes. Nevertheless, AGM + inorganic fertilizer in season 2015 B was significantly higher (8.4 t ha⁻¹ increment on average, p < 0.05) than all treatments in season 2016 A due to an interaction between treatment and season.

Determination of nutrient agronomic efficiency

The applied leaf biomass of A. acuminata accounted for, on average, 3.1% of N, 0.09% of P, and 1.4% of K. The dry matter content observed was 40%. The results of AE for N, P, and K had different trends for three treatments (inorganic fertilizer, AGM, and AGM + inorganic fertilizer) along the gradient of fertility assessed by the level of the nutrient in the soil (Fig. 3). The N-AE ranged from 4.0 to 53.0 kg, from − 9.8 to 17.6 kg, and from − 98 to 620.9 kg of yield per kg of N applied for maize, climbing beans, and potato, respectively. The K-AE varied similarly, with yield increase ranging from 7.4 to 61.4 kg, from − 11.9 to 21.4 kg, and from − 118.6 to 751.3 kg of yield per kg of K applied. Moreover, P-AE happened to have the most varying index (from 24.5 to 396.1 kg, from − 87.5 to 250.0 kg, and from − 736.1 to 1597.2 kg kg⁻¹ of P applied).

Our findings indicated that the lowest AE was observed in the case of the AGM application alone. The highest N and K AEs were achieved using MF for the three crops, although K was not added in the applied MF recommendation for maize. The N-AE index of the AGM + MF was higher by 27and 66% as compared with inorganic fertilizer for maize and climbing beans, respectively. The average of the index in potato was quasi the same for AGM and AGM + inorganic fertilizer. The K-AE index for AGM + MF was higher by 42 and 68% for maize and potato, respectively, as compared with AGM. Regarding potato, the index average for AGM was 10% higher than the AGM + inorganic fertilizer combination.

On the other hand, P application resulted in lower AE when treatment consisted of a combination of AGM and MF. The P-AE was higher by 51, 6, and 63% higher for AGM, with respect to AGM + inorganic fertilizer, for maize, climbing beans, and potato, respectively.

Financial profitability analysis of Alnus acuminata green manure with or without mineral fertilizers under various soil fertility classes

Financial profitability concerned the evaluation of the cost of inputs involved in crop production and outputs (crop yields) in high and medium soil fertility classes.

The analysis of the financial performance of the three fertilizer options (AGM, inorganic fertilizer, and AGM + inorganic fertilizer) per hectare revealed that potato had the highest net return income (RWF − 1,941,500 to 7,455,850), whereas climbing beans yielded the lowest net return (RWF − 268,750 to 593,050). For the three crops, sole application of AGM consistently resulted in the highest VCR and marginal return, while the highest net benefit was achieved with the combination of MF and AGM in particular during the best season (season A for maize and season B for potato and beans) as shown in Table 3.

Table 3 Net returns, value cost ratio, and marginal returns for climbing beans, maize, and potato in Kadahenda, Nyabihu District, Rwanda

Discussion

Convergence between farmers’ criteria and soil analysis for soil fertility description

The results identified medium and high soil fertility classes that helped discriminate the effect of AGM biomass with or without inorganic fertilizer on crop productivity. These findings are supported by research conducted in the high-altitude region of Gishwati (an area with similar characteristics as those of our experiment site in Kadahenda) as reported by Mukashema (2007) who emphasized that soils are dominantly favorable for crop production. Therefore, no low soil fertility class was reported. As demonstrated in this study, the combination of farmers’ knowledge on soil fertility and scientific soil fertility classification provides useful information for understanding soil behavior and serves as guidance for appropriate management. Based on color and texture, farmers classified soils as having medium and high fertility, thereby generating more useful information, although a slight difference from the standard soil fertility classification was observed in this study (Table 1). The MLFI provided useful metrics to measure the changes in soil fertility, while SFI used by Mukashema (2007) were not as detailed as the farmers’ classification.

The results of this study on the usefulness of local indicators (farmer-defined) on soil quality supported the research conducted by Rushemuka et al. (2014), Kuria et al. (2019), and Abera et al. (2020). Soil color, depth, consistency, and texture are the recurrent assessment criteria in farmers’ classification. The combination of both mineral and organic input sources was adopted to abide by farmer setting requirement in a bid to optimize production as well as organic application. Such a combination has potential to amend soil fertility and its capacity to retain nutrients–amidst losses and leaching occurrences–as well as increase water storage potential.

Crop responses (beans, maize, and potato) to A. acuminata green manure

Our findings reveal that the effective combination of AGM + inorganic fertilizer significantly increased the yield of maize dry grain, climbing beans, and fresh tubers of potato in the Kadahenda study area. Chivenge et al. (2011) also reported that the meta-analysis of a combination of organic input with inorganic fertilizer significantly improved crop yields in several parts of SSA than did sole application of MF.

Overall, crop response to fertilizer application was comparable in soils with medium and high soil fertility levels in various seasons. These findings seem to contradict the interpretation of crop response as established by Vanlauwe et al. (2011), whereby authors categorized two types of soils based on their responsiveness to fertilizer application: (i) responsive soils that exhibit improved crop productivity upon receiving additional nutrients and (ii) non-responsive soils in which crop productivity is minimally affected or does not respond to fertilizer (poorly responsive soils) due to other limiting factors besides the nutrients contained in the fertilizer. In the context of the current study, both soils with medium and high fertility levels responded in similar proportions, as highlighted by Fig. 2, for maize, beans, and potato. For all crops, yield was comparable across soil fertility levels, indicating that both types of soils did not differ significantly in terms of crop productivity. Although soils were categorized into two different groups by farmers, their response to fertilizers was similar. Farmer categorization was only limited to soil color and texture. Also, the effect of other factors, such as climate (prevailing weather conditions), could have influenced the soil response to fertilizers.

Fig. 2

Comparison of model-predicted means per season and soil fertility classes for yield of climbing beans, maize, and potato in Kadahenda, Nyabihu District, Rwanda, for treatments of sole MF, combination of AGM + MF, sole AGM, and control (C). Means with the same letter up the bars are not significantly different. SED is for the interaction season*soil fertility class*treatment. ha-1 should be ha⁻¹

Categorizing soil fertility into medium and high soil fertility classes revealed differences in crop response to different inputs applied with or without GM.

Soil fertility classes influenced the response to treatments for beans (interaction of treatment and soil fertility classes with p < 0.05), but this was less clear for maize and potato. Beans have been reported to be sensitive to soil fertility status (Chekanai et al. 2018). Across the two soil fertility classes, AGM + MF was the best option with significant interaction of seasons. Interaction of season and treatment was observed for the three crops, confirming that location and rainfall pattern influence the response of fertilizer (Ichami et al. 2019). In the most favorable seasons (A for maize and B for beans and potato), AGM + MF was still the best treatment in nutrient AE as it highlighted the magnitude of response across soil fertility levels.

A. Acuminata green manure influencing nutrient agronomic efficiency (nitrogen, phosphorus, and potassium)

The soil category was higher (p < 0.05) for each fertilizer option (treatment) considered separately only for beans in the 2015 season B. Going beyond the means, exploring the boundary lines of AE along the individual soil fertility nutrient value in soils, an agreement was observed between farmers’ overall characterization and Vanlauwe et al.’s (2011) raised trend. N-AE for maize is declining, as suggested by the scholar. Nevertheless, except for K-AE (for maize) and P-AE (for beans), other trends are either ascending or having two trends (an ascending followed by a descending phase). The soils in Kadahenda–classified into medium to high fertility classes–did not show any non-responsive behavior to the applied nutrients (Vanlauwe et al. 2011).

The differential trends of N-AE, K-AE and P-AE for maize, beans, and potato may suggest that the threshold of fertile soils proposed by Vanlauwe et al. (2011) does not coincide for the three elements and vary for each element by crop, as asserted by research on corroborating a recommendation for N, P, and K for potato by Westermann (2005). The findings also suggest that the fertility categories as defined by farmers in Kadahenda can be relevant for a large area (probably larger than the site) where low soil fertility is involved in fine-tuning fertilizer recommendations.

The combination of AGM and inorganic fertilizer for potato showed a different trend of nutrient AE for N and K. The AE of these two elements increased with soil fertility for the application of 300 kg NPK (17:6:14.2), with boundary lines N-AE of potato yield increasing from 34.5 to more than 620 kg kg⁻¹ of N applied (in soils with 0.19–0.36% total N). It stemmed from the fact that its K-AE increased potato yield from 59.04 to over 751.31 kg per kilogram of K applied (with 0.05 to 0.6 ppm of P in soils) when inorganic fertilizer was applied alone (Fig. 3). This may be due to the low application rate of nutrients recommended in Rwanda as compared with nutrient uptake. In other areas, the N recommended rates are 235 and 336 kg ha⁻¹ in USA and Canada, respectively Westermann (2005).

Fig. 3

Agronomic efficiency of nitrogen (N-AE), potassium (K-AE), and phosphorus (AE-P) for individual farm (different replicates and different seasons) of maize, climbing beans, and potato in Kadahenda, Nyabihu District, along fertility gradient. AGM A. acuminata green manure, MF mineral fertilizer, AGM + MF mineral fertilizer + A. acuminata green manure, b. line = boundary line. MF (red triangle) is not presented for K-AE for maize because there is no K applied in the mineral fertilizer treatment

The P-AE for the three crops tested (maize, climbing beans, and potato) was higher in the MF application due to the easy nutrient release of P in the DAP and NPK fertilizers as compared with Alnus GM, which was poor in P. However, a reduction was observed in soil fertility gradient with MF application. Westermann (2005) indicated that P uptake of potato was 31 kg ha⁻¹, while Šrek et al. (2010) suggested 63 kg of P ha⁻¹ as the optimum fertilizer dose. For potato, the P-AE trending along the fertility gradient was different compared with N and K. This may be due to the lower P requirement for potato as reported by several authors. Currently, 50 kg of N, P, and K are recommended countrywide. Our findings suggest the need to rethink those rates and the use of NPK (17:17:17).

N-AE, considered the most environmentally sensitive nutrient AE (Basosi et al. 2014), was highest for sole MF treatment for the three crops. Nevertheless, reduction in N-AE as induced by AGM could be disputed because trees on-farm are generally considered to be a major environmental rehabilitation measure (Nyaga et al. 2015) and that biomass will supply nutrients to the soil and water regardless of whether trees were planted or not. The falling leaves of A. acuminata could supply up to 50 kg of N ha⁻¹ (Ndoli et al. 2017). The lowest AE was observed in the case of GM application, which was in line with results ofprevious studies on organic fertilizer. A. acuminata leaves are rich in nutrients but their nutrient concentration varies according to location (Alvarado-Hernández et al. 2022).

Non-responsive soils (due to low soil fertility) were not observed while the decline of the boundary line was observed for all elements and all nutrient AE indices (N-AE, P-AE, and K-AE), except for potato (Vanlauwe et al. 2011). This also indicates that soils in Kadahenda are dominated by medium to high fertility types.

Financial profitability of Alnus acuminata versus mineral fertilizer depends on crop type

The financial analysis indicated that, for the three crops, sole application of AGM consistently resulted in the highest VCR and marginal return, while the highest net benefit was achieved with the combination of MF and AGM during the best season (season A for maize and season B for potato and beans). However, assuming that VCR above 4 is the threshold that defines attractive technologies (Morris et al. 1999), only fertilizer options tested for potato may be self-propagating.

This study found that the best financial option, considering only the net benefit, is the combination of inorganic fertilizer and AGM, particularly during the favorable season for maize (season A) and potato and beans (season B). The sole application of AGM is preferred in less favorable seasons (season B for maize and season A for potato and beans) as it is supported by the highest VCR and marginal return. Farmers have been reported using topdressing application of urea on potato. This is not without risks as marketable tuber yield may decline more quickly (100–150 kg N ha⁻¹) than tuber yield (200–250 kg N ha⁻¹) (Zebarth et al. 2012). This work proposes the combination of inorganic fertilizer and AGM as VCR was above the threshold of 4. Hence, all fertilizer options for potato crop and AGM for maize and climbing beans are profitable and can be promoted as attractive technologies (Morris et al. 1999).

Several authors stress that the financial valuation of biomass-based on labor time⁻¹ may be misleading because the action of fetching biomass is tedious (Ngetich et al. 2012). Considering and evaluating all the steps show that collecting biomass is not particularly difficult or costly compared to using other inputs such as seeds, fertilizer, or labor for tilling, weeding, spraying pesticides, and harvesting. It is the least expensive. However, for a farmer who does not have A. acuminata trees on the farm, his participation in the on-farm trials was constrained by the non-availability of enough biomass. In Kadahenda, A. acuminata met the condition necessary to be a desirable source of GM set by Ngetich et al. (2012). In addition, Ngetich et al. (2012) stated that the use of GM should be linked to a tree value chain with other benefits apart from the biomass applied for the technology to be successfully adapted to the needs of the farmers.

Conclusion

This study contributes to closing the gaps in evidence of the benefits of using A. acuminata as GM, particularly for highland soils where the tree is abundant. We concluded that combining MF and GM was the best option among the tested treatments. We demonstrated that P-AE is superior for sole AGM, followed by the combination of GM and inorganic fertilizer application in the P-constrained soils of Kadahenda for the three crops. Agronomic use efficiency of N and K for AGM treatments was lower than that for MF. We also conclude that the higher mineral application rates of the two elements are worth exploring, given the trend of the agronomic use efficiency of N and K for potato. Although GM is affordable, crop prices can be a disincentive. The net benefit of the combination of AGM and inorganic fertilizer was the most attractive for farmers, while the MR and VCR indicated that, only for potato, the adoption of AGM and AGM + inorganic fertilizer can be easily done.

In the medium soil fertility category, AGM + inorganic fertilizer should be encouraged as the best option for increasing yields of maize and beans at the rate used in this study. Long-term analysis of continuous AGM application could be envisaged as adoption of AGM increases at scale. The A. acuminata biomass is freely available in large parts of the eastern African highlands. This highlights the fact that strategies are in place to optimize the opportunities offered by trees on or near the farm.