Abstract:
East African highland bananas (Musa spp., AAA-EA genome group) are a major staple and
income-generating fruit crop in the highlands of eastern and central Africa, grown across the
countries of the Great Lakes region (i.e. Rwanda, Burundi, Uganda, eastern Democratic
Republic of Congo and North-West Tanzania). Despite its importance, farmers and
researchers are reporting that yields are declining, most notably in areas with low soil
fertility. Although numerous studies have been conducted on yield constraints of bananas in
the East African highland region, there is virtually no understanding of the impact of plant
density management on the yields of these low-input banana systems. The productivity and
profitability of various plant densities was studied in contrasting agro-ecological sites of
Rwanda (Ruhengeri, Rusizi, Karongi, Butare, Ruhango, Kibungo and Bugesera), that differed
distinctly in terms of altitude (1400-1960 m a.s.l), temperature (17-20°C), annual rainfall
(950-1400 mm yr-1) and soil types (Nitisols, Ferralsols, Acrisols and Andosols). Under those
cropping systems, the plant density is one management factor that resource poor farmers have
some control over.
An on-farm survey was conducted in all sites to determine the influence of climatic and
edaphic factors on variations in on-farm plant density practices and bunch mass. In addition,
three researcher-managed banana density experiments were conducted in contrasting agroecological
sites (Kibungo low rainfall with medium soil fertility, Rubona high rainfall with
low soil fertility and Ruhengeri high rainfall with high soil fertility) to (i) investigate the
influence of plant density on the vegetative growth and yield parameters of AAA-EA bananas
for typical highland agro-ecological zones, (ii) to assess the effect of plant density on nutrient
deficiencies and imbalances, and (iii) to assess the magnitude and variability of nutrient
depletion in the smallholder banana systems that are characterized by low external input use.
Three different local EA highland banana varieties (i.e. “Ingaju”-cooking type, “Injagi”-
cooking type, “Intuntu”-beer type) were each planted at five different plant densities (plants
ha-1) of 1428, 2500, 3333, 4444 and 5000. Agronomic data (growth and yield traits) were
collected over two cropping cycles (plant and ratoon crops). Soil, plant and climate data were
also collected. Approaches such as compositional nutrient diagnosis (CND), boundary line
functions and yield gap analysis were used to quantify the contribution of each identified
yield limiting factor to yield gap. A first order magnitude of nutrient depletion was
determined using partial nutrient balance calculations.
Plant density positively correlated with water supply (i.e. difference between rainfall and
evapotranspirative demand of bananas), with highest plant densities (>1500 mats ha-1) found
in high rainfall areas (>1200 mm yr-1) with water surplus (218-508 mm yr-1) and lowest plant
densities (1000-1400 mats ha-1) found in lower rainfall areas (1000-1200 mm yr-1) with water
deficit (from -223 to -119 mm yr-1). Bunch masses were significantly higher at the lowest
plant densities (18.1-20.8 kg fresh mass plant-1) when compared to the highest plant densities
(14.7-15.5 kg). Lower soil and banana leaf nutrient contents were observed on weathered
soils (Acrisols) and were associated with smaller bunch mass in comparison to fertile soils
(Andosols, Nitisols). Farmers tended to reduce mat densities (i) if they wanted to intercrop,
and (ii) to increase bunch mass to adapt to market preferences for large bunches. The plant
densities generally recommended by extension bodies (3 × 3 or 2 × 3 m; i.e. 1111 and 1666
mats ha-1, respectively) are seldom practiced by farmers, nor do they seem to be very
appropriate from an agronomic or economic perspective.
Per hectare bunch and above ground biomass yields increased with increasing plant density,
but maximum yield strongly depended on agro-ecological site. Bunch yields of beer bananas continued to increase with density, but maximum yields for the cooking cultivars were
observed at 4444 plants ha-1 at Kibungo and Rubona, whereas yields continued to increase
linearly beyond this level at Ruhengeri. Relationships between bunch yield, the total above
ground dry matter yields and soil chemical properties suggest that nutrient deficiencies were
larger at Kibungo (i.e. notably K) and Rubona (i.e. K, P, Ca, Mg) when compared with
Ruhengeri. With increasing densities, leaf area index (LAI) continues to increase up to a
value of 4 with 95% of photosynthetic active radiation (PAR) intercepted by the crop canopy.
This suggests that further density and LAI increases would probably have little additional
positive effect on total per hectare production.
Compositional nutrient diagnosis (CND) indices showed that K, Mg and P were the most
deficient elements in areas with low inherent soil fertility (Kibungo and Rubona) compared
with relatively fertile areas (Ruhengeri). The boundary line functions and yield gap analysis
also confirmed that K was the most limiting factor, contributing to an expected yield gap of
55.3% at Kibungo, while P and Mg contributed to a 35% yield gap at Rubona. An increase in plant density resulted in an increase in average yield gap from 45.6 % to 70.2% at Kibungo,
whilst average yield gap decreased significantly from 47.5% to 30.2% at Rubona, and 76.6 to
53.7% at Ruhengeri. Nutrient uptake increased with plant density. Partial N and K balances
(kg ha-1 yr-1) were estimated to be strongly negative at Rubona and Ruhengeri, while Ca and
Mg were positive at Kibungo and Ruhengeri, but negative at Rubona. The results of this
study indicate that, generally, soil fertility is a more limiting factor than water, but both CND
norms and boundary line analysis showed that expected yield gaps seem to be high for plant
density due to low inherent soil fertility. Partial nutrient balances provide a first order
magnitude of nutrient depletion. Nutrient mining is significant, particularly for K. The current
extraction rates will not allow farmers to sustain their yields, and options should be
developed to improve the productivity of EAH banana cropping systems. The limited
availability of manure and inorganic fertilizers is a real threat to the food and income security role that banana production plays in smallholder systems.
In summary, the results from this study suggest the optimal density for bananas depends on
water availability, soil fertility and cultivar. The agronomic optimal plant density is lower (<
4444 plants ha-1) in low rainfall (< 1000 mm yr-1) and less fertile areas, but seems to be
higher (> 5000 plants ha-1) in areas with high fertility, which receive high rainfall (> 1300
mm yr-1). Improved plant density management can serve as an important entry point for
resource poor farmers to maximize yield potential of EAH bananas in the various production
zones. Blanket density recommendations do not make sense. While farmers can significantly
improve their banana production, increased densities will put significant additional stress on
limited nutrient resources, and region-specific integrated soil fertility recommendations
should be developed and adopted to ensure sustained improvements of banana production and
smallholder livelihoods.