11-Year Rainfall and Maize Yield Variation (2003-2013) in Four Northern Districts of Malawi

Rainfall in the maize cropping season (Oct-Apr) in the four northern districts of Malawi was examined in terms of seasonal fluctuation and spatial distribution, and data spanning 11 years were analyzed. Rainfall fluctuations in the 11-year period differed considerably among the four districts and the Extension Planning Areas (EPAs) showed high coefficients of variance (CVs) (16.9-93.7). The equation with the three-month rainfall (October, February, and April), i.e., Maize yield (kg/ha) in SH = 2.29 + 0.0042 × Oct rainfall – 0.0009 × Feb rainfall + 0.00045 × Apr rainfall (r = 0.41), better explained maize yield in the 2013/14 season than the equation with total rainfall in the cropping season. Rainfall accounted for more than 41% of the total variation in maize yields of smallholder farmers (SHs). Rainfall in April was the most critical factor influencing maize and other crop yields. After the Farm Input Subsidy Programme (FISP) was implemented in 2005/06, maize yield became more dependent on rainfall. CV was higher in maize than in groundnut and sweet potato, indicating that maize is susceptible to rainfall fluctuations, and groundnut and sweet potato should be incorporated in farming as a countermeasure against unpredictable rainfall.


Introduction
Increasing inorganic fertilizer prices, decreasing farm size, and declining soil fertility are the major constraints of current agricultural practice in Malawi.Malawi has been experiencing high seasonal variability in maize yield in the last few decades.A 43% national food deficit was recorded in 2005 and a 53% surplus, in 2007 (Denning et al., 2009).A major factor affecting maize yield is chemical fertilizer application.To this end, the government of Malawi has been offering chemical fertilizer subsidies to farmers since 2006.For one bag of fertilizer (50 kg) sold at the market price of around 15,000 MKT (60 US dollars as of Oct 2014), farmers could purchase it at 500 MKT (= 1 US dollar).
Maize is the most important staple food crop in Malawi (JAICAF, 2008) and in Africa (Byerlee & Heisey, 1996).Almost all maize crops are grown without irrigation during the single rainy season that starts in October and ends in April; thus, the crops are subject to rainfall variability that can be particularly damaging when short dry spells occur during the critical flowering and early grain filling stages (Famine Early Warning Systems Network, 2007).Several studies have indicated that the availability of adequate rainfall is by far the greatest limiting factor in maize production in sub-Saharan Africa (CIMMYT, 1988;Diallo et al., 1989;Ammani et al., 2012).As drought during the flowering and grain filling period may lead to 40-90% crop loss (Grant et al., 1989;NeSmith & Ritchie, 1992;Menkir & Akintunde, 2001), total crop failure due to drought is experienced once every ten years in semi-arid sub-Saharan Africa (Ngigi, 2003), and 80% of the cultivated maize crops have reduced yield due to drought stress (Bolonos & Edmeades, 1993).
Global warming is expected to intensify drought problem in Africa (Edmeades, 2008) and rising temperatures will be associated with greater rainfall variability and increase the frequency of severe weather events such as droughts and floods (Boko et al., 2007).Rainfed agriculture accounts for more than 95% of farmed land in sub-Saharan Africa, meaning that the effects of drought are likely to be more damaging than in developed nations where irrigation is more prevalent (Rockstrom & Falkenmarkac, 2000).In order to attain sustainable food security under uncertain climatic conditions, it is necessary to examine how much rainfall contributes to maize yield at the regional level.This will lead to better farm management in order to sustain or improve maize yield.
In this study, we examined rainfall patterns in the northern region of Malawi during the last 11 years from 2003 through 2013 and determined rainfall influence on maize yields of smallholder farmers (SHs).We also clarified which months of the cropping season (October to April) are important for enhancing the yields of different maize varieties.Furthermore, other common crops, such as cassava, sweet potato, finger millet, and groundnuts, were examined in terms of adaptability to rainfall fluctuations.

Description of Study Sites
The study sites were located in the four northern districts of Malawi, namely, Mzimba North (N), Mzimba South (S), Nkhatabay, and Rumphi (Figure 1).Those districts were composed of Extension Planning Areas (EPAs).The EPAs had two lower sublevels, Section and Block.The number of EPAs was 7 in Mzimba N, 12 in Mzimba S, 8 in Nkhatabay, and 6 in Rumphi as of 2013.
The mean annual rainfall over a 22-year period between 1989 and 2011 was 1,129 mm in Mzimba N, 702 mm in Mzimba S, 612 mm in Rumphi, and 1,610 mm in Nkhatabay.Rainfall patterns of the four districts were almost identical (Figure 2): high rainfall from November to April and low rainfall from May to October.Nkhatabay had very high rainfall in March and April relative to the other districts.

Information Collection and Statistical Analysis
Rainfall and maize yield data of the northern districts for the period between 2003 and 2013 were obtained from the Planning Department of Mzuzu Agricultural Development Division (ADD).Regarding rainfall data, descriptive statistics, including the mean, maximum, and minimum values and the coefficient of variance (CV), were calculated.The Planning Department collects maize and other crop yield data every year according to the methodology devised by the Ministry of Agriculture and Food Security (2008).25% of the Blocks were randomly selected from each Section.In each selected Block, all households were listed and 20% of them, namely, more than 15 households, were identified.The overall sampling number represented approximately 5% of all agricultural households.
Although hybrid and composite maize crops are characterized by their remarkable seed yields, seed prices are high and fertilizers are required.Therefore, they are planted only when the government introduces a support scheme.The unit yields of the hybrids are higher than those of traditional varieties, but so is the cost incurred.For this reason, a production area would register a temporary increase when chemical fertilizer and improved seeds are provided free of charge.Generally, farmers in remote areas grow traditional varieties (local), whereas farmers living near the capital cultivate hybrid and/or composite varieties.
Statistical analysis was conducted using the software JMP 8.0.2 version for Windows (SAS Inc., 2009).Besides general statistical analysis, multiple regression analysis was carried out for rainfall and maize yield to understand which month(s) affected maize yield in a particular year.Note. 1) Th

Season
The effect (Table 5).Nevertheless, November, January, February and April were likely to be more critical than the other months.
Hybrid and composite were more strongly influenced by rainfall than the local variety.Multiple regression analysis showed that R 2 was lower for the local variety than for hybrid and composite (Table 6).Maize yields of all types were correlated with rainfall in April and the regression coefficient was the highest in hybrid.February was another month affecting maize yield in composite.The negative regression coefficients for December rainfall in hybrid and composite and January rainfall in local variety indicated that rainfall in those months adversely affected maize yield.Note. 1) This is partial regression coefficient in multiple regression analysis; 2) Statistics degree of freedom adjusted coefficient of determination.Note. 1) This is partial regression coefficient in multiple regression analysis; 2) Statistics degree of freedom adjusted coefficient of determination.
The coefficient of determination (R 2 ) was measured for the other crops (Table 8).All crops except sweet potato showed low coefficients of determination.Sweet potato requires high rainfall as shown by its high regression coefficient (14.6).Finger millet, cassava, tobacco also require relatively high rainfall in April.

Table 1 .
Statistical values for rainfall during the period between 2003 and 2013 in the four districts and their EPAs Note.EPA: Extension Planning Area; N: Number of data; Min: Minimum; Max: Maximum; Stdv: Standard deviation; CV: Coefficient of Variance.

Table 2 .
Means and coefficients of variance of maize yields (kg/ha) of SHs for the period between 2003 and 2013 Note.N: Number of data; Stdev: Standard deviation; CV: Coefficient of Variance.Table 4. Changes in maize yield (kg/ha) for the 10-year period between 2004/2005 and 2013/2014 www.ccsen

Table 7 .
Coefficients of variance (CVs) of maize and other crops produced during the period between 2003 and 2013 in the four districts would be promising as a countermeasure against drought.CV of soybean in Nkhatabay was 65.1, and such a high CV could be due to fungal infection as a result of high rainfall. thus