«JM Thamaga-Chitja March 2008 Submitted in fulfilment of the requirements for degree: Doctor of Philosophy (Food Security), African Centre for Food ...»
The choice of moisture as an environmental risk factor was based on the premise that the presence or lack of moisture at a satisfactory level is a key requirement for diseases to initiate (Agrios, 2004). Rainfall was used because it is a correlate of moisture.
In order to determine the risk of disease occurrence, monthly rainfall levels (source of moisture) were modelled over twelve months using equation 7. A reference database of diseases that could affect each of the listed crops and their corresponding predisposing climatic conditions was created. The disease database consisted of three most important diseases associated with moisture and three others associated with lack thereof. The importance of the disease was based on economic importance and extent of devastation. The diseases were separated into two categories: those that set due to moisture presence and abundance (Appendix D, high moisture diseases) and those that set due to low moisture or lack thereof (Appendix D, low moisture diseases).
It was recognised that the mean annual rainfall reported in the Bioresource Database of the KwaZulu-Natal Department of Agriculture and Environmental Affairs (Camp,
1999) includes rainfall distribution. However, it is an over-expectation of any one without records to know the monthly rainfall and, even when these are available, provision of input area in the user-interface will render the interface very
In order to determine parameters in this function (mean annual rainfall) the mean annual rainfall values from five random locations of varying agro-ecologies in KwaZulu-Natal were used to develop a deterministic pattern based on the monthly rainfall and its variation across locations. The monthly rainfall in each of the five locations was expressed as a proportion of the annual rainfall. The mean proportion (ux) and its standard deviations (σx) were calculated for each month. Random values (120) ranging from 0 to 1 were generated using a random Excel function.
The estimated random proportion was multiplied by the mean annual rainfall to derive a rainfall estimate for each month. The monthly risk of disease onset risk was based on how much rainfall was predicted per month. The following rainfall ranges were used to define the degree of diseases’ risk in Table 5.2.
Each month was assigned a disease risk profile based on the range of rainfall (Table 5.2). This information could function as an early warning system to determine planting periods for various crops by checking which prevailing moisture level (low,
Figure 5.2: The process followed in determining the risk of disease onset based on crop choice The decision support process was then employed to assign an appropriate disease risk profile (L, M or H) to each month depending on whether the predicted rainfall values were less than 50ml, between 50–100ml and more than 100ml.
The last of the assumptions related to production made in this study related to the fact that rainfall can be predicted and that the rainfall based moisture can determine the onset of disease. However, it is accepted that rainfall is not the only contributor of moisture where diseases are concerned. Humidity and condensation, among other factors, are important determinants of moisture for different environments. However, rainfall data was easier to access and apply in the development of the decision support tool.
5.4 Validation of model inputs Due to the study’s multidisciplinary nature, which involved the use of data from disciplines such as Horticultural Science; Agronomy; Soil Science; Plant Pathology;
Simulation Modelling; Sustainable Livelihoods; Extension; Rural Development and
Three seminars were conducted at the proposal, model input development stage and output stage. The objective of the first seminar was to receive critical analysis and input on the proposed design, methodologies and sources of information for the model. Once the proposal for the study was developed, the expert panel was invited to participate in the consultative process of this study, interrogate the proposal and to make an input. They provided inputs on the type of model proposed and relevant outputs, and verified relevant science included in the model. The second seminar’s objective was to critically review the identified inputs for the model. Once the decision support tool development took shape, the experts gave their input on the model development approach chosen and verified statements, assumptions and explanations provided through the overall approach. The last seminar’s objective was to discuss the output and receive critical review of the tool. The experts also guided the researcher by pointing out areas of potential concern, such as ensuring that the decision support tool can be applied to any location.
This innovative approach to trans-disciplinary research ensured the experts were able to verify their inputs in the presence of specialists from different disciplines. This reduced gaps in knowledge and interpretation, and cut down on inaccuracies. Their input contributed to an integrated design and holistic approach to the study. The researcher also consulted experts individually during the course of the study when necessary. Farmers also had an important role in the validation of desired outputs and the developed tool. Their experiences and impressions of the tool are separately reported in Chapter 7.
5.5 Results and discussion In stage one, once the list of crops was finalised twenty crops, growth conditions for each crop on the list was established. A decision was made that single values instead of ranges would be used when capturing plant growth data as it was easier to work 66 with a single entry during data-capturing. In the case of rainfall, absolute values relating to rainfall were used because many smallholder farmers practice rain-fed agriculture and are found in low rainfall areas. Aliber et al (2006) explains that many smallholder farmers in South Africa are located in poor parts of the country (former homelands), which are also less favourable agricultural areas. On the other hand, using optimal ranges is supportive of obtaining better yields. Nevertheless, currently smallholder farmers is South Africa do not experience optimal conditions which is likely to result in difficulties in organic production. In the case of temperature, mean values for temperature were used due to the variation related to the nature of temperature.
In the first sub-problem absolute values relating to rainfall were used because many smallholder farmers practice rain-fed agriculture and are found in low rainfall areas.
Mean values for temperature were used due to the variation in the nature of temperature. On the other hand, it is accepted that the use of mean means that certain crop growth will occur for values that fall below the mean.
In stage two, several decisions on crop nutrition requirements and soil analysis were applicable. One of the most important factors in organic farming is soil fertility. In organic farming, the practices for improving soil fertility must be aligned with approved organic standards (BDOCA, 2006). The use of commercial inorganic fertiliser to provide crop nutrients is prohibited in certified organic farming.
Crop nutrient requirements and soil analyses were based only on three key nutrients, namely, nitrogen (N), phosphorous (P) and potassium (K), as these are the basic nutrients required by all plants in high quantities for good growth (van Averbeke & Yoganath, 2003). It was assumed that all other nutrients were in adequate supply.
Adequate water and good maintenance of soil health is also important for good organic production (OFRF, 2001).
A ‘one-size-fits-all’ soil nutrition programme would be misleading due to variation in soils and local climates. Soil nutrition improvement recommendations would need to be farm-specific. Due to the apparent lack of uniformity and the generalised nature of the soil fertility data, a decision was taken to use nutrient removal norms to indicate
As some nutrients are required in higher quantities than others, it is quite possible that a particular nutrient could accumulate and perhaps reach an undesirable concentration in the soils, causing an imbalance and affecting the availability of other nutrients.
Raw manure use is frequently associated with imbalances in soil fertility because manure is often rich in specific nutrients such as phosphate (Kuepper, 2003).
Continued applications of manure may lead to a detrimental nutrient build-up.
Excessive nutrient levels affect the uptake of other minerals in the soil. This may be avoided by conducting continuous soil analyses, crop rotation, cover-cropping and addition of other natural fertilisers (Kuepper, 2003).
Manure produced by one beast and one sheep/goat can total 8.85 wheelbarrows per annum. The calculation below demonstrates how the amount of manure was calculated by the model. Depending on the crop, this can fertilise varying parcels of land ranging from 0.01 ha to 0.3 ha resulting in yields ranging from of 0.6 t/ha (mint, basil and coriander) to 6.4 t/ha (peach) respectively as illustrated in figure 5.3. The method of storage and application of the indicated manure is important in determining the quality and level of nutrients available.
The study recognised that although manure is the common organic nutrition source, other sources such as compost are relevant. Therefore, total nutrition from available organic sources was calculated by adding animal manure and compost.
Large amounts (7–10 ton/ha) of manure would be required to obtain near-maximum yields. This may pose a real challenge for farmers who do not have livestock, as is the case with many smallholder farmers. Even those with livestock will require unrealistically large amounts of manure to meet yield demands. This may not be sustainable if livestock numbers drop from current tables (Table 3.4) the few there currently are. Farina (2005) argues that it is barely possible for farmers to make up their nitrogen inputs using only organically-acceptable manures or compost.
Many studies have been undertaken to validate the potential benefits of manure application as a means of sustaining soil fertility and have shown improvement in soil structure and water retention in the smallholder farming environment (Mkhabela, 2006). It is accepted that cattle and chicken manure cannot be used as a substitute for inorganic N fertilisers but these manures can be helpful in augmenting nitrogen supply to crop production and thereby reduce the cost of purchasing inorganic fertilisers (Mkhabela, 2006).
It is a known fact that most smallholder farmers in South Africa do not use large amounts of commercial fertiliser due to the cost (Mkhabela, 2003). The economics of manure usage versus no usage of manure among smallholder in KwaZulu-Natal was studied by Mkhabela (2003) and revealed that for smallholder farmers, there was improved profitability in using manure compared to no manure usage. The study further indicated that although manure usage was beneficial to smallholder farmers, greater benefit was derived when manure was supplemented with inorganic fertiliser.
This finding is supported by Farina (2005) who proposes that organic inputs alone may not meet crop nitrogen needs.
Stage three was concerned with ascertaining if the climatic conditions of the study areas are conducive or detrimental to organic farming. It was important to check how close the observed and the predicted rainfall moisture values were as a way of validation of the rainfall distribution function. Computations of the rainfall distribution for the three study areas are illustrated in Figure 5.4 while rainfall patterns are given in figure 5.5.
The observed rainfall values and the predicted rainfall values are strongly correlated.
With the exception of Muden, the correlation is strong throughout most of the year.
This suggests that given the mean value, the rainfall distribution can be predicted with reasonable accuracy over 12 months.
Evidently, risk is low during low-rainfall months (winter). However, with the increase in rainfall (moisture), the risk of disease also increases. It is to be noted that rain is not the only contributor of moisture but that mist and humidity can also play a role in disease development. Nevertheless, data on mist and humidity levels of rural areas is hard to find. According to the geographic location of the study areas, Mbumbulu is closest to the sea and is located in a humid area compared to Muden and Centocow, which are drier, although Centocow’s higher rainfall than Muden and Centocow may lead to a higher presence of moisture.
72 160 140 120
40 20 0 12 3 4 5 67 8 9 10 11 12 120 100 80
40 20 0 160 140 120
40 20 0 12 3 4 5 6 7 8 9 10 11 12 Figure 5.5: Observed and predicted monthly rainfall at Mbumbulu, Muden and Centocow
The user interface as the interactive element of the model is the ‘visible’ element.
Other cells are locked to prevent interference with data. The model was developed with the aim of establishing a user interface that farmers and other users (extension staff) can use with ease. An example of a user interface is presented in Figure 5.6.
The output was demonstrated in figure 5.7 and 5.8 as illustrated earlier. This is the screen that allows the user to supply the inputs and obtain a printout. Other sheets responsible for the computation are protected to discourage inadvertent manipulation.
Figure 5.6: An example of the user interface.