SOIL QUALITY ASSESSEMENT OF SOUTH AFRICAN HOME GARDENS THE CASE OF THE VILLAGES MUTSHENZHENI, TSHIDZINI AND DZINDI Word count: 19.872 Elien Haverbeke Student number: 01405565 Promotors: Prof. dr. Geert Baert, Prof. dr. Wim Van Averbeke A dissertation submitted to Ghent University in partial fulfilment of the requirements for the degree of Master of Science in Bioscience Engineering Technology: Agriculture and Horticulture (Tropical Plant Production) Academic year: 2017-2018
80
Embed
SOIL QUALITY ASSESSEMENT OF SOUTH AFRICAN HOME … · 2018. 8. 29. · DSS Decision Support System EC Electrical Conductivity EC e ... WTD Weighted Additive Score Zn Zinc . 5 List
This document is posted to help you gain knowledge. Please leave a comment to let me know what you think about it! Share it to your friends and learn new things together.
Transcript
SOIL QUALITY ASSESSEMENT OF
SOUTH AFRICAN HOME GARDENS
THE CASE OF THE VILLAGES MUTSHENZHENI, TSHIDZINI AND
DZINDI
Word count: 19.872
Elien Haverbeke Student number: 01405565
Promotors: Prof. dr. Geert Baert, Prof. dr. Wim Van Averbeke A dissertation submitted to Ghent University in partial fulfilment of the requirements for the degree of Master
of Science in Bioscience Engineering Technology: Agriculture and Horticulture (Tropical Plant Production)
Academic year: 2017-2018
SOIL QUALITY ASSESSEMENT OF
SOUTH AFRICAN HOME GARDENS THE CASE OF THE VILLAGES MUTSHENZHENI, TSHIDZINI AND
DZINDI
Word count: 19.872
Elien Haverbeke Student number: 01405565
Promotors: Prof. dr. Geert Baert, Prof. dr. Wim Van Averbeke
A dissertation submitted to Ghent University in partial fulfilment of the requirements for the degree of
Master of Science in Bioscience Engineering Technology: Agriculture and Horticulture (Tropical Plant
Production)
Academic year: 2017-2018
“The author and the promoter give the permission to use this thesis for consultation and to
copy parts of it for personal use. Every other use is subject to the copyright laws, more
specifically the source must be extensively specified when using the results from this thesis”.
Promotor 1 Promotor 2 Author
Prof. Dr. G. Baert Prof. Dr. W. Van Averbeke Elien Haverbeke
Preface
I would like to express my gratitude to the following:
The VLIR-UOS travel grant, who supported me financially and gave me the opportunity. It was
an amazing experience to participate in the extension of a former project entitled Improving
home garden soil fertility management to enhance nutritional security among rural homesteads
in Vhembe (Limpopo, South Africa), that was done in collaboration between Ghent University
and the Crop Sciences Department of Tshwane University of Technology.
My promotors Prof. Dr. G. Baert and Prof. Dr. W. Van Averbeke, for guiding me patiently
through the process of writing a thesis and for providing me the required equipment to see this
project through.
More specific I want to thank Prof. W. Van Averbeke for guiding me through the villages of
Thulamela and for his help collecting the samples. I am deeply grateful for allowing me to stay
at his home as a guest when I arrived. I could not have accomplished this without his support.
Gumani, Patho and others who assisted me in the field. They made the fieldwork a lot easier
and helped me to understand the interviewed people. I would like to thank Gumani for teaching
me a lot about her culture and life. I am also grateful for the gardeners who let us take samples
and provided the needed information.
Lastly, I would like to thank my parents, family and friends, who supported my choices and
made me the person I am this day. They taught me how to be independent and gave me the
love and encouragement to accomplish this work.
Abstract
In the context of the growing concern about global food security and inequality in South Africa,
soil quality of South African home gardens in Limpopo province was evaluated. Animal manure
is not fully utilized by home gardeners while often financial resources aren’t available to use
chemical fertilizers. Different soil fertility management practices (application of chemical
fertilizer, animal manure, a combination of animal manure and chemical fertilizer and no
application of fertilizer or manure) were compared. 68 samples were taken from 45 gardens.
This means that in multiple gardens, 2 samples were taken to use for pairwise comparison.
Samples were analyzed on: pH(H2O), EC(1:5), ECe, NO3-N, NH4-N, total mineral N, Olsen-P,
exchangeable K and Ca, total C, total N and C/N-ratio. A soil quality index was compiled to
construct an overall opinion about the soil quality but did not give an adequate insight. This
was due to high variability in the data as a result of differences in soil types. Exchangeable K,
exchangeable Ca, EC and ECe were better when animal manure was applied. Mineral N and
P levels in soils that received animal manure were comparable with soils that received a
combination of chemical fertilizer and animal manure or chemical fertilizer alone. Total N and
C levels were higher in soils that received animal manure than in soils that received chemical
fertilizer. Animal manure is an important source of organic matter and is a better soil fertility
management practice than chemical fertilizer for home gardens in South Africa.
Key words: South Africa, animal manure, chemical fertilizer, chemical soil properties, home
gardens
Samenvatting
In het kader van de groeiende bezorgdheid rond globale voedselzekerheid en ongelijkheid in
Zuid-Afrika, werd het thema rond bodemkwaliteit in Zuid-Afrikaanse moestuinen in de Limpopo
provincie behandeld. Door gebrek aan financiële middelen voor chemische meststoffen bij
particulieren, is dierlijke mest vaak het enige alternatief. Dierlijke mest wordt echter niet ten
volle benut. Verschillende toepassingen ter verbetering van de bodemvruchtbaarheid werden
vergeleken. Deze toepassingen waren toediening van chemische mest, dierlijke mest, geen
mest of een combinatie van dierlijke en chemische mest. Er werden 68 stalen genomen uit 45
verschillende tuinen. In meerder tuinen werden 2 stalen genomen die paarsgewijs vergeleken
werden. Alle stalen werden geanalyseerd op: pH(H2O), EC(1:5), ECe, NO3-N, NH4-N, totale
minerale N, Olsen-P, uitwisselbare K en Ca, totale C, totale N en C/N-ratio. Er werd een
bodemkwaliteitsindex samengesteld om de toepassingen te beoordelen. De
bodemkwaliteitsindex bleek echter geen nuttige manier om uitsluitend chemische parameters
te beoordelen. Hoge variabiliteit in de data, veroorzaakt door verschillen in de bodemtypes,
maakte deze resultaten twijfelachtig. De chemische parameters werden afzonderlijk bekeken.
Uitwisselbare K en Ca, EC(1:5) en ECe waren beter wanneer dierlijke mest werd toegediend.
Mineral N en P in bodems waar dierlijke mest werd gebruikt waren gelijkwaardig aan bodems
waar chemische mest of een combinatie van chemische en dierlijke mest werd toegediend.
Totale N en C waren hoger in bodems waar dierlijke mest werd toegediend dan in bodems met
chemische bemesting. Dierlijke mest is een belangrijke bron van organisch materiaal en is een
betere bemestingsvorm dan chemische mest voor moestuinen in Zuid-Afrika.
Application rates in Table 8 show at what level of specific animal manure type, vegetables
have a maximum biomass peak, except for application rates of goat and cattle manure for
pumpkin and nightshade which did not reach the maximum biomass with this rate (Okorogbona
and Adebisi, 2012).
2.5 Soil quality assessment
As a response to the increased global focus on sustainable land use, the concept of soil quality
became more important during the 1990s. The individual soil management goals were
combined in a framework with management goals to prevent erosion, reduce contamination of
soil and water… Education about- and the assessment of the soil quality where put in the
spotlight. Soil quality assessment focusses on inherent and dynamic soil properties and
processes. The inherent properties are mainly based on the whole soil profile (0 to ±2m) while
dynamic soil quality focusses on the surface layer (0 to 20 or 30 cm). The dynamic soil
properties are focused on more recent land use and management practices. Still, no
universally accepted method for soil quality assessment determination exists, but in general,
soil quality evaluation is based on mathematical equations which show the relation between a
soil function and soil quality indicator. This is also known as indexing and contains three steps
as shown in Figure 3. The first step is to select appropriate soil indicators to evaluate critical
soil functions in function of specific management goals. These indicators form the minimum
dataset (MDS). The next step is to score the indicators to make it possible to evaluate
characteristics with different measurement units. This can be done in different ways (linear,
non-linear, optimum, more is better, more is worse). Same values can be used multiple times,
for example: a higher nitrate level can be positive for plant growth but at the same time can
give a problem with leaching. Lastly, all these unitless values are combined into an index of
soil quality (Andrews et al., 2002; Cherubin et al, 2016; Karlen et al., 2003).
23
Figure 3: Model indexing dynamic soil quality (Wienhold et al., 2004)
2.5.1 Selection of the minimum dataset
To reduce redundant information of the original dataset, the principal component analysis
(PCA) can be used. Variables showing no significant differences between groups are dropped
and from these principle components the factors with a high eigenvalue (≥1) and a high factor
loading are selected. These factors represent soil properties with the highest influence on
variance in the data. A second way to compile the MDS is to use expert opinion (Andrews et
al., 2002; Cherubin et al., 2016; Zhan-jun et al., 2013).
2.5.2 Scoring of the indicators
As mentioned before, scoring the indicators is needed to normalize indicator observations to
become values between 0 and 1 as a result of linear or non-linear scoring. When using linear
systems, the highest values for indicators that are scored as ‘high is better’ are scored as 1.
For indicators with ‘lower is better’ the lowest value is scored as 1. For values like pH or
phosphorus ‘higher is better’ is used together with a threshold value. The non-linear scoring
system uses three standard scoring functions (SSF). The first one is the sigmoid shaped curve
with a higher asymptote used for ‘higher is better’ indicators. The second one is also a sigmoid
shaped curve but this one has a lower asymptote and is used for ‘lower is better’ indicators.
Finally, the third used curve is bell shaped and is used for indicators with an optimal value like
pH. Andrews et al. (2002) conclude that the non-linear method gives more meaningful results
(Andrews et al., 2002, Cherubin et al., 2016).
2.5.3 Soil quality index (SQI)
No universal way to calculate the SQI exists, but overall 3 ways to obtain a meaningful result
are used. The first one is the least complicated one and is the additive SQI (ADD SQI). This
SQI is calculated by counting all the indicator scores together and dividing the result by the
number of factors. The ADD SQI has a risk of misinterpretation because the weight of the
24
different indicators is not included. The second way is more precise because the indicators are
weighted. This SQI is known as the weighted additive scores SQI (WTD SQI) and the weighted
scores are obtained from the PCA. Lastly, the hierarchical decision support system (DSS SQI)
is compiled from an additive value function method used in solving hierarchical multi-attribute
problems. Generally, it can be concluded that a higher SQI is better or in other words shows a
better performance of the soil function (Andrews et al., 2002; Cherubin et al., 2016).
25
3 Methods and materials
3.1 Overall approach
The soil quality status of home gardens of three different villages was examined. The villages
are Mutshenzheni, Tshidzine and Dzindi. Prior to the interviews and soil sampling the
extension officers were asked for permission and asked to locate the different home gardens
with different soil fertility practices. The interviews with home gardeners were conducted to
gain a clear vision on the used fertility management practices and to count in all variabilities
for the differentiation between the different fertility management practices. After this the soil
samples were taken to the soil laboratory of the Technical University of Tshwane to be
analyzed. Finally, all the results were put into a database to be processed for the study and for
the home gardeners.
3.2 Description of the study area
Mutshenzheni, Tshidzini and Dzindi are all part of the Thulamela municipality in Vhembe
district, Limpopo province, where rural activity is predominant. Vhembe district is situated in
the northern part of Limpopo province and is generally semi-arid with rainfall between 300 and
1000 mm per year. Soils in Vhembe are sandy in the west and are higher in loam and clay
content in the east. Basalt, sandstone and biotite gneiss are mainly the base on which the soils
are developed and are generally low in soil fertility. A general map of the places where samples
were taken can be found on Figure 4. More detailed maps of the villages can be found in
Appendix 3, Appendix 4 and Appendix 5. Mutshenzheni and Dzindi have predominantly Haplic
Lixisols. These soils are highly weathered, leached and have an argic B horizon, >50% base
saturation and the CEC of the clay is <24meq/100g. Tshidzini has predominantly Rhodic
Nitosols which contain >30% clay and a thick argic B horizon that is red to dusky red. This soil
has >0.2% oxalate extractable iron (FAO, 1997).
26
Figure 4: Soil map sample locations Mutshenzheni, Tshidzini and Dzindi
3.3 Field work
The field work consists of the interviews with home gardeners and the soil sampling.
3.3.1 Interviews
45 different home gardeners were interviewed. The interview started with filling in
administrative information and was continued by enquiring about all information concerning
the fertility management practices in the past 5 years. An example of the questionnaire can be
found in Appendix 6 and Appendix 7. When fertilizers were used, a sketch of the gardens was
made, completed by the measurements of the garden to calculate the rates of the used
fertilizer(s). A distinction was made between summer and winter practices.
3.3.2 Soil sampling
Homogenous samples were taken by collecting a minimum of 5 sub-samples. The sub-
samples were taken as shown in Figure 5. The minimum distance from the edge was 5m. The
gardener was askedif the planting rows changed every year. If not, the sub-samples were
taken in the row only. If the planting rows changed every year, subsamples were taken
between and in the rows. Figure 6 shows how the samples were taken. In gardens that were
split up into different parts with different soil fertility management practices were used, 2
samples were taken. This helps to exclude hidden variables, like nature of the soil and ensures
more reliable results in the end.
27
3.4 Analysis
All analytical methods are described below. First general preparation and physical
characteristics were determined, followed by chemical analysis.
3.4.1 Preparation of the soil samples
A general preparation of the soil samples was needed before further determination of the
chemical and physical characteristics could be performed. This preparation consists of air
drying during two days, grinding and sieving the soil through a sieve with mesh size of 2mm.
3.4.2 Physical characteristics
The determined physical characteristics are soil colour, moisture content and sand percentage.
3.4.2.1 Soil colour
The soil colour was determined using the Munsell soil color charts (1975) in direct sunlight.
3.4.2.2 Moisture content
The moisture content of the soil samples was determined by weighing the soil before and after
drying them for 48 hours at 105°C.
3.4.3 Chemical characteristics
Description of following analysis methods is very detailed because some difficulties with
procedures and interpretation of original literature were experienced.
3.4.3.1 pH(H2O)
To determine the pH(H2O) of the soil samples a 1:2.5 soil to water ratio was used according to
the Non-affiliated Soil Analysis Work Committee (1990). 25ml distilled water was added to 10g
of soil and the mixture was stirred with a glass rod for 5 seconds. After 50 minutes the mixture
Figure 5: Sketch soil sampling
Figure 6: Soil sampling
28
was stirred again and was then left for another 10 minutes. The pH(H2O) was read with a pH-
meter.
3.4.3.2 EC(1:5)
To determine the EC of the soil samples a 1:5 soil to water ratio was used. 50ml Distilled water
was added to 10g of soil and was shaken 8 hours at 132 rpm according to He et al. (2013).
After this the mixture was filtered through a Whatman No. 1 filterpaper. The EC values were
read by an EC-meter.
3.4.3.3 ECe
According to He et al. (2013) a predictive model can be used to predict the ECe from EC(1:5)
data. The used equation is shown below:
ECe = -13.87 x (EC1:5)2 + 13.62 x (EC1:5) – 0.3
3.4.3.4 Available phosphorus
To determine P in the soil samples, a method recommended by Anderson and Ingram (1993)
was used. This method is also known as the Olsen method. This method consists of extraction
of P from the soil and determine the values by spectrophotometry. First the extraction of P from
2.5g of soil was done by adding 50ml 0.5 M NaHCO3 solution that was adjusted to pH 8.5.
Extraction was stimulated by shaking the bottles for 30 minutes at 180 rpm. After this the
mixture was filtered through 2 Whatman filter papers No. 1. Besides this, a series of working
standards was needed to obtain the values of the soil samples. To acquire a series of 0, 1, 2,
3, 4 and 5 ppm P, a KH2PO4 stock solution was used. A 1% ascorbic acid solution and
molybdate reagent were prepared. The molybdate reagent consists of 4,3g ammonium
molybdate dissolved in 400ml distilled water, complemented by 0,4g antimony sodium tartrate
dissolved in 400ml distilled water and 54ml H2SO4. After cooling, the solution was dilluted with
distilled water to 1000ml. 1ml of both samples, standards and blank were pipetted in test tubes
and 4ml of 1% ascorbic acid was added. After this 3ml of molybdate reagent was added to all
test tubes and they were mixed using a vortex. It takes one hour for the colour to develop. The
test tubes were mixed again before reading with a spectrophotometer at 880nm.
3.4.3.5 Total mineral nitrogen
To determine total mineral nitrogen in the soil samples a method recommended by Anderson
and Ingram (1993) was used. This method consists of extraction of nitrate and ammonium from
soil and determine the values by spectrophotometry. First the extraction of nitrate and
ammonium from 5g of soil was done by 20ml 0.5 M K2SO4 solution. Extraction was stimulated
by shaking the bottles for 30 minutes at 160 rpm. After this the mixture was filtered through 2
Whatman filter papers No. 1. Ammonium-N and nitrate-N were determined in a different way
as described below. To determine total mineral nitrogen, ammonium-N and nitrate-N were
counted together.
29
3.4.3.5.1 Ammonium-N
A series of working standards was needed to obtain the values of the soil samples. To acquire
a series of 0, 1, 2, 3, 4 and 5 ppm NH4+, a KH2PO4 stock solution was used. Reagent N1 was
prepared 24 hours before use and needed to be stored at room temperature in a dark place.
This was done by dissolving 34g sodium salicylate, 25g sodium citrate and 25g sodium tartrate
in 750ml of distilled water. Then 0.12g sodium nitroprusside was dissolved and the solution
was diluted to 1000ml with distilled water. Reagent N2 was also prepared 24 hours before use
and also needed to be stored at room temperature in a dark place. The reagent was made by
dissolving 30g sodium hydroxide in 750ml of distilled water. After the solution was cooled
down, 10 ml sodium hypochlorite solution was added and the mixture was diluted to 1000ml.
Then 0,1ml of both samples, standards and blank were pipetted into test tubes and 5ml of
reagent N1 was added. Each test tube was mixed using a vortex and was left for 15 minutes.
Hereafter 5ml of reagent N2 was added to each test tube. After mixing the test tubes were left
for 1 hour for full colour to development. The test tubes were read with a spectrophotometer
at 655nm.
3.4.3.5.2 Nitrate-N
A series of working standards were needed to obtain the values of the soil samples. To acquire
a series of 0, 2, 4, 6, 8 and 10 ppm NO3-, a K2NO3 stock solution was used. A 4M Sodium
hydroxide solution and 5% salicylic acid reagent were made. The salicylic acid needed to be
made the day before use. Then 0,5ml of both samples, standards and blank were pipetted into
test tubes and 1ml of salicylic acid solution was added. Each test tube was mixed using a
vortex and was left for 30 minutes. After this 10ml of sodium hydroxide solution was added to
each test tube. After mixing the test tubes were left for 1 hour for full colour to development. It
was important that the temperature of the test tubes stayed above 25°C to avoid crystallization.
This could be done by keeping the test tubes in a water bath around 30°C. The test tubes were
read with a spectrophotometer at 410nm.
3.4.3.6 Exchangeable potassium
To determine exchangeable potassium in the soil samples, a method recommended by
Anderson and Ingram (1993) was used. This method consists of extraction of potassium from
soil and determine the values by flame photometry. First the extraction of potassium from 5g
of soil was done by adding 50ml 1 M ammonium acetate solution. Extraction was stimulated
by shaking the bottles for 30 minutes at 180 rpm. After this the mixture was filtered through 2
Whatman filter papers No. 1. Beside this, a series of working standards was needed to obtain
the values of the soil samples. To acquire a series of 0; 2.5; 5; 7.5 and 10 ppm K, a KCl stock
solution was used. 10ml of samples were pipetted into 25ml volumetric flasks and filled to the
mark with distilled water. After calibrating and reading the standard solutions with the flame
photometer, the sample solutions were read.
30
3.4.3.7 Exchangeable calcium
To determine exchangeable calcium in the soil samples, a method recommended by Anderson
and Ingram (1993) was used. This method consists of extraction of calcium from soil and
determine the values by flame photometry. First the extraction of calcium from 5g of soil was
done by adding 50ml 1 M ammonium acetate solution. Extraction was stimulated by shaking
the bottles for 30 minutes at 180 rpm. After this the mixture was filtered through 2 Whatman
filter papers No. 1. Beside this, a series of working standards was needed to obtain the values
of the soil samples. To acquire a series of 0, 25, 50, 75 and 100 ppm Ca, a CaCO3 stock
solution was used. 5ml of samples were pipetted into 25ml volumetric flasks and filled to the
mark with distilled water. After calibrating and reading the standard solutions with the flame
photometer, the sample solutions were read.
3.4.3.8 Total Nitrogen and carbon
Total nitrogen and carbon is determined with the Dumas method with the Vario Max CNS-
analyzer. This analysis method is based on complete catalytic combustion at 900 °C with an
excessive amount of oxygen. Combustion gasses are send into a helium-flow over multiple
absorption tubes to exclude unwanted gasses. The grade of thermic conductivity helps to
measure the amounts of N2 and CO2.
3.5 Statistical analysis
3.5.1 Interviews
The first step to analyze the collected data during the interviews was a general exploration in
excel. The use of the gardens and soil fertility management practices in winter and summer
were compared and visualized. Mean values for application rates of animal manure and
chemical fertilizer were explored by SPSS.
3.5.2 Chemical analysis
The chemical soil properties (pH(H2O), EC(1:5), ECe, NO3-N, NH4-N, total mineral N, Olsen-P,
exchangeable K and Ca, total C, total N and C/N-ratio) were explored per soil fertility
management practice. The Shapiro-Wilk test showed if the data was distributed normal.
Homogeneity of variance was tested by the Levene’s test. If the variance was homogenous,
the soil property was tested with the One-way ANOVA. If not, the Kruskal-Wallis test was used.
This showed significant differences (p ≤ 0.05) for each chemical soil property per soil fertility
management practice. Parallel with the previous tests, a pairwise test was used for data
coming from the same gardens. This test is called the Wilcoxon Signed Rank Test. Lastly, All
the data was split based on the 2 main soil types and was analyzed again.
3.5.3 Soil quality index
Based on the chemical soil properties, a soil quality index was compiled from a minimum
dataset. To compile the minimum dataset, a principal component analysis was used. The
31
factors representing the minimum dataset were scored. Values in the optimum range,
according to literature, were given a score 1. Values outside the optimum range were given a
score 0. The average score per soil fertility management practice was determined which is
also known as the ADD SQI. Again, normality and homogeneity of variance was tested
followed by a One-Way ANOVA to find significant differences (p ≤ 0.05) between the soil fertility
management practices. Parallel to this, a pairwise comparison was done on the paired data
from the same gardens to exclude variability caused by soil type. For the same reason, the
test was repeated after splitting the data for the main 2 soil types.
32
4 Results and discussion
4.1 General findings interviews
4.1.1 Use of home garden in summer and winter
68 soil samples were taken from 45 different gardens. In some gardens, in winter, only a certain
part of the garden was used while in summer the whole garden was used, so the garden was
split up in a winter and summer plot. Figure 7 below shows that almost all plots (97%) are used
in summer while in winter only 66% of the plots are used for cultivation of crops.
Figure 7: Plots used in winter and summer
Parts of the home garden not used in winter were often used in summer for growing maize and pumpkin while winter plots were often used for growing green vegetables (Table 9).
0
10
20
30
40
50
60
70
80
90
100
winter summer
% o
f th
e p
lots
use
d
Use of plots
33
Table 9: Crops grown in winter and summer
Winter Summer
• Sweet potato (Ipomea batatas)
• White cabbage (Brassica oleracea)
• Chinese cabbage (Brassica rapa subsp.
chinensis),
• Spinach (Spinacia oleracea,)
• Green beans (Phaseolus vulgaris)
• Green mustard (Brassica juncea (L.)
czern.)
• Muxe (Solanum retroflexum Dun.)
• Tomatoes (Solanum lycopersicum)
• Spring onion (Allium cepa)
• Chilis (Capsicum annuum)
• Swiss chard (Beta vulgaris)
• Carrot (Daucus carota)
• Lettuce (Lactuca sativa)
• Maize (Zea mays)
• Pumpkin (Cucurbita pepo, C. maxima
and C. moschata)
• Groundnut (Arachis hypogaea)
In almost all home gardens fruit bearing trees were present that produced mango (Mangifera
Results from Table 11 about ECe are visualized in Figure 14. Both EC and ECe are significantly
higher when animal manure is used as the soil fertility management practice. The paired
comparison in Table 12 shows that soils that received animal manure have indeed a
significantly higher EC(1:5) and ECe than soils that received no fertilizer or animal manure.
Knowing that chemical fertilizers can also be a source of salts and that the data has a high
40
variability, the finding that no significant difference between soils that received no fertilizer or
animal manure and soils that received chemical fertilizer or a combination of chemical fertilizer
and animal manure remains questionable. Application rates of animal manure show a
significant positive correlation (p ≤ 0.05) with the EC(1:5) and the ECe with correlation
coefficients of respectively 0.606 and 0.518. Azeez et al. (2010) shows that animal manure
increases the soil’s electrical conductivity due to high salt contents. When the ECe is
considered, it can be concluded that only some soils where animal manure was applied
exceeded the limit of 2 dS.m-1 which is the threshold value between non-saline and low salinity
according to Shirokova et al. (2000) although the mean ECe was 1.48 dS.m-1. All gardens that
received chemical fertilizer (mean 0.60 dS.m-1), a combination of chemical fertilizer and animal
manure (mean 0.54 dS.m-1) or no fertilizer (mean 0.57 dS.m-1) can therefore be classified as
non-saline. Again, the lower application rates for animal manure should be considered when
the soil fertility management practice ce/m is compared with the application of only animal
manure. The higher application of chemical fertilizer does not equipoise EC increasing effect
of animal manure.
Figure 14: Boxplot ECe - fertilizer category
4.2.3 Total mineral N, nitrate and ammonium
As expected, Table 11 shows that TMN (total mineral N) and nitrate were significantly higher
for gardens that received animal manure (mean TMN 39.13 mg.kg-1 and mean nitrate 36.53
mg.kg-1) than for gardens that received no fertilizer (mean TMN 13.99 mg.kg-1 and mean nitrate
11.53 mg.kg-1) which shows that soil quality improves by adding animal manure (Figure 15).
This finding was confirmed by the pairwise comparison in Table 12. Figure 15 also shows that
gardens that received chemical fertilizer or a combination of chemical fertilizer and animal
manure did not differ significantly on TMN and nitrate-N levels from gardens that received
animal manure or no fertilizer. Although animal manure provides organic and inorganic-N and
41
chemical fertilizer only provides inorganic-N, the differences are not big enough to be
significant. Again, the high variability in the data due to the differences in the nature of the soils
make these findings questionable. The level of ammonium-N was highest for soils that received
only animal manure (mean 2.59 mg.kg-1) but no significant difference existed between different
soil fertility management practices. TMN and nitrate are highly correlated (p ≤ 0.05) as shown
in Table 15 with a correlation coefficient of 0.997. They both show a significant positive
correlation (p ≤ 0.05) with application rates of animal manure of respectively 0.632 and 0.628.
The high TMN levels of outliers in the gardens that received animal manure were due to high
application rates. The outlier with 189 mg.kg-1 total mineral N received up to 66 t.ha-1 animal
manure per year, of which 39 tonnes were chicken manure and 27 tonnes goat manure.
Chicken manure is known to be very rich in nitrogen.
Figure 15: Boxplot total mineral N - fertilizer category
*Outlier for manure with value 189 mg.kg-1 not shown
4.2.4 P(Olsen)
Figure 16 shows that Olsen-P or plant available P levels significantly differ between soils that
received animal manure and no fertilizer (means respectively 14.19 and 7.19 mg.kg-1). This
conclusion is based on 2 different statistical tests (Table 11 and Table 12). Soils that received
chemical fertilizer (mean 8.37 mg.kg-1) or a combination of chemical fertilizer and animal
manure (mean 8.26 mg.kg-1) did not differ significantly from soils that received animal manure
or no fertilizer. P requirements differ strongly between plant species. Plants that require low
amounts of available P are grass, maize and soybeans and are already satisfied if the available
P level exceeds 7 mg.kg-1. This is why values below 7 mg.kg-1 are recognized to be deficient.
Almost all gardens that received no fertilizer are deficient in P while gardens that received
42
chemical fertilizer or a combination of chemical fertilizer and animal manure score a bit better
but don’t really stand out. The outlier with more than 50 mg.kg-1 P received almost 16 t.ha-1 of
chicken manure annually which is mainly known to be rich in N but also contains relatively high
P levels (Landon, 1984).
Figure 16: Boxplot P(Olsen) - fertilizer category
4.2.5 Exchangeable K
Soils that received animal manure (mean 500.63 mg.kg-1) were significantly higher in
exchangeable K than soils that received chemical fertilizer (mean 111.70 mg.kg-1) or a
combination of chemical fertilizer and animal manure (mean 154.99 mg.kg-1). Soils that
received no fertilizer (mean 305.83 mg.kg-1) did not differ significantly from soils that received
animal manure or a combination of animal manure and chemical fertilizer, which is remarkable
(Figure 17). The finding that soils that received animal manure did not differ from soils that
received no fertilizer or animal manure is confirmed in the pairwise comparison in Table 12.
The other findings remain questionable due to high variability in the data as a result of the
differences in the nature of the soils. Figure 17 shows some remarkable high values in soils
that received no fertilizer and soils that received animal manure. These values can be the
result of termite activity that was noticed in some om the soils. One of the gardeners even
mentioned that the garden was built on an old termite mound. Deke et al. (2016) show that
exchangeable K levels can be three times as high in the center of a termite mound than in the
surrounding soil. High K levels can also be the result of a long dry period with buildup of applied
nutrients. The extreme levels for soils that received animal manure can therefore be seen in
the context of these influence parameters put together.
43
Figure 17: Boxplot exchangeable K - fertilizer category
It is worth mentioning that 28 out of the 68 samples were taken from Haplic Lixisols which are
known to be poor in plant nutrients but have a base saturation of >50%. The results of
comparison of soil fertility management practices on Haplic Lixisols can be seen on Figure 18.
The differences are less clear than when both soil types are included in the statistical analysis.
In addition, another source of variation exists which includes the intensity of use of the plots.
Plots where no fertilizer or animal manure was applied were often only used once a year while
plots where any nutrients were applied, were often used 2 times a year. A note that should be
made for the gardens with extremely high K levels, in terms of crop growth, is that high K/Mg
ratios can lead to Mg deficiency. Unfortunately, Mg could not be analyzed with the flame
spectrometer and further conclusions cannot be made (Landon, 1984).
44
Figure 18: Boxplot exchangeable K - fertilizer category (Haplic Lixisol)
4.2.6 Exchangeable Ca
Exchangeable Ca levels are significantly higher for soils that received animal manure (mean
2034.09 mg.kg-1) and soils that received no fertilizer (mean 1525.86 mg.kg-1) compared with
values for soils that received chemical fertilizer (mean 535.71 mg.kg-1) or a combination of
animal manure and chemical fertilizer (mean 656.25 mg.kg-1) (Figure 19). The pairwise
comparison in Table 12 confirms the finding that no significant difference between Ca-levels
in soils that received no fertilizer and soils that received animal manure exists. Exchangeable
Ca levels can increase as a result of basic amendments present in chicken manure as
mentioned by Mokolobate and Haynes (2002). It also need to be said that 28 out of the 68
samples were taken from Haplic Lixisols which are known to be poor in plant nutrients but have
a base saturation of >50%. In addition, another source of variation which includes the intensity
of use of the plots needs to be taken into account. Plots where no fertilizer or animal manure
was applied were often only used once a year while plots where any nutrients were applied,
were often used 2 times a year. Table 15 shows that a significant positive correlation (p ≤ 0.05)
between Ca levels and application rates of animal manure of 0.313. A significant negative
correlation (p ≤ 0.05) between application rates of chemical fertilizer and Ca levels of -0.375
also is found. Keeping in mind that pH values ranged between 4.92 and 8.95 and the highest
K level was around 1600 mg.kg-1 some important interactions can occur. In combination with
a pH below 5.5 and low CEC, Ca deficiencies can occur in tropical soils. High K levels may
also obstruct Ca uptake by plants. A P deficiency can occur when the soil has a high pH (> 8)
in combination with high Ca levels. Once the pH exceeds 8.5, high Na levels increase the P
availability but plants can be deficient in Ca. Below 40 mg.kg-1 most crops have a strong
response to exchangeable Ca as a plant nutrient. Crops with high Ca requirements show
45
deficiencies when the exchangeable Ca level is below 160 mg.kg-1 but almost all values exceed
these limits (Landon, 1984; Pernes-Debuyser and Tessier, 2004).
Figure 19: Boxplot exchangeable Ca - fertilizer category
4.2.7 Total C, total N and C/N ratio
Figure 20 shows that soils that received animal manure had a significantly higher carbon level
(mean 2.40 %) than soils that received chemical fertilizer (mean 1.36 %) but it does not differ
from soils that received no fertilizer (mean 1.93 %) or soils that received a combination of
chemical fertilizer and animal manure (mean 1.54 %). The pairwise comparison in Table 12
confirms that no significant difference exists in carbon levels between soils that received animal
manure and soils that received no fertilizer or animal manure. Table 15 shows a significant
negative correlation (p ≤ 0.05) between chemical fertilizer application rates and total C with a
correlation coefficient of -0.249. Overall, the C levels are rather low and indicate a poor level
of organic matter in the soil (Landon, 1984).
46
Figure 20: Boxplot total C - fertilizer category
The total N level is significantly higher for soils that received animal manure (mean 1728.18
mg.kg-1 or 0.17 %) compared with soil that received chemical fertilizer (mean 924.29 mg.kg-1
or 0.09 %), no fertilizer (mean 1285.86 mg.kg-1 or 0.13 %) or a combination of chemical fertilizer
and animal manure (mean 1057.50 mg.kg-1 or 0.11 %) (Figure 21). The pairwise comparison
in Table 12 confirms that no significant difference occurs in total N levels between soils that
received animal manure and soils that received no fertilizer or animal manure. Table 15 also
shows the significant positive correlation (p ≤ 0.05) of 0.383 between the application rate of
animal manure and the total N level in the soil. According to Landon (1984) N levels below 0.2
% are rather low. Soils that received animal manure scored the best, but nitrate-N levels are
more relevant to estimate soil fertility because total N is strongly influenced by pH. A low pH
results in low microbial activity and thus causes lower plant available N levels because of the
lower mineralization rate (Landon, 1984).
47
Figure 21: Boxplot total N - fertilizer category
The C/N ratio for the different soil fertility management practices did not differ significantly. The
highest ratio was found in soils that received a combination of chemical fertilizer and animal
manure (mean 15.43), followed by soils that received no fertilizer (mean 15.27). soils that
received animal manure had a mean C/N ratio of 14.35 and soils that received chemical
fertilizer had a mean C/N ratio of 14.22.
4.3 Soil quality determination
4.3.1 Compilation of the minimum dataset
To compile the minimum dataset, dimensions needed to be reduced. One way to accomplish
the dimension reduction is to execute the principle component analysis. For this analysis,
indicators that differed significantly (p ≤ 0.05) for the different soil fertility management
practices were used. PC’s (principle components) with an eigenvalue higher than 1, best
represent the variation in the data. These principle components are shown in Table 14. For
each principle component the percentage of variance that can be explained by this component
is given and when these four components are put together, 81.43% of the variance can be
explained.
48
Table 14: Principal components
*Ex. K = Exchangeable K, Ex. Ca = Exchangeable Ca
The component loadings are given for each indicator, these component loading represent the
correlations between the principle component and the indicator. Under PC1 all indicators had
relatively high loadings, except pH(H2O). pH(H2O) has the highest loading for the PC2 and will
therefore be used in the MDS. As seen in the correlation matrix in Table 15 some indicators
are highly correlated (bold/red). Strongly correlated indicators are: EC(1:5) and ECe, Total N
and C. EC(1:5) and total N had the highest factor loading and will be used in the MDS together
with K, Ca, NO3 and P (Andrews et al., 2002).
Principal components PC1 PC2 PC3 PC4
Eigen value 5.29 1.48 1.14 1.05
Variance % 49.08 13.44 10.38 9.54
Cumulative variance % 49.08 61.52 71.90 81.43
EC(1:5) 0.89 0.34 -0.12 -0.13
ECe 0.88 0.36 -0.04 -0.02
Total N 0.81 -0.55 0.07 -0.13
Total C 0.74 -0.64 0.14 0.01
Ex. K* 0.77 0.10 0.01 0.23
Ex. Ca* 0.80 -0.30 0.25 0.35
NO3 0.71 0.17 -0.39 -0.39
P(Olsen) 0.66 0.05 -0.22 0.15
pH(H2O) 0.41 0.61 0.47 0.39
49
Table 15: Correlation matrix (p ≤ 0.05)
1: TMN = Total mineral N, P= Olsen P, Ex. K = Exchangeable K, Ex. Ca = Exchangeable Ca, ARC= Application rates chemical fertilizer, ARM = application rates animal manure