GROWTH AND YIELD RESPONSE OF COWPEA (Vigna unguiculata [L] Walp) TO NPK FERTILIZER AND RHIZOBIA INOCULATION IN THE GUINEA AND SUDAN SAVANNA ZONES OF GHANA
A thesis presented to the Department of Crop and Soil Sciences, Faculty of Agriculture, College of Agriculture and Natural Resources, Kwame Nkrumah University of Science and Technology, Kumasi Ghana, in partial fulfilment of the requirements for the award of degree of
DOCTOR OF PHILOSOPHY IN SOIL SCIENCE by
EMMANUEL OBIANUJU CHIAMAKA M.Sc Soil Science (Ibadan) 2005.
September, 2014
CERTIFICATION I hereby declare that this submission is my own work toward the PhD and that, to the best of my knowledge, it contains no material previously published by another person nor material which has been accepted for the award of any other degree of the University, except where due acknowledgement has been made in the text.
Emmanuel Obianuju Chiamaka ............................... Student Signature
............................... Date
Certified by:
Prof. E.Y. Safo (Principal Supervisor)
................................... Signature
..
................................... Date
Dr. F.M Tetteh (Co - Supervisor)
Signature
...................................... Date
Dr. Andrews Opoku (Co – Supervisor)
................................... Signature
...................................... Date
................................... Signature
...................................... Date
Certified by: Dr. Charles Kwoseh (Head of Department)
ii
DEDICATION This thesis is dedicated to the Almighty God, to whom all hearts are open, all desires known and from whom no secrets are hidden. Also to my late parents, Sir and Lady J.E Ojimadu who did not live to see me realize my dream.
iii
ACKNOWLEDGEMENTS
My deepest gratitude goes to Alliance for Green Revolution in Africa (AGRA) for providing the financial support for my entire Ph.D work. I lack words to express my profound gratitude to my supervisors Prof. E.Y. Safo, Dr. F.M Tetteh and Dr. Andrew Opoku whose patience, understanding and constructive criticisms made the completion of this work possible. I won‟t also forget the efforts of Prof. A.O Ogunkunle who became my mentor and helped me in my quest to acquire a higher degree. I acknowledge the encouragement I received from my pastor and former supervisor Prof. G.E Akinbola. I am also greatly indebted to Guillaume Ezui of IFDC Togo for his assistance in running the DSSAT model and also to Dr. J. Naab for his contributions in running the DSSAT model. Even though we never met physically, he was always replying my mails with suggestions. The assistance of Mr. Godwin Opoku and Mr. Aziz Abdul Rahman during my field work was highly appreciated. Mr. Azizz, thanks for the „Okada‟ ride to the field several times. Special thanks to Mr Anthony and Mr. Johnbosco, the farmers I worked with whose expertise were remarkable. My colleagues here in Ghana also made my stay worthwhile, Bright, Patrick, Dorcas, Blessing, Ruth, Mary, Garba, Idris and others, we have all come this far but the best is yet to come. Mrs Nana-Adwoa Insaidoo of the Provost Office is also appreciated for her constant words of encouragement and support. I deeply appreciate the support I received from fellowshipping with brethren at the Victory Baptist Church, Ayigya Kumasi. Rev. Ampofo, Rev. Mrs Comfort Otoo, Mr Ansah, Mrs Beatrice Solaga, Dr. Faustina Wireko-Manu and my ever dependable mother and friend Mrs. Gina Sarkodie. Thank you all for making me feel loved.
iv
I specially thank my brother Maduka for his unwavering support to me through out my academic pursuit and assuming the role of father in my life. My sisters (Aunties Ngozi, Ulumma, Uche and Patience) and in-laws in Nigeria are appreciated for their continuous support manifested through phone calls and the warmest welcome I receive each time I go back home. Finally, I want to thank someone who is largely responsible for my being where I am and working on what I do. Jewel, you meant a lot to me all these years, thank you for your unconditional love and having to bear with my absence from home for a long time with the children. Special thanks to my children Chichi, Chisom and Chinwendu for keeping me company in Ghana. Also to the „little one‟ whose kicks are a constant reminder that I need to increase my pace and submit.
v
TABLE OF CONTENTS CERTIFICATION
ii
DEDICATION
ii
ACKNOWLEDGEMENTS
iv
TABLE OF CONTENTS
vi
LIST OF TABLES
xi
LIST OF FIGURES
xiii
LIST OF APPENDICES
xiv
LIST OF ACRONYMNS
xv
ABSTRACT
xvi
CHAPTER ONE
1
1.0
1
INTRODUCTION
CHAPTER TWO
5
2.1
5
LITERATURE REVIEW
2.2 Soil fertility depletion 2.2.1 Extent of soil fertility depletion in Africa 2.3 Soil fertility replenishment
5 5 6
2.3.1 Fertilizer use
7
2.3.2 Constraints to mineral fertilizer use
8
2.3.3 Fertilizer use for cowpea
9
2.4 Cowpea production 2.4.1 Constraints to cowpea production
10 11
2.4.2 The contribution of N, P, K fertilizers to growth and yield of cowpea 11 2.4.2 Biological nitrogen fixation in legume systems
12
2.4.3 BNF benefits to succeeding crop in rotation
13
2.4.4 Factors that limit biological nitrogen fixation
15
2.4.4.1
Phosphorus availability
15
2.4.4.2
Soil nitrogen availability
17
2.4.2.3 Potassium
17
2.4.2.4
18
Populations of rhizobia.
2.4.5 Inoculation of cowpea with rhizobium strains 2.4.3.1
Factors affecting the response to inoculation
2.5 Decision Support Tools in agriculture
19 20 21
vi
2.6 Crop Simulation Models (CSMs)
23
2.6.1 Decision Support System for Agro-technology Transfer)
25
2.6.2 DSSAT - CROPGRO
28
2.6.2.1
CROPGRO - Cowpea analysis of yield gap
30
2.7 Knowledge Gaps
32
2.8 Summary
33
CHAPTER THREE
35
3.0
35
MATERIALS AND METHODS
3.1 Study area
35
3.1.1 Soil type
36
3.1.2 Climate
36
3.2 Study 1: Response of cowpea to Rhizobia inoculant and mineral fertilizer application
37
3.2.1 Pot experiment
37
3.2.2 Field experiment
38
3.2.3 Cowpea variety used
38
3.2.4 Experimental design and treatments
39
3.2.5 Nodule number per plant and nodule dry weight
39
3.3 Study 2: The effect of NPK fertilizer application on growth and yield of cowpea
40
3.3.1 Cowpea variety used and Land preparation and sowing
40
3.3.2 Experimental design and treatments
40
3.4 Measurement of crop variables
41
3.4.1 Biomass yield
41
3.4.2 Grain yield
41
3.4.3 Number of seeds pod-1
41
3.4.4 Mean 100 seed weight
41
3.5 Nutrient uptake
41
3.6 Economic analysis
42
3.7 Laboratory soil analyses
42
3.7.1 Soil sampling and profile characterization
42
3.7.2 Determination of soil chemical properties
43
3.7.2.1
Soil pH
43
3.7.2.2
Available phosphorus
43 vii
3.7.2.3
Soil organic carbon
44
3.7.2.4
Total nitrogen
45
3.7.2.5
Determination of exchangeable cations
46
3.7.2.6
Determination of NH4+ - N
49
3.7.2.7
Determination of NO3--N
51
3.7.2.8 Soil physical analyses
52
3.7.2.9 Determination of soil particle size
52
3.7.2.10 Determination of soil moisture content
53
3.7.2.11 Determination of soil bulk density
53
3.7.2.12 Determination of volumetric moisture content
54
3.7.2.13 Porosity
54
3.8 Data analysis 3.9 Model simulation and analyses procedures
55 55
3.9.1 Weather data
55
3.9.2 Model calibration
56
3.9.2.1
Soil parameters
56
3.9.2.2
Genetic Coefficients
57
3.9.2.3
Experiment
59
3.9.3 Seasonal analysis
60
3.9.4 Estimation of potential yield and yield gap
61
3.9.5 Model validation
61
3.9.6 Statistical evaluation of model performance
62
CHAPTER FOUR
64
4.0
64
RESULTS AND DISCUSSION
4.1 Selected initial physical and chemical soil properties of the experimental site. 64 4.1.1 Results
64
4.1.2 Discussion
65
4.2 Study 1: Pot experiment
66
4.2.1 Shoot dry weight
66
4.2.1.1
Results
66
4.2.1.2
Discussion
68
4.2.2 Nodule number and Nodule dry weight 4.2.2.1
Results
69 69 viii
4.2.2.2
Discussion (Nodule number and nodule dry weight)
4.3 Study 1: Field experiment 4.3.1 Grain yield
72 73 73
4.3.1.1
Results
73
4.3.1.2
Discussion
75
4.3.2 Biomass yield
76
4.3.2.1
Results
76
4.3.2.2
Discussion
78
4.3.3 Nodule number and nodule dry weight
79
4.3.3.1
Results
79
4.3.3.2
Discussion
80
4.3.4 Biomass and grain nitrogen, phosphorus and potassium uptake at Lawra and Nyoli in 2013 cropping season 4.3.4.1
Results
4.3.4.2 Discussion
81 81 83
4.3.5 Returns on investment of combined application of rhizobia inoculant and fertilizer
84
4.3.5.1
Results
84
4.3.5.2
Discussion
86
4.4 Study 2: Field experiment
86
4.4.1 Biomass yield of cowpea at Lawra during the 2012 cropping season 86 4.4.1.1
Results
86
4.4.2 Grain yield of cowpea at the Ferric Luvisol in 2012 and 2013 cropping season 4.4.2.1
88 Results
88
4.4.3 Grain yield of cowpea at Lawra and Nyoli during the 2013 cropping season
90
4.4.3.1
Result
90
4.4.3.2
Discussion
92
4.4.4 Biomass yield of cowpea at Nyoli during the 2012 and 2013 cropping season 4.4.4.1
94 Results
4.4.5 Biomass yield of cowpea in 2013 at Lawra and Nyoli 4.4.5.1
Result
94 96 96 ix
4.4.5.2
Discussion
98
4.4.6 Effects of fertilizer rates on nitrogen, phosphorus and potassium uptake at Lawra and Nyoli during the 2013 cropping season 4.4.6.1
Results
4.4.6.2
Discussion
99 99 101
4.4.7 Associations among group of variables at Lawra and Nyoli in 2012 102 4.4.7.1
Results
102
4.4.7.2
Discussion
104
4.4.8 Relationship between cowpea grain yield, grain phosphorus and potassium uptake 4.4.8.1
Regression plot of the relationship between grain yield and grain
P/K uptake at Lawra and Nyoli 4.4.9 Multivariate analysis of variance 4.5 Model Simulation 4.5.1 Model validation and evaluation 4.5.1.1
104
Results and discussion
4.5.2 Seasonal analysis
105 107 107 107 107 109
4.5.2.1
Results
110
4.5.2.2
Discussion
112
4.5.3 Simulation of best sowing dates for cowpea at Lawra and Nyoli
112
4.5.4 Yield gap analysis for cowpea production at Lawra and Nyoli
117
4.5.4.1
Results
118
4.5.4.2
Discussion
118
CHAPTER FIVE
119
5.0
119
SUMMARY, CONCLUSIONS AND RECOMMENDATIONS
5.1 SUMMARY
119
5.2 Conclusions
121
5.3 Recommendations
123
LIST OF REFERENCES
124
APPENDICES
154
x
LIST OF TABLES Table 3.1 Genetic coefficients of Omondaw cowpea used for model simulations
59
Table 3.2 Dates used to estimate the best sowing dates for cowpea at Lawra and Nyoli
60
Table 4.1 Selected initial soil physical and chemical properties of the experimental sites
64
Table 4.2 Effect of fertilizer application and inoculation on cowpea shoot dry weight at 60 DAS
67
Table 4.3 Effect of fertilizer application rates on cowpea nodule number at 60 DAS 69 Table 4.4 Effect of treatments on cowpea nodule dry weight at 60 DAS
71
Table 4.5 Variance component analysis of nodule dry weight
72
Table 4.6 Effects of inoculant and fertilizer rates on grain yield of cowpea at Lawra and Nyoli in 2013 cropping season.
74
Table 4.7 Effects of inoculant and fertilizer treatments on biomass yield of cowpea at Lawra and Nyoli in 2013 cropping season
77
Table 4.8 Effects of inoculant and fertilizer treatments on the number of nodules and nodule dry weight of cowpea at Lawra and Nyoli in 2013 cropping season. 79 Table 4.9 Biomass and grain nitrogen, phosphorus and potassium uptake by cowpea at Lawra and Nyoli in 2013 cropping season
82
Table 4.10 Effect of fertilizer rates on biomass yield of cowpea at Ferric Lixisol (Lawra) in 2012 cropping season
87
Table 4.11 Effect of fertilizer rates on grain yield of cowpea at Nyoli in 2012 and 2013 cropping seasons
89
Table 4.12 Effect of fertilizer rates on grain yield of cowpea at Lawra and Nyoli in 2013 cropping season
91
Table 4.13 Effect of fertilizer rates on biomass yield of cowpea at Nyoli in 2012 and 2013 cropping season
95
Table 4.14 Variance component analysis of biomass yield of cowpea at Nyoli in 2012 and 2013.
96
Table 4.15 Effect of fertilizer rates on biomass yield of cowpea at Lawra and Nyoli in 2013 cropping season
97
xi
Table 4.16 Effects of fertilizer rates on Biomass and Grain Nitrogen, Phosphorus and Potassium uptake at Lawra and Nyoli in 2013 cropping season
100
Table 4.17 Canonical correlations and percentage correlations between yield related variables and nutrient related variables
102
Table 4.18 Canonical loadings for the yield related set of variables
103
Table 4.19 Cannonical loadings for the nutrient related set of variables
103
Table 4.20 Multiple regression of grain yield of cowpea with grain P uptake and grain K uptake at Lawra and Nyoli (2013) Table 4.21 Wilk‟s lambda and F statistics of multivariate analysis of variance
105 107
Table 4.22 Simulated and observed grain yield for 2012 and 2013 cropping season 108 Table 4.23 Climatic potential yield, research yield, farmers yield and yield gap of cowpea at Lawra and Nyoli
117
xii
LIST OF FIGURES
Figure 1.1 Locations of study areas
35
Figure 4.1 Value cost ratios of rhizobia inoculant and Fertilizer application at Lawra 85 Figure 4.2 Value cost ratios of rhizobia inoculant and fertilizer application at Nyoli86 Figure 4.3. Rainfall in August and September, 2012 at Lawra
88
Figure 4.4 Regression plot of the relationship between grain yield and grain P uptake at Lawra and Nyoli
106
Figure 4.5 Regression plot of the relationship between grain yield and grain K uptake at Lawra and Nyoli
106
Figure 4.6 Comparison between simulated and observed yield at Lawra and Nyoli. 109 Figure 4.7 Box plot of cowpea yield over 20 years at Lawra
110
Figure 4.8 Box plot of cowpea yield over 20 years at Nyoli
111
Figure 4.9 Cumulative probability of cowpea yield as affected by different sowing dates (June 01 – August 30) at Lawra for 20 years
113
Figure 4.10 Cumulative probability of cowpea yield as affected by different sowing dates (July 01 – August 16) at Lawra for 20 years.
114
Figure 4.11 Cumulative probability of cowpea yield as affected by different sowing dates (July 01 – August 16) at Lawra for 20 years
115
Figure 4.12 Cumulative probability of cowpea yield as affected by different sowing dates (July 01 – August 16) at Nyoli for 20 years.
116
xiii
LIST OF APPENDICES
Appendix 1. Profile description of soil at Lawra
154
Appendix 2. Profile description of soil at Nyoli
155
Appendix 3. Selected chemical and physical properties of the soil profile
157
Appendix 4. Soil parameters at Nyoli used for DSSAT model
160
Appendix 5. Soil parameters at Lawra used for DSSAT model
161
Appendix 6. Experimental details for Nyoli simulation
163
Appendix 7. Experimental details for Lawra
167
Appendix 8 Weather data for Lawra and Nyoli used in running the model
171
xiv
LIST OF ACRONYMNS ANOVA
Analysis of variance
BNF
Biological Nitrogen Fixation
CCA
Canonical correlation analysis
CSM
Crop simulation model
CV
Coefficient of variation
d
Willmott index of agreement
DAS
Days after sowing
DSSAT
Decision Support System for Agro-technological Transfer
DST
Decision Support Tools
ECEC
Effective cation exchange capacity
FAO
Food and Agriculture Organization
FC
Field capacity
IITA
International Institute of Tropical Agriculture
K
Potassium
kg
Kilogram
MANOVA
Multivariate analysis of variance
MoFA
Ministry of Food and Agriculture
N
Nitrogen
NRMSE
Normalized Root Mean Square Error
NS
Not significant
P
Phosphorus
R2
Coefficient of determination
RCBD
Randomized complete block design
RMSE
Root mean square error
SED
Standard error of differences
SSA
Sub Sahara Africa
VCR
Value cost ratio
xv
ABSTRACT The attendant low yield of cowpea among smallholder farmers has increased the need for site specific fertilizer recommendation and integration of biological materials to increase the yield of the crop. The effectiveness of applied fertilizer is constrained by the use of the in appropriate rate and improper timing of sowing. Rhizobia inoculant, on the other hand, needs a balanced nutrient application to increase crop yield. The study was designed to: i) determine the effects of inoculant, P and K mineral fertilizers on N,P,K uptake, growth and grain yield of cowpea, ii) evaluate the effect of NPK fertilizer application on the growth and yield of cowpea in the Guinea and Sudan Savanna zones of Ghana, and iii) simulate the potential yield, yield gap, best sowing date, growth and yield of cowpea using CROPGRO – cowpea DSSAT model. The response of cowpea to rhizobia inoculation and fertilizer application revealed that sole application of inoculant was not sufficient to raise cowpea yield except when combined with 30 kg P2O5 ha-1 and 20 kg K2O ha-1. The yield response to mineral fertilizer also showed that K is very essential in determining cowpea yield and should therefore, not be omitted in any fertilizer recommendation rate. CROPGRO - Cowpea model was used to simulate the response of cowpea to different N rates, potential yield and the best sowing dates for cowpea. The model was calibrated for Omondaw cowpea cultivar using data from the experiment carried out at Lawra (Ferric Lixisol) and Nyoli (Ferric Luvisol) during the 2012 and 2013 cropping seasons. The model performance was evaluated statistically using RMSE (0.13 tons ha-1), CV (RMSE) (9.9%) and Wilmott index of agreement d (0.97). A long term seasonal analysis using the model was able to detect that 15 kg N ha-1 is optimal for Ferric Lixisol while 20 kg N ha-1 was optimal for Ferric Luvisol. To avoid crop failure, sowing window for Lawra ranged from July 11
xvi
to July 21st while that of Nyoli was extended to 26th July with consideration for August 5th. The yield gap analysis revealed that a wide gap exists between climatic potential yields and yields obtained from farmers‟ fields. The gaps between potential and research station yields (yield gap 1) ranged from 1.05 to 18 % of potential yield, while that of the farmers field was 84.21 % for Lawra soil and 79.55 % for Nyoli soil. It is necessary to reduce the yield gap by using site specific fertilizer recommendation and appropriate timing of sowing dates.
xvii
CHAPTER ONE 1.0
INTRODUCTION
Cowpea as a grain legume is an important source of food, income and livestock feed and forms a major component of tropical farming systems because of its ability to improve marginal lands through nitrogen fixation and as cover crop (Sanginga et al., 2003). It is a valuable and dependable commodity that earns income for many smallholder farmers and traders in sub Saharan Africa (Langyintuo et al., 2003). Cowpea is widely cultivated in Ghana under rainfed conditions
mainly in the
savanna and transitional agro ecological zones (CRI, 2006), but yields are among the lowest in the world, averaging 310 kg ha-1 (Ofosu-Budu et al., 2007). Consequently, efforts have been made to improve cowpea production in Ghana through various means including the introduction of new varieties (Addo-Quaye et al. 2011). None of these improved varieties could achieve the optimum yield without appropriate and site specific fertilizer recommendations. On the other hand, there is limited use of fertilizer for agriculture in Ghana (less than 8 kg ha -1) while N, P and K depletion rates in the country range from about 40 to 60 kg ha-1 yr-1 of (FAO, 2005a), which is among the highest in Africa. Farmers use organic manure in soil fertility management; however, the benefits derived from the use of organic materials have not been fully realized due to their low nutrient content and the huge quantities required in order to satisfy the nutritional needs of crops (Ewusi-Mensah, 2009). Transportation as well as handling costs also constitute major constraints (Ayoola, 2006). The addition of organic amendments by smallholder farmers corresponds in most cases to a recycling process, which cannot compensate for nutrients exported through crop harvest (Bationo et al., 2002); as a result, the use of inorganic fertilizer remains a key input for increased productivity for smallholder farmers. One
1
significant problem to effective utilization of fertilizers by farmers has been blanket recommendation which fails to take into account differences in resource endowment (soil type, labour capacity, climate risk, and finances) of the farmers. Various blanket fertilizer recommendations exist for cowpea but are not site specific as they were formulated on a national basis. Food and Agriculture Organization (FAO) in conjunction with the Ministry of Food and Agriculture (MoFA) and the Council for Scientific and Industrial Research (CSIR) carried out intensive fertilizer use studies throughout Ghana from 1962 to 1969 and recommended 20-40-20 NPK for cowpea (FAO/UN, 1974). However, soil conditions have changed over the years and these rates may no longer be valid; therefore, there is the need for an updated and site specific fertilizer recommendation for the major food crops. Estimating the amount of fertilizer needed for optimal crop growth and yield requires many years of experimentation in order to be able to make meaningful deductions and for the results to be valid for use across soils of different agro-ecologies. (Fosu et al., 2012). This can be expensive and very time consuming. Crop simulation models (CSMs), however, provide an excellent alternative approach to years of experimentation and they also offer the opportunity to study „what if‟ type of situations in which various options are compared (MacCarthy et al., 2009). Decision Support System for Agro-technological Transfer (DSSAT) model has been used to access yield gaps in peanut production in the Guinea Savanna zone of Ghana (Naab et al., 2004) but it is yet to be calibrated for cowpea simulation in the Guinea and Sudan savanna zones of Ghana. DSSAT CROPGRO-cowpea, once tested and validated for an area, can be used with long term historical weather data to simulate
2
crop performance under varying cultural practices such as sowing dates, sowing densities, cultivar selection, soil fertility and diseases. In addition to updating fertilizer recommendations, there is also the need to optimize Biological Nitrogen Fixation (BNF) which provides a cheaper source of nitrogen for resource poor farmers. This can be achieved by inoculating the cowpea with effective rhizobia strains. The BR3267 rhizobium strain developed in Brazil but produced in Ghana by the Savanna Agricultural Research Institute (SARI) could have the potential of improving the yield of cowpea. It has been tested in the Northern region but it is yet to be evaluated in the Upper West region of Ghana. The ability of cowpea to fix nitrogen is largely influenced by the population of rhizobia in the soil, effectiveness of the rhizobia and availability of phosphorus. Where these conditions are not met, nitrogen fixation may not be optimal thereby reducing yield. Fening and Danso (2002) indicated that most indigenous cowpea bradyrhizobia isolates in the soils of the Guinea Savanna zone of Ghana were moderately effective. Soils deficient in phosphorus (P) limit the extent of nodulation, N fixation and seed yield of legume crops (Ankomah et al., 1995). Phosphorus, apart from its effect on the nodulation and plant growth, has been found to exert some direct effects on soil rhizobia. Since phosphorus deficiency is common in West African soils (Nwoke et al., 2005; Kisinyo et al., 2011), it is important to determine the level of P that will make the inoculants effective. The overall objective of this study was therefore, to improve yield of cowpea in the Guinea and Sudan savanna zones of Ghana using rhizobia inoculant and site specific fertilizer recommendation.
3
The specific objectives were to: i.
assess the influence of inoculant, P and K fertilizers application on nutrient uptake, growth and grain yield of cowpea,
ii.
determine the effect of NPK fertilizer application on the growth and yield of cowpea in the Guinea and Sudan Savanna zones of Ghana
iii.
to simulate potential yield, yield gap, best sowing date, growth and yield of cowpea using CROPGRO – cowpea DSSAT model.
Hypotheses The above specific objectives were formulated to test the following null hypotheses: i.
nutrient uptake, growth and grain yield of cowpea is not influenced by
inoculant, P and K fertilizer application, ii.
growth and yield of cowpea in the Guinea and Sudan Savanna zones of Ghana is not affected by the application of NPK fertilizer,
iii.
potential yield, yield gap, best sowing date, growth and yield of cowpea cannot be simulated using CROPGRO – cowpea DSSAT model
4
CHAPTER TWO 2.1 2.2
LITERATURE REVIEW
Soil fertility depletion
Alley and Vanlauwe (2009) defined soil fertility as the capacity of soil to retain, cycle and supply essential nutrients for plant growth over extended periods of time. It is important to crop production as the soil is the nutrient base for plants but low soil fertility, particularly N and P deficiencies, is one of the major biophysical constraints affecting African agriculture (Mokwunye et al., 1996). The import of these is that no matter how other factors of production are remedied, food production in Africa will continue to decrease unless there is a conscious effort to address soil fertility depletion. Nutrient depletion rates are field specific, depending on the way each particular field has been managed over decades (Sanchez et al., 1997) hence the replenishment should also be field/site specific for effective results. Nutrient depletion rates vary with soil properties. The proportion of nutrients loss is normally greater in sandy soils because soil organic matter (SOM) particles are less protected from microbial decomposition in sandy soils than in loamy or clayey soils (Swift et al., 1994).
2.2.1 Extent of soil fertility depletion in Africa Africa covers an area of about 3.01 × 109 hectares, out of which about 230 × 106 hectares represents natural water resources (FAO, 1976). Most soils of Africa are poor compared to most other parts of the world. In addition to low inherent fertility, nutrient balances in African soils are often negative indicating that farmers mine their soils. During the last 30 years, soil fertility depletion has been estimated at an average of 660 kg N, 75 kg P and 450 kg K ha -1 from about 200 million ha of
5
cultivated land in 37 African countries (Smaling et al., 1997). This is equivalent to 1400 kg urea ha-1, 375 kg triple superphosphate ha-1 and 896 kg muriate of potash ha1
. These figures represent the balance between nutrient inputs as fertilizer, manure,
atmospheric deposition, Biological Nitrogen Fixation (BNF) and sedimentation and nutrient outputs as harvested products, crop residue removals, leaching, gaseous losses, surface runoff and erosion. Africa loses $4 billion per year due to soil nutrient mining (Smaling et al., 1997). Considerable nutrient losses at a district scale are -1
-1
evident revealing amounts of 112 kg N, 30 kg P and 70 kg K ha yr (Smaling et al., 1997). Estimates in Sub-Saharan Africa show that 320 million hectares of land are affected by human induced soil degradation, 124 million of which are highly degraded (Crosson and Anderson, 1995). The amounts of nutrients extracted from soil through harvesting of crops, leaching and water erosion normally outweigh those imported naturally through atmospheric deposition, biological nitrogen fixation, and artificially through organic manure and mineral fertilizers (Smaling and Janssen, 1993).
2.3 Soil fertility replenishment Soil fertility replenishment should be considered as an investment in natural resource capital (Sanchez et al., 1997). Reversing soil fertility depletion is one of the requirements for increasing per capital agricultural production in Africa (Sanchez and Leakey, 1997). Options aimed at improving soil fertility should rely on soil nutrient-supplying capacity, available soil amendments, and judicious use of mineral fertilizers to achieve balanced nutrient-management systems. Such an approach is usually referred to as Integrated Soil Fertility Management (ISFM) and should be embedded in a framework that includes aspects such as: weather, the presence of
6
weeds, pests and diseases, crop management, and socio-economic aspects such as input and output prices and labour availability (Bontkes and Wopereis, 2003 a). Bekunda et al. (1997) observed that small amounts of crop residues are produced where soil fertility depletion is high making mineral fertilizers the principal sources for building up nutrients in soils. Farmers must however be aware of the fertilizer forms, methods of application, and potential benefits accruing from their use. Other pathways include, using microbial products (rhizobia inoculants), technologies such as soil conservation and sound agronomic practices.
2.3.1 Fertilizer use Approaches for soil-fertility management range from recurring fertilizer applications to low external input agriculture based on organic sources of nutrients (Sanchez et al., 1997). Although both extremes work well in specific circumstances, they pose major limitations for most smallholder farmers in Africa. Fertilization in tropical agriculture has the potential to dramatically increase production due to the highly weathered soils and the limited reserves of nutrients (Stewart et al., 2005) and therefore should be at the core of strategies to restore soil fertility and raise crop productivity. Over the years, there has been a rapid increase in fertilizer application worldwide as a result of the favourable policies which were created by introducing fertilizer subsidies and crop price-support programme and investment in distribution systems (Bumb, 1989). However, there has been a reduction in the use of fertilizer especially in sub-Saharan Africa (Sanchez et al., 1997) due to several reasons ranging from high cost to availability. In Africa approximately 1.38 million tons of fertilizer per year is applied to cultivated lands resulting in an average fertilizer
7
consumption of 8.3 kg ha-1. This consumption represents only 2% of worldwide demand and the lowest in the world (Morris et al., 2007).
2.3.2 Constraints to mineral fertilizer use Although mineral fertilizers can improve crop nutrition, they are sparingly used by farmers in Ghana, as in many regions in sub-Saharan Africa, partly due to the prohibitive cost as a result of removal of government subsidies (Gerner et al., 1995). Sanginga and Woomer (2009) stated that fertilizer consumption pattern within nations in Africa are often sketchy and inconsistent. Most smallholder farmers in Africa use fertilizers, but they are seldom able to apply them at the recommended rates and at the appropriate time because of high cost, lack of credit, delivery delays, and low variable returns (Heisey and Mwangi, 1996; Larson and Frisvold, 1996). Such constraints are largely due to the lack of an enabling policy environment in rural areas caused by the poor road and market infrastructure typical in most African countries. The price of fertilizers in rural areas of Africa is usually at least twice the international price (Bumb and Baanante, 1996). African countries subsidized fertilizers; however, the removal of fertilizer subsidies by most African governments has increased fertilizer prices in relation to crop prices in many of these countries (Bumb and Baanante, 1996).
Fertilizer recommendations disregard variations in
crop demand and soil properties and farmers‟ access to inputs and commodity markets with scales that are too large to capture soil heterogeneity (Smaling et al., 2002). Farmers on their part lack information about the best fertilizer to use for their particular fields and cropping practices, making the crop response to fertilizers more erratic and less profitable. Even within more localized recommendation domains, households operate at different stages of economic development leading to misuse
8
and associated economic (Chase et al., 1991) and environmental risks (Bundy et al., 2001). The decline in the use of mineral fertilizer in Ghana can be attributed to policy changes by the Government of Ghana since 1988. CSIR – NARP (1998) identified privatization of the importation and distribution of fertilizers and the removal of subsidies as one of the causes of low fertilizer use. Obeng et al. (1990) also showed that economic response to fertilizer use by farmers varies with the type of farming systems and level of fertilizer application. Fertilizer supply and availability at the right time also affects its usage. Moreover, the little fertilizer available is often not the correct type required for various crops and farmers are unfamiliar with its correct usage.
2.3.3 Fertilizer use for cowpea In many African countries including Ghana, the main use of fertilizer is on maize, sorghum/millet and rice (Camara and Heinemann, 2006) with cowpea receiving little attention from farmers in terms of fertilizer application. As farmers lack adequate nutrient resources to fertilize all crops, they prefer to apply fertilizers to cereals and rarely target fertilizers directly to grain legumes which are mostly grown on residual fertility (Zingore et al., 2008). It is generally believed by most cowpea growers that the production of legumes do not require inorganic fertilizer application (Kan‟ankuk‟a, 1999). This is due to the excessive vegetation at expense of grain production of this crop under fertilized fields. Cowpea can fix about 40 kg N ha -1 from nodules in the presence of right rhizobia strain which can satisfy the crop nitrogen requirements (Singh et al., 1997). There are some reports indicating that in
9
poor soils, cowpea hardly satisfies N requirements but the crop performance is improved by fertilization (Chiezey et al., 1990; Kan‟ankuk‟a, 1999; FAO, 2005b). The following are various fertilizer recommendations for cowpea in Ghana. i.
FAO (1974) based on soil test recommended 22 kg N, 28 – 67 kg P2O5 and 22 – 45 K2O ha-1.
ii.
The Grains Development project (1988) recommended the application of 5060 kg P2O5 ha-1.
iii.
CSIR - NARP (1998) suggested 20-50-20 kg N - P2O5 – K2O ha-1.
iv.
SARI (2013) recommended 20 kg N ha-1 on old land (continuously cropped land) where organic matter content may be as low as 1% and 40 kg P205 ha-1.
2.4
Cowpea production
Cowpea is a major component of traditional cropping systems in many parts of the tropics and it is important because of its multiple uses. Common uses of cowpea include soil fertility improvement through Biological Nitrogen Fixation, green manure, forage, production of high quality hay and silage, synthesis of nutritional products, suppression of weeds, food, and a source of protein and income generation (Saha and Muli, 2002). Cowpea is one of the economically important indigenous African legume crops (Langyintuo et al., 2003) especially in the dry
regions
covering 12.5 million hectares (FAO, 2005b) with annual production of about 5.2 million tons of the nearly 5.4 million tons produced worldwide (IITA, 2003). Nigeria is the world‟s largest producer with 2.1 million tons produced annually followed by Niger with 650,000 tons and Mali with 110,000 tons (IITA, 2003) while Ghana produces about 57,000 tons. The demand for cowpea in Ghana is estimated to be
10
169,000 tons thereby giving a deficit of 112,000 tons making importation inevitable (Langyintuo et al., 2003).
2.4.1 Constraints to cowpea production The major constraints to cowpea production in Ghana are insect pests, diseases, drought and low soil fertility (ICRISAT, 2013). Chiezey et al. (1990) and Kan‟ankuk‟a (1999) also identified absence of right strains of rhizobia in the soil as one of the constraints to cowpea production. Lack of inputs such as fertilizer, insecticides and improved seeds, poor cultural practices and lack of appropriate machinery for expanding planted area are other constraints experienced. Most cowpea crops are rain fed and although it is drought tolerant, cowpea farmers in the dry areas of sub-Saharan Africa obtain low yields, estimated at about 350 kg per hectare.
2.4.2 The contribution of N, P, K fertilizers to growth and yield of cowpea In the Southern Guinea Savanna zone of Nigeria, Abayomi et al. (2008) reported that the application of 30-15-15 kg NPK ha-1 which gave a yield of 1.29 tons ha-1 is beneficial for cowpea production in the area. There was no significant increase in yield with application rate of 60-30-30 kg NPK ha-1 which gave a yield of 1.23 tons ha-1. Hasan et al. (2010) conducted a study to investigate the effect of nitrogen on the biomass yield of cowpea forage and found that application of N at the rate of 25 kg ha-1 gave a biomass yield of 5.47 ton ha-1 and increasing the application to 30 kg ha-1 did not differ significantly with a yield of 5.49 tons ha-1. Ayodele and Oso (2014) working in Ekiti, sub-humid Nigeria found that application of P at 20 kg ha-1 in the presence of basal 20 kg N and 30 kg K2O ha-1 is optimum P rate for cowpea
11
production with a grain yield of 1.26 tons ha-1 compared to 0.78 tons ha-1 for the control. Similarly, Magani and Kuchinda (2009), reported that grain yields realized with 35.5 kg P ha-1 did not differ significantly from that of 75 kg P ha-1 giving a yield of 1.85 tons ha-1 and 1.91 ton ha that 40 kg P ha
-1
-1
respectively. Ndor et al. (2012) also suggested
fertilizer could be the best P fertilizer level for cowpea in the
Southern Guinea Savanna zone of Nigeria. Osunde et al. (2007) evaluated the response of 15 cowpea lines to low (0 kg P ha-1), intermediate (20 kg P ha-1) and high (40 kg P ha-1) phosphorus level in the Southern Guinea zone of Nigeria. With a basal application of 20 kg N ha-1 and 30 kg K2O ha-1 the grain yields of the cowpea lines increased as applied P rates were increased. Yields ranged from 17 tons ha -1 to 3.0 tons ha-1. The addition of 20 kg P ha-1 significantly increased the yield of 5 lines while 3 other lines had yield increases due to application of 40 kg P ha -1. Seven lines however did not show yield increase due to P application which was attributed to genotypic variations. Azarpour et al. (2011) reported that 45 kg P ha-1 gave the highest grain yield of 1.57 tons ha-1 against the yield of 0.88 tons ha-1 from the control plot in Iran.
2.4.2 Biological Nitrogen Fixation in legume systems Biological Nitrogen fixation is the process whereby a number of species of bacteria use the enzyme nitrogenase to convert atmospheric N2 into ammonia (NH3) and NO3, a form of nitrogen (N) that can then be incorporated into organic components, e.g. protein and nucleic acids, of the bacteria and associated plants (Unkovich et al., 2008). In this way, unreactive N2 enters the biologically active part of the global N cycle. These bacteria form nodules in association with grain legumes. Cowpea can fix about 240 kg ha-1 of atmospheric nitrogen and make available about 60-70 kg ha-1
12
nitrogen available for succeeding crops grown in rotation with it (Aikins and Afuakwa, 2008). Adjei-Nsiah et al. (2008) evaluated 5 cowpea varieties for N2 fixation with no fertilizer application. The amounts of N2 fixed in the above-ground biomass were 41, 43, 32, 34 and 67 kg ha–1. Yusuf et al. (2009) on the other hand reported that the amount of N2 fixed by cowpea genotypes ranged from 13.9 to 40.3 kg ha-1 in the Northern Guinea Savanna zone of Nigeria. The differences observed in the amount of N2 fixed by the legume genotypes were attributed to the number of days required to attain maturity. Higher amounts of fixed N2 were found in longer duration genotypes (Sanginga et al., 2002). Sanginga et al. (2000a) also reported a range of 13.1–31.9 kg ha-1 fixed N for eight cowpea genotypes in the derived savanna of West Africa. Vesterager et al. (2008) stated that cowpea fixed around 60% of its N from the atmosphere amounting to 70 kg N ha -1 under sole cropping and 36 kg N ha-1 when intercropped with maize in the semi–arid zone of Tanzania.
2.4.3 BNF benefits to succeeding crop in rotation Grain legumes are either grown as sole crops, intercropped or rotated with cereals and other staple foods, such as cassava and even cotton. Legumes included in the cropping system improve the fertility of the soil, which largely depends on how their residues are utilized (whether incorporated which is of more benefit, totally removed from the field or burned) (Giller, 2001). They help in solubilising insoluble P in soil, improving the soil physical environment, increasing soil microbial activity, restoring organic matter and smothering weeds (Ghosh et al., 2007). Furthermore, its ability to fix atmospheric nitrogen through symbiosis with rhizobia in the soil (AsumingBrempong, 2013) makes it a good option for rotation with cereals as it enriches the soil with nitrogen. Carry-over of N from BNF, e.g. in roots and stover, can supply
13
the N demand of subsequent non-N2 fixing crops (Van Kessel and Hartley, 2000). For example, sorghum yields increased when sown after groundnuts and cowpea (Ghosh et al., 2007). Cowpea can provide an equivalent of 60 kg N ha-1 to the subsequent non-legume crop or cereal. The estimates of N fertilizer replacement value for cowpea range from 10 kg N ha-1 (Carsky and Iwuafor, 1999) to 60 kg N ha1
(Ghosh et al., 2007). Part of the N requirement of cereal crops can be satisfied by
cowpea crop rotation. Bationo et al. (2002) stated that yields of cereals succeeding cowpea could, in some cases, double compared to continuous cereal cultivation. Also with efficient soil fertility management, cowpea can fix up to 88 kg N ha-1 and this result in an increase of nitrogen use efficiency on the succeeding cereal crop from 20% in the continuous cereal monoculture to 28% when cereals are in rotation with cowpea. Furthermore, benefits of cowpea rotation are sometimes higher than expected based on the N content of the cowpea crop alone. Reasons for this include substantial root biomass and N, substantial N-sparing by the legume, and other benefits such as reduction in Striga hermonthica, pests and diseases, and possibly access to sparingly soluble P grown in one season (Horst and Hardter, 1994). Ncube et al. (2007) reported that 2 varieties of cowpea in Zimabwe fixed 64 and 109 kg N ha-1 respectively while sorghum grain yield after the legume reached up to 1.62 tons ha-1 compared to 0.42 tons ha-1 when following sorghum in the rotation. Bado et al. (2006) reported that cowpea fixed about 50–115 kg N ha-1 in Burkina Faso which increased the yield of succeeding sorghum by 290%. Rusinamhodzi et al. (2006) intercropped cotton and cowpea in Zimbabwe. Results showed that cowpea suppressed cotton yields but the reduction in yield was compensated for by high yields obtained from cowpea and the succeeding maize. Cotton lint yield was reduced from 2.5 Mg ha–1 (sole cotton) to 0.9 Mg ha–1 (1:1 14
intercrop) and 1.5 Mg ha–1 (2:1 intercrop). Cowpea grain yield was also significantly reduced in the intercrop as follows: 1.6 Mg ha–1 (sole cowpea), 1.1 Mg ha–1 (1:1 intercrop), and 0.7 Mg ha–1 (2:1 intercrop) respectively. Remarkably, cowpea was beneficial to subsequent maize planted as the yields were as follows: after sole cotton (1.4 Mg ha–1), sole cowpea (4.6 Mg ha–1), 1:1 intercrops (4.4 Mg ha–1) and 2:1 intercrops (3.9 Mg ha–1).
2.4.4 Factors that limit biological nitrogen fixation 2.4.4.1 Phosphorus availability The N2 fixing legume plants usually require more P than plants dependent on mineral N fertilizer making P nutrition a crucial factor for optimal BNF. Soil P availability during plant seedling development is an important determinant for plant growth, N 2 fixation, and grain formation of legumes (Vance, 2001) and can prevent nodulation (Giller, 2001). Nodule establishment and function are important sinks for P and nodules usually have the highest P content in the plant (Sinclair and Vadez, 2002). Obviously, P deficiency conditions will result in reduced BNF potential and P fertilization usually leads in enhanced nodule number and mass, as well as greater N fixation activity per plant. Yakubu et al. (2010) stated that low phosphorus content of the soil may restrict rhizobia population and legumes root development, which in turn, can affect their N2 fixing potential. Also an increase of whole plant growth and plant nitrogen concentration responded to increased soil P supply from 20 to 40 kg P2O5 ha-1. Magani and Kuchinda (2009) argued that most farmers in Africa do not have access to P-fertilizer, hence the need to select cowpea with high P use efficiency as a low input approach to addressing this constraint. After conducting the trials, application of 37.5 P2O5 kg ha-1 was the most economic level for maximum
15
grain yield and crude protein content. Singh et al. (2011) in the Sudan Savanna zone of Nigeria obtained significant response to applied P up to 60 kg P2O5 ha-1 with all the varieties of cowpea used. Higher grain yield was recorded in plots that received 60 kg P2O5 ha-1 (1353 kg ha–1) than 0 kg P2O5 ha-1 (1017 kg ha–1), 20 kg P2O5 ha-1 (1067 kg ha–1) and 40 kg P2O5 ha–1 (951 kg ha–1). Application of 20 and 40 kg P2O5 ha–1 was not significantly different from those plots that were not applied with P (0 kg P2O5 ha–1). This indicated that P became a limiting factor at later stage of plant growth. Higher yield recorded when 60 P2O5 kg ha–1 was applied was attributed to higher availability of P that is responsible for effective nodulation in cowpea and nitrogen fixation. Similarly, with the different phosphorus and potassium levels, Amaral et al. (2013) obtained the highest grain yield of cowpea with 67.73 kg of P2O5 ha-1 and 35 kg K2O ha-1. In the Guinea Savanna zone of Nigeria, Uzoma et al. (2006) noted that application of 20 and 40 kg P2O5 ha-1 improved nodulation, N accumulation and grain yield of cowpea. The best performance was obtained with 40 kg P2O5 ha-1. Asuming-Brempong (2013) conducted a pot experiment and reported that growing cowpea at the coastal savanna zone of Ghana might require phosphorus at 90–120 kg P2O5 ha-1. All these studies confirm that P recommendation for cowpea should be location specific. Vesterager et al. (2008) noted that the amount of N2 fixed in the sole crops increased from 58 to 77 kg N ha-1 and in the intercrop from 30 to 43 kg N ha-1 by P application whereas cowpea yield was unaffected in the semi-arid zone of Tanzania. Abayomi et al. (2008) on the other hand observed that nodulation was significantly reduced by successive application of NPK fertilizer from 0-0-0 to 60-30-30 kg N-P-K ha-1. Nodulation decreased with increasing P application suggesting an antagonistic effect of readily available N on the functions of P, thereby
16
indicating that application of P may not optimize the BNF in presence of readily available N in the soil.
2.4.4.2 Soil nitrogen availability Biological Nitrogen Fixation, mineral soil and fertilizers are the main sources of meeting the N requirement of legumes (Salvagiotti et al., 2008). However, antagonism between nitrate concentration in the soil and the N2 fixation process in the nodules is the main constraint the crop faces in terms of increasing N uptake (Streeter, 1985) when no other abiotic stress that reduce BNF activity occurs, e.g. soil moisture (Dadson et al., 2005), soil pH and temperature (Salvagiotti et al., 2008). Many reports exist on the effects of mineral N on nodulation and N2 fixation by legumes as reviewed by Salvagiotti et al. (2008); however it is generally accepted that small quantities of mineral N available during early growth can promote nodulation and N2 fixation. Additional available mineral N substitutes for N2 fixation, and if the mineral N is supplied in abundance, it can completely suppress the symbiosis (Subasinghe, 2001).
2.4.2.3 Potassium Potassium has no direct role in nodulation, but its addition can increase nodulation on deficient soils (Giller, 2001). Sangakkara et al. (2001) while studying the influence of soil moisture and K on the growth of cowpea in Switzerland noted that K fertilizer can be considered a significant factor in overcoming soil moisture stress in Cowpea and Mungbean. Potassium was found to be important for cowpea in dry conditions by promoting vegetative growth and optimizing physiological parameters which influence 17
subsequent grain yields. Oliveira et al. (2009) further stated that K is the nutrient extracted and exported in larger quantities by cowpea. Fugger (1999) suggested that nutrient replacement fertilization based on nutrient removal by crop yields should be the farmers‟ general practice in order to avoid a decline in exchangeable K.
2.4.2.4 Populations of Rhizobia Rhizobia form a symbiotic relationship with cowpea roots to fix atmospheric nitrogen. The higher the population of rhizobia, the more chances for nodule infection. Leite et al. (2009) stated that microbial diversity in the soil involves a variety of species, which can either be effective or not effective. Martins et al. (2003) also confirmed that the rhizobial population in the soil can be extremely variable, both in the composition and the symbiotic characteristics of a species. Furthermore, native and inefficient rhizobium strains can compete with efficient ones, introduced by inoculation, for sites of infection on the host plant roots. Nodules formed by different strains and even by different species may occur in the same plant (Moreira and Siqueira, 2006). Legume-Rhizobium symbiosis can sustain tropical agriculture at moderate levels of output, provided all environmental constraints to the proper functioning of the symbiosis have been alleviated (CIAT-TSBF, 2004). Singleton et al. (1992) reported that in 305 soil samples collected from 17 different countries in the tropics, 60% of the soils had fewer than 1,000 rhizobia g-1 soil (47% with 100 rhizobia g-1 soil) belonging to the „„cowpea‟‟ cross-inoculation group and the results of a number of on-farm inoculation trials around the world indicate that even promiscuous cowpea often benefit from inoculation. Similarly, Fening and Danso (2002) examined 100 isolates of cowpea Bradyrhizobia in 20 soils in Ghana. The isolates varied from ineffective (6%) to highly effective (26%), but with a
18
predominance (68%) of all isolates being ranked as moderately effective. The results obtained raise an important question, whether the 26% highly effective bradyrhizobia are capable of satisfying optimum N requirements from symbiotic N2 fixation.
2.4.5 Inoculation of cowpea with rhizobium strains The need for inoculation depends on the presence of compatible rhizobia in the soil and their effectiveness (Giller, 2001). Until recently in Brazil and in Africa it was assumed that rhizobium capable of nodulating cowpea were found in the soils of these regions, it was therefore not worth inoculating it with selected strains (Fosu 2012). However, Singleton et al. (1992) and Fening and Danso (2002) indicated that inoculating cowpea with effective rhizobia strain may increase BNF and subsequently, the yield of cowpea. Awonaike et al. (1990) reported that amounts of nitrogen fixed among the 3 varieties of cowpea studied were not appreciably influenced by inoculation in an Oxic Paleutstalf of Ibadan Nigeria. The average values of N2 fixed by the 3 varieties were 116.87, 87.0 and 73.7 kg N ha-1 for the 3 varieties. The amount of N2 fixed between inoculated and uninoculated suggested that adequate numbers of effective cowpea bradyrhizobia existed in the soil of the study area. Theuri et al. (2003) stated that commercial inoculant was more effective in terms of shoot biomass yield and number of nodules than the indigenous soil rhizobia while studying the abundance of indigenous rhizobia nodulating cowpea central Kenyan soils. Nodules count for inoculated plant was 38 nodules per plant while uninoculated averaged 8 nodules per plant. Mean shoot biomass was 0.68 g plant-1 for inoculated and 0.36 g plant-1 for uninoculated. Nyoki and Ndakidemi (2013) inoculated cowpea with B. japonicum in Tanzania and the results showed that inoculation increased the number of pods per
19
plant by 13.7%, number of seeds per pod (11.6%), mean pod weight (24.6%) and 100 seed weight (8.5%). Grain yield in kg per hectare was increased by 12.44% in the field experiment over control. There was interactive effect between B. japonicum and phosphorus on the number of nodules counted at 50% pod formation in the field experiment at 40 kg P2O5 ha-1. The result shows that phosphorus is important in enhancing nodulation, symbiotic association between rhizobia and host plant and consequently improved N2 fixation. This is buttressed by the fact that the control treatment (0 kg P2O5 ha-1) with or without B. japonicum produced fewer number of nodules over other phosphorus treatments. Singh and Syamal (2011) studied the number of nodules per plant of cowpea as influenced by Rhizobium inoculation, phosphorus application and their interaction effect on cowpea. The treatment combinations exhibited significant effect on number of nodules per plant with the maximum number of nodules per plant i.e., 61.65 and 65.60 recorded at phosphorus application rate of 100 kg P2O5 ha-1 along with Rhizobium inoculation.
2.4.3.1 Factors affecting the response to inoculation Environmental and management variables influence legume yield (AsumingBrempong et al., 2013) and, as a result, the requirement for atmospheric nitrogen. Nitrogen fixation and the potential response to inoculation, therefore, are also necessarily affected by the environment. Crop management and nutrient deficiencies may reduce the population of free-living rhizobia in the rhizosphere, limit the growth of the host plant, restrict nodulation itself, and have an adverse effect on the functioning of the nodule (Sanginga et al., 2002). In some agricultural environments, the magnitude of response to inoculation will be mainly influenced by general crop adaptation rather than specific direct impacts of
20
environment on symbiotic processes. Inoculation success/failure is highly site specific. It depends on a lot of interacting factors that it would make it virtually impossible to make generalizations based on data gathered by different workers without standardization. Date (1982) has proposed a similar specificity classification scheme for tropical forage legumes that include three categories: (i) Group PE, Promiscuous and Effective; (ii) Group PI, Promiscuous and Ineffective; and (iii) Group S, Specific. Even among the promiscuous group, there are those that nodulate with a subgroup of bacteria.
2.5
Decision Support Tools in agriculture.
Making on-farm and land management decisions can be challenging as farmers decision making ranges from practical day-to-day decision making, to decision making that affects the productivity of the land for several growing seasons to a number of years Bontkes et al. (2001). Examples of day-to-day decisions includes: when to apply fertilizer and how much, when to weed, when to apply pesticide, etc. Medium-term, season-to-season decision making may involve choice of production system (mixed legume and cereal cropping or sole cropping), tillage system (ploughing or conservation tillage), cultivar choice (short or medium duration) and sowing date (early or late). Farmers also are strongly dependent on, and constrained by, what is often one of their most important assets: the soil and are faced with the decision of how to effectively manage the soil (Bontkes and Wopereis, 2003a.). Factors that play a role in soil fertility management do not only pertain to those directly related to soil fertility such as inherent soil characteristics, history of land use, spatial differences in soil fertility, application of inorganic and organic fertilizer, but also to weather, the presence of weeds, pests and diseases and crop management, and beyond that to socio-economic aspects such as in- and output prices and labour
21
availability (Bontkes and Wopereis, 2003a). With these complexities, there is need for tools that can help decision makers (farmers, extension officers, researcher, policy makers or traders) improve the quality of their decision. It also implies that effective solutions of the past may not work in the present due to climate uncertainties and declining soil fertility. Under such conditions the traditional prescriptive approach will not work rather by an ability to analyse and understand specific situations thereby offering alternative options to solve problems or exploit opportunities in a sustainable manner (Bouma and Jones, 2001). Such ability can be obtained through the use of DSTs. They can assist with the identification of options for alternative crop and soil management interventions and impact evaluation studies. The use of DSTs can, therefore, greatly facilitate participatory development and dissemination of ISFM options. DSTs that have been developed in the past 20 years range from sophisticated computer models to simple tables that help provide answers to questions such as “What are best-bet options related to cultivar choice and use of mineral fertilizer for cowpea production on a degraded sandy soil?” Some decision tools are very simple to use and require a very limited amount of data, whereas others are more complex and can only be used by trained researchers (Matthews and Stephens 2002). Even though these tools differ in data requirements and potential users, the ultimate beneficiary will always be the farmer. Bontkes and Wopereis (2003b) stated that for DSTs to be useful to the farmers, information generated must be transformed in such a way that it is easy to use and relevant for them. To be relevant, it should address the peculiarities of the problem and provide the farmer with alternative options to solve his/her problem. For example, the information may include fertilizer recommendations adapted to the soil, crop, and the amount of money the farmer is
22
willing to invest in fertilizer, or advice on the best combination of sowing date and variety. 2.6
Crop Simulation Models
Crop Simulation Models (CSMs) study crop performance under wide range of conditions and provide means of quantifying dynamics in crop yield responses over a given time within a given location (Fosu-Mensah, 2011). It further helps to identify options for crop management and assessing the risks associated with each option (Jones et al., 2003). Nix (1984) criticized the predominance of a “trial and error” approach in agricultural research for evaluating management practices. He emphasized the need for a systems approach in which: i) experiments are conducted over a range of environments; ii) a minimum set of data is collected in each experiment; iii) cropping system models are developed and evaluated; and iv) models are used to simulate production technologies under different weather and soil conditions so as to provide a broad range of potential solutions for farmers. The basic concept of crop modelling is that simulating crop growth and yield using dynamic crop models will produce results that represent how a real crop growing under specific environment and management conditions would perform. Many papers have been published on research, demonstrating that cropping system models perform adequately for the intended purposes, such as to study impacts of different cultivars, irrigation, fertility, and cultural management practices on yield and other predicted outputs (Cheryl et al., 2009; Dzotsi et al., 2010; White et al., 2011). They can only be used successfully if the newly introduced varieties are well described (both genetically and phonologically) and their respective genetic coefficients made available.
23
Nix (1984) referred to the high cost of field experiments in addition to their limited extrapolation domain because results are site-specific. However, there are practical limitations that must be considered before making use of this approach in any study. Jones et al. (2012) conducted a global sensitivity analysis using DSSAT cropping system model in three contrasting production situations: irrigated high nutrient inputs (Florida, USA), rainfed crops with manure application (Damari, Niger) and no nutrient inputs (Wa, Ghana) and concluded that the use of crop models under lowinput, degraded soil conditions requires accurate determination of soil parameters for reliable yield predictions. According to Bationo et al. (2012), one main limitation is that crop models do not contain all of the factors in the field that may influence crop yield. For example, crop diseases, weeds, and spatial variability of soils and management implementation can cause large differences in yield, and these factors are seldom included in crop simulation analyses. Another limitation is that inputs must be accurate or else simulated outputs are unlikely to match observations from the field. Crop simulation models have been used to access yield gaps in peanut production in the Guinea Savanna zone of Ghana (Naab et al., 2004). Moeller (2004) used a crop simulation model (CSM) as a tool to analyse the sustainability of wheat-chickpea rotation. Nutrient use efficiency and water productivity were also analysed for a semi-arid region in Ghana (MacCarthy et al., 2010) using CSM. CSMs require information on soil, crop management, crop cultivar specific coefficients and climatic information (daily maximum and minimum temperature, solar radiation and rainfall) to simulate crop and soil processes and to predict yield. Cropping systems are highly complex due to the many biological, chemical, and physical processes that affect the productivity of a crop in response to its 24
environment and management. One can generally consider a cropping system to be composed of a crop and the soil on which it is grown, but the environment (physical, chemical, and biological) as well as management actions are also integral determinants of a cropping system‟s performance (Porter et al., 2010). Because of these inherent complexities and a lack of knowledge about how many of these factors interact to affect crop production, modelling gained fame during the last 40 years. These models, such as DSSAT (Jones et al., 2003; Hoogenboom et al., 2004), APSIM (McCown et al., 1996; Keating et al., 2003); STICS (Brisson et al., 2003) and QUEFT (Janssen et al., 1990; Smaling and Janssen 1993) are usually able to approximate crop and soil dynamics for a rather narrow range of factors that influence soil and crop growth processes under limited conditions. Although crop models were not originally developed for use in climate change research, they have been widely used for this purpose (Rosenzweig et al., 1995). They are well suited for these studies because they incorporate the effects of daily weather conditions on crop growth processes, predicting daily growth and development and ultimately crop yield. By simulating a crop grown in a particular soil, under specified management practices, and using a number of years of daily historical weather data at a site, one obtains an estimate of how a particular management system would perform under current and changed climate conditions.
2.6.1 Decision Support System for Agro-technology Transfer DSSAT is a computational system that includes a data base management system, crop models, and application programmes. It was originally developed by an international network of scientists, cooperating in the International Benchmark Sites Network for Agro-technology Transfer project (Uehara, 1998 and Jones et al., 1998),
25
to facilitate the application of crop models in a systems approach to agronomic research. It was specifically designed to answer „„what if‟‟ questions frequently asked by policy makers and farmers concerned with sustaining an economically and environmentally safe agriculture (Tsuji et al., 1994) and the need to integrate knowledge about soil, climate, crops, and management for making better decisions about transferring production technology from one location to others where soils and climate differed (IBSNAT, 1994; Uehara and Tsuji, 1998). It simulates growth, development and yield of a crop growing on a uniform area of land under prescribed or simulated management as well as the changes in soil water, carbon, and nitrogen that take place under the cropping system over time thereby helping decision-makers reduce the time and human resources required for analysing complex alternative decisions (Tsuji et al., 1998). MacCarthy et al. (2010) used CERES-sorghum model to evaluate impact of varied weather conditions on crop productivity and also mineral fertilizer and water productivity of sorghum in both the homestead and the bush fields in Navrongo, Upper East of Ghana. The DSSAT model is quite complex requiring the minimum data set for operation which encompasses data on the site where the model is to be operated, on the daily weather during the growing cycle, on the characteristics of the soil at the start of the growing cycle or crop sequence, and on the management of the crop (e.g. seeding rate, fertilizer applications, irrigations). Dzotsi et al. (2003) used DSSAT to identify the optimum combination of sowing date and cultivar choice of maize for two agroecological zones in Southern Togo. They went further to extrapolate these results for the whole of southern Togo using Geographic Information System (GIS). The required weather data encompass daily records of total solar radiation incident on the top of the crop canopy, maximum and minimum air temperature above the 26
crop, and rainfall. All aspects of crop management including modifications to the environment (e.g. photoperiod extension) as imposed in some crop physiology studies, are needed. Typical crop management factors include planting date, planting depth, row spacing, plant population, fertilization, irrigation and inoculation. Model evaluation and testing involves comparison of model outputs with real data and a determination of suitability for an intended purpose. Fosu et al. (2012) evaluated the response of maize to nitrogen in the northern Guinea Savanna agroecology of Ghana using the seasonal analysis component. The model accurately predicted maize grain yield up to 90 kg ha-1 nitrogen application but failed to accurately predict maize grain yield when nitrogen was applied at 120 kg ha-1. This was attributed to excessive water stress induced by high N application and therefore concluded that DSSAT-CSM can be used to accurately predict maize growth, development and yield in Ghana if well calibrated. After calibrating and validating for a specific environment, the model can be used to assess alternative management choices. Prior to the development of the DSSAT, crop models were available, but these were used mostly in laboratories where they were created (Jones et al., 2003). Most crop models implemented in DSSAT, the CERES models for maize and wheat; SOYGRO for soybean and PNUTGRO peanut models, were already enjoying early successes. Those models required different file and data structures and had different modes of operation. The decision to make these models compatible led to the design of the DSSAT and the ultimate development of compatible models for additional crops, such as potato, rice, dry beans, sunflower, and sugarcane (Hoogenboom et al., 1994; Hoogenboom et al., 1999a). In addition, the programmes contained in DSSAT allow users to simulate options for crop management over a number of years to assess the 27
risks associated with each option. DSSAT therefore includes models for several cereal crops (maize, wheat, sorghum, millet, rice and barley), legume crops (soybean, peanut, dry bean, chickpea, cowpea, velvet bean and faba bean) and cassava. The grain legume models operate using a generic grain legume model, called CROPGRO.
2.6.2 DSSAT - CROPGRO The CROPGRO is one of the crop simulation models that is included in the Decision Support System for Agro-technology Transfer (DSSAT) (Hoogenboom et al., 1999b; Jones et al., 2003) and has been used in many applications around the world (Tsuji et al., 1994). CROPGRO model was created due to the need to have one common programme with values from files providing information for each species to be modelled rather than having separate models like SOYGRO, PNUTGRO and BEANGRO which have different modes of operation (Hoogenboom et al., 1994). CROPGRO was then developed as a generic approach for modelling crops in the sense that it has one common source code, yet it can predict the growth of a number of different crops. Currently, it simulates ten crops; including seven grain legumes (soybean, peanut, dry bean, chickpea, cowpea, velvet bean and faba bean), and nonlegumes such as tomato (Scholberg et al., 1997; Boote et al., 1998.). The model is physiologically based and simulates the productivity of legumes under various management and environmental conditions (Singh et al., 1994; Kaur and Hundal, 1999). Examples of the simulation efficiency of CROPGRO were presented by Ruíz-Nogueira et al. (2001). The model was used to evaluate soybean growth and yield under water limitation conditions for three cultivars in different climatic conditions of northeast Spain. Meireles et al. (2002) calibrated the CROPGRO-Dry bean model to quantify decreases in the yield of beans sown on 36 sowing dates in
28
Central Brazil. In a study conducted in Paraná state also in Brazil, Dallacort et al. (2005) used the model‟s simulation of the growth and development of the bean to determine the best sowing date for beans from 1980 through 2000. Boote et al. (2002) tested CROPGRO-legume model for chickpea to simulate phenology, growth and yield. The dates of physiological maturity, total dry matter (TDM) and final grain yield were predicted well using the model. Wajid et al. (2013) conducted a study with the objective to evaluate the performance of CROPGRO-chicken pea model to simulate growth, development and seed yield for promising lentil hybrids in semi-arid conditions of Punjab, Pakistan. The model was able to simulate the development and yield of promising cultivars of lentil at different nitrogen doses under local climatic conditions of Pakistan. CROPGRO chickpea model can be successfully used as a research tool to explore the effects of complex and alternate management decisions to sustain lentil production and evaluate the risks associated with adopting such decisions. Wang et al. (2003) evaluated the performance of CROPGRO – soybean model for simulating crop growth and grain yield of soybean on claypan soils of Missouri. Simulated leaf area index and grain yield agreed well with measured values during average precipitation years, but were under-estimated during extremely dry years. In the model validation, the average Mean Deviation (MD) was 64 kg ha-1, and Root Mean Square Error (RMSE) was 112 kg ha-1. Bastos et al. (2002) conducted an experiment to simulate cowpea growth and development under soil and climatic conditions in Brazil. The results showed good fit for dry matter and leaf area index reflected by the high determination coefficients (R2), which varied from 0.92 to 0.98. The difference between observed and simulated values of plant phenology varied from 0 to 3 days.
29
2.6.2.1 CROPGRO - Cowpea analysis of yield gap Identifying the yields at different production levels and quantifying the yield gaps through field experiments may involve many years of data collection on which to make meaningful inferences. Besides being time consuming and expensive, total elimination of factors other than the ones controlling growth and development and their interactions for a given production level may not be possible in field experiments (Bhatia et al., 2008). Determination of the potential yield and gaps between potential and actual yields requires a thorough understanding of crop growth and development, which in turn depends on several climatic, edaphic, hydrological, physiological and management factors. Crop simulation model predicts the crop growth, development and yield using a systems approach involving integrated knowledge of the underlying processes and interaction of different components of crop production (Boote et al., 1996). It can very well supplement the field trials in determination of potential yield and yield gap analysis. Naab et al. (2004) evaluated the CROPGRO-peanut model for its ability to simulate growth, yield, and soil water balance of peanut crop and to quantify yield losses caused by biotic and abiotic factors with different dates of sowing. The model successfully quantified the yield potential and yield gaps associated with yieldreducing stresses and crop management in the Guinea Savanna zone of Ghana. Simulated yield losses caused by water deficits were small (averaging 5–10%) for early sowing dates (late May to mid-July) and increased with later sowing dates (20 and 70% for third and fourth sowings). Yield losses due to diseases and pests were simulated as a percentage of potential yields under water-limited environments and averaged 40%, also increasing with later sowing dates.
30
Bhatia et al. (2008) analyzed the potential yield and yield gap of soybean in India using CROPGRO-soybean model. The district yield was used as actual (farmers) yield which was compared with the simulated potential yields.
The model
simulations showed 28% reduction in yield due to adverse soil moisture conditions while actual yield gap in the rainfed environment was attributed essentially due to non-adoption of improved crop management practices and could be reduced if proper interventions are made. Pathak et al. (2004) used DSSAT- CERES model to establish the yield gap for rice production in North West India. The gap between potential and on-station yields ranged from 28 to 38% and that between the potential and on-farm yields widened (48 to 68%) in all the locations studied. The yield gap was attributed to low adoption of suitable management strategies for water and Nitrogen losses. Silva (2012) studied the best sowing date for sunflower production using DSSAT OILCROP – sunflower in Brazil. The results showed that the yield gap will be nearly closed if appropriate sowing dates were used for the various genotype of sunflower. There is therefore a direct relationship between appropriate sowing date and reducing yield gap in crop production. This is to reduce the effect of climatic factor in crop production. Bhatia et al. (2006) further confirmed this while quantifying the yield gap for four legumes in India. Sowing dates for soybean was identified as an important factor which determines productivity. Sowing time for obtaining optimum yield was limited by the erratic nature of rainfall in major soybean growing regions. Bhatia et al. (2006) also estimated the total yield gaps between potential yield and farmers‟ yields of soybean, groundnut, pigeon pea and chicken pea using CROPGRO model. Total yield gap of the legumes were considered to be in similar range in the production zones (47 – 85%) with a positive relationship with rainfall. To abridge the yield gaps,
31
integrated watershed management approach comprising in-situ soil and water conservation, water harvesting and groundwater recharging for supplemental irrigation and improved crop management technologies was suggested. However, before a model is put to use, it needs to be thoroughly tested and validated for given site/region to establish its credibility (Boote et al., 1996).
2.7 Knowledge Gaps
Most existing blanket fertilizer recommendations fail to take into account the farmer‟s resource endowment which affects their ability to apply the fertilizer at the recommended rate. Providing a basket of options for farmers where fertilizer rate will be determined by the target yield will necessitate the pragmatic allocation of scarce mineral fertilizer. Nyamangara et al. (2011) suggested that agro-ecology based fertilizer recommendation can be made more effective by incorporating farmer‟s resource endowment to encourage adoption by farmers. Secondly, savanna soils in Ghana have been reported to have enough potassium to support legume production. This explains why some fertilizer recommendations for cowpea are without potassium rates focusing more on nitrogen and phosphorus.
Most farmers use past experience to determine the best sowing dates for cowpea which is comparable to a game of chance. Irregularity of rainfall is one constraint that constitutes an important risk factor which may lead to total crop failure. Accurate determination of best sowing dates requires proper use of historical rainfall data to determine the rainfall pattern. Currently there is no report of best sowing dates for cowpea production in the Guinea and
32
Sudan savanna zones of Ghana. This will help farmers target sowing so as to harvest when prices are high.
CROPGRO model has been extensively used to assess the effects of management practices and environmental conditions of legume growth and development (Wang et al., 2003). However, the model has not been evaluated for cowpea in Ghana, particularly the Guinea and Sudan Savanna zones. The reported use of CROPGRO model in Ghana was to quantify yield gaps of peanut in the Guinea Savanna zone of Ghana (Naab et al., 2004). Calibrating and evaluating CROPGRO – Cowpea model for use in the Guinea and Sudan Savanna zone of Ghana will facilitate its use to enhanced productivity. CROPGRO-Cowpea model has also not been used to quantify the yield gap associated with cowpea productions in the Guinea and Sudan savanna zones of Ghana.
BR 3267 rhizobia strain is a newly introduced inoculant in Ghana that has not been evaluated in the Upper West region. There is the need for P and K fertilizer rates that will optimize the efficiency of this inoculant.
2.8 Summary of literature review Cowpea yield in Ghana is low due to infertile soils and low fertilizer use. The main threat is the diminishing ability of the soil to support crop growth without effective soil fertility management. Although fertilizer use is paramount for increased crop yield, lack of a harmonized and site specific recommendation for farmers remains a major challenge. Crop simulation models can be used to generate fertilizer recommendation for crops using short term field trials. They are an effective tool to quantify yield gap and determine the best sowing dates for agricultural crops.
33
DSSAT - CROPGRO model has been in use in Africa to simulate crop performance but its use for cowpea fertilizer recommendation will help give a long term picture of the yield of the different fertilizer application rates. Furthermore to reduce the cost of input for the farmers, combined use of rhizobia inoculant and mineral fertilizer could be beneficial but there is need to understand the complementary role that fertilizer plays to enhance BNF and increase yield.
34
CHAPTER THREE 3.0 MATERIALS AND METHODS 3.1
Study area
The study was carried out in two locations at Dondori (Sudan Savanna zone) and Nyoli (Guinea Savanna zone) both in the Upper West region of Ghana (Figure 3.1). Dondori is located in the Lawra district of Upper West region of Ghana and lies geographically between latitudes 10o40ʹ and 11°00ʹ N and longitude 2o51ʹ and 2°45ʹ W. Nyoli on the other hand is located in the Wa West District of the Upper West region approximately between longitudes 9o 40ʹ and 9o 46ʹ N and Latitudes 2o 30ʹ and 2o 32ʹ W.
Study location
Figure 1.1 Locations of study areas
35
3.1.1
Soil type
At Dondori, the study was conducted on Dorimon soil series which is classified as Ferric Lixisol according to IUSS World Reference Base (2006). The soil is moderately well drained and located at the middle slope with Lower Birrimian rock as the parent material. The soil of the Nyoli site belongs to Varempere soil series which is classified as Ferric Luvisol according to IUSS World Reference Base (2006). The soil is moderately well drained and located at the upper slope. The parent material is Cape Coast Granite. The soil profile description and chemical properties of the profile pit are presented in Appendices. 1, 2 and 3.
3.1.2 Climate Dondori is characterized by a unimodal rainfall pattern with 2 distinct seasons (dry and wet). The dry season starts from October - April and the wet season starts from May – September. The mean annual rainfall is between 800 to 1100 mm while the annual average temperature is about 35° C for maximum and 21° C for minimum. Relative humidity is generally high (80 – 90%) in the wet season (from May to MidOctober) and low (10 – 11%) in the dry season especially in the harmathan period from November to February. Nyoli enjoys two marked seasons. The rainy season begins in May and ends in September while the dry season also begins in October and ends in April. The mean annual rainfall figures vary between 840 mm and 1400 mm. A very important feature of rainfall in the district is that it is erratic in nature i.e. it is torrential and poorly distributed. Temperatures are high in most part of the year, ranging between 22.5 ºC to 45 ºC, low between December and January, and high between March and April. Average monthly maximum temperature is 33 ºC whereas the daily highest is 35 ºC.
36
3.2
Study 1: Response of cowpea to rhizobia inoculant and mineral fertilizer application
The study consisted of a pot and field experiments.
3.2.1
Pot experiment
The pot experiment was carried out to compare the inoculated and the uninoculated treatments after which the promising treatments were selected for confirmation trials in the field. The pot experiment was conducted in the greenhouse at the Soil Research Institute, Kwadaso, Kumasi in August to September 2012. Topsoil (0 - 15 cm) collected from Lawra (Ferric Lixisol) and Nyoli (Ferric Luvisol) was air-dried and sieved. Five (5 kg) soil was weighed into pots of 4710 cm3. Thirty pots were used for each location comprising of 10 treatments and 3 replications and arranged in a Complete Randomized Design (CRD). The rhizobium inoculation was done at the rate of 5 g of inoculants per one kg of cowpea seed. The cowpea variety used was Omondaw which is an early maturing variety (65). Rhizobium inoculated seeds were spread on flat plywood to air dry for one hour before sowing. Two cowpea seeds were sowed per pot. Fertilizer was applied at planting in the form of urea, triple superphosphate and muriate of potash. The soil water content was maintained at field capacity (FC) by weighing the pots every other day and adding water to maintain the weight of the soil at FC. Harvesting was done at 60 days after sowing (DAS) by cutting the shoot at the soil level. Roots were removed, washed and the nodules collected. The treatments used were as follows:
37
N-P2O5-K2O kg ha-1 T1
0–0-0
T2
IN* – 0 – 0
T3
IN* – 45 – 20
T4
IN* – 30 – 20
T5
20 – 45 – 20
T6
10 – 15 - 10
T7
30 – 45 – 30
T8
0 – 45 – 30
T9
30 – 0 – 30
T10
30 – 45 – 0
* Rhizobia inoculant
3.2.2
Field experiment
A field experiment was conducted during the 2013 cropping season at both study locations. The field was ploughed with a tractor and then harrowed after which the field was laid out. Plot sizes measuring 4 x 6 m were demarcated. Fertilizer was applied at planting while the control plots did not receive fertilizer. Cowpea seeds were planted at 2 seeds per hill at a spacing of 60 x 20 cm. Refilling was done 7 days after planting.
3.2.3
Cowpea variety used
Omondaw cowpea variety was also used for the experiment. It was obtained from the seed unit of the Ministry of Food and Agriculture (MOFA) Wa regional office.
38
3.2.4
Experimental design and treatments
The treatments were laid out in a Randomized Complete Block Design (RCBD) with 4 replications at both study locations. Four out of the ten treatments with promising results from the pot experiment were selected for verification in the field while one of the treatments (20 – 30 – 20 kg N, P2O5 and K2O ha-1) was selected from study 2 first season results. Treatment IN*/20 – 30 – 20 kg N, P2O5 and K2O ha-1 was added to evaluate the effect of N fertilizer on the inoculant. The treatments used for this study were as follows: N-P2O5-K2O kg ha-1 T1
0–0-0
T2
IN* – 0 – 0
T3
IN* – 45 – 20
T4
IN* – 30 – 20
T5
20 – 30 – 20
T6
IN*/20 – 30 – 20
* - Rhizobia inoculants
3.2.5
Nodule number and nodule dry weight
At 50 % flowering, ten consecutive plants were randomly harvested from the row next to the border row and roots gently dug out. The plants were cut at about 5 cm above the ground. The plants roots were then washed through a 1 mm mesh sieve in water to remove soil particles after which nodules were detached. The number of nodules on each plant was then counted and recorded. The nodules were then put in an envelope and oven dried at 60 °C for 72 h and the dry weights recorded.
39
3.3
Study 2: The effect of NPK fertilizer application on the growth and yield of cowpea
3.3.1
Cowpea variety used and Land preparation and sowing
Cowpea variety (Omondaw) was used for this field study at Dondori and Nyoli respectively. Land preparation and sowing were carried out as described in section 3.2.2. The experiment was carried out during the 2012 and 2013 cropping season. 3.3.2
Experimental design and treatments
The treatments were laid out in a Randomized Complete Block Design (RCBD) with 4 replications at both locations.
3.3.2.1 Treatments The treatments were selected based on FAO (1976) cowpea fertilizer recommendation of 22 kg N, 28 – 67 kg P2O5 and 22 – 45 kg K2O ha-1 and SARI (2013) recommendation of 20 kg N ha-1. The treatments were as follows: N-P2O5-K2O kg ha-1 T1
0–0-0
T2
0 – 45 – 30
T3
30 – 0 – 30
T4
30 – 45 – 0
T5
10 – 15 – 20
T6
20 – 45 - 30
T7
20 – 30 – 20
T8
20 – 45 – 20
T9
10 – 15 – 10
T10
20 – 30 – 10
40
3.4
Measurement of crop variables
3.4.1
Biomass yield
From the two border rows on each of the treatment plots, five plants were randomly chosen and cut to the ground level for shoot dry matter determination at 50% flowering. Plant materials were then put in large brown envelopes and oven dried at 60 °C for 72 hours. The dried plant materials were then weighed and biomass dry weight determined 3.4.2
Grain yield
Cowpea was harvested at physiological maturity. Pods were removed from the plants, after harvesting. The pods were then air dried and threshed. The grains were oven dried at 60 °C for 72 h and the dry weights recorded. The dry weights were then extrapolated to estimate the grain yield per hectare. 3.4.3
Number of seeds pod-1
The pods harvested from five consecutive plants of each treatment were shelled and the seeds counted. The number of seeds for each treatment was divided by the number of respective pods to obtain the number of seeds pod-1. 3.4.4
Mean 100 seed weight
Hundred seeds from each treatment were randomly picked and weighed. This was replicated five times and the average seed weight determined. The average weight of five counts was then taken as the weight of hundred seeds for each treatment. 3.5
Nitrogen, Phosphorus and Potassium Uptake
Nitrogen, phosphorus and potassium uptake by cowpea crop was determined by multiplying the grain and biomass yields with the N, P and K concentration in the specific components.
41
3.6
Economic analysis
The economic viability of fertilizer recommendation was assessed using value cost ratio (VCR). Crop prices, fertilizer prices and the operational cost were the average prices prevailing in the study area during the trial. Gross benefit accruing from each treatment was calculated as the product of the grain yield from the treatment and the average unit price of the grains. The cost of land preparation, planting and fertilizer application did not differ among treatments and were ignored in the partial budget. The VCR was calculated as follows: (
)
Where: YF = Grain yield from plots with fertilizer YC = Grain yield from control plots PG = Unit price of grains yield PF = Unit price for fertilizer QF = Quantity of fertilizer 3.7 3.7.1
Laboratory soil analyses Soil sampling and characterization
A composite soil sample was taken from 15 different spot across each field using an auger from a depth of 0–15 cm. The soil samples were air dried and taken to the laboratory where they were sieved with a 2 mm mesh sieve. The sieved soil was thoroughly mixed and labelled for analysis. Fresh soil samples were transported in an ice pack and then stored in a refrigerator for nitrate nitrogen and ammonium nitrogen determination. A 1.5 m profile pit was dug close to the side of the experimental field and described. Soil samples were taken from each layer (0 - 10, 10 - 20, 20 - 30, 30 42
40, 40 - 50, up to 150 cm) and taken to the laboratory for chemical and physical analysis. The result of analysis of the profile samples was used for DSSAT calibration. Except where otherwise stated, all laboratory analyses reported in the following sections were carried out in duplicates. 3.7.2
Determination of soil chemical properties
3.7.2.1 Soil pH This was determined using the glass electrode pH meter in a 1: 1 soil to distilled water (soil: water) ratio. A 10 g soil sample was weighed into a 100 ml beaker. To this 25 ml distilled water was added from a measuring cylinder, stirred thoroughly and allowed to stand for 30 minutes. After calibrating the EUTECH pH meter with buffer solutions at pH 4.0 and 7.0, the pH was read by immersing the electrode into the upper part of the suspension.
3.7.2.2 Available phosphorus Available P was determined by the Bray and Kurtz (Bray P-1) method (1945). Five grams of soil was weighed and transferred into a 50 ml centrifuge tube. Thirty ml of Bray-1 extracting solution (0.025 N HCl + 0.03N NH4F) was added. Soil suspension was shaken for five minutes via a mechanical reciprocating shaker and allowed to stand for 2 minutes and then centrifuged for 10 minutes at 3000 rpm. Working standards in Bray 1 extractant using 5 clean 250 ml volumetric flasks were prepared. 0, 2, 4, 8, 12, 16 and 20 ml of stock 250 mg P ml-1 of KH2PO4 (A.R. grade) solution were pipetted into each 250 ml volumetric flask and made up to the 250 ml mark using Bray-1 solution. The working standards contain respectively 0, 1, 2, 4, 6, 8 and 10 mg P ml-1 in 250 ml volumetric flasks.
43
One ml of the clear supernatant solution (sample), blank and the standard solutions were pipetted into a set of clean 15 ml centrifuge tubes. Six ml of distilled water was added and mixture shaken vigorously followed by the addition of 2.0 ml of molybdate-HCl reagent. Finally, 1.0 ml of 1.76 % solution of ascorbic acid (reducing reagent) was added to the mixture and was vigorously shaken. The mixture was allowed to stand undisturbed for 6 minutes for development of the blue coloration after which the percent transmittance values were recorded at 650 nm wavelength on a colorimeter or visible range spectrophotometer. A graph of absorbance against concentration (mg kg-1) P was plotted. The unknown samples were read and mg kg-1 P obtained by interpolation on the graph plotted. The P content was determined by comparing the recorded values to a standard curve plotted using standard P solutions after the percent transmittance (% T) was converted to absorbance by the formula: Absorbance = 2- log T. Calculation:
where: C = Concentration derived from the standard curve
3.7.2.3 Soil organic carbon The modified Walkley and Black procedure as described by Nelson and Sommers (1982) was used to determine organic carbon. The procedure involved a wet combustion of the organic matter with a mixture of potassium dichromate and sulphuric acid after which the excess dichromate was titrated against ferrous sulphate. Two grams of soil sample was weighed into a 500 ml erlenmeyer flask. A
44
blank was included. Ten millilitres of 0.1667 M (1.0 N) potassium dichromate solution was added to the soil and the blank flask followed by 20 ml of concentrated sulphuric acid. The mixture was swirled and allowed to stand for 30 minutes on an asbestos sheet. Distilled water (200 ml) and 10 ml concentrated orthophosphoric acid were added and allowed to cool. One millilitre of diphenylamine indicator was added and titrated with 1.0 M ferrous sulphate solution. Calculation: Organic
(
) wt. of soil
100
where: m.e = milliequivalent = Normality of solution × ml of solution used 0.003 = m.e. wt of C in grams (12/4000) f = correction factor = 1.32 3.7.2.4 Total nitrogen The macro Kjeldahl method involving digestion and distillation as described by Soil Laboratory Staff (1984) was used in the determination of total nitrogen. A 10 g air dried soil sample was weighed into a Kjeldahl digestion flask and 10 ml distilled water added to it for 10 minutes to moisten. Thirty millilitres concentrated sulphuric acid and one spatula full selenium mixture were added, mixed carefully and digested for 2 hours until a clear and colourless digest was obtained. The digest was allowed to cool and then decanted into a 100 ml volumetric flask which was made up to mark with distilled water. A 10 ml aliquot of the digest was transferred to the Kjeldahl distillation apparatus and 20 ml of 40% NaOH solution was added followed by distillation. The distillate was collected in 10 ml of 4% boric acid. Using bromocresol green and methyl red as indicator, the distillate was titrated with 0.1 N HCl till blue colour changes to grey and then suddenly flashes to pink. A blank
45
distillation and titration was also carried out to take care of traces of nitrogen in the reagents as well as the water used. Calculation: (
)
Where: N = Normality of HCl used in titration A = ml HCl used in sample titration B = ml HCl used in blank titration 1.4 = 1.4 × 10-3 × 100% (14 = atomic weight of nitrogen) V = total volume to digest s = mass of air dry soil sample taken for digestion in g (10 g) t = volume of aliquot taken for distillation (10 ml)
3.7.2.5 Determination of exchangeable cations Exchangeable bases (calcium, magnesium, potassium and sodium) content in the soil were determined in 1.0 N ammonium acetate (NH4OAc) extract (Black, 1986) and the exchangeable acidity (hydrogen and aluminium) was determined in 1.0 N KCl extract (Page et al., 1982). 3.7.2.5.1
Extraction of the exchangeable bases
A 10 g soil sample was weighed into an extraction bottle and 100 ml of 1.0 N ammonium acetate solution was added. The bottle with its contents was shaken for one hour. At the end of the shaking, the supernatant solution was filtered through No. 42 Whatman filter paper.
46
3.7.2.5.2
Determination of exchangeable calcium and magnesium
A 10 ml portion of the extract was transferred into a conical flask and 5 ml of ammonium chloride – ammonium hydroxide buffer solution was added followed by 1 ml of triethanolamine. Few drops of potassium cyanide and Eriochrome Black T solutions were then added. The mixture was titrated with 0.02 N EDTA solution from a red to a blue end point. Calculations: (
)(
)
(
)
where: g = mass (g) of air dry soil used in the extraction va = ml of 0.02 N EDTA solution used in the sample titration vb = ml of 0.02 N EDTA solution used in the blank titration 0.02 = concentration of EDTA 1000 = conversion factor of g to Cmol+ kg-1 3.7.2.5.3
Determination of exchangeable calcium and magnesium
For the determination of calcium, a 10 ml portion of the extract was transferred into a conical flask. To this, 10 ml of potassium hydroxide solution was added followed by 1 ml of 30 % triethanolamine. Three drops of potassium cyanide solution and a few crystals of cal-red indicator were then added. The mixture was titrated with 0.02 M EDTA (ethylene diamine tetra-acetic acid) solution from a red to a blue end point. Exchangeable magnesium was calculated by subtracting the value obtained from calcium alone from the calcium + magnesium value.
47
3.7.2.5.4
Determination of exchangeable potassium and sodium
Potassium and sodium in the soil extract were determined by flame photometry. Standard solutions of 0, 2, 4, 6, 8 and 10 ppm K and Na were prepared by diluting appropriate volumes of 100 ppm K and Na solution to 100 ml in volumetric flask using distilled water. Flame photometer readings for the standard solutions were determined and a standard curve constructed. Potassium and sodium concentrations in the soil extract were read from the standard curve. Calculation:
(
)
(
)
where: W = weight of air – dried sample soil in grams 39.1 = mole of potassium 23 = mole of sodium 3.7.2.5.5
Determination of exchangeable acidity
This consists of aluminium (Al3+) and hydrogen (H+). Five grams of soil sample was put into a shaking bottle and 100 ml of 1.0 M KCl solution added. The mixture was shaken for 2 hours and then filtered. Fifty millilitres portion of the filtrate was transferred into an Erlenmeyer flask and 3 drops of phenolphthalein indicator solution added. The solution was titrated with 0.05 M NaOH until the colour just turned permanently pink. The amount of base used was equivalent to total acidity (H+ + Al3+). A few drops of 0.05 M HCl were added to the same mixture to bring the solution back to colourless condition and 10 ml of 1.0 M sodium fluoride (NaF)
48
solution added. The solution was then titrated with 0.05 M HCl until the colour disappeared permanently. The milliequivalents of acid used are equal to the amounts of exchangeable Al. The amount of H was determined by difference. Calculation:
where: 0.05 = molarity of NaOH or HCl used for titration V = ml NaOH or HCl used for titration W = weight of air – dried soil sample in grams 3.7.2.5.6
Calculation of effective cation exchange capacity (ECEC) and
percent base saturation Effective cation exchange capacity was calculated by the sum of exchangeable bases (Ca, Mg, K, and Na) and exchangeable acidity (Al and H). Percent base saturation was calculated from the sum of exchangeable bases as a percent of the ECEC of the soil.
3.7.2.6 Determination of NH4+ - N The Berthelot procedure as outlined by Kempers and Zweers (1986) was used. The procedure is based on the reaction in which a phenol derivative forms an azo dye in the presence of ammonia and hypochlorite. In this method salicylic acid is used as the phenol source. The end product is an indophenol derivative which in the presence of an alkaline medium is a greenish-blue colour which can be measured at 660 nm wavelength on a visible wavelength range spectrophotometer. The intensity of the colour depends on the quantity of ammonium ion or ammonia present. 49
Working standards of 0, 5, 10, 15, 20, and 25 mg NH4+-N l-1 were prepared from a 1000 mg NH4+-N l-1 stock standard. A solution called colour reagent 1 (R1) was prepared by measuring out 50 ml sodium salicylate [prepared by dissolving 110g salicylic acid in 10 M NaOH] plus 100 ml of 0.5% sodium nitroprusside and 5 ml of 4% Na2EDTA. Colour reagent 2 (R2) was prepared by weighing 0.2g of sodium dichloroisocyanurate in 5 ml of distilled water and transferring it into 200 ml volumetric flask and making it up to the mark with di-sodium hydrogen phosphate (Na2HPO4.12H2O) buffer solution of pH 12.3. The buffer was made by dissolving 26.70 g of Na2HPO4.12H2O in a two litre of volumetric flask and making up to mark with distilled water after adjusting it to pH 12.3. One millimetre of sample and standard series were pipetted into 5 ml volumetric flask and then 3 ml of R1 was added followed by 5 ml of R2 and distilled water added to the mark. This was left to stand for two hours for maximum colour development. The colour intensity of the solution was measured at 660 nm wavelength on spectrophotometer. Calculation: (
)
where: a = NH4+-N L-1 of sample b = NH4+ N L-1 blank V = volume of extract df = dilution factor g = weight of soil used for the extraction.
50
3.7.2.7 Determination of NO3--N The colorimetric method of Cataldo et al. (1975) was used. Salicylic acid was reacted with the nitrite in the presence of NaOH to form a yellow colour. The intensity of the colour is a measure of the nitrite content in solution. A stock standard of 1000 mg NO3- -N L-1 was prepared by dissolving 7.223 g of potassium nitrate in a litre of volumetric flask with distilled water. A sub-standard solution of 50 mg NO3- N L-1 was prepared from the 1000 mg NO3- -N L-1 stock solution and from this a standard series of 0, 2, 5, and 10 mg NO3- -N L-1 was prepared. Other solutions prepared were 5% salicylic solution (by dissolving 5 g of salicylic acid in 95 ml of concentrated sulphuric acid) (R1) and 4 M NaOH (R2). One millimeter each of the standard series and sample extracts were pipetted into 25 ml volumetric flask, then 1 ml of R1 was added and left to stand for 30 minutes. Ten (10) ml of R2 was then added and left to stand for 1 hour for full colour development. Colour intensity was measured at 410 nm wavelength on spectrophotometer. Calculation: (
)
where: a = NO3- - N L-1 of sample b = NO3- - N L-1 blank V = volume of extract df = dilution factor g = weight of soil used for the extraction
51
3.7.2.8 Soil physical analyses 3.7.2.9 Determination of soil particle size This was determined by the Bouyoucos hydrometer method (Bouyoucos, 1963). A 40 g soil was weighed into 250 ml beaker and oven dried at 105oC overnight. The sample was removed from the oven and placed in a desiccator to cool, after which the oven dry weight was taken. A 100 ml of dispersing agent (sodium hexametaphosphate) was added to the soil. It was then placed on a hot plate and heated until the first sign of boiling was observed. The content of the beaker was weighed into a shaking cap and fitted to a shaking machine and shaken for 5 minutes. The sample was sieved through a 50 μm sieve mesh into a 1.0 L cylinder. The sand portion was dried and further separated using graded sieves of varying sizes into coarse, medium, and fine sand. These were weighed and their weights taken. The 1.0 L cylinder containing the dispersed sample were placed on a vibration - less bench and then filled to the mark. It was covered with a watch glass and allowed to stand overnight. The hydrometer method was used to determine the silt and the clay contents. The cylinder with its content was agitated to allow the particles to be in suspension. It was then placed on the bench and hydrometer readings taken at 40 seconds and 6 hours interval. At each hydrometer reading, the temperature was also taken. The percent sand, silt and clay were calculated as follows: % Clay = corrected hydrometer reading at 6 hours x 100/weight of sample % Silt = corrected hydrometer reading at 40 seconds x100/weight of sample % clay. % Sand = 100 %- % silt - % clay The texture was determined using the textural triangle.
52
3.7.2.10
Determination of soil moisture content
The soil moisture content (θm) was determined by the gravimetric method (Marshall and Holmes, 1988). In this method, the loss in weight after oven – drying as a fraction of the oven – dried soil represents the moisture content. A moisture can with its lid was dried in oven at 105℃ to a constant weight, allowed to cool and its weight recorded. A 10 g soil was put into the can, covered and the weight taken. The lid was removed and placed under the can. The can with soil and the lid was placed in the oven and dried at 105 ℃ for about 24 hours to a constant weight. The can was removed from the oven, covered with the lid, allowed to cool and the weight taken. Calculation:
(
)
where: W1 = Weight of empty can + lid W2 = Weight of can + lid + fresh soil W3 = Weight of can + lid + oven – dried soil
3.7.2.11
Determination of soil bulk density
The bulk density (ℓb) was determined using the metal core sampler method (Blake and Harte, 1986). The core sampler was driven into the soil with the aid of a mallet. Soil at both ends of the tubes was trimmed and the end flushed with a straight-edged knife. The core sampler with its content was dried in the oven at 105 oC to a constant weight, removed, allowed to cool and its weight taken. The weight of the core cylinder and its volume was determined.
53
Calculation: ℓ (
)
where: W2 = Weight of core cylinder + oven-dried soil W1 = Weight of empty core cylinder V
Volume of core cylinder (πr2h), where:
π
3.142
r = radius of the core cylinder h = height of the core cylinder
3.7.2.12
Determination of volumetric moisture content
The volumetric moisture content (θv) was calculated by multiplying the gravimetric moisture content by the dry bulk density. Calculation: (θ )
θ ℓ
ℓ
where: θm
gravimetric moisture content
ℓb = dry bulk density ℓs = Particle density with value of 2.65 gcm-3
3.7.2.13
Porosity
Porosity (ƒ) was computed from the relation: ( )
ℓ ℓ
where: 54
ℓb = dry bulk density ℓs = particle density, with a value of 2.65 g cm
3.8
-3
Data analysis
All data collected was analysed with Genstat 9th edition (2007) using general linear model (GLM) analysis of variance (ANOVA) with randomized blocks procedure. Analysis of variance was combined over the two locations with replications nested in locations. Location therefore became a blocking component while fertilizer rates define the treatment structure. Where the CV was high, variance component analysis was performed using the method of Residual Maximum Likelihood (REML). REML estimates the treatment effects and variance components in a linear mixed model with both fixed and random effects. Means were separated using the Standard Error of Difference (SED) of the means at P = 0.05. Pre-conceived orthogonal contrast was used for treatment comparison. Canonical correlation analysis (CCA) was used to determine associations among group of variables. Relationship between the associated variables was determined with Multiple regression. Multivariate analysis of variance (MANOVA) was also carried out to further establish if there were statistically significant differences among the group of variables. 3.9
Model Simulation and Analyses Procedures
3.9.1
Weather data
Daily data for rainfall and minimum and maximum air temperature and sunshine hours for Lawra and Nyoli were collected from the Wa regional meteorological agency WA. For Nyoli, data from 2002 – 2013 was used. Rainfall (2008 – 2013) for Lawra was however collected from the Babile Sub- station of the Wa metrological agency which was about 1 km to the study site. The collected data were evaluated for
55
errors and missing values through a concise data check. The data was loaded into the WEATHERMAN utility program of the DSSAT and SIMMETEO (weather generator program component of WEATHERMAN) was used to generate weather data for 20 years at both sites. The WEATHERMAN was also used to estimate solar radiation (MJ/m2) from bright sunshine hours obtained from the weather station. The generated data was used to run the seasonal analysis program on the model. 3.9.2
Model calibration
A calibration of a model is the adjustment of some parameters and functions of a model so that the predictions are the same or at least very close to data obtained from field experiments (Penning de Vries et al., 1989). Calibration was first conducted for soil parameters, for the genetic coefficient for the cowpea variety and then for the experimental data.
3.9.2.1 Soil parameters For the calibration of soil parameters, soil data collected from each soil horizon was entered into the soil file creation utility (SBuild). In the DSSAT model, volumetric water content in each soil layer varies between a lower limit (LL) to which plants can extract soil water and a saturated upper limit (SAT) as described by Ritchie (1985). The water content at field capacity or drained upper limit (DUL), lower limit of plant available soil water (LL), saturated upper limit (SAT), saturated hydraulic conductivity (Ksat) and root growth factor (RF) for each soil layer were estimated by the model. Data used for the estimation included; soil particle size distribution, bulk density and organic carbon contents of the soil. The soil fertility factor (SLPF) was set to 0.8 (Pedersen et al., 2004) which is less than the standard SLPF value of 1 used
56
for soils without fertility limitations. SLPF is an input variable (constant for a given site) that affects crop growth rate by modifying daily canopy photosynthesis.
3.9.2.2 Genetic coefficients The calibration procedure of the CROPGRO-COWPEA model consisted of making initial estimates of the genetic coefficient and running the model interactively, so that simulated values match as closely as possible the measured data. The genetic coefficients describe the durations of phases of the crop life cycle, vegetative growth traits, and reproductive traits unique to a given cultivar (Boote et al., 1998). As these were not available in the model, cowpea cultivar TVU3644 originating from Nigeria was used as a starting point from which to calibrate the Omodaw cowpea cultivar. Genetic coefficient for omondaw was collected from the seed unit of the Ministry of Food and Agriculture Wa regional office. Other genetic coefficients (Table 3.1) were estimated from the field experiment. The coefficients used for the modelling are presented in Table 3.1. First, coefficients for duration to flowering (EM-FL), beginning pod (FL-SH) beginning seed (FL-SD), and maturity (SD-PM) were adjusted to predict the observed life cycle from data collected. FL-LF is the time between first flower (R1) and end of leaf expansion (photo thermal days). Photo thermal day requirements are equivalent to calendar days, if the temperature is at the optimum 28 oC for the entire 24 h day. The coefficients for flower to first seed (FL - SD) and first seed to maturity (SD - PM) were adjusted to reproduce duration of growth stages observed by cowpea breeders for the cultivar. A parameter for rate of early leaf area growth (SIZELF) was changed to mimic rate of early - season dry matter gain as Omondaw is an early maturing cultivar (60 – 65 days). Reproductive coefficients for the timing of pod set,
57
seed set, rate of pod addition, maximum partitioning intensity to pods - seeds, and single seed fill duration (FL-SH, FL-SD, PODUR,XFT, SFDUR) were modified to simulate pod and seed growth, especially the time series of pod harvest index. The genetic coefficients that determine the vegetative and reproductive growth duration were set to the default value. These are LFMAX-Maximum leaf photosynthesis rate and high light; and SLAVR- Specific leaf area of cultivar under standard growth conditions). Other genetic coefficients were determined from the observed value at the field which includes: WTPSD-Maximum weight per seed (g), SDPDV-Average seed per pod under standard growing conditions and THRESH-the maximum ratio of seed to shell at maturity (Causes seed to stop growing as their dry weights increase until shells are filled in a cohort) The coefficients determining the nutritional value SDPRO
(Fraction protein in seeds)
and SDLIP (Fraction oil in seeds) was left at the default value as this was the standard practice.
58
Table 3.1 Genetic coefficients of Omondaw cowpea used for model simulations Genetic coefficient
Ab
Values
Critical Short Day Length below which reproductive Development
CSDL
14.80
Relative response slope of development to photoperiod with time
PPSEN
.1576
Time between plant emergence and flower appearance (R1) (Ptd)
EM-FL
33.00
Time between first flower and first pod (R3) (Ptd)
FL-SH
2.50
Time between first flower and first seed (R5) (Ptd)
FL-SD
5.500
Time between first seed (R5) and physiological maturity (R7) (Ptd)
SD-PM
13.20
Time between first flower (R1) and end of leaf expansion (Ptd)
FL-LF
15.00
Maximum leaf photosynthesis rate (mg CO2m s )
LFMAX
1.000
Specific leaf area of cultivar standard growth conditions (cm2 g-1)
SLAVR
300.0
Maximum size of full leaf (three leaflets) (cm2)
SIZLF
270.0
Maximum fraction of daily growth partitioned to seed + shell
XFRT
1.000
WTPSD
0.1120
SFDUR
12.5
Average seed per pod under standard growing conditions (no.pod )
SDPDV
15.0
Time required for cultivar to reach final pod load under
PODUR
8.5
The maximum ratio of seed to seed + shell at maturity.
THRSH
66.0
Fraction protein in seeds (g protein g-1 seed)
SDPRO
.300
Fraction of oil on seeds (g oil g-1 seed)
SDLIP
.065
progresses with no day length effect
-2 -1
Maximum weight per seed (g) Seed filling duration for pod at standard growth conditions (Ptd) -1
Ptd = Photo thermal days, Ab = Abbreviation
3.9.2.3 Experiment calibration The model was calibrated using the water limited yield in addition to data from 2 treatments; 0-45-30 and 20-45-30 N P2O5 K2O respectively. The model assumes that P and K were sufficiently supplied to the soil. Observed water limited values were obtained by taking the maximum values obtained from the mean value of each treatment (mean of all replicates) per site and per year. The yield data was used as input parameters to calibrate the cowpea model. To further calibrate the model,
59
rhizobium effectiveness (under initial condition tab in the experimental file) was adjusted from its default value of 1 (used for best soils in the world) to 0.70. The Mann - Whitney U test was used to test for statistical differences between the observed and the simulated data. 3.9.3
Seasonal analysis
Seasonal analysis in DSSAT is used to examine the year to year variation in crop productivity due to climate. Biophysical analysis component of the seasonal analysis program was used to simulate the optimal N rate for the yield of cowpea and also to determine the best sowing dates. It was assumed that growth is limited by shortage of nitrogen and water for some part of the plant life. This is due to the limitation of the model in simulating phosphorus and potassium for cowpea. The dates used to determine the best sowing dates for the first simulation was from June 01 – August 30 at ten days interval. The sowing dates were later narrowed down to July 01 and August 16 (Table 3.2). Table 3.2 Dates used to estimate the best sowing dates for cowpea at Lawra and Nyoli June 01 – August 30
July 01 and August 16
June 01
July 01
June 11
July 06
June 21
July 11
July 01
July 16
July 11
July 21
July 21
July 26
July 31
July 31
August 10
August 05
August 20
August 11
August 30
August 16
60
All the data on climate, soil and crop growth collected from the two sites were entered in the standard file formats (*.PNX, *.PNA, *.PNT, *.WTH, and SOIL.SOL) needed for execution of the CROPGRO-cowpea model. 3.9.4
Estimation of potential yield and yield gap
CROPGRO-cowpea model was used to simulate the potential yield of Omondaw cowpea cultivar for various sowing dates and also to estimate the yield gap. Potential yield is the maximum yield of a crop restricted only by climatic conditions (Pathak et al., 2004). This assumes that other inputs (nutrient, water, pests, etc.) are not limiting and cultural management is optimal. Thus the potential yield of a crop is dependent upon temporal variation in solar radiation and maximum and minimum temperatures during the cropping season, and physiological characteristics of the variety. This was simulated by switching “Off” the water and N in the simulation control section of Sbuild. The average cowpea grain yield for Nyoli and Lawra was obtained from the extension department of the Ministry of Food and Agriculture, Wa regional office. This consists of average yield of cowpea from farmers‟ field with no intervention from researchers. The research yield was obtained from the researcher managed experimental plot laid on a sub - sample of the farmers‟ field used for the study. Yield gaps were estimated for each site by subtracting the district farmers‟ yield from research yield and potential yield. 3.9.5
Model validation
The CROPGRO – cowpea model was validated by comparing the observed data with the simulated data for the grain yield in 2012 and 2013 cropping season. The data used for the validation were the potential and water limited yield, 0-45-30 and 20-4530 N, P2O5 and K2O respectively. For water limited yield scenario, growth is assumed to be limited by water availability at least for a part of the plant life thus
61
decreasing crop growth rate and yield (Bhatia et al., 2008). Rainfed or partially irrigated crops with adequate nutrients are examples of this production system. For the simulation of water limited yield, only the water balance switch of the model was activated. 3.9.6
Statistical evaluation of model performance
An analysis of the degree of coincidence between simulated and observed values was carried using statistical methods which include: 1. Root Mean Square Error - This reflects the magnitude of the mean difference between predicted and observed values and it is a good measure of how accurately the model predicts the response. The value of the RMSE should approach zero (Pathak et al., 2004). (
)
where: Yield sim = simulated yield Yield meas = Measured yield 2. Coefficient of variation of the root mean square error - is the RMSE normalized to the mean of the observed values. It is the same concept as the coefficient of variation except that RMSE replaces standard deviation. It is expressed as a percentage, where lower values indicate less residual variance.
62
(
)
where: CV = Coefficient of variation RMSE = Root Mean Square Error m.obs = Mean observation
3. Wilmott index of agreement - Index of agreement (d) was chosen instead of Pearson‟s product-moment correlation coefficient. Wilmot (1981) observed that product-moment correlation coefficient often is unrelated to the sizes of the differences between the observed and the predicted observations. A value of 1 for the index of agreement (d) indicates a good agreement between the simulated and observed data while values closer to 1 indicate better prediction. A d value of zero indicates no predictability (Ortiz et al., 2009).
(
( (| |
) ) | |)
where: Pi = Predicted/simulated value Oi = Observed values n = Number of observations
63
CHAPTER FOUR 4.0 RESULTS AND DISCUSSION 4.1
Selected initial physical and chemical soil properties of the experimental sites
4.1.1
Results
The initial soil physical and chemical properties of the experimental sites are shown in Table 4.1. Results showed that soils from both study sites were sandy loam textured. Table 4.1 Selected initial soil physical and chemical properties of the experimental sites Soil Parameter
Ferric Lixisol
Sand (%)
(Lawra) 57.00
Ferric Luvisol (Nyoli) 60.00
Silt (%)
38.00
38.00
Clay (%)
5.00
2.00
pH (1:1 H2O)
5.78
6.52
Org. C (%)
0.46
0.90
Total N (%)
0.03
0.08
1.79
16.98
Ca2+
3.60
9.00
Mg2+
1.7
3.60
K+
0.11
0.23
Na+
0.15
0.18
Total exchangeable bases
5.56
12.90
1.20
1.60
H+
1.00
1.20
ECEC
7.76
15.70
Available P (mg kg-1) -1
Exchangeable cations (cmol+ kg )
Al
3+
64
Total nitrogen values for the Ferric Lixisol (Lawra) and the Ferric Luvisol (Nyoli) were 0.07 and 0.08% while the available P values were 1.79 mg kg-1 and 16.98 mg kg-1 respectively. Both values are below the critical levels (N < 0.10 % and P
