| dc.description.abstract | Poor agricultural productivity and food security remain challenging problems for the majority of smallholder famers in Sub-Saharan Africa, including semi-arid Eastern Kenya. However, there is a general consensus that there is urgent need to significantly increase food production to meet the growing demand aligned with the continuing population growth. Furthermore, the intensification and stabilization of agricultural productivity of small-scale farming systems in Sub-Saharan Africa holds a key position to contribute to the economic development and reduce poverty. The major driver for declining or stagnating agricultural productivity in many parts of Sub-Saharan Africa, such as semi-arid Eastern Kenya, is the decline in soil fertility. Food production is not keeping pace with rapid population growth, forcing farmers to change their traditional farming systems characterized by shifting cultivation, fallow and the use of animal manure. Land and labour restrictions, as well as an increased limited resource endowment further impose the mainly smallholder farmers to focus on the production of staples, such as maize in Eastern Kenya. The investment in soil fertility management strategies remains low and the change from traditionally diverse farming systems to cereal-based monocultures has further increased the susceptibility of the fragile production systems, in particular, to impacts of climate change and variability. The predicted increase in temperature as well as inter- and intra-seasonal rainfall variability will additionally challenge the largely rainfed smallholder farming systems to sustain their productivity in the future. The integration of legumes within the farming system has been part of traditional soil fertility management strategies since legumes are able to fix atmospheric nitrogen and yields of cereal crops are generally better if grown in rotation or intercropped with legumes. In particular grain legumes are highly valued components in smallholder farming systems due to their direct contribution to food and nutrition security. Moreover, legumes display a great agro-morphological diversity with great potential for challenging environments. Challenges aligned with climate change, such as increased rainfall variability, and restricted short growing periods, make short-season grain legumes a viable option as their adaption strategy of completing their life cycle before the onset of terminal drought seems to be advantageous for cropping with frequent droughts in semi-arid areas. However, to understand the temporal and spatial resource use and use efficiency of potential short-season grain legumes, especially in respect to light and water, it is of fundamental importance to design strategies for climate smart agriculture in risky environments, including areas of semi-arid Eastern Kenya. Furthermore,quantifying possible magnitudes of yield increase of different grain legumes can be useful in identifying niches in smallholder farming systems to increase overall farm productivity and sustainability. In order to explore the potential of certain crops and cropping strategies in diverse smallholder farming systems, the development and application of crop growth simulation models proved to be an excellent tool. Since African farming systems are highly heterogeneous and dynamic simulation models manage to address the complexity of these systems which is difficult to address through classical agronomic experiments alone. Simulation models are able to capture interactions between climatic conditions, soil type and nutrient dynamics. One of the most applicable models to better understand the complexities of plant growth in response to the environment has been the Agricultural Production System sIMulator (APSIM) framework, which has been successfully used for numerous farming system analyses in semi-arid areas in the past already. Against this background the objectives of this PhD thesis were, first, to compare growth and development of three promising short-season grain legumes (common bean, cowpea and lablab) in response to plant density and water regime to evaluate their production potential and resource capture in semi-arid environments (research chapter II). This was undertaken by the implementation and analysis of comprehensive field experiments carried out over two season 2012/13 and 2013/14 in Machakos, Eastern Kenya. Additionally to this comparative study of three legume species, the photo-thermal response of early-flowering lablab types were examined in a more detail from a combination of field experiments in South Africa and controlled environments studies conducted in Göttingen, Germany with the aim to evaluate their potential adaption to (sub)-tropical environments as a climate smart farming practice (chapter I). During the field experiments conducted in Machakos Kenya crop development, biomass and yield accumulations as well as leaf area index (LAI) were measured intensively throughout the growing period to determine import agronomic and physiological parameters, such as biomass partitioning coefficient, harvest index (HI) and radiation use efficiency (RUE) for the short-season legumes common bean, cowpea and lablab (chapter II). The output derived from the field experiments was further used to quantify essential cultivar-specific parameters to better calibrate (and later validate) APSIM to simulate growth and development of short-season grain legumes under semi-arid conditions (chapter III). Finally the agro-climatic conditions and changes as well as associated risk for rainfed crop production along the Machakos-Makueni transect in semi-arid Eastern Kenya was characterized in detail to identify possible niches for short-season grain legumes. For that purpose growth and development, as well as water use and use efficiency were simulated along the environmental gradient using APSIM (chapter IV). Within the first research chapter (chapter I) a comprehensive analysis of three datasets derived from field experiments in South Africa (different sites and sowing dates) and growth chamber experiments in Germany with a combination of two temperature and four daylength regimes were analysed to evaluate the response of temperature and photoperiod on flowering time of ten promising short-season lablab accessions (CPI 525313, CPI 52533, CPI 52535, CPI 52535, CPI 52552, CPI 52554, CPI 60795, CPI 81364, CQ 3620, Q 6880B). Hence, knowledge of phenological development and, in particular, time to flowering is crucial information needed for estimating the possible production success of new accessions in new and challenging environments, such as semi-arid Eastern Kenya. Therefore, the photoperiod sensitivity was quantified using the triple-plane rate model of flowering response with time to flowering expressed in thermal time (Tt, °Cd). Additionally, piecewise regression analysis was conducted to estimate the critical photoperiod (𝑃𝑐) above which time to flowering was delayed significantly. Relatively high variation of time to flowering among and within accessions in days after planting (DAP) was observed, ranging from 60 to 120 DAP depending on the site, sowing date or daylength/temperature regime. Furthermore, a clear positive effect of temperature on growth and development of the tested accessions was found and time to flowering, expressed as thermal time, were relative consistent for the tested accessions, ranging from 600 to 800 °Cd for daylength <13.5 h. Only at daylength of ≥13.5 h and temperatures above 28 °C development towards flowering was delayed significantly for accessions CPI 52513, CPI 52535, CPI 52554 and CPI 60795 with vegetative growth continuing for >110 DAP. The tested lablab accessions are, therefore, considered only weak photoperiod responsive and are classified as short-day plants (SDP). Since daylength does not exceed 13 h between latitude 30°N to 30°S covering the semi-arid tropical regions, these lablab accessions can be recommend for further evaluation of their adaption to, and productivity under, on-farm conditions. However, not only lablab offers a great potential for farming in semi-arid areas, legumes in general have proved to be a promising option in small-scale farming systems by combining benefits for the farmer, soil and environment. Therefore, effects of plant density and drought on growth and development of three promising short-season grain legumes including common bean, cowpea and lablab were quantified in detail to evaluate their agricultural production potential for semi-arid areas (chapter II). Two comprehensive field experiments; a plant density trial (three different plant densities; low, medium, high) and a water response trial (three different irrigation level: rainfed, partly irrigated (total 50 mm of water per week with supplementary irrigation till bud formation, i.e., onset of flowers), fully irrigated (total of 50 mm of water per week with supplementary irrigation throughout the growing period) were conducted to quantify the effect of plant density and water availability on canopy development, biomass accumulation and partitioning to evaluate resource use and use-efficiency of the different legumes. Therefore, biomass accumulation, leaf area index (LAI) and fractional radiation interception were measured repeatedly during the growing period while grain yield were measured at maturity. From the data collected, harvest index (HI), biomass partitioning coefficient and radiation use efficiency (RUE) were calculated. It was found that clear differences in temporal and spatial development and growth among the evaluated grain legumes are the major drivers for the observed variance in the fraction of intercepted radiation, biomass accumulation and grain yield. Moreover, the response of RUE to plant density and moisture availability differed among the three legumes. Common bean had a very short growing period (10 weeks), limiting the overall production potential (1000-1900 kg ha-1) under favourable conditions through limited source-sink dynamics in terms of time and space. Nevertheless, the short life cycle and the comparatively high RUE of common bean could be advantageous in environments with very short cropping windows. Cowpea showed a high phenological plasticity and potential to respond to favourable water supply in wet years by out-yielding the other legumes and reaching yields up to 3000 kg ha-1 under non water limited conditions. However, leaf development was observed to be sensitive to drought leading to decreased biomass development and consequently yield accumulation. The RUE of both common bean and cowpea was relatively low under rainfed conditions reaching only 0.49 and 0.54 g MJ-1, respectively, but more than doubled with supplementary irrigation. In contrast, lablab displayed stable RUE values (0.76 - 0.92 g MJ-1), and was not affected by limiting water availability, resulting in yields of 1200 to 2350 kg ha-1 across all water regimes. Nevertheless the growing period length of lablab was by far the longest (~100 days) compared to common bean and cowpea. The information revealed from the field experiments conducted in semi-arid Eastern Kenya was used to determine genetic coefficients and site-specific soil characterization to parameterize APISM for short-season legumes and semi-arid conditions (chapter III). The models were validated against data from the plant density and water regime trial conducted for two season (2012/13 and 2013/14) including observed data on soil moisture, phenology, biomass accumulation and yield development. Further, the adapted APSIM legume models were used in a long-term simulation experiment to evaluate the yield potential of the different short-season legumes under various management practices. The model accuracy to predict flowering time and time of physiological maturity was excellent and with a mean root squares of derivation (RMSD) of 5 days and less. For the different plant density and water regime treatments model predictions of biomass and grain yield were satisfactory reaching RMSD values expressed in % of the observed mean of about 12 for common bean biomass and grain yield and 23.5 and 26.0 and 20.8 and 25.1 for cowpea and lablab biomass and grain yield respectively. A good relationship between simulated yield and in-crop rainfall highlighted the importance of taking a water-limited potential yield into account when management practices are designed. To further quantify the potential of different short-season grain legumes in semi-arid areas where water is the most limiting factor for agricultural production the fourth research chapter aimed to examine the water use and water-use efficiency of short-season grain legumes along an environmental gradient in semi-arid Eastern Kenya (chapter IV). First, the climate variability along this transect was characterized in great detail including the analysis of annual and seasonal temperature development, inter- and intraseasonal rainfall variability as well as the analysis of the dry spell probability throughout the year. Second, growth and development of the short season grain legumes was simulated along the transect using APISM to assess the overall performance of the short-season legumes at different sites (potential rainfall areas) and evaluate the impact of various soil types to estimate their overall agricultural production potential. The analysis of long-term weather data from the Machakos – Makueni transect in semi-arid Eastern Kenya revealed large inter-annual as well as inter- and intra-seasonal variation in rainfall. Further trends showed that the growing season rainfall slightly decreased within the last decades. A decrease in mean rainfall intensity (rainfall per rain day) was observed for the past years as well. Regarding temperature development a slight increase in mean minimum and maximum temperatures was observed over the last decades, associated with an increase in days with maximum temperatures over 25 °C. Further analysis indicated an increased probability of long dry spells within the growing periods along the Machakos - Makueni transect and highly variable start and length of growing periods - creating a risky production environment. The observed variability of determined WUE of the different short-season grain legumes in terms of dry matter and grain yield production from the long-term simulations can be attributed to the effects of both the amount of rainfall and its distribution through the growing period. Water-potential yield of common bean was relatively stable (1000 kg ha-1), independent of total in-crop rainfall and soil conditions. Cowpea growth and development was, however, very responsive to in-crop rainfall. This is obvious as in wet years cowpea yield is very high (3000 kg ha-1), whereas in drier years grain yields (>500 kg ha-1) are even lower than common bean grain yields. Lablab yields instead, were fairly robust (1000 – 3000 kg ha-1) and higher than those observed for common bean, even at low in-crop rainfall levels. Determined WUE in terms of biomass production was highest for cowpea and lablab (8 – 12 kg ha-1 mm-1 Et) in comparison to common bean (6 – 8 kg ha-1 mm-1 Et), but in terms of grain yield production only lablab (4 - 6 kg ha-1 mm-1 Et) achieved higher values compared to common bean (3 - 5 kg ha-1 mm-1 Et) and cowpea (2 - 4 kg ha-1 mm-1 Et). The magnitude of the soil impact on crop growth and development as well as water use and use efficiency differed with texture and water-holding capacity of the soil, soil evaporation and the interaction between these factors, rainfall pattern, crop canopy architecture and management. The current results revealed that resource capture of the studied legumes was primarily outlined by their characteristic phenological development and further determined by phenological plasticity related to water deficit and the ability to respond to environmental conditions. Pronounced spatial and temporal differences in water use and use efficiency of the studied legumes were therefore first driven by the varying phenological development and secondly by species-specific morphological and physiological characteristics and mechanisms. However, the ability of the legumes to respond to environmental conditions and the degree of phenological plasticity have evolved different strategies to cope with challenging conditions in semi-arid areas. To consider the pronounced temporal and spatial differences in resource use and growth characteristics is fundamental to better design strategies for climate smart agriculture in the smallholder farming systems of Eastern Kenya. The calibrated and validated APSIM legume models can be used to make appropriate management decisions to provide smallholder farmers in semi-arid with alternative options to better integrate short-season legumes to improve the overall farm productivity and sustainability. Crop models such as APSIM allow to account for necessary complexity but at the same time manage to address high location specificity. This is particular important in diverse smallholder farming systems in semi-arid areas to adequately address their individual needs and opportunities. The variability in phenological development and resource use and use efficiency observed for the different legumes and their different adaption mechanism to semi-arid areas offer great potential for small-scale farming systems in challenging environments. APSIM seems to be a great tool to explore their site-specific agricultural production potential and the impact of different management strategies is semi-arid Eastern Kenya. However, socio-economic constraints including labour requirements and market opportunities need to be assed in more detail to better channel agricultural recommendations to increase the possible adaption among farmers. Furthermore, long-term aspects of better integrated legumes towards improved farm sustainability and increased eco-efficiency need to be determined with the help of multidimensional whole farm analysis tools in order to proceed beyond crop and plot level in the future. | |