Issues surrounding water scarcity will become topical in future as global fresh water resources become more limited thus threaten crop production. Predicted climate change and increasing population growth will place more pressure on agriculture to produce more food using less water. As such, efforts have now shifted to identifying previously neglected underutilised species (NUS) as possible crops that could be used to bridge the food gap in future. Taro (Colocasia esculenta L. Schott) and bambara groundnut (Vigna subterranea L. Verdc) currently occupy low levels of utilisation in South Africa. Both crops are cultivated using landraces with no improved varieties available. Information describing their agronomy and water–use is limited and remains a bottleneck to their promotion. The aim of this study was to determine the drought tolerance and water–use of selected landraces of taro and bambara groundnut from KwaZulu-Natal, South Africa. In order to meet the specific objectives for taro and bambara groundnut management, an approach involving conventional and modelling techniques was used. Three taro landraces [Dumbe Lomfula (DL), KwaNgwanase (KW) and Umbumbulu (UM)] were collected from the North Coast and midlands of KwaZulu-Natal, South Africa, in 2010. The UM landrace was classified as Eddoe type taro (C. esculenta var. antiquorum) characterised by a central corm and edible side cormels. The DL and KW landraces were classified as Dasheen (C. esculenta var. esculenta), characterised by a large edible main corm and smaller side cormels. A bambara groundnut landrace was collected from Jozini, KwaZulu- Natal, and characterised into three selections (‘Red’, ‘Light-brown’ and ‘Brown’) based on seed coat colour. Seed colour was hypothesised to have an effect on seed quality. Field and rainshelter experiments were conducted for both taro and bambara landraces at Roodeplaat in Pretoria and Ukulinga Research Farm in Pietermaritzburg, over two growing seasons (2010/11 and 2011/12). The objective of the field trials for taro and bambara groundnut was to determine mechanisms associated with drought tolerance in taro and bambara groundnut landraces. Experiments were laid out in a split-plot design where irrigation [fully irrigated (FI) and rainfed (RF)] was the main factor and landraces (3 landraces of either taro or bambara groundnut) were sub-factors. Treatments were arranged in a randomised complete block design (RCBD), replicated three times. Rainfed trials were established with irrigation to allow for maximum crop stand. Thereafter, irrigation was withdrawn. Whilst experimental designs and layouts for taro and bambara groundnut were similar, differences existed with regards to plot sizes and plant spacing. Trials were planted on a total land area of 500 m2 and 144 m2, for taro and bambara groundnut, respectively. Plant spacing was 1 m x 1 m for taro and 0.3 m x 0.3 m for bambara groundnut. Irrigation scheduling in the FI treatment was based on ETo and Kc and was applied using sprinkler irrigation system. Separate rainshelter experiments were conducted for taro and bambara groundnut landraces at Roodeplaat, to evaluate growth, yield and water-use of taro and bambara groundnut landraces under a range of water regimes. The experimental design was similar for both crops, a RCBD with two treatment factors: irrigation level [30, 60 and 100% crop water requirement (ETa)] and landrace (3 landraces), replicated three times. Irrigation water was applied using drip irrigation system based on ETo and Kc. Data collection in field and rainshelter trials included time to emergence, plant height, leaf number, leaf area index (LAI), stomatal conductance and chlorophyll content index (CCI). For taro field trials, vegetative growth index (VGI) was also determined. Yield and yield components (harvest index, biomass, corm number and mass) as well as water–use efficiency (WUE) were determined at harvest. Intercropping of taro and bambara groundnut was evaluated under dryland conditions using farmers’ fields at Umbumbulu, KwaZulu–Natal, South Africa. The experimental design was a RCBD replicated three times. Intercrop combinations included taro and bambara groundnut sole crops, a 1:1 (one row taro to one row bambara groundnut) and 1:2 intercrop combinations. The taro UM landrace and ‘Red’ bambara groundnut landrace selection were used in the intercropping study. Lastly, data collected from field and rainshelter experiments were used to develop crop parameters to calibrate and validate the FAO’s AquaCrop model for taro and bambara groundnut landraces. The UM landrace was used for taro while the ‘Red’ landrace selection was used for bambara groundnut. AquaCrop was calibrated using observed data from optimum (FI) experiments conducted during 2010/11. Model validation was done using observations from field and rainshelter experiments conducted during 2011/12 as well as independent data. Results showed that all taro landraces were slow to emerge (≈ 49 days after planting). Stomatal conductance declined under conditions of limited water availability (RF, 60% and 30% ETa). The UM landrace showed better stomatal regulation compared with KW and DL landraces under conditions of limited water availability. Plant growth (plant height, leaf number, LAI and CCI) of taro landraces was lower under conditions of limited water availability (RF, 60% and 30% ETa) relative to optimum conditions (FI and 100% ETa). The UM landrace showed moderate reductions in growth compared with the DL and KW landraces, suggesting greater adaptability to water limited conditions. The VGI showed a large reduction in growth under RF conditions and confirmed the UM landrace’s adaptability to limited water availability. Limited water availability (RF, 60% and 30% ETa) resulted in lower biomass, HI, and final yield in taro landraces relative to optimum conditions (FI and 100% ETa). For all trials, the DL landrace failed to produce any yield. WUE of taro landraces was consistent for the three irrigation levels (30, 60 and 100% ETa); however, on average, the UM landrace was shown to have a higher WUE than the KW landrace. Bambara groundnut landraces were slow to emerge (up to 35 days after planting). ‘Red’ and ‘Brown’ landrace selections emerged better than the ‘Light-brown’ landrace selection, confirming the effect of seed colour on early establishment performance. Plant growth (stomatal conductance, CCI, plant height, leaf number, LAI and biomass accumulation) was lower under conditions of limited water availability (RF, 60% and 30% ETa) relative to optimum conditions (FI and 100% ETa). The ‘Red’ landrace selection showed better adaptation to stress. Limited water availability resulted in early flowering and reduced flowering duration as well as early senescence and maturity of bambara groundnut landrace selections. The ‘Red’ landrace selection showed delayed leaf senescence under conditions of limited water availability. Yield reductions of up to 50% were observed under water limited conditions (RF, 60% and 30% ETa) relative to optimum conditions (FI and 100% ETa). Water use efficiency increased at 60% and 30% ETa, respectively, relative to 100% ETa, implying adaptability to limited water availability. The ‘Red’ landrace selection showed better yield stability and WUE compared with the ‘Brown’ and ‘Light-brown’ landrace selections suggesting that seed colour may be used as a selection criterion for drought tolerance in bambara groundnut landraces. The intercropping study showed that intercropping, as an alternative cropping system, had more potential than monocropping. Evaluation of growth parameters showed that taro plant height was generally unaffected by intercropping but lower leaf number was observed as compared with the sole crop. Bambara groundnut plants were taller and had more leaves under intercropping relative to the sole crop. Although not statistically significant, yield was generally lower in the intercrops compared with the sole crops. Evaluation of intercrop productivity using the land equivalent ratio (LER) showed that intercropping taro and bambara groundnut at a ratio of 1:1 was more productive (LER = 1.53) than intercropping at a ratio of 1:2 (LER = 1.23). The FAO’s AquaCrop model was then calibrated for the taro UM landrace and ‘Red’ bambara groundnut landrace selection. This was based on observations from previous experiments that suggested them to be drought tolerant and stable. Calibration results for taro and bambara groundnut landraces showed an excellent fit between predicted and observed parameters for canopy cover (CC), biomass and yield. Model validation for bambara groundnut showed good model performance under field (FI and RF) conditions. Model performance was satisfactory for rainshelters. Validation results for taro showed good model performance under all conditions (field and rainshelters), although the model over-estimated CC for the declining stage of canopy growth under RF conditions. Model verification using independent data for taro showed equally good model performance. In conclusion, the taro UM landrace and ‘Red’ bambara groundnut landrace selection were shown to be drought tolerant and adapted to low levels of water–use. The mechanisms responsible for drought tolerance in the taro UM landrace and ‘Red’ bambara groundnut landrace selection were described as drought avoidance and escape. The taro UM landrace and ‘Red’ bambara groundnut landraces avoided stress through stomatal regulation, energy dissipation (loss of chlorophyll) as well as reducing canopy size (plant height, leaf number and LAI), which translates to minimised transpirational water losses. This indicated landrace adaptability to low levels of water–use. The ‘Red’ bambara groundnut landrace selection showed phenological plasticity and escaped drought by flowering early, delaying leaf senescence, and maturing early under conditions of limited water availability. Performance of the ‘Red’ landrace selection lends credence to the use of seed coat colour as a possible selection criterion for drought tolerance in bambara groundnut, and possibly for other landraces with variegated seed. The taro UM landrace escaped drought by maturing early under conditions of limited water availability. The FAO’s AquaCrop model was successfully calibrated and validated for taro UM and ‘Red’ bambara groundnut landraces. The calibration and validation of AquaCrop for taro is the first such attempt and represents progress in the modelling of neglected underutilised crops. The calibration and validation of AquaCrop for taro requires further fine-tuning while that for bambara groundnut still needs to be tested for more diverse landraces. / Thesis (Ph.D.)-University of KwaZulu-Natal, Pietermaritzburg, 2011.

Issues surrounding water scarcity will become topical in future as global fresh water resources become more limited thus threaten crop production. Predicted climate change and increasing population growth will place more pressure on agriculture to produce more food using less water. As such, efforts have now shifted to identifying previously neglected underutilised species (NUS) as possible crops that could be used to bridge the food gap in future. Taro (Colocasia esculenta L. Schott) and bambara groundnut (Vigna subterranea L. Verdc) currently occupy low levels of utilisation in South Africa. Both crops are cultivated using landraces with no improved varieties available. Information describing their agronomy and water–use is limited and remains a bottleneck to their promotion. The aim of this study was to determine the drought tolerance and water–use of selected landraces of taro and bambara groundnut from KwaZulu-Natal, South Africa. In order to meet the specific objectives for taro and bambara groundnut management, an approach involving conventional and modelling techniques was used. Three taro landraces [Dumbe Lomfula (DL), KwaNgwanase (KW) and Umbumbulu (UM)] were collected from the North Coast and midlands of KwaZulu-Natal, South Africa, in 2010. The UM landrace was classified as Eddoe type taro (C. esculenta var. antiquorum) characterised by a central corm and edible side cormels. The DL and KW landraces were classified as Dasheen (C. esculenta var. esculenta), characterised by a large edible main corm and smaller side cormels. A bambara groundnut landrace was collected from Jozini, KwaZulu- Natal, and characterised into three selections (‘Red’, ‘Light-brown’ and ‘Brown’) based on seed coat colour. Seed colour was hypothesised to have an effect on seed quality. Field and rainshelter experiments were conducted for both taro and bambara landraces at Roodeplaat in Pretoria and Ukulinga Research Farm in Pietermaritzburg, over two growing seasons (2010/11 and 2011/12). The objective of the field trials for taro and bambara groundnut was to determine mechanisms associated with drought tolerance in taro and bambara groundnut landraces. Experiments were laid out in a split-plot design where irrigation [fully irrigated (FI) and rainfed (RF)] was the main factor and landraces (3 landraces of either taro or bambara groundnut) were sub-factors. Treatments were arranged in a randomised complete block design (RCBD), replicated three times. Rainfed trials were established with irrigation to allow for maximum crop stand. Thereafter, irrigation was withdrawn. Whilst experimental designs and layouts for taro and bambara groundnut were similar, differences existed with regards to plot sizes and plant spacing. Trials were planted on a total land area of 500 m2 and 144 m2, for taro and bambara groundnut, respectively. Plant spacing was 1 m x 1 m for taro and 0.3 m x 0.3 m for bambara groundnut. Irrigation scheduling in the FI treatment was based on ETo and Kc and was applied using sprinkler irrigation system. Separate rainshelter experiments were conducted for taro and bambara groundnut landraces at Roodeplaat, to evaluate growth, yield and water-use of taro and bambara groundnut landraces under a range of water regimes. The experimental design was similar for both crops, a RCBD with two treatment factors: irrigation level [30, 60 and 100% crop water requirement (ETa)] and landrace (3 landraces), replicated three times. Irrigation water was applied using drip irrigation system based on ETo and Kc. Data collection in field and rainshelter trials included time to emergence, plant height, leaf number, leaf area index (LAI), stomatal conductance and chlorophyll content index (CCI). For taro field trials, vegetative growth index (VGI) was also determined. Yield and yield components (harvest index, biomass, corm number and mass) as well as water–use efficiency (WUE) were determined at harvest.Intercropping of taro and bambara groundnut was evaluated under dryland conditions using farmers’ fields at Umbumbulu, KwaZulu–Natal, South Africa. The experimental design was a RCBD replicated three times. Intercrop combinations included taro and bambara groundnut sole crops, a 1:1 (one row taro to one row bambara groundnut) and 1:2 intercrop combinations. The taro UM landrace and ‘Red’ bambara groundnut landrace selection were used in the intercropping study. Lastly, data collected from field and rainshelter experiments were used to develop crop parameters to calibrate and validate the FAO’s AquaCrop model for taro and bambara groundnut landraces. The UM landrace was used for taro while the ‘Red’ landrace selection was used for bambara groundnut. AquaCrop was calibrated using observed data from optimum (FI) experiments conducted during 2010/11. Model validation was done using observations from field and rainshelter experiments conducted during 2011/12 as well as independent data. Results showed that all taro landraces were slow to emerge (≈ 49 days after planting). Stomatal conductance declined under conditions of limited water availability (RF, 60% and 30% ETa). The UM landrace showed better stomatal regulation compared with KW and DL landraces under conditions of limited water availability. Plant growth (plant height, leaf number, LAI and CCI) of taro landraces was lower under conditions of limited water availability (RF, 60% and 30% ETa) relative to optimum conditions (FI and 100% ETa). The UM landrace showed moderate reductions in growth compared with the DL and KW landraces, suggesting greater adaptability to water limited conditions. The VGI showed a large reduction in growth under RF conditions and confirmed the UM landrace’s adaptability to limited water availability. Limited water availability (RF, 60% and 30% ETa) resulted in lower biomass, HI, and final yield in taro landraces relative to optimum conditions (FI and 100% ETa). For all trials, the DL landrace failed to produce any yield. WUE of taro landraces was consistent for the three irrigation levels (30, 60 and 100% ETa); however, on average, the UM landrace was shown to have a higher WUE than the KW landrace. Bambara groundnut landraces were slow to emerge (up to 35 days after planting). ‘Red’ and ‘Brown’ landrace selections emerged better than the ‘Light-brown’ landrace selection, confirming the effect of seed colour on early establishment performance. Plant growth (stomatal conductance, CCI, plant height, leaf number, LAI and biomass accumulation) was lower under conditions of limited water availability (RF, 60% and 30% ETa) relative to optimum conditions (FI and 100% ETa). The ‘Red’ landrace selection showed better adaptation to stress. Limited water availability resulted in early flowering and reduced flowering duration as well as early senescence and maturity of bambara groundnut landrace selections. The ‘Red’ landrace selection showed delayed leaf senescence under conditions of limited water availability. Yield reductions of up to 50% were observed under water limited conditions (RF, 60% and 30% ETa) relative to optimum conditions (FI and 100% ETa). Water use efficiency increased at 60% and 30% ETa, respectively, relative to 100% ETa, implying adaptabilityto limited water availability. The ‘Red’ landrace selection showed better yield stability and WUE compared with the ‘Brown’ and ‘Light-brown’ landrace selections suggesting that seed colour may be used as a selection criterion for drought tolerance in bambara groundnut landraces. The intercropping study showed that intercropping, as an alternative cropping system, had more potential than monocropping. Evaluation of growth parameters showed that taro plant height was generally unaffected by intercropping but lower leaf number was observed as compared with the sole crop. Bambara groundnut plants were taller and had more leaves under intercropping relative to the sole crop. Although not statistically significant, yield was generally lower in the intercrops compared with the sole crops. Evaluation of intercrop productivity using the land equivalent ratio (LER) showed that intercropping taro and bambara groundnut at a ratio of 1:1 was more productive (LER = 1.53) than intercropping at a ratio of 1:2 (LER = 1.23). The FAO’s AquaCrop model was then calibrated for the taro UM landrace and ‘Red’ bambara groundnut landrace selection. This was based on observations from previous experiments that suggested them to be drought tolerant and stable. Calibration results for taro and bambara groundnut landraces showed an excellent fit between predicted and observed parameters for canopy cover (CC), biomass and yield. Model validation for bambara groundnut showed good model performance under field (FI and RF) conditions. Model performance was satisfactory for rainshelters. Validation results for taro showed good model performance under all conditions (field and rainshelters), although the model over-estimated CC for the declining stage of canopy growth under RF conditions. Model verification using independent data for taro showed equally good model performance. In conclusion, the taro UM landrace and ‘Red’ bambara groundnut landrace selection were shown to be drought tolerant and adapted to low levels of water–use. The mechanisms responsible for drought tolerance in the taro UM landrace and ‘Red’ bambara groundnut landrace selection were described as drought avoidance and escape. The taro UM landrace and ‘Red’ bambara groundnut landraces avoided stress through stomatal regulation, energy dissipation (loss of chlorophyll) as well as reducing canopy size (plant height, leaf number and LAI), which translates to minimised transpirational water losses. This indicated landrace viii adaptability to low levels of water–use. The ‘Red’ bambara groundnut landrace selection showed phenological plasticity and escaped drought by flowering early, delaying leaf senescence, and maturing early under conditions of limited water availability. Performance of the ‘Red’ landrace selection lends credence to the use of seed coat colour as a possible selection criterion for drought tolerance in bambara groundnut, and possibly for other landraces with variegated seed. The taro UM landrace escaped drought by maturing early under conditions of limited water availability. The FAO’s AquaCrop model was successfully calibrated and validated for taro UM and ‘Red’ bambara groundnut landraces. The calibration and validation of AquaCrop for taro is the first such attempt and represents progress in the modelling of neglected underutilised crops. The calibration and validation of AquaCrop for taro requires further fine-tuning while that for bambara groundnut still needs to be tested for more diverse landraces. / Thesis (Ph.D.)-University of KwaZulu-Natal, Pietermaritzburg, 2011.