Sustaining Alfalfa Forage Production with Limited Water Resources

Variability in climatic conditions and uncertainty in water supply brings challenges for sustainable irrigated forage production. Alfalfa (Medicago sativa L.), one of the major forage crops worldwide, is affected due to its high water demands for successful irrigated production. To address the challenge of sustaining forage yields in a changing climate, we designed three studies to understand the yield response of alfalfa under different irrigation systems and with varying water deficits. We also examined the usefulness of aerial imagery to understand the yield and quality variability in the field as affected by drought. In study 1, the potential of sub-surface drip irrigation (SDI) combined with reduced irrigation for improving forage yield, quality, and water productivity of alfalfa under water deficit conditions in California’s Central Valley was explored. In study 2, we examined the two overhead irrigation systems: low elevation spray application (LESA) and mobile drip irrigation (MDI) for producing alfalfa and the impacts of deficit irrigation using these two technologies on alfalfa production, quality, and productivity. In study 3, we examined the ability of aerial imagery, specifically, multispectral and LiDAR imaging technologies to understand yield and quality variability on a field scale with differential drought treatments. Study 1 was conducted at Kearney Agricultural Research and Extension Center (Parlier, CA) from 2016-2019 using randomized complete block (RCB) design on sandy loam soil while study 2 and 3 were conducted during 2019 and 2020 at Davis, CA using split plot randomized complete block design on a clay-loam soil. In Study 1, we found that there were no significant differences between SDI and flood irrigation systems over two years of study. In small-plot studies, it is likely that flood irrigation techniques are advantageous due to smaller checks, an advantage that may disappear when larger fields are considered- thus advantages of these systems should be viewed at scale. In addition, the SDI plots had maintenance issues in the later part of the study, a common problem with SDI. But SDI had the advantage of applying water to more closely match crop evapotranspiration demands while flood seems slightly stressed during the growth cycle due to less frequent irrigation. Over the three years of study, deficit irrigation utilizing SDI resulted in yields that were 82%, 84%, and 87% of fully irrigated treatments for the 50%, 75% (sudden cutoffs), 75% (gradual deficits), respectively (percent of full ET requirement). There were slight improvements in forage quality using deficits, but differences were not great. Higher water productivities were found in deficit irrigation compared with full irrigation under SDI. Under water uncertainty, deficit irrigation using subsurface drip irrigation (SDI) could be beneficial to address the challenges of climate change, if such systems can be economically managed and maintained. In study 2, we found few differences in yield results between LESA and MDI systems in overhead irrigation of alfalfa over two years. Yields were sustained or were higher under MDI 60% cutoff compared with LESA 60% summer cutoffs, likely due to superior sub-soil moisture recharge with the dragging drip lines. Higher water productivity and irrigation water use efficiency was found in MDI-60% ET- Cutoff in 2020 (21.2 kg ha-1 mm-1) while there were no significant differences found in 2019. In general, LESA performed better in all other treatments than MDI over the two years of study period. Both LESA and MDI systems minimized wind losses and improve deep moisture availability. MDI systems do not have the rodent or maintenance issues as does SDI but have the disadvantage of requiring some filtration. In study 3, we utilized the drought affected research field and tried to understand the spatial-temporal variability in alfalfa yield and quality using aerial imagery. Aerial flights were conducted at harvest using multispectral and LiDAR camera in separate flights. We trained the models on the field data for plant height and dry matter yields and it was found that model performed well when an unknown dataset was provided for model testing. The model was created using a step-wise regression model and was compared with random forest (RF) and support vector machine (SVM) for multispectral imagery. It was found that step-wise regression model performed somewhat better than RF and SVM with an R2 of 0.82, 0.79 and 0.81 respectively. The model also performed well for predicting yield in a separately measured area of 11.15 m2 with an R2 0.83. LiDAR also performed well and predicted the yield with slightly lower R2 (0.67) but successfully predicted yields in 11.15 m2 area (R2 0.91). Both multispectral and LiDAR imagery were able to predict the dry matter yield for alfalfa based on the trained models, and each system has advantages and disadvantages. LiDAR is more demanding in terms of cost and analysis requirements. Some of the predicted results exhibited bias, these may have resulted from differences in sample size or sampling protocols, but such biases can be corrected mathematically. Aerial imagery should be considered a useful tool to create yield maps, to estimate the impacts of drought on yield, and to understand sources of field variability to guide management decisions. For model applicability, further work at different scales should be conducted to predict yields on a farm or regional scale.Keywords: Alfalfa, SDI, Flood, LESA, MDI, deficit irrigation, forage quality, irrigation efficiency, water productivity, UAV remote sensing, yield prediction, drought, field diagnostics, Multispectral, LiDA