We have begun merging and comparing the collected data from the three different sampling locations, Ridge, South Florida and Indian River. Using neural networks to analyze the data is underway and a few compelling results have been obtained and need further validations. Objective 1: Leaf nutrient thresholds At this point, only the Ridge data, which is complete through March 2017, has been analyzed using the neural network software Easy-NN. We are looking at the sample dates as snapshots in time and combined for any possible connection or correlation with HLB severity. First, we found that the parameters of mean leaf area and mean leaf perimeter were highly correlated with severity of HLB symptoms. As has been noticed in the past, trees with severe symptoms of HLB have smaller leaves, leading to lower average leaf areas. Also with the use of ImageJ software for analysis of leaf size and shape, we have found perimeter to also be correlated with the severity of tree dieback from HLB. Using mean leaf perimeter and leaf area as outputs in generating neural networks, we have found that some of our various measurements are of more important in explaining the means for leaf area and perimeter. Looking at the Ridge area data across all dates (including all data except soil data), we found the leaf nutrients Calcium and Magnesium are of high importance. When we run the neural network with Soil parameters, we find that the organic matter content in the soil plays a role in determining the mean leaf area. As the soil L* parameter increases, which is the lightness in color of the soil, the mean leaf area decreases. However on its own this variable is not well suited for predicting HLB incidence or severity, it does help to show the importance of organic matter in tree health. Other variables of importance in the data set include, Soil Calcium and Magnesium content as well as soil pH. Looking further into the soil parameters as inputs and using Leaf Thickness (grams of leaf dry mass / meter square leaf area) as an output, we find the most important soil variables to be the soil organic matter content and the soil color variable a*, the redness of the soil. These are only preliminary results and more investigation is necessary as well as increasing the data set over the next few quarters. Including the other two areas, Indian River and South Florida, we should be able to strengthen our data set, neural network and conclusions. Objective 2: Determine soil conditions that favor root hair and VAM proliferation i. We have discussed further soil analysis that we would like to work on, including data about permanent wilting coefficient as well as possibly quantifying the silicon content of the soil. Soils from the South Florida area will be included into the data set and will be measured for all of variables the other two regions have been measured for, including organic matter content, and color analysis. ii. Using rooted cuttings and the aeroponics tanks did not yield any results. We believe this is because the micro-jet spray in the tanks causes too much disturbance and damage to the root hair development. Next, we will try with true hydroponics, growing seedlings in solutions for minimal root disturbance. Seeds are in trays for germination now.