Non Technical Summary
Coffee leaf rust (CLR) is a devastating fungal pathogen of coffee plants. When infected, coffee trees experience a reduction in photosynthesis and a loss of leaves. These symptoms can lead to reduced coffee yields, an estimated global value of $1-2 billion in losses. Due to the complex life cycle of the fungal agent responsible for CLR (Hemileia vastatrix), the disease is challenging to manage when outbreaks occur and can spread rapidly. This work aims to understand whether the community of other microbes (particularly fungi and bacteria) present on the leaf surfaces of coffee can help or hinder the establishment of the fungus that causes CLR. To pursue this research, I will use DNA sequencing to identify any differences in the fungal and bacterial communities on the surfaces of healthy and CLR-infected coffee leaves. I will also investigate if different cultivars of coffee can harbor different communities of fungi and bacteria on healthy and CLR-infected leaves. Changes in these communities associated with the progression of the CLR disease will be monitored in greenhouse experiments over a 4-week period. Lastly, the effect of fungicides and biological control of CLR on the communities of leaf surface fungi and bacteria will be studied. Combined, the results of this work can lead to understanding the microbiological indicators that are associated with CLR disease. This insight can aid in the selection of cultivars grown and the pathogen management regime employed to combat CLR. Results of this work will be communicated to coffee growers and farm laborers through workshops, with the scientific community through research articles and conference presentations, and with the general public through open-access publishing and collaborating with extension agents.

Classification

Knowledge Area (KA)	Subject of Investigation (SOI)	Field of Science (FOS)	Percent
212	4099	1070	100%

Knowledge Area
212 - Pathogens and Nematodes Affecting Plants;

Subject Of Investigation
4099 - Microorganisms, general/other;

Field Of Science
1070 - Ecology;

Goals / Objectives
The overall goal of this project is to understand microbial community dynamics associated with coffee phylloplanes to support management efforts of coffee leaf rust (CLR). The causal agent of CLR is Hemileia vastatrix, a rust fungus that is capable of decimating coffee trees. The objectives of this study are as follows: 1) Characterize the epiphytic fungal and bacterial diversity and community composition on healthy coffee leaves and CLR-infected leaves; 2) elucidate the differences in the epiphytic microbiome of cultivars that have different levels of susceptibility to CLR; 3)characterize shifts in taxonomic and functional diversity as CLR infection progresses; and 4) determine the impact of fungicides on the microbial community associated with coffee trees.

Project Methods
Objectives 1 and 2:On Hawaii Island, two conventional farms in the Kona district and two conventional farms in the Kau district will be selected for sampling. Farms will be chosen based on the criterion of having neighboring blocks of one CLR susceptible and oneresistant cultivar. Leaf sampling will be conducted at the peak of CLR infection in July-August. Ten trees per cultivar will be randomly sampled. From each coffee tree, one healthy leaf and one CLR-infected leaf will be sampled. Leaves at the 3rd node down from the branch tip will be selected to minimize differences associated with leaf age. Leaves will be swabbed with a sterile cotton swab moistened with phosphate-buffered saline on the abaxial surface, where H. vastatrix is able to enter plant tissue via stomata. Swabs will be transported to the laboratory in sterile tubes containing phosphate-buffered saline. High-throughput DNA extractions will be done using a Kingfisher nucleic acid sampling system (Thermo Fisher Scientific) in 96-well plates. Following DNA extraction, fungal metabarcoding of the full internal transcribed region (ITS) of the rRNA gene will be performed, amplified with primer pair ITS1Fngs/ITS4-Fun. The fungal samples will be sequenced using PacBio to sequence the entire ITS region to be able to more accurately identify unique taxa and overcome challenges with intragenomic variability of sections of the ITS region known to occur in fungi. The bacterial community will beamplified using the primer pair 515Fand 806RBto amplify the V4 region of the rRNA (16S) gene. Bacterial samples will be sequenced using the Element AVITI short-read platform (2x300 bp). The fungal dataset resulting from long-read sequencing will first have flanking 18S and 28S gene portions sequences removed to avoid erroneous clustering for taxonomic identification, along with any chimeras that may have arisen during sequencing using ITSxpress.Inference of operational taxonomic units (OTUs) will be performed using PIPITS, which is designed for fungal ITS sequences. Fungal communities are best analyzed by OTU, due to the overestimation of diversity when attempting analysis with amplicon sequencing variants (ASVs). The bacterial dataset resulting from short-read sequencing will first be subject to quality control, chimera removal, and inference of ASVs with DADA2, followed by the removal of mitochondrial and non-bacterial reads and the collapsing of ASVs to a 97% similarity threshold using the LULU algorithm. The differences in approach for analyzing fungal and bacterial communities is to capture the diversity of each such that the results are attributable to biological reality and not algorithmic or computational artifact.Alpha diversity between communities from different cultivars will be compared using the Wilcoxon rank sum statistic. For beta diversity, an ordination analysis using non-metric multidimensional scaling (NMDS) will be constructed to visualize any differences between communities. Differences between the cultivar communities will be calculated using Bray-Curtis distance dissimilarity, significance tested using permutational multivariate analysis of variance (PERMANOVA).Objective 3:Twelve plants of the Kona Typica cultivar (S) and twelve plants of the Geisha cultivar (T) will be grown in a greenhouse for six months. Two individual humidifying chambers will be set up in the greenhouse: one control chamber and one infectious chamber. The relative humidity, temperature, and daylight hours of both chambers will be controlled. A spore suspension of H. vastatrix urediniospores will be produced by collecting spores from field sites and suspended in Tween 20. Viability of the spore suspension will be live/dead assayed using trypan blue. Half of the plants (6 Kona Typica plants, 6 Geisha plants) will serve as control plants (i.e., no infection). The other half of the plants will be inoculated with the H. vastatrix spore suspension. Prior to inoculation, leaf swabs of plants will be collected and sequenced to ground-truth the microbiome community. Whole plants will be sprayed to inoculate the plants with the CLR agent. Plants will be monitored on a weekly basis to ensure no establishment of rust on controls plants occurred, and to observe signs and/or symptoms of CLR in the infectious chamber. Once CLR lesions are observed in the infectious chamber, plants will be monitored daily. Day 0 for longitudinal sample collection is the day on which the first symptoms/signs are observed.Leaves will be swabbed weekly for communities of fungi and bacteria over four weeks. Sterile cotton swabs will be swirled in phosphate-buffered saline solution, followed by DNA/RNA extraction using a Kingfisher nucleic acid extraction system. The fungal and bacterial communities will be sequenced as described in Objectives 1 and 2. Transcriptomics will be used to assess putative function and will be sequenced using RNASeq. The fungal and bacterial datasets will be processed like those in Objective 1, including an alpha diversity analysis supplemented with an ordination analysis using Bray-Curtis distances. Significance tests for these diversity analyses are Wilcoxon rank sum statistics and PERMANOVA on NMDS results, respectively.Objective 4:Farms that fit the following categories will be selected as field sites: 1) contact fungicides (copper-based) are applied to combat CLR; 2) systemic fungicides are applied; 3) both contact and systemic fungicides are used;4) a biological control or agroecological approach is used; 5) no chemical prophylactic or preventative measures are taken. Three farms of each type will be selected for replication. Ten trees will be sampled from each farm. From each tree, ten leaves will be swabbed on the abaxial leaf surfaces. Collection swabs will be swirled in phosphate-buffered saline solution and transported to the laboratory for DNA extraction and sequencing. Each tree will be sampled twice. The first sample will occur 24-72 hours prior to first application of CLR control agent. The last sample will occur 24-72 hours after final application of the CLR control agent. Time windows are necessary to accommodate the restricted entry interval (REI) as labeled for each fungicide used. These sampling regimes will likely occur at the beginning and end of the peak of CLR outbreak. DNA from these samples will be extracted using the Kingfisher nucleic acid extraction system. Sequencing and data processing. Shotgun metagenomics will be used to sequence the entire microbial community using an Illumina platform to obtain 2x150bp reads. By using shotgun metagenomics, sequences from microorganisms spanning multiple domains can be obtained. The reads generated from sequencing will be subject to FastQC for quality assessment. The quality report will be used to guide trimming using Trimmomatic. Reads will be assembled into contigs using SPAdes run with the "metaspades" option to configure the tool to a metagenomic dataset. Reads will be mapped to contigs using Bowtie2. Taxonomic assignment will be performed using DIAMOND against nucleotide databases. All of these processing steps will be run using the SqueezeMeta metagenomics pipeline.Alpha and beta diversity will be computed using the Wilcoxon statistic and Bray-Curtis distance dissimilarity represented by NMDS, respectively. Differences between communities pre- and post-application for each site will be analyzed using non-parametric multivariate analysis of variance (NP-MANOVA). To compare communities across different CLR control regimes, NP-MANOVA will also be used. Co-occurrence and differential interactions between sequenced microorganisms will be analyzed using sparse inverse covariance estimation for ecological association and statistical inference (SPIEC-EASI). Identifying negative and positive edges (interactions) and potential hub nodes can identify any key interactions or key taxa that are lost when fungicides are applied.