Skip to main content
Contact Us
Research

Phyto-threats project team meeting November 2019

November 14th 2019 held at APHA, Sand Hutton, York

The aim of this meeting was to bring the entire project team together to share and discuss research progress since the last all-project team meeting on November 20th 2018, to outline planned outputs and to discuss next steps for continuing collaborations now that the project is finishing.

The science updates for each work package were presented the day before at the stakeholder workshop and a full report on each work package is available in the stakeholder meeting report published online:

https://www.forestresearch.gov.uk/research/global-threats-from-phytophthora-spp/phytothreats-meetings-and-events/.

Therefore, the report here focuses on additional information not presented at the stakeholder meeting and on the resulting discussion.

 

WP1 Phytophthora distribution, diversity and management in UK nursery systems – David Cooke (JHI) and Peter Cock (JHI)

David Cooke started off by emphasising the need to get back to the different organisations with an interest in the outcomes of this work package and to think about how best to engage. He planned to show today the Phytophthora species findings by nursery and for Peter Cock to show progress on the bioinformatics pipeline. They also need to link Phytophthora findings to nursery management structure with Beth’s team.

While David was running through the slides a question was asked regarding clades;

Q: Are you going to give a breakdown of clade when reporting back to nursery managers?

A: We had decided not to report clade, especially as sometimes clades change.

Comment: Nursery managers don’t need to know clade.

Q: But isn’t there a clade-specific treatment or approach? i.e. clade 6 species tend to be less damaging than clade 8 species.

A: Grouping by clade might be risky, for example sending out a message that all clade 6 species are ubiquitous in water courses and not problematic. What about P. pinifolia for example? That one is a damaging needle pathogen of radiata pine in Chile.

Comment: We have this unknown clade 5 sequence cropping up in water samples at several nurseries. Clade 5 are regarded as tropical species so this is most likely to represent a tropical Phytophthora and we think about it differently. Basically, if it affects how a species is dealt with then we should report clade.

Comment: Should we then include information on traits for each species found? i.e. provide a link in each nursery manager’s report to the traits database online. This means we put a simplified version of the traits database on to the Phyto-threats website and in our report to nursery managers we say ‘For more information on each species found and clade, follow this link…..’ It would have to be digestible for general use so should say e.g. oospores, chlamydospores, soilborne, airborne, damaging to which host etc.

Comment: Clade 11 has to be included somehow.

FR are currently pulling together distribution maps based on UK findings of Phytophthora species and it was suggested that these maps also link in to the Phytophthora traits database when that goes online. This would be useful as it would allow nursery managers to see the distribution of those species in the UK.

Comment: The species distribution maps should differentiate records from cultures vs DNA findings. There is a danger of reporting a species based on DNA findings only when Statutory Plant Health policy is currently based on the need for a viable culture.

Comment: We need to look at technology and how it relates to policy – does Plant Health policy need to catch up with technology and start to base actions on DNA findings too?

Comment: The sequences of all the species will go on GenBank; a few species can’t be differentiated using ITS1.

Q: Ultimately, could you offer this metabarcoding technology as a service for the public good? i.e. as a testing service?

A: That could be possible. This metabarcoding method would really lend itself to being developed and used as part of statutory plant health surveillance of nurseries. There are already some commercial companies specialising in eDNA barcoding so this could be an extension of the services they offer. Water in nurseries could be tested, and then re-tested every few years. This might be a tool to assist in the issue of phytosanitary certificates.

Moving on with the presentation, David showed a slide listing host genus sampled by survey type, i.e. fine scale and broad scale, to illustrate the different range of hosts sampled. In the broad scale survey, the hosts were largely ornamentals with viburnum the most commonly sampled host, followed by rhododendron, camellia and pieris. David then showed the percentage of Phytophthora-positive samples for the broad scale and fine scale surveys – all around 40-50% positive.

Q: Can we relate broad scale survey results to practice?

A: We only asked for 5-10 samples per broad scale nursery – with each sample taken from different batches of plants. On average we received 6.3 samples per broad scale nursery which is possibly too few to link to practice. Also, associated metadata were not collected for the broad scale survey, only host symptoms description. However, APHA and SASA have knowledge of the background of each nursery, for example import history might be useful – a plant may have arrived with a Phytophthora and not acquired it in that nursery. We have scanned copies of each broad scale sampling sheet and those can be referred to when interpreting results.

Back to the presentation, David showed that over 70% of DNA sequence reads produced so far have matches to known Phytophthora species on the database. However, 7% of reads match unclassified Phytophthora species. So, it appears that there are quite a few unknown species in the samples.

Q: What about the ypt gene which has been used as an alternative barcode for Phytophthora?

A: There have been some issues with the ypt gene mainly due to it being only single copy. We are planning to look at other barcoding options as a separate project.

David then went on to show Phytophthora species diversity by nursery and this generated much discussion on how management practice affects diversity. There were clear examples of how certain species predominate at certain nurseries, with nursery size, practice (in particular importation of stock) and geographical location appearing to have an effect on species diversity. E.g. P. cryptogea (broad host range) was top, followed by P. gonapodyides, P. cinnamomi and P. plurivora. Records of P. pseudocryptogea and P. cryptogea are put together. One nursery has lots of P. cryptogea. Another has almost no Phytophthoras and operates with very tight biosecurity.

Comment: The Phytophthora species that we find on trees as part of our FR advisory service do seem to match species found in nurseries. For example, P. plurivora is our most common species on trees and P. cryptogea/pseudocryptogea is very common in agricultural soils and we find it causing serious root rot of trees planted on former agricultural land. We should try to match nursery findings to FR’s Tree Health Diagnostic Advisory Service findings in trees.

David moved on to discuss ongoing work/challenges, which, in addition to the need to complete sample processing, report to nursery managers and liaise with Plant Healthy Assurance Scheme over results, included continued refinement of the detection tool – the very high sensitivity of metabarcoding is both a blessing and a curse! For example, the need to prevent and account for field and lab contamination through the use of synthetic control sequences. Also, the computational pipeline development, coping with reads beyond 1-2bp thresholds, species boundaries and visualising data outputs.

David concluded his presentation with final thoughts/conclusions; metabarcoding is a very powerful targeted method for exploring microbial diversity. A classifier has been developed although results are interpreted with caution. When confident, the data will go to GenBank. With expanded primer sets we could look at other oomycetes. The sample bank of eDNA offers huge potential and experiments are now needed to advance biology and ecology.

Peter Cock presented on the bioinformatics pipeline Thapbi-pict now published online https://pypi.org/project/thapbi-pict/. He talked about technical variation in metabarcoding, determined through the use of four synthetic DNA control sequences alongside real samples. Illumina produces 1000s of sequence variants generated in very low abundance. These 1bp variants are most likely PCR artefact as single base change by PCR can happen and we need to account for that in metabarcoding diagnostics. Lots of these variants are coming through at low level. These variants were used to set a minimum threshold, which was 100. This means that any unique sequence needs to be present at least 100 times before going through the pipeline; i.e. any unique sequence at abundance of less than 100 is dumped early on in the pipeline. Sequence thresholds are then set for each plate for reporting species in a sample based on the level of contamination of synthetic control sequences in environmental samples. So, this is a plate-by-plate threshold set above the highest level of known sequence contamination.

Peter ran through how sequence data are prepared, by quality trimming FASTQ reads, merging the overlapping FASTQ reads into single sequences, discarding reads without both primers, converting reads into a non-redundant FASTA file and then filtering reads with hidden Markov models of ITS1 and synthetic control sequences so that non-matching sequences are discarded. This results in big data reduction!

Peter then showed edit graphs to illustrate the number of ITS1 variants observed for each species. Many at low sample abundance are likely to be PCR variants, but those occurring at higher sample abundance and with higher read numbers could be true ITS1 variants within the genomes, for example in P. austrocedri and P. nicotianae. There are also lots of Peronosporas and Phytopythium. Complex ITS1 clusters with up to 3 bp edits are observed for P. rubi and P. cambivora (the latter which is likely to be a hybrid complex?) but the most complex cluster occurs with P. gonapodyides, P. megasperma, P. chlamydospora, P. lacustris. All of these species are hard to distinguish by ITS1. So, the cut-off very much depends upon the species, sometimes 3bp difference works, for others we have to use just one bp difference.

In terms of interpreting sequence space, there seem to be some novel species in our sequence data. Species classification starts with a 100% match to a database sequence. If there is no perfect match, then the classifier looks for a 1bp difference. The database has ~170 individual sequences curated to species (by David) plus a much larger set of sequences from the NCBI database. Any sequences not matching the curated database, but which match the NCBI database based on 2bp difference, are reported to genus only. If it doesn’t match anything then it is marked as ‘unknown’. The pipeline does not use Swarm anymore as it is too ‘fuzzy’. Looking at complex sequence clusters, these could be curated manually if we want to know what the non-matching species or genera are.

Comment: Sometimes we only manage short reads for an organism so it can’t be identified but we have isolated it and we can see it’s a Phytophthora. The point is that we can still miss things using metabarcoding methodology.

Peter showed the outputs produced by the software tool Thapbi-pict, for example the Excel spreadsheet of results. He has tested the classifier on other Phytophthora datasets as well as a nematode ITS1 dataset and found that it works well. The team plans to publish a paper on the Thapbi-pict software, another on the use of synthetic controls, the nursery sampling dataset and an environmental monitoring paper. They hope to try and culture some of the novel Phytophthora species that the data hints at.

There was some discussion of the method and also on the Phytophthora community analyses planned for the nursery data. It was decided that there are enough data now to start planning these analyses, i.e. categorising factors to include in the analyses. This needs to link in with the nursery surveys that Mike and Mariella did, i.e. to categorise nurseries based on management practice/behaviour. Mariella has lots of information on partner nursery type and practice and if Beth gave her a list of what classification criteria they want, Mariella will pull it out and pass it on. The challenge with management factors is that some are easy to categorise, but others are qualitative and hard to rate, for example level of biosecurity practice. David is to give Mike a framework of what he needs, and they will provide it. Beth said she was planning to provide details on the information she needs after Xmas.

WP2 Feasibility analyses and development of ‘best practice’ criteria – Mariella Marzano (FR) and Gregory Valatin (FR)

Gregory Valatin presented on the costs and benefits of implementing nursery best practice from a nursery perspective, based on data pooled from 75 nurseries. The data come from three sources; an initial FR survey of nine nurseries, joint visits by FR and Fera to eleven nurseries and the work package 2 nursery and garden centre survey of 55 nurseries. Nursery responses were based on the perceived costs and benefits of 12 best practices (water testing for pathogens, water storage in fully enclosed tanks, water treatment facility, clean/covered storage of growing media, installation of drains or free draining gravel beds, raised benches, disinfectant stations for tools, quarantine holding area for imported plants, composting/incineration facility for diseased/unwanted plants, boot washing station, car washing station, buy only from trusted or accredited UK suppliers).

One main issue with collecting such data are the high number of non-responses when each nursery was asked to estimate the costs of installing each best practice. For most of the best practices, the majority of nurseries did not provide an estimate. In addition, the numerical responses that had been provided were dominated by a few high estimates, resulting in a skewed distribution of cost estimates. Quite a lot of the nurseries are already using some of the best practice measures, so the analysis needed to consider a baseline of what measures are already being implemented. The assumed baseline of practice included five of the most commonly used best practices – those currently used by the majority of the nurseries indicating whether they used a specific practice, or not. These were; water storage in fully enclosed tanks, clean/covered storage of growing media, installation of drains or free-draining gravel beds, raised benches and tool disinfestation stations. For the purposes of comparing the costs of establishing and maintaining the best practices with the anticipated benefits, a ten-year time horizon was selected, reflecting both the length of time before many of the best practice infrastructure would need to be replaced and the investment decision time horizon used in practice by the one nursery for which this information was available. Nursery respondents were also asked to estimate the costs to their business of a Phytophthora outbreak and (in the case of the nursery and garden centre survey) of a Xylella outbreak.

Gregory asked the question, how many Phytophthora outbreaks could nurseries avoid by adopting best practice? This created some discussion as generally it is very hard to predict with any accuracy how many disease outbreaks can be avoided.

Q: Isn’t it better to take a measure of the price benefit of being part of accreditation, i.e. by adopting best practice the nursery can charge more for its plants than other nurseries?

A: But if every nursery adopts best practice and joins an accreditation scheme then (apart from that associated with increased plant health/quality) there is no price advantage! It’s better therefore to stick with the data we’ve got (also given there was no time for further survey work).

Gregory suggested that if reduction of risk of an outbreak is too complex to consider at the moment (perhaps it requires a new project to address this, by looking at nursery characteristics and environmental factors and modelling the frequency of outbreaks) then perhaps we need to think in terms of how many outbreaks you avoid during a certain period until the benefits outweigh the costs? He then presented an exploratory cost-benefit analysis based on the implementation costs for the seven measures which are not common best practice currently and the perceived benefits (avoided cost) of avoiding outbreaks. Based upon mean costs and benefits, this indicated that the costs outweighed the benefits where avoiding Phytophthora outbreaks alone is considered and there are one or fewer avoided outbreaks a year over a ten-year period. As these are widely considered relatively infrequent, the analysis suggested that the costs outweigh the benefits of implementing best practice based on avoiding Phytophthora outbreaks alone. (The costs of the best practices amounted to roughly nine times the costs saved by avoiding one outbreak). However, if the potential benefit of avoiding an outbreak of Xylella was considered in the analysis, then the benefits of nursery best practice outweighed the costs if one outbreak of Xylella were avoided over an eight-year period, or one outbreak of both Phytophthora and of Xylella avoided every 9 years. These estimates were based on using the mean costs and benefits from the survey data, and assuming a 3.5% discount rate (standard in government appraisals). Considering instead median costs and benefits – which are arguably more typical of nurseries given the skewed distribution of estimates, suggested for the benefits to outweigh the costs, there would need to be at least one avoided Phytophthora outbreak every 4 years, or an outbreak of Xylella avoided every 7 years, or one Phytophthora outbreak and one outbreak of Xylella avoided every 11 years.

A further factor that could also usefully be considered is the impact best practices have on reducing risks of other pests and diseases. However, no estimates were available in the survey data for these benefits. This analysis focuses upon the perspective of nursery costs and benefits, without accounting for the costs of outbreaks in the wider environment, for example, in forest and amenity landscapes (which is the focus of modelling work by Colin Price).

There was some discussion on the costs of disease outbreaks. Nursery managers expect to lose some plants each year. What threshold of disease/mortality causes nursery managers to take action? It might be that an ‘all or nothing’ loss due to an outbreak is unrealistic and that a general erosion of loss (quantity/quality) might occur over a longer time until it gets critical.

Gregory explained that he considered this analysis to be exploratory, as risk reduction is dependent on measures implemented by the whole plant trade sector and not just nurseries. It was also difficult to get enough quantitative responses from nursery managers on costs of implementing best practice and costs of avoiding outbreaks of Phytophthora and Xylella to be confident in the analysis. Some of the responses to the outbreak questions were qualitative. Gregory concluded by emphasising the nurseries’ divergent perspectives on best practice and accreditation; some nurseries want stricter regulation and others prefer voluntary measures, particularly ‘traders’ who import many of their plants. Some nurseries adopting best practice consider other plant traders to be exposing them to unacceptable risks – an aspect that from an economic perspective could be characterised as an ‘externality’ (uncompensated effect of activities over which they have no control by another company).

For those nurseries already investing in accreditation, questions arise as to how they perceive risks and pass on accreditation costs.

Discussion then focused on frequency of disease outbreaks in UK nurseries and how to quantify this? We could look at interception data and look at the number of positives found at different nurseries for each pathogen. A future project needs to pull together all the different data to get a good bottom line. Gregory’s study has found just how hard it is to get good data in order to estimate costs.

There are also risks of disease outbreaks imposed on society, and the societal perspective will be provided by scenario analysis using Colin Price’s CARBBROD model. Colin was too unwell to attend the meeting, but he put together a PowerPoint presentation available here on the economic costs of tree diseases, with the predominant impact being on carbon fluxes. Some of the presentation is a recap of that presented at the 2018 Phyto-threats project team meeting, and some of it an update. Colin was looking for feedback on a number of inputs for the CARBBROD model which will be used for a cost-benefit analysis of introducing nursery best practice from a societal perspective, in which the carbon impacts of tree disease epidemics are considered key. Gregory posed the questions for Colin, which ranged around the likely reduction in risk of an hypothetical Phytophthora outbreak by introducing best practice, the percentage of total UK area of Sitka spruce and oak susceptible to introduced Phytophthora, and the likely rate of spread of Phytophthora in both hosts. There was much discussion of these questions, the first one being particularly hard to estimate! Some estimates were pulled together finally for Gregory to pass on to Colin, with further discussion to be had by email.

It was suggested that Colin could find useful a forestry model for felling larch for P. ramorum control. Jane can help with this. Beth and Louise also have references to forward to Gregory.

Mariella gave a summing up of the work package 2 research which looked at the feasibility of accreditation through exploring the perspectives of a range of consumers. Over the course of the project they have surveyed or interviewed 75 nurseries, 61 garden centres, 183 landscape architects or designers, 18 landscaping contractors, 15 local authorities and 1500 members of the plant buying public. They are now busy coding all the qualitative data for analysis and working on improving landscape architect and contractor data. This is being done as part of the BRIGIT project on Xylella. Progress this year includes publication of two technical reports on the consumer and nursery/garden centre/landscaper/surveys which have been very popular. A journal paper on the public consumer survey has also been published in ‘Plants, People, Planet’. Future publications will include a collaborative article on the role of landscape architects in mitigating the spread of plant pests and diseases, another paper on the sector appetite for accreditation – opportunities and challenges (with Tim Pettitt) plus other papers on the role of tourism in spreading plant pathogens, biosecurity practices and the cost-benefit of biosecurity measures led by Gregory.

Mariella asked what other outputs should be produced this year? They have loads of data, where could it be used? The outcomes need to go out to growers but how do we do this? How can it change their behaviours? It was suggested that Mariella’s team produce two sided fliers outlining key results for producers and which could go out to Plant Health inspectors. Their results should link to the Plant Healthy team. Articles in trade magazines are another good way of reaching out to the industry. Possibly soundbites on changes introduced by participants already e.g. I’ve made small changes at not much cost by doing this and this is the benefit.

 

WP3 Global Phytophthora risks to the UK – Louise Barwell (CEH)

Louise Barwell focused her presentation on global niche models being developed by Dan Chapman as they are looking for feedback. She showed examples of the global niche models being developed using P. cinnamomi. Their aim was to train the models using distribution data from outside the UK and then use the models to predict suitability within the UK. The approach is then validated by using distribution data from within the UK. For P. cinnamomi the match is really good between predicted and actual distributions. Importantly, the UK data were not used to train the model so that a completely independent dataset was used to validate the predictions.

Dan was able to fit models for a further nine Phytophthora species with sufficient distribution data (more than 75 records outside of the UK) to make a niche model. All of these species are in the UK so are useful for validating the methods to be used for species not yet in the UK. The main issue here though is that for most species not yet recorded in the UK there are insufficient records outside of the UK to produce a model.

Louise listed the drivers which the project team thought would be important in risk of establishment – plant host distribution came out top, followed by pathogen traits (spore/inoculum density). She then showed some correlations between environmental temperature and Phytophthora species’ cardinal temperatures for growth. The correlations are good, showing that species tend to be temperature-constrained. Accounting for the invasion process used a model developed by Dan and co-authors in a paper published earlier this year (J Biogeography 2019 46 1029-40). They looked at Phytophthora known occurrences and defined a 100 km buffer around known presences based on a literature review of Phytophthora dispersal distances conducted by a summer student, Katy Roy. Model performance is good for non-UK Phytophthora in terms of discriminating presences and absences.

Louise showed the visual outputs for global niche models for P. ramorum and P. cinnamomi, and then presented a figure showing the model outputs for relative importance of different environmental drivers affecting risk of establishment among Phytophthora species. For all species the climate variables seemed to be more important that land cover, although this should be broken down by forest type to better reflect the hosts. For P. cambivora, P. gonapodyides, P. lacustris, P. ramorum and P. alni the minimum winter and mean summer temperatures are the most important drivers, but for P. cactorum, P. cinnamomi, P. cryptogea and P. plurivora, the more important driver is precipitation seasonality. Based on these drivers the model poorly predicted the UK distribution of P. plurivora but gave a good prediction for P. ramorum. Louise asked whether the project team thought the drivers being used in the model were realistic? For example, is there a more appropriate predictor of establishment than precipitation seasonality for P. plurivora?

Comments:

Prior to 2009, P. plurivora was known as P. citricola, which might explain the poor prediction. The findings of P. plurivora seem to be concentrated around south/south-east of England and the Midlands, but we are finding it more frequently in Scotland over recent years. There have also been outbreaks of P. plurivora on lime trees in Cornwall, which don’t appear as locations on the model. The most updated distribution data for this species needs to be shared. It is possible that pathogen introductions and outbreaks are mediated by human activity so do not necessarily correlate with climate suitability.

True that for P. cinnamomi the seasonal precipitation is very important. It was only possible to bait for P. cinnamomi at a particular site in Spain following rainfall.

P. ramorum outbreaks followed when there was a lot of summer rainfall. So, spring 2013 and 2018 had outbreaks after wet summers in 2012 and 2017.

David will send Louise distribution maps for P. infestans;

The P. ramorum prediction fits well with where a species is present. Louise was directed to look at data from Northern Ireland for her distribution.

It was pointed out that sometimes climate preferences don’t accurately map occurrences, for example chestnut blight which was found outside predicted areas because it was introduced to those areas.

Louise then asked where the Phytophthora traits and global occurrence databases should be hosted. There was some discussion about having them hosted on the new IDphy tool which is replacing the Phytophthoradb website. However, Beth felt we should try to look for further funds to maintain the database on our Phyto-threats website and try to include intra-species variability. Beth also thought that as part of a grant proposal we could look at bringing the Phytophthora community together to link the different Phytophthora databases. It was decided that Beth’s team should contact the collaborators in Australia and New Zealand and discuss how to take the traits and occurrence databases forwards and where/who to host them.

Louise updated everyone on papers with two currently in submission on the traits-based analyses of Phytophthora impact; more are planned, including one on trade pathways, one on the niche models based on UK data, i.e. what makes a species able to establish outside nurseries? Other outputs include a list of Phytophthoras which affect current and future forestry species. Ana offered to send Beth a list of all Phytophthoras the FR THDAS has found infecting plantation forestry species in the UK.

Beth and Louise finished up by demonstrating the tools and model outputs showing global trade pathways and risk of importation. They have 35 species with presence in exporter databases and source countries based on data since 2000. There was lots of discussion around how the tool could be improved and how to broaden its relevance to a range of stakeholders. It was suggested that the model is circulated to the project team to have a play with. It needs coding for each Phytophthora/host record to rank in terms of reliability of the record, with isolation of the species as a viable culture better than PCR + sequencing which is better than metabarcoding.

A suggestion was made that the tool could be used to explore what pathogen might be coming, with which host it is associated, and an estimate of risk i.e. if you increase your trade in this area the risk might increase to this! Some practices have risk to the environment. It would be useful for every sector but the big players moving large quantities of plants should come on board. It could be a useful learning tool.

Q: Based on which criteria does a species make it into the model?

A: We looked at data since 2000. If there was a reference, source and distribution to connect it was added. So, there are 35 species.

Q: Does the model assume free trade?

A: It is based on actual observed trade patterns.

Q: Can you model the risk if the supplier is ‘UK-grown’ only.

A: The problem is that trade flows may not reflect the real pathway. Netherlands (NL) to UK may be Italy to NL to UK. And ‘home grown’ labels are applied when an import has been in a UK nursery for as little as 6 weeks in some cases.

Q: Can you interpret data to show closely related species behave similarly?

A: Yes, some do but some behave differently! Relatedness is only part of the story.

Comment: Nursery data will be removed before making it accessible to all because currently it may be possible to zoom in to an area and correlate an identification with a nursery.

 

WP4 Predicting risk via analysis of Phytophthora genome evolution – Ewan Mollison (University of Edinburgh)

Ewan Mollison updated the team on the genome sequencing of three Phytophthora species currently regarded as less damaging but closely related to highly damaging species. These were; i) P. europaea, a clade 7 species which was first described from soil associated with European oaks, but which is closely related to P. alni, a hybrid species killing riparian alder across Europe, ii) P. foliorum, a clade 8 species currently known only to cause a minor foliage blight of azalea and rhododendron but which is closely related to P. ramorum which is killing larch in the UK and tanoak in the USA, and iii) P. obscura, another clade 8 species first associated with horse chestnut and pieris but closely related to P. austrocedri which is killing juniper in the UK and Chilean cedar in Argentina. The rationale behind the work is that by comparing genes present in closely related damaging and less-damaging species we can identify genes or gene families involved in Phytophthora virulence.

Ewan ran through the genome assembly process; the genomes were sequenced at the University of Exeter Sequencing Centre using PacBio and two ‘SMRT’ cells to ensure high genome coverage. The assembly was done using ‘CANU’ with a 100 Mbp estimate. Ewan presented a table with the assembly statistics, essentially showing high N50 values and a low number of contigs for all three genomes. The contigs were scaffolded with ‘SSPACE-Longread’ and gaps filled with ‘gapfinisher’ over several rounds, then run through ‘Arrow’ through three rounds of polishing. The resulting assembly statistics looked pretty good, with genome sizes of 77 Mbp, 61 Mbp and 62 Mbp for P. europaea, P. foliorum and P. obscura, respectively, assembled in 39, 71 and 29 scaffolds, respectively. However, the larger number of scaffolds for P. foliorum raised concern and these were checked further using ‘blobtools’ to identify the taxonomic origin of the scaffolds in the assembly. In P. foliorum they found 49 very short scaffolds with a lower GC content than the rest of the assembly; in P. obscura there was one such scaffold and none in P. europaea. It was determined that these scaffolds were likely bacterial contamination (firmicutes) and they were manually removed from the assemblies.

The final assembly statistics are very good indeed, with the number of scaffolds for P. europaea, P. foliorum and P. obscura reduced to just 39, 22 and 28, respectively. Gene prediction was done using ‘Augustus’ and training sets based on the closest available relative; the three species were found to have ~19,500 predicted genes. Looking at the N50 and scaffold counts across all 30 available Phytophthora genomes, it is clear that the three species assembled in this project have the most complete Phytophthora genome assemblies to date, with the closest published assembly being P. sojae. Looking at completeness of coverage using ‘BUSCO’, which provides a measure of genome completeness based on an analysis of near-universal single copy orthologs, it is clear again that our three genomes have among the lowest levels of missing orthologs, being rated by ‘BUSCO’ as 98-99% complete.

To compare gene complement, Ewan used protein sets from 26 Phytophthora genomes assessed as greater than 85% complete using ‘BUSCO’. From ~55,000 total gene clusters containing sequences from any of the 26 species, a core set of ~8,700 clusters common to at least 75% (20 or more) of the genomes was identified and, of these, over 5,000 can be classed as single-copy orthologue clusters, containing a single gene from each of the species represented by that cluster (i.e. no paralogues or duplicates, etc.). Looking at gene content differences for clade 7, Ewan compared genes from damaging woody-host infecting species and non-woody host-infecting species together with the less-damaging P. europaea. His analyses pulled out 101 genes present in the other selected pathogens but not in P. europaea. Further assessment found that 36 of these genes are in fact present in the P. europaea genome but are disrupted by stop codons or indels that might affect expression or function.

Looking at gene content differences in clade 8, Ewan selected three aggressive pathogens that can all infect woody hosts (P. cryptogea, P. ramorum and P. austrocedri) and compared gene complement in these species with the sequences of less-aggressive P. foliorum and P. obscura. Downstream filtering found 40 genes common to all three damaging pathogens but not present in P. foliorum or P. obscura. The next steps are to dig deeper into the presence/absence gene sets and assess the putative functions of genes found to be absent in the less-damaging species. In terms of papers planned, they are drafting one paper on the assemblies of P. europaea, P. foliorum and P. obscura and comparative genome analyses, and a second paper on the P. austrocedri genome and isolate resequencing analyses.

Ewan received a number of questions following his talk;

Q: Are you seeing telomeres – or centromeres – in the genome data?

A: Yes, we are seeing some kind of repeats that look like a signature of this.

Q: How good are your rDNA assemblies? Can you determine rDNA copy number? If so, this could be added to the genome papers.

A: I can send you the ITS data for these three species so we can look at rDNA assembly in relation to copy number.

Q: Has P. foliorum been found again in the UK? – so far it has only been found in Scotland on a rhododendron leaf. However, inoculation experiments have found it to cause lesions on larch.

Comment: We are picking up DNA signatures of P. foliorum in our rainfall and wind vane traps in Scotland so it might be more widely distributed than we think. So far, we have not had any findings of P. foliorum causing significant damage to any host through natural infections.

WP5 Synthesis and integration – Sarah Green (FR)

Sarah Green rounded off the work package sessions by giving a brief overview of project management, i.e. three board meetings held over the last year and minutes posted on Huddle, the last all-project team meeting report written and posted on the Phyto-threats website. In terms of stakeholder engagement members of the project team have attended the National Plant Show and Four Oaks show, also various scientific conferences including two IUFRO meetings and a Phytophthora symposium in New Zealand where Mariella and Sarah did a double act presenting on Phyto-threats work. Researchfish will open in Feb/March again for outputs. Sarah wasn’t sure when a final project report is due and will get in touch with Debbie Harding of BBSRC to ask this. In terms of publications, Sarah is still planning to co-ordinate the writing of a project overview paper for which she has a draft available, will keep plugging away at this, perhaps circulating for inputs in the new year. David Cooke also promised to put out an interim report on the broad scale sampling for SASA and APHA.

Discussion of future options to continue collaboration – Sarah Green (FR)

Various leads were discussed as potential future work areas, for example following up on the growing media as bark-based (peat-free) growing composts are not always heat-treated. There is also the issue of whether anything is coming in on coir. These components could be tested for Phytophthora.

Sarah mentioned a proposed EUPHRESCO topic that she is coordinating on early detection of Phytophthora in EU nurseries and traded plants. The idea is to roll out the nursery testing and metabarcoding protocols to EU and third countries. So far eight different countries have expressed interest in the topic. Discussions are ongoing. However, the funding levels are not high and might only be £25K (if Defra commits).

Assessing different metabarcoding tools (i.e. barcodes) for Phytophthora would be very useful and David thought there might be some JHI seed corn funding for this – December application, usually 10-20K.

Beth mentioned a NERC call ‘Constructing the Digital Environment’ which might cover our activities collating data on Phytophthora outbreaks and pathogen behaviour. Beth will look more closely at the call and what it means. Deadline is January.

David mentioned the upcoming eDNA conference, there might be funding there for environmental monitoring.

Mariella thought there could be mileage in exploring funding for evaluating impacts of science and novel information on biosecurity-related behavioural changes and how it influences policy.

NERC/BBRSC might also have money to look at interactions of Phytophthora, in particular at whether species compete, whether they hybridise: NERC treescapes call is coming! This is more ecological so more about restoration, provenances, climate change, woodland expansion etc.

Following the end of this discussion the meeting was closed and Sarah thanked everyone for their contributions to what has been a very enjoyable and productive collaboration.

 

List of Participants

Contact

Organisation

Jane Barbrook

Animal and Plant Health Agency

Louise Barwell

NERC Centre for Ecology & Hydrology

Peter Cock

James Hutton Institute

David Cooke

James Hutton Institute

Mike Dunn

Forest Research

Debbie Frederickson Matika

Forest Research

Sarah Green

Forest Research

Kelvin Hughes

Animal and Plant Health Agency

Mariella Marzano

Forest Research

Ewan Mollison

University of Edinburgh

Ana Perez-Sierra

Forest Research

Beth Purse

NERC Centre for Ecology & Hydrology

Alexandra Schlenzig

SASA

Tim Pettitt

University of Worcester

Gregory Valatin

Forest Research